View Full Version : which processor for avisynth encoding
Terka
7th January 2007, 20:15
hope this is the right thread
which processor do you recommend for avisynth encoding-
DV 50i -> xvid 50p
DV 50i -> x264 50p
most used filters:
deinterlacers (tdeint, mvbob, maybee mcbob)
denoise (fft3d,..)
plz include overclocking possibilities
sh0dan
7th January 2007, 20:29
Athlon 64 and Core 2 Duo are both very good. You seem to get very similar performance per $ for both platforms. Avisynth has no preference.
For overclocking, you should ask in the Hardware section.
Terka
7th January 2007, 21:31
and semptron vs Athlon 64 -similar performance or big difference?
respective
celeron vs P4-similar performance or big difference?
Pookie
7th January 2007, 22:21
What about a horse vs a steam powered train ? :D
Can you even buy a Sempron still? The Athlon64 3800 sells for $60.
If you were to put together a new system today and wanted a good price vs performance ratio, the Intel Core2 Duo e6400 would be a decent choice. The Athlon X2 4600+ is good as well.
Think about dual core platforms, especially if you plan on transcoding with h264. Certain Avisynth plugins will also take advantage of dual core as well.
DonQ
7th January 2007, 22:34
Rough order (from best to worst, not looking at price):
Core2Duo
Athlon64 X2, Opteron X2
Pentium4 D8xx, D9xx
Athlon64, Opteron
Pentium4
Sempron
Celeron
First three are dualcores, thereby recommended. Celerons are for office use only ;)
Didée
7th January 2007, 22:36
One thing I'd like to see are some comparative figures about the practical impact of processor's L1 & L2 cache sizes, especially when dealing with scripts that build long/complex filter chains. Complex scripts often seem to run much slower than one would expect from the sum of the parts they're made of.
Pookie
7th January 2007, 23:00
Didée - Can you post an example script ? If we all use the same source clip, we can roughly determine which proc would be optimal.
rfmmars
8th January 2007, 01:41
Rough order (from best to worst, not looking at price):
Core2Duo
Athlon64 X2, Opteron X2
Pentium4 D8xx, D9xx
Athlon64, Opteron
Pentium4
Sempron
Celeron
First three are dualcores, thereby recommended. Celerons are for office use only ;)
Take a look at this link,
http://www.pcper.com/article.php?aid=192
I have now 6 workstations some running 24/7. 5 of the six are AMD, i have found that with the best motherboards with both processors AMD is always faster. My seveth will be a AMD 5000 Dual core overclocked based on the above link. My first Duo Core was a AMD 4200 AMD which blew everything away.
Richard
photorecall.net
Pookie
8th January 2007, 05:00
Here's a benchmarking script which might provide some info.
* Modified to run 3 benchmarks - 25% Total RAM, 50% Total RAM, 100% Total RAM-64MB
* Small modification to work with XP/64
1) Download Wget.exe from http://users.ugent.be/~bpuype/cgi-bin/fetch.pl?dl=wget/wget.exe
2) Copy this batch file, save as Runme.bat
::v1.00
if not exist 7za.exe wget http://markosf.ath.cx/mSoftware/7za.exe
if not exist cpu-z-138.zip wget http://www.cpuid.com/download/cpu-z-138.zip
if not exist pthreadGC2.dll wget http://sourceware.mirrors.tds.net/pub/sourceware.org/pthreads-win32/dll-latest/lib/pthreadGC2.dll
if not exist ffmpeg.rev7375.7z wget http://tirnanog.fate.jp/mirror/ffmpeg/ffmpeg.rev7375.7z
if not exist Motion1.exe wget http://www.streamcrest.com/Motion1.exe
if not exist yuv2avi.exe wget http://www.streamcrest.com/yuv2avi.exe
if not exist cat.exe wget http://www.ufridman.com/cat-src.zip
if exist %SystemRoot%\system32\findstr.exe copy %SystemRoot%\system32\findstr.exe /y
7za e -y cpu-z-138.zip
7za e -y ffmpeg.rev7375.7z
7za e -y cat-src.zip
motion1 test.yuv
yuv2avi test.yuv test.avi 525c RGB235
del test.yuv
cpuz.exe -txt=cpuinfo
type cpuinfo.txt | findstr "Specification" >result.txt
type cpuinfo.txt | findstr "L1" >>result.txt
type cpuinfo.txt | findstr "L2" >>result.txt
type cpuinfo.txt | findstr "Memory" | findstr "Size">>memory.txt
for /f "Tokens=3" %%a in (memory.txt) do echo set tmem=%%a >tmem.bat
call tmem.bat
set /a qmem=%tmem%/4
set /a hmem=%tmem%/2
set /a fmem=%tmem%-64
echo setmemorymax(%qmem%) >%qmem%.txt && cat %qmem%.txt test.avs >%qmem%test.avs
echo setmemorymax(%hmem%) >%hmem%.txt && cat %hmem%.txt test.avs >%hmem%test.avs
echo setmemorymax(%fmem%) >%fmem%.txt && cat %fmem%.txt test.avs >%fmem%test.avs
echo. >>result.txt
echo Total Memory Installed=%tmem%MB >>result.txt
echo. >>result.txt
echo %qmem%MB memory >>result.txt
ffmpeg -benchmark -i %qmem%test.avs -an -vcodec rawvideo -y out.avi >>result.txt
echo. >>result.txt
echo %hmem%MB memory >>result.txt
ffmpeg -benchmark -i %hmem%test.avs -an -vcodec rawvideo -y out.avi >>result.txt
echo. >>result.txt
echo %fmem%MB memory >>result.txt
ffmpeg -benchmark -i %fmem%test.avs -an -vcodec rawvideo -y out.avi >>result.txt
notepad result.txt
rem del test.avi
del out.avi
3) Save the following as TEST.AVS (thanks, Didée)
BTW - You'll need to have the plugins below installed. Modify script if your plugins are located elsewhere.
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\fluxsmooth.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\MedianBlur.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\mt_masktools.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\MVTools.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\RemoveGrain.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Repair.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\TTempSmooth.dll")
aviSource("test.avi").converttoyv12()
trim(100,129)
Bob(0,0.5)
bobbed = last
base = last.minblur(2).FluxSmoothT(9)
bw4_vec = base.MVAnalyse(isb=true, truemotion=false,delta=4,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
bw3_vec = base.MVAnalyse(isb=true, truemotion=false,delta=3,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
bw2_vec = base.MVAnalyse(isb=true, truemotion=false,delta=2,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
bw1_vec = base.MVAnalyse(isb=true, truemotion=false,delta=1,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
fw1_vec = base.MVAnalyse(isb=false, truemotion=false,delta=1,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
fw2_vec = base.MVAnalyse(isb=false, truemotion=false,delta=2,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
fw3_vec = base.MVAnalyse(isb=false, truemotion=false,delta=3,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
fw4_vec = base.MVAnalyse(isb=false, truemotion=false,delta=4,lambda=512,pel=2,sharp=2,blksize=16,overlap=16/2,pnew=32,idx=1)
bw4_comp = bobbed.MVCompensate(bw4_vec,idx=2)
bw3_comp = bobbed.MVCompensate(bw3_vec,idx=2)
bw2_comp = bobbed.MVCompensate(bw2_vec,idx=2)
bw1_comp = bobbed.MVCompensate(bw1_vec,idx=2)
fw1_comp = bobbed.MVCompensate(fw1_vec,idx=2)
fw2_comp = bobbed.MVCompensate(fw2_vec,idx=2)
fw3_comp = bobbed.MVCompensate(fw3_vec,idx=2)
fw4_comp = bobbed.MVCompensate(fw4_vec,idx=2)
Interleave(bw4_comp,bw3_comp,bw2_comp,bw1_comp,bobbed,fw1_comp,fw2_comp,fw3_comp,fw4_comp)
MedianBlurT(0,0,0,4)
SelectEvery(9,4)
base2 = last
bobbed2 = last
bw4_vec2 = base2.MVAnalyse(isb=true, truemotion=false,delta=4,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
bw3_vec2 = base2.MVAnalyse(isb=true, truemotion=false,delta=3,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
bw2_vec2 = base2.MVAnalyse(isb=true, truemotion=false,delta=2,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
bw1_vec2 = base2.MVAnalyse(isb=true, truemotion=false,delta=1,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
fw1_vec2 = base2.MVAnalyse(isb=false, truemotion=false,delta=1,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
fw2_vec2 = base2.MVAnalyse(isb=false, truemotion=false,delta=2,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
fw3_vec2 = base2.MVAnalyse(isb=false, truemotion=false,delta=3,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
fw4_vec2 = base2.MVAnalyse(isb=false, truemotion=false,delta=4,lambda=256,pel=2,sharp=2,blksize=8,overlap=8/2,pnew=32,idx=3)
bw4_comp2 = bobbed2.MVCompensate(bw4_vec2,idx=3)
bw3_comp2 = bobbed2.MVCompensate(bw3_vec2,idx=3)
bw2_comp2 = bobbed2.MVCompensate(bw2_vec2,idx=3)
bw1_comp2 = bobbed2.MVCompensate(bw1_vec2,idx=3)
fw1_comp2 = bobbed2.MVCompensate(fw1_vec2,idx=3)
fw2_comp2 = bobbed2.MVCompensate(fw2_vec2,idx=3)
fw3_comp2 = bobbed2.MVCompensate(fw3_vec2,idx=3)
fw4_comp2 = bobbed2.MVCompensate(fw4_vec2,idx=3)
Interleave(bw4_comp2,bw3_comp2,bw2_comp2,bw1_comp2,bobbed2,fw1_comp2,fw2_comp2,fw3_comp2,fw4_comp2)
ttempsmooth(4,8,8,2,2,strength=5,pfclip=last.RemoveGrain(11))
SelectEvery(9,4)
maxx = bobbed.mt_logic(bw1_comp,"max")
\ .mt_logic(fw1_comp,"max")
minn = bobbed.mt_logic(bw1_comp,"min")
\ .mt_logic(fw1_comp,"min")
#.mt_logic(bw2_comp,"max").mt_logic(bw3_comp,"max")
#.mt_logic(fw2_comp,"max").mt_logic(fw3_comp,"max")
#.mt_logic(bw2_comp,"min").mt_logic(bw3_comp,"min")
#.mt_logic(fw2_comp,"min").mt_logic(fw3_comp,"min")
ssharp = mt_lutxy(last,last.RemoveGrain(11), \
"x x y - abs 12 / 1 4 / ^ 12 * x y - 2 ^ x y - 2 ^ 2.3 + / * x y - x y - abs 0.0001 + / * 1 x y - abs 16 / 1 4 / ^ + / +",U=2,V=2)
ssharp.mt_Clamp(maxx,minn,0,0,U=2,V=2)
SeparateFields().SelectEvery(4,0,3).Weave() # before or after,
interleave( selectevery(5,0).FluxSmoothT(7),selectevery(5,1).FluxSmoothT(7),
\ selectevery(5,2).FluxSmoothT(7),selectevery(5,3).FluxSmoothT(7),
\ selectevery(5,4).FluxSmoothT(7) )
# SeparateFields().SelectEvery(4,0,3).Weave() # that's a question
return(last)
#=================================
# Nifty Gauss/Median combination
function MinBlur(clip clp, int r, int "uv")
{
uv = default(uv,3)
uv2 = (uv==2) ? 1 : uv
rg4 = (uv==3) ? 4 : -1
rg11 = (uv==3) ? 11 : -1
rg20 = (uv==3) ? 20 : -1
medf = (uv==3) ? 1 : -200
RG11D = (r==1) ? mt_MakeDiff(clp,clp.RemoveGrain(11,rg11),U=uv2,V=uv2)
\ : (r==2) ? mt_MakeDiff(clp,clp.RemoveGrain(11,rg11).RemoveGrain(11,rg11),U=uv2,V=uv2)
\ : mt_MakeDiff(clp,clp.RemoveGrain(11,rg11).RemoveGrain(20,rg20).RemoveGrain(20,rg20),U=uv2,V=uv2)
RG4D = (r==1) ? mt_MakeDiff(clp,clp.RemoveGrain(4,rg4),U=uv2,V=uv2)
\ : (r==2) ? mt_MakeDiff(clp,clp.medianblur(2,2*medf,2*medf),U=uv2,V=uv2)
\ : mt_MakeDiff(clp,clp.medianblur(3,3*medf,3*medf),U=uv2,V=uv2)
DD = mt_lutxy(RG11D,RG4D,"x 128 - y 128 - * 0 < 128 x 128 - abs y 128 - abs < x y ? ?",U=uv2,V=uv2)
(r==0) ? clp : clp.mt_MakeDiff(DD,U=uv,V=uv)
return(last)
}
#end
4) Create a folder to place these three files. It should look like this:
Runme.bat
Test.avs
wget.exe
5) Double Click on Runme.bat. It will download the necessary files to run the benchmark. When it is done, notepad will pop up with your CPU info and processing time. Post the info in this thread.
Specification Intel(R) Pentium(R) 4 CPU 2.40GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1226.932s
check
8th January 2007, 10:07
Specification AMD Athlon(tm) 64 Processor 3000+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=572.281s
Blue_MiSfit
8th January 2007, 11:41
Specification AMD Athlon(tm) 64 Processor 3500+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=277.906s
I'm a little surprised that mine was so much faster than check's. We're only a couple hundred MHz off, and both have 512k L2 cache. I guess I have the midas touch :D 1 GB DC DDR400 @ CAS3 2T, Avisynth 2.5.7 RC2
Nice benchmark app by the way - I will have to try this on the MacPro!
~MiSfit
krieger2005
8th January 2007, 12:20
Can you all post also your size of the RAM (and the used Avisynth-Version would be also usefull, i think).
mooms
8th January 2007, 15:12
this don't work for me.
here is my results:
Specification AMD Athlon(tm) 64 Processor 3500+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=0.109s
made with avs 2.5.6
maybe it's a bad version of a (couple) of plugin(s).
could you post links the download urls of all the plugins?
here is an image of my plugins directory:
http://img156.imageshack.us/img156/3821/testvj4.jpg
Mr_Odwin
8th January 2007, 15:30
From my work's computer (1GB of RAM):
Specification Intel(R) Pentium(R) 4 CPU 3.20GHz
L1 Data cache 16 KBytes, 8-way set associative, 64-byte line size
designation SERIAL1 (internal)
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
bench: utime=547.406s
Note that when I added SetMTMode(2) to the top of the AVS script:
Specification Intel(R) Pentium(R) 4 CPU 3.20GHz
L1 Data cache 16 KBytes, 8-way set associative, 64-byte line size
designation SERIAL1 (internal)
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
bench: utime=412.359s
This is just a machine with Hyper Threading, not dual core.
Pookie
8th January 2007, 16:59
Specification Intel(R) Xeon(TM) CPU 3.06GHz
Specification Intel(R) Xeon(TM) CPU 3.06GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=958.922s
------------------------------------------------
Specification AMD Opteron 248
L1 Data cache 64 KBytes, 4-way set associative, 64-byte line size
L2 cache 1024 kBytes
bench: utime=613.969s
DonQ
8th January 2007, 19:14
Specification AMD Athlon(tm)
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=832.906s
CPU is old AMD Athlon 2800+ @2GHz @FSB167Mhz
1GB RAM DDR333 dual
Abit NF7
Avisynth 2.5.7
-----------------------------------------------------------
Specification Dual Core AMD Opteron(tm) Processor 170
L1 Data cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 2 x 1024 KBytes, 16-way set associative, 64-byte line size
bench: utime=620.431s
1GB RAM DDR400 dual
Abit AV8
Avisynth 2.5.7
Couldn't test with latest SetMTMode() - ffmpeg crashed.
foxyshadis
8th January 2007, 20:06
You might want to replace the beginning of your batch file with:
if not exist 7za.exe wget http://markosf.ath.cx/mSoftware/7za.exe
if not exist cpu-z-138.zip wget http://www.cpuid.com/download/cpu-z-138.zip
if not exist pthreadGC2.dll wget http://sourceware.mirrors.tds.net/pub/sourceware.org/pthreads-win32/dll-latest/lib/pthreadGC2.dll
if not exist ffmpeg.rev7375.7z wget http://tirnanog.fate.jp/mirror/ffmpeg/ffmpeg.rev7375.7z
if not exist Motion1.exe wget http://www.streamcrest.com/Motion1.exe
if not exist yuv2avi.exe wget http://www.streamcrest.com/yuv2avi.exe
7za e -y cpu-z-138.zip
7za e -y ffmpeg.rev7375.7z
Will post benchmarks as they finish.
Specification Genuine Intel(R) CPU @ 2.16GHz (Engineering Sample)
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=492.344s
(It's a Core2 Merom T7400 pre-release. I was impatient, what can I say.)
Pookie
8th January 2007, 20:46
Thanks, foxyshadis. Silly of me to think that folks wouldn't test performance a few times in a row....
I'm finding the initial results to be... perplexing. My Xeons are getting killed.
CPUid doesn't like command line mode for the faster Athlons/Opterons :(
BTW, I intentionally set a low SetMemoryMax(128) statement in order to keep the benchmarks focused on CPU.
Specification AMD Opteron 275
L1 Data Cache 64K
L2 Cache 1024 Kbytes
bench: utime=623.563s
Specification Intel(R) Xeon(TM) CPU 2.80GHz
Specification Intel(R) Xeon(TM) CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1126.891s
foxyshadis
8th January 2007, 20:55
That matches the (single-thread) results for my Dual 275s to within a second. =o I guess that means it's repeatable, at least. I tried enabling MT for both systems, but I think mvtools is just plain breaking it and causing extra slowdown from extra processing(?), because there's almost no speedup. I'll try with ELDER or one of the other cluster tools later.
Pookie
8th January 2007, 20:56
Oh, great. Good to know. Thx
mooms - Download Msvcr71.dll from http://www.dll-files.com/dllindex/dll-files.shtml?msvc71
Put it in your C:\windows\system32 folder
If that doesn't work, Google for vcredist_x86.exe
Specification AMD Opteron(tm) Processor 242
Specification AMD Opteron(tm) Processor 242
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 1024 KBytes, 16-way set associative, 64-byte line size
L2 cache 1024 KBytes, 16-way set associative, 64-byte line size
bench: utime=793.500s
Specification Intel(R) Pentium(R) 4 CPU 3.00GHz
L1 Data cache 16 KBytes, 8-way set associative, 64-byte line size
L2 cache 1024 KBytes, 8-way set associative, 64-byte line size
bench: utime=853.563s
squid_80
9th January 2007, 01:21
Surely I'm not the only one with an E6600:
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=319.922s
with setmtmode(4) (the lowest mode that didn't crash):
bench: utime=186.375s
Pookie
9th January 2007, 01:25
Nice one, squid_80 http://www.lengua-translations.de/anthems/australia.mid
All it takes is a 4K L2 Cache..... :D
Anyone with an e6400 ready to try this ?
*.mp4 guy
9th January 2007, 06:24
WithSetMemoryMax(128)
Specification AMD Athlon(tm) 64 Processor 3800+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Ram: 2-2-2-6 1t 400mhz dual channel
bench: utime=534.187s
With SetmemoryMax(896)
Specification AMD Athlon(tm) 64 Processor 3800+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Ram: 2-2-2-6 1t 400mhz dual channel
bench: utime=83.000s
The test is flawed, with the original scrip hard drive speed will actually influence the results... I think ram speed is more relevant then hard drive speed for avisynth usage.
Pookie
9th January 2007, 06:46
Not really "flawed", per se. Just leveling the playing field to determine CPU related factors on Avisynth performance. The raw number results are ridiculously low, but we do get to see ratio-wise the difference between a Xeon and a P4 3.0 or an Athlon64 3000 and an Opteron 248. Opening up the RAM introduces too many variables into the mix, namely speed rating of the memory. We would have the same CPU/motherboard being benchmarked with PC2100, 2700, 3200, 3500, etc.
*.mp4 guy
9th January 2007, 07:14
My issue isn't that you wan't to exclude ram influences, thats fine, the problem is that you replace it a hard drive bias, this test will be heavily effected by hard drive latency, and to a lesser extent hard drive bandwidth.
Pookie
9th January 2007, 07:39
You mean pagefile swapping ? Okay, I can image that being a factor, but doesn't the CPU have an influence on how frequently the data is being swapped ? I'll try it at work tomorrow - I've got 2 machines with the same procs, but 1 with a 7200RPM Seagate and 1 with a Raid 0 pair of WD 10000RPMs.
*.mp4 guy
9th January 2007, 08:10
The issue isn't bandwidth, its latency, I just switched to tighter ram timings to test how big an issue lateny is, here are my results.
Ram: 2-2-2-6 1t 400mhz dual channel
Specification AMD Athlon(tm) 64 Processor 3800+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=83.000s
Ram: 2-2-2-5 1t 400mhz dual channel
Specification AMD Athlon(tm) 64 Processor 3800+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=61.344s
That speed gain is ~2 times more then I was expecting, the benchmark apears to be almost completely CPU<->RAM IO latency limited, which means it should also be HD<->CPU IO latency limited aswell, if it doesn't have enough ram to use. However since it is latency limited and not bandwidth limited (as far as I can tell) in two Identicle systems the system with the faster cpu would always score a set amount higher then the slower one. So if we all had the same HD and motherboard the results would be an accurate way of determining the fastest cpu for this situation. That said the results would be worthless since the cpu plays very little part in the speed of that script since it isn't the limiting factor on my system, which has a rather slow cpu, and rather fast ram (a configuration I choose to get better performance on slow avisynth monster-scripts, and better performance in games).
Terka
9th January 2007, 09:26
as i see above, the results are not compareable, if the
SetmemoryMax(896) has so biggggg influence on results.
What about to retest with SetmemoryMax(896)?
Terka
9th January 2007, 10:19
So maybee it is better to spent money on bigger RAM, then faster processor.
Is it clever to have 2GB ram instead of 1GB ?
*.mp4 guy
9th January 2007, 10:30
1 GB is enough for all but the most insane scripts (the benchmark script needs less then 768MB). Before I recommend you spend more on ram It would be good to have an example of what you wan't to have good performance on, low latency ram helps a lot with ram bound scripts like this, but doesn't do much for other stuff. It would also be good to have some comparable benchmarks from other peoples systems (along with ram specs) to see how things stack up, my system could end up being the slowest in a fair comparison.
Terka
9th January 2007, 10:37
as i mentioned above
DV -> x264 or xvid encoding
using mvbob, tdeint or mcbob
denoised via fft3d or dust or ...
maybee some limitedsharpen...
*.mp4 guy
9th January 2007, 11:32
Well, in that case ram speed is definately important, basically I would recommend getting 1GB of the lowest latency memory you can afford, bandwidth isn't as important for your usage (bandwidth is the memories speed in mhz). Some other people will have to post benchmarks before it becomes clear which cpu family you should get.
squid_80
9th January 2007, 12:44
That makes it tricky. If you go for a core2 you're pretty much stuck with ddr2 ram which has high latency. But on the other hand, if I run the benchmark with setmtmode(4) and setmemorymax(896): utime=23.563s
bkman
9th January 2007, 14:14
IIRC SetMTMode(4) will produce corrupted output if you are using MVTools.
foxyshadis
9th January 2007, 19:22
SetMTMode anything will, but even if it's corrupt it's still valid for timing in the context of this test.
I was pretty sure the test was I/O bound when I compared initial results, but thanks to mp4 guy for proving it. (HT vs FSB makes a big difference.) HD reading probably isn't a big problem, because Windows itself will cache disk reads, so you get a little extra cache in addition to avisynth's.
But this was only ever meant to answer one question: How much does more cache help within a family? That's the reason it's I/O bound, really.
A full "benchmark" would be multiple variants on the test. ;) Avisynth can be extremely demanding on your components in very different ways depending on your script and plugins used, if you really wanted to test with it.
Pookie
9th January 2007, 20:41
Ran these tests 3 times in a row on each system. These are the fastest results, although they only varied by a second or two on each run.
Seagate 7200RPM Drive
Specification Intel(R) Xeon(TM) CPU 2.80GHz
Specification Intel(R) Xeon(TM) CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1126.891s
WD 10000RPM Raid 0
Specification Intel(R) Xeon(TM) CPU 2.80GHz
Specification Intel(R) Xeon(TM) CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1088.922s
-
Specification Intel(R) Pentium(R) 4 CPU 2.40GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1280.656s
The P4 2.4 is 1/4 the performance of the E6600.
Pookie
11th January 2007, 02:17
Hope I got them all, I'm bad at repetition
http://img90.imageshack.us/img90/5305/20070110171101cj9.png (http://imageshack.us)
Terka
11th January 2007, 08:56
!!!!!!!!!!!!!!! important !!!!!!!!!!!!!!!
as we saw above the testing is till now worthless, but we know, that most important is the RAM.
When SetmemoryMax(128), we are testing our harddisks, not processors
1. the size of ram is most important.
!!! 700% !!! difference for*.mp4 guy - 7x faster !!!
So please repeat the tests with higher SetmemoryMax
I suggest SetmemoryMax(512) or higher.
2. the speed of ram is the second important (a bit less important)
please type the speed of ram, if possible try different settings
3. the third is the processor
if we have no influence of harddisk speed - the setmemorymax is higher enough, we can get the infuence of processor to the encoding speed
Pookie
11th January 2007, 11:33
===> http://members.aol.com/chiprowe/gotoeleven2.wav <====
Seagate 7200RPM Drive
Specification Intel(R) Xeon(TM) CPU 2.80GHz
Specification Intel(R) Xeon(TM) CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1126.891s
WD 10000RPM Raid 0
Specification Intel(R) Xeon(TM) CPU 2.80GHz
Specification Intel(R) Xeon(TM) CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
bench: utime=1088.922s
YOU ARE NOT TESTING YOUR HARD DRIVE
Does the title of YOUR THREAD say "which processor/ram combination for avisynth encoding" ? No, it doesn't.
*.mp4 guy
11th January 2007, 12:06
1 - Aslong as you have anough ram for the script to "fit" in it extra won 't get you any more speed, so 1GB of ram is enough and you won't get any more speed by getting more.
2 - Ram speed isn't more important then the processor, whats important is having ram fast enough to keep your processor happy, slow ram slows down your processor, but fast ram won't make a bad processor any faster.
3 - pookies results are accurate enough, if you have enough money get a lowend c2d (core 2 duo) otherwise get a cheep athlon 64 X2, in either case get good ram to help you get the most out of your processor, otherwise you will be throwing performance away. I didn't mean to give you the impression that processor speed was secondary to ram speed, I was trying to show how ram speed can limit your processor speed at certain things.
4 - I was mostly wrong about HD speed influencing the test as it was originally.
Terka
11th January 2007, 14:02
are the results with SetmemoryMax(128)?
if yes, we are not testing the processor speed only.
i want only say that if harddisk or something else is limiting the processor, we cant compare the results.
Pokie: yes, the difference is only 0.5% but the processor must wait for the data.
if the processor spends 10s and HDD 1000s solving the problem in summary it is 1010s
if you got 2 time faster processor, but the same disk, you have 10s/2 + 1000s =1005s
so it would be best add bigger SetmemoryMax() to decrease the hdd influence
krieger2005
11th January 2007, 16:25
You can't get a "only CPU-Usage" Result. There would always be some influence. If you set SetMemoryMax low you have more influence by Harddisk. With higher values you get influence of RAM. Then you have in all cases the influence of the Mainboard, Cache-Size, how full the RAM while testing is. Which Programs run when you test. Which OS you use is also a factor.
And while you set MemoryMax higher you can't exclude, than the Hard-Disk will not affect on the results, because page-faults occour all the way (see above: OS, other programs etc.). So the only reason, why Pookie set the value that low is to reduce one factor of X others on a system. Best would be to reduce all other factors also. But you can't do it.
Terka
11th January 2007, 16:30
yes of course, but the slowest thing has the most influence on results
krieger2005
11th January 2007, 17:13
yes. but it is allmoust constant influence on most computers (most have a 7200 rpm HD).
Terka
11th January 2007, 21:01
as i posted above:
if the processor spends 10s and HDD 1000s solving the problem in summary it is 1010s
if you got 2 time faster processor, but the same disk, you have 10s/2 + 1000s =1005s
the speedup you got is 5s but from 1000 =0.5%,
-and the processor is by 100% faster
-and in reality you would not use the setmemorymax(128) or would you? :D :D if you would not, then it is like
10s+200s=210s
10s/2+200s=205s you got 5s from 200 = 2.5%
its 5times the difference or more as we see
by mp4 guys example !
WithSetMemoryMax(128)
Specification AMD Athlon(tm) 64 Processor 3800+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Ram: 2-2-2-6 1t 400mhz dual channel
bench: utime=534.187s
SetmemoryMax(896)
Specification AMD Athlon(tm) 64 Processor 3800+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Ram: 2-2-2-6 1t 400mhz dual channel
bench: utime=83.000s
*.mp4 guy
11th January 2007, 22:43
Latency only slows down a processor, it doesn't completely stop it from working faster then a given speed. so instead of 1010s and 1005s it would be 1010s and ~755s, whereas if there were no latency (including ram and cache latency) it would be 85s v 44s or thereabout.
foxyshadis
12th January 2007, 04:25
If you open the avs script while notepad's still open (so the files don't get deleted yet), you'll probably find a script error. Or it might be an ffmpeg error, you'd have to check the command output for that.
Pookie
12th January 2007, 04:32
Jeremy - you might need this: vcredist_x86.exe
Google around for it
*BTW, 1300 views and nobody has a E6400 to test ?*
squid_80
12th January 2007, 05:07
This is the scripts I'm using:
I think he meant load the script in virtualdub or mpc to make sure it's being processed correctly. We already know what the script consists of. Also load out.avi to make sure it's not full of garbage.
Pookie
12th January 2007, 05:52
Hmm.... I like squid_80s suggestion, but you'll have to make a change to the batch file:
2nd to last line(s):
rem del test.avi
rem del out.avi
That way, you can load test.avs in VirtualDub and it will point to test.avi. Otherwise, the file gets deleted when the batch file ends.
squid_80
12th January 2007, 06:51
There's still no way those results are accurate.
squid_80
12th January 2007, 08:15
I've tried to load the test.avs the same way I loaded the plugins.
I can't get a .txt result when I run Virtual Dub, it just closes after running the test.avs, even with Pookies tweak.
I can't get a result above utime=30.157s
We said load test.avs with virtualdub to make sure the script is working properly and not simply returning an error statement.
You can't run the benchmark with virtualdub.
FFmpeg is used by the batch file to encode test.avs to out.avi; that's where the benchmark figure comes from.
Execute runme.bat and load out.avi with virtualdub, it should look the same as test.avs.
Jeremy Duncan
12th January 2007, 08:42
Ok, I deleted everything and started over.
Specification Intel(R) Pentium(R) M processor 1.60GHz
L1 Data cache 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
bench: utime=804.203s
Ha ha. That looks more like it. :D
I'll delete my other posts about this now.
Terka
12th January 2007, 09:07
please change the line setmemorymax(128) to higer value
at least setmemorymax(512)
Jeremy Duncan
12th January 2007, 09:22
please change the line setmemorymax(128) to higer value
at least setmemorymax(512)
I only have 512 MB ram in my PC.
Blue_MiSfit
13th January 2007, 01:15
Hey everyone... I thought I would post another score, this time with heavy overclocking and tight RAM timings. From a stock speed of 2.2 GHz, I am now running at 2.6 GHz, with CAS2.5 and 1T. Memory speed is reduced (divider), but overall performance is higher.
2.6 GHz CAS2.5 1T results:
Specification AMD Athlon(tm) 64 Processor 3500+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=243.469s
2.2 GHz, CAS3 2T results:
Specification AMD Athlon(tm) 64 Processor 3500+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=277.906s
Pookie
13th January 2007, 02:14
What do you have your SetMemoryMax() set to ?
Blue_MiSfit
13th January 2007, 02:46
128. I have not modified the avisynth script.
It seems too fast to me, almost 2x faster than terka's results, which seems a bit strange...
~MiSfit
Adub
13th January 2007, 02:48
Using Avisynth 2.5.7 MT version by TSP rev 2.0
SinlgeThreaded,128 mb memory
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=356.219s
Multithreaded,128 mb memory
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=326.203s
Mutlthreaded, 1024 mb mem
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=20.953s
Mutlthreaded, no mem restrictions
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=70.313s
Edit: Oh, why does ffmpeg crash when SetMemoryMax(2000) is set? Or Higher?
Pookie
13th January 2007, 03:05
You possibly aren't.... Try it with 128MB. If the numbers are the same, the script isn't being read. See above for Jeremy's issue and solution.
Adub
13th January 2007, 04:03
Actually, it wasn't loading my plugins. I will load the new results up when they are done.
Pookie
13th January 2007, 05:05
I'm going to modify the batch file to determine installation locations so the loadplugin commands are more specific.
edit - On second thought, I decided not to. The only way to do it was to query the registry with a command line util, and that seems too risky and invasive. Yes, it is only a read operation, but why even take the slightest chance....
squid_80
13th January 2007, 05:39
I'm going to modify the batch file to determine installation locations so the loadplugin commands are more specific.Maybe add a hashing command on out.avi to determine if it's been created properly?
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=356.219s
That's weird, why is your overclocked e6600 performing worse than my stock system?
Adub
13th January 2007, 05:53
Maybe because the memory was limited. were you using limited memory on yours?
I will run a new instance and see if that changes anything.
My memory is not overclocked, only my processor.
When I increase the SetMemoryMax(1024) and keep it single threaded, then I get this;
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
bench: utime=43.219s
squid_80
13th January 2007, 06:30
With the original script (setmemorymax(128)): 319.922s
Original script with setmtmode(4): 186.375s
Original script with setmemorymax(896): 42.297s
setmemorymax(896) and setmtmode(4): 23.563s
BigDid
13th January 2007, 06:40
Grmbl :mad: :scared: :sly:
After 2 days, finally got it running, the plugins were not loading!
Proc (see my signature) is an athlon64 running @ 2420
Ram is 2*512 DDR3200 Corsair- 2.2.2.5.1 @ 201.6Mhz
With setmemorymax(128)
Specification AMD Athlon(tm) 64 Processor 3000+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
bench: utime=584.594s
With setmemorymax(384)
bench: utime=278.047s
With setmemorymax(768)
bench: utime=88.735s
Did
Adub
13th January 2007, 06:47
setmemorymax896 and setmtmode(4): 23.047s
This is wierd.
Edit: Wait, what speed are you running your ram at? My is at stock DDR2 667 Mhz.
foxyshadis
13th January 2007, 10:01
Edit: Oh, why does ffmpeg crash when SetMemoryMax(2000) is set? Or Higher?
32-bit address limitations, no single process can access more than 2gb without the /3gb startup switch (and proper coding) or segmented addressing.
Avisynth is good and will complain about being out of memory, but if another plugin that doesn't check for allocation failure gets there's first, it goes boom.
krieger2005
13th January 2007, 12:31
I'm going to modify the batch file to determine installation locations so the loadplugin commands are more specific.
Is it possible to integrate a more than one encoding (at one time) and take than the average result? I think this would be usefull to get more proper results.
Pookie
14th January 2007, 01:24
Modified to run 3 benchmarks - 25% Total RAM, 50% Total RAM, 100% Total RAM-64MB. The results will now be affected by at least 7 variables -> 1) CPU speed 1a) Number of CPUs 2) L2 cache 3) Amount of RAM 3a) Speed of RAM 4) Version of Windows 5) Version/Build of Avisynth. We're going to learn something that's completely profound and original- a fast CPU with lots of fast RAM will give us fast avisynth performance.:D
You'll need to copy the newly modified Runme.bat and test.avs files from the first page. I removed the setmemorymax line, as it is dynamically re-added later in the batch file, depending on the amount of RAM is installed on your system.
I had to hardwire the loadplugin commands at the start of the script. I didn't want to risk doing a registry query with reg.exe.
Change the script to reflect the location of your plugins if it is different. Hopefully, 90% of everyone won't need to touch it.
If you wish to test Dual/Duo Core functionality, make whatever modifications you wish to the script and enjoy. As we all know, mvtools is not compatible with MT in actual use, so the results aren't going to reflect real life performance.
Specification Intel(R) Pentium(R) 4 CPU 2.40GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed=1024MB
256MB memory
bench: utime=317.546s
512MB memory
bench: utime=244.561s
960MB memory
bench: utime=260.415s
BigDid
14th January 2007, 05:45
Hi,
I have a script error (when testing test.avs alone) with this line in the new script:
base = last.minblur(2).FluxSmoothT(9)
Beside that the 3time script seems to work...
Did
Pookie
14th January 2007, 07:58
Try this version of fluxsmooth.dll 73,728 bytes
http://www.bestsharing.com/files/YbncY203554/fluxsmooth_.rar.html
And add a few carriage returns to the bottom your test.avs file. buzzqw pointed out the batch was cutting off the bottom line, causing the encode to fail.
buzzqw
14th January 2007, 08:52
@Pookie
please edit the runme.bat
it will download the ffmpeg 7375 but un7zip another version
BHH
EDIT: when creating 256test.avs the last } is missing
squid_80
14th January 2007, 09:38
EDIT: when creating 256test.avs the last } is missing
I got that too. Just add a new line to the end of test.avs.
Also win64 doesn't like lmod - I guess because it's a 16-bit program. I made a file called memory.bat:
set tmem=%2
and modified runme.bat:type cpuinfo.txt | findstr "L1" >>result.txt
type cpuinfo.txt | findstr "L2" >>result.txt
type cpuinfo.txt | findstr "Memory Size">>memory.txt
type memory.txt | findstr "Size" | findstr "Memory" > tmem.bat
call tmem.batand the results:Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
512MB memory
bench: utime=104.250s
1024MB memory
bench: utime=41.937s
1984MB memory
bench: utime=41.563s
@Merlin7777: I'm using stock DDR2 PC5300.
Pookie
14th January 2007, 10:24
Many thanks, buzzqw and squid_80. Done. :o
Weirdness... I just added a #end remark which gets amputated during the concatenation, but now it preserves the } for the script to run properly.
Specification Intel(R) Xeon(TM) CPU 3.06GHz
Specification Intel(R) Xeon(TM) CPU 3.06GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed= 2048MB
512MB memory
bench: utime=218.922s
1024MB memory
bench: utime=100.156s
1984MB memory
bench: utime=101.234s
Looks like diminishing returns after 50% RAM allocation.
Moitah
14th January 2007, 10:49
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed= 2048MB
512MB memory
bench: utime=93.766s
1024MB memory
bench: utime=38.140s
1984MB memory
bench: utime=38.266s
Using official 2.5.7, 2x1GB PC6400 DDR2.
check
14th January 2007, 14:14
It seems the script uses less than 1gb. It would be interesting to do tests for every 64mb increase :)
squid_80
14th January 2007, 14:23
Damn, I've been de-throned!
Using official 2.5.7, 2x1GB PC6400 DDR2.
Which cost more, the RAM or the CPU?
BigDid
14th January 2007, 18:05
... And add a few carriage returns to the bottom your test.avs file. buzzqw pointed out the batch was cutting off the bottom line, causing the encode to fail.
Grmbl again
I believe it was the truncated end of script that made the error.
I now have no error but here is the result:
"Total Memory Installed=MB
MB memory
bench: utime=0.063s
MB memory
bench: utime=0.063s
-64MB memory
bench: utime=0.062s"
And here is the test.avs after being run:
"setmemorymax()
˙˙˙˙˙˙˙˙"
I have alot of files in the directory I made for the circumstance so I will wipe out anything but what is needed and try again
Edit: wiped out extra files, no good.
Did
ThEdEaDLiVe
14th January 2007, 19:07
Specification AMD Athlon(tm) 64 X2 Dual Core Processor 5000+
L1 Data cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 2 x 512 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed= 2048MB
512MB memory
bench: utime=234.297s
1024MB memory
bench: utime=59.328s
1984MB memory
bench: utime=59.797s
Moitah
14th January 2007, 19:15
Which cost more, the RAM or the CPU?
CPU was $315, RAM was $280 - $50 mail-in rebate (which I received) = $230. USD.
Adub
14th January 2007, 19:17
It looks like ram speed is highly influential. Moitah is using 800mhz ram while I am only using 667mhz ram. Will post results when finished.
Moitah
14th January 2007, 22:33
I'm not sure the RAM speed makes that much difference, I tried underclocking it and here's what I came up with:
http://www.moitah.net/misc/tmem1.png
(underclocked)
1st run:
1024MB memory
bench: utime=37.562s
2nd run:
1024MB memory
bench: utime=37.344s
http://www.moitah.net/misc/tmem2.png
(normal speed)
1st run:
1024MB memory
bench: utime=37.204s
2nd run:
1024MB memory
bench: utime=37.250s
The reason why the time is better than what I posted last night is because I rebooted right before each set of tests and had nothing else running.
At both speeds, the 1984MB always failed at the same place, ffmpeg just exited silently. I tried the script in AVS2AVI and it errored with "AVIReadStream failed", and in VirtualDub with "Run video analysis pass" I got this: The instruction at "0x004bdc4d" referenced memory at "0x0000000d". The memory could not be "read". Only with nothing running though, once I start Firefox to use a bit more RAM, it works fine.
Pookie
14th January 2007, 23:22
Moitah & squid_80 - You guys are running different versions of XP (32 vs 64), right ? That could account for the performance difference.
Bigdid - It is almost like you don't have environment space to hold variables. Findstr.exe might be corrupt or missing. It comes with Windows, do a search. It might also not be in your path. Copy it to the same folder that you're running the benchmark in. (Sorry to end a sentence with a preposition):D
ThEdEaDLiVe - Thanks for posting, and welcome to the forums.
Fizick
15th January 2007, 06:07
Hm..., why test script is too slow?
may be more simple regular script would be more useful (and fast).
Moitah
15th January 2007, 06:44
@Pookie: Yeah, that could be it. I'm running Windows Server 2003 Standard, 32 bit.
BigDid
15th January 2007, 06:57
... Findstr.exe might be corrupt or missing. It comes with Windows, do a search. It might also not be in your path. Copy it to the same folder that you're running the benchmark in...
Correct,
It was missing from Win/System32
It is now in both places but result is same as in my previous post
Test incomplete and test.avs being truncated ...
I can live without the test results, but if other ideas, I am willing to try it.
Did
Pookie
15th January 2007, 07:45
Fizick - 30 frames only(x3), but very stressful on the cpu. My P4 2.4 with 1GB of RAM is in last place. :D I had to find a script that would be difficult even for the fastest PCs out there, and we're testing machines that are 6 times faster at processing than the slowest one. It is also the slowest result that you can expect on a system (unless you combine it with a h264 1080p transcode), and with 30 frames it is easy to make your own calculations. I think it is a good script to test, because many people want to use the newest functions (mcbob, mvbob, mvdegrain2) and now they can see what kind of performance they can expect if they buy a new system.
Bigdid - Try this one. I tested it and it works for me.
BTW, the Runme.bat I included in that (now deleted) download is bad. I modified the one on the first page to work with more "native" windows apps.
BigDid
15th January 2007, 17:40
Bigdid - Try this one. I tested it and it works for me.
http://www.bestsharing.com/files/WL1WJm204176/avstest.zip.html
Thanks,
Worked for me too. I'll wait the actual encode to finish to redo the test and post results.
Did
Jeremy Duncan
16th January 2007, 00:05
Fizick - 30 frames only(x3), but very stressful on the cpu. My P4 2.4 with 1GB of RAM is in last place. :D I had to find a script that would be difficult even for the fastest PCs out there, and we're testing machines that are 6 times faster at processing than the slowest one. It is also the slowest result that you can expect on a system (unless you combine it with a h264 1080p transcode), and with 30 frames it is easy to make your own calculations. I think it is a good script to test, because many people want to use the newest functions (mcbob, mvbob, mvdegrain2) and now they can see what kind of performance they can expect if they buy a new system.
Bigdid - Try this one. I tested it and it works for me.
http://www.bestsharing.com/files/WL1WJm204176/avstest.zip.html
This test takes too long. I quit after over running the test more than a hour.
Pookie
16th January 2007, 00:40
Jeremy said "This test takes too long. I quit after over running the test more than a hour"
http://frogstar.soylentgeek.com/wav/scoobhuh.wav Modify TEST.AVS from trim(100,129) to trim(100,114) and multiply the results x2
Specification Dual Core AMD Opteron(tm) Processor 275
Specification Dual Core AMD Opteron(tm) Processor 275
L1 Data cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Data cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 2 x 1024 KBytes, 16-way set associative, 64-byte line size
L2 cache 2 x 1024 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed= 3072MB
768MB memory
bench: utime=99.734s
1536MB memory
bench: utime=66.906s
3008MB memory
bench: utime=66.953s
- Edit - Tested in 128MB Increments all the way to 3008MB
128MB
bench: utime=624.109s
256MB
bench: utime=331.000s
384MB
bench: utime=303.172s
512MB
bench: utime=264.140s
768MB
bench: utime=92.375s
896MB
bench: utime=70.047s
1024MB
bench: utime=66.953s
1152MB
bench: utime=66.797s
1152-3008 utimes stayed between 66.79 and 67.79s
BigDid
16th January 2007, 05:38
Here are the results :(
I'm afraid something is wrong cause the numbers do not reflect those of the first test: http://forum.doom9.org/showthread.php?p=933583#post933583
Same parameters
"Total Memory Installed=c MB
0MB memory
bench: utime=2881.938s
0MB memory
bench: utime=2878.078s
-64MB memory
bench: utime=2884.218s
"
Will try again with trim(100,114) and multiply the results x2 :rolleyes:
Did
Pookie
16th January 2007, 06:21
Nope, you're doing everything right. It was my fault. I'm probably driving you nuts with all of these issues. The "Runme.bat" File I posted in the link I gave you was bad. My apologies. Insomnia.
Replace the one you're using with one:
http://www.bestsharing.com/files/QJjwoO204794/runme.zip.html
Note: this link above is specific to BigDid's issues.
Jeremy Duncan
16th January 2007, 08:07
Modified TEST.AVS from trim(100,129) to trim(100,114) and multiplied the results x2
Specification Intel(R) Pentium(R) M processor 1.60GHz
L1 Data cache 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed=504 MB
126MB memory
bench: utime=590.438s
252MB memory
bench: utime=275.906s
440MB memory
bench: utime=385.094s
Jeremy Duncan
16th January 2007, 08:12
Here's a link to the necessary software. You'll still need to get the plugins yourself though.
Link (http://www.bestsharing.com/files/vd7YK204826/Pookies%20Test.zip.html)
BigDid
16th January 2007, 17:36
Hurray, Working,
Specification AMD Athlon(tm) 64 Processor 3000+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=1024 MB - about 750Mb free -
With latest revised script:
with 15 frames*2 (trim 100,114) and MVTools 1.4.13.0 :
256MB memory
bench: utime=85.641s *2=171.282s
512MB memory
bench: utime=41.656s *2=83.312s
960MB memory
bench: utime=49.875s *2=99.75s
Complete test with 30 frames (trim 100,129) and MVTools 1.4.13.0 :
256MB memory
bench: utime=271.516s (MVT 1.3.0.0=301.625s, MVT 1.6.2=321.547s )
512MB memory
bench: utime=209.062s (MVT 1.3.0.0=232.828s, MVT 1.6.2=255.329s )
960MB memory
bench: utime=153.640s (MVT 1.3.0.0=160.734s, MVT 1.6.2=102.390s)
Proc (see my signature) is an athlon64 running @ 2420
Ram is 2*512 DDR3200 Corsair- 2.2.2.5.1 @ 201.6Mhz
Avisynth 2.57 final
Plugins:
# MVTools 1.4.13.0
# MedianBlur 0.8.4.1
# MaskTools 2.0.30.0
# TTempSmooth 0.9.4.0
# FluxSmooth ? Pookie's rev
# RemoveGrain/Repair 1.0 SSE2
Did
zilog jones
16th January 2007, 17:45
Using files from the link above:
Specification Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed=1024 MB
256MB memory
bench: utime=94.266s *2 = 188.532s
512MB memory
bench: utime=32.953s *2 = 65.906s
960MB memory
bench: utime=33.281s *2 = 66.562s
(With every process and service I could think of turned off, leaving 177MB used by Windows when idle)
256MB memory
bench: utime=93.953s *2 = 187.906s
512MB memory
bench: utime=32.828s *2 = 65.656s
960MB memory
bench: utime=34.234s *2 = 68.468s
Chipset
------------------------------------------------------------------------------
Northbridge Intel i975X rev. C0
Southbridge Intel 82801GB (ICH7) rev. 01
Graphic Interface PCI-Express
PCI-E Link Width x16
PCI-E Max Link Width x16
Memory Type DDR2
Memory Size 1024 MBytes
Memory Frequency 267.0 MHz (1:1)
CAS# 4.0
RAS# to CAS# 4
RAS# Precharge 4
Cycle Time (tRAS) 12
Bank Cycle Time (tRC) 16
Memory SPD
------------------------------------------------------------------------------
DIMM #1
General
Memory type DDR2
Manufacturer (ID) Corsair (7F7F9E0000000000)
Size 512 MBytes
Max bandwidth PC2-4300 (266 MHz)
Part number VS512MB533D2
Attributes
Number of banks 1
Data width 64 bits
Correction None
EPP no
Timings table
Frequency (MHz) 200 266
CAS# 3.0 4.0
RAS# to CAS# delay 3 4
RAS# Precharge 3 4
TRAS 9 12
TRC 11 15
- DIMM 2 is identical - they're both Corsair Value Select PC2-4200 512MB.
- MSI 975X Platinum PowerUp Edition motherboard - I wish I bought one with a less convoluted name :)
- AviSynth 2.5.7
- WinXP SP2 32-bit
- WD Caviar SE16 7200RPM 320GB hard disk.
[edit]Forgot to multiply them! Results are a lot less impressive now compared to Moitah's :(
@Moitah: Could you try underclocking your RAM to 266MHz?
BigDid
16th January 2007, 18:17
... I don't know how I'm beating Moitah's times with a slower CPU and cheapskate Corsair Value Select PC2-4200 RAM (512MB pair) O_o...
Hi,
Maybe the timings? Would you mind to share your cpu-z memory tab screenshot?
Did
Jeremy Duncan
16th January 2007, 19:24
Your supposed to multiply the results by 2.
zilog jones
16th January 2007, 19:26
Oh :(
I was wondering what was going on with people doing that above...
Pookie
16th January 2007, 20:52
Oi, your system is fast enough to do a full test. Use the Test.avs on the first page. Thanks for testing, I was interested in particular at the E6300/E6400/E6600.
BTW, Intel is going to add 4MB L2 cache on all of their E series procs starting in March.
zilog jones
16th January 2007, 21:19
Ah, it all makes sense now :D
Here's the full test:
256MB memory
bench: utime=303.313s
512MB memory
bench: utime=156.204s
960MB memory
bench: utime=103.546s
<2nd try>
256MB memory
bench: utime=303.812s
512MB memory
bench: utime=156.860s
960MB memory
bench: utime=79.391s
...OK, it's significantly worse :confused:
alfixdvd
16th January 2007, 21:19
Specification Intel(R) Pentium(R) 4 CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed=2048 MB
512MB memory
bench: utime=317.172s
1024MB memory
bench: utime=86.078s
1984MB memory
bench: utime=75.843s
With SetMTMode(2)
Specification Intel(R) Pentium(R) 4 CPU 2.80GHz
L1 Data cache 8 KBytes, 4-way set associative, 64-byte line size
L2 cache 512 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed=2048 MB
512MB memory
bench: utime=225.907s
1024MB memory
bench: utime=69.609s
1984MB memory
ffmpeg crashes : Runtime Error at frame21
Pookie
16th January 2007, 22:22
You can't MT with MVtools. This test will probably run with MT 3, but in a real world application it won't. Maybe that should be a sticky somewhere.
Ah, it all makes sense now :D
...OK, it's significantly worse :confused:
Sigh... too many variables are getting changed. I think the 2 things we've learned from this benchmark are - 1) The E6000 series is the proc of choice if you're planning on taking advantage of the latest advancements in denoising and deinterlacing - namely motion compensated processing. 2) 50% of total RAM seems to be the optimal value to setmemorymax()
Jeremy Duncan
16th January 2007, 22:37
I ran the test on the first page.
Specification Intel(R) Pentium(R) M processor 1.60GHz
L1 Data cache 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed=504 MB
126MB memory
bench: utime=809.188s
252MB memory
bench: utime=432.422s
440MB memory
bench: utime=621.157s
Taurus
16th January 2007, 22:38
Just for grins:
I used the "stripped off" BigDid Version of the script on an old
AMD 2200XP out of curiosity.
What makes me wonder, after fiddling around with the avs plugins, there's quite a difference in speed by changing plugins around. Even holes in the encodes happen as you can see in the first lineup.
So I think everybody should use the same plugins and versions to make a fair test. Maybe somebody can throw a bundle of the "right" ones into this thread.
What amazes me the most, how memory allocation and bus speed pushes the numbers.
Multiply numbers by x2 please.
Specification AMD Athlon(tm) XP 2200+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 256 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=512 MB 133Mhz:D
-------------------------------------------------------
with new removegrain.dll:
128MB memory
bench: utime=626.971s
256MB memory
bench: utime=0.000s
448MB memory
bench: utime=241.718s
--------------------------------------------------------
with older removegrain.dll
128MB memory
bench: utime=644.537s
256MB memory
bench: utime=261.616s
448MB memory
bench: utime=252.723s
Pookie
16th January 2007, 22:51
Just for grins:
Maybe somebody can throw a bundle of the "right" ones into this thread.
That would be considered not adhering to the GNU General Public License. You can't just throw the needed ones into a zip file. And most of all, I have too much respect for the plugin developers to even seem like I'm dissing them. It really ruffles their feathers.
Then there's the version of Windows, Avisynth, etc...
ThEdEaDLiVe
17th January 2007, 02:53
Using modified test.avs from trim(100,129) to trim(100,114) and tested after fresh reboot.
Specification AMD Athlon(tm) 64 X2 Dual Core Processor 5000+
L1 Data cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 2 x 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 2 x 512 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=2048 MB
512MB memory
bench: utime=48.875s *2 = 97.75s
1024MB memory
bench: utime=28.297s *2 = 56.594s
1984MB memory
bench: utime=27.859s *2 = 55.718s
Moitah
17th January 2007, 03:12
@Moitah: Could you try underclocking your RAM to 266MHz?
1024 MB Test
266 MHz:
1st: 37.735s
2nd: 37.797s
400 MHz:
1st: 37.375s
2nd: 37.032s
I am using MVTools 1.4.13. I noticed when I tried MVTools 1.6.2 it took more like 50 seconds. I'll list the versions of everything:
Windows Server 2003 Standard (32 bit)
Tritical's Modified AviSynth 2.5.7 CVS 2006-Nov-25 - I switched back to this after upgrading to 2.5.7 final official because it was crashing not only during the 1984 MB test but also during my normal encodes.
MVTools 1.4.13
MedianBlur 0.8.4
MaskTools 2.0a30
TTempSmooth 0.9.4
FluxSmooth 1.1b
RemoveGrain/Repair 0.9 SSE3 - I tried it with the regular (non-SSE3) versions, didn't seem to make a noticeable speed difference on this script.
Pookie
17th January 2007, 03:23
Now THAT is good information. Perhaps it'll save someone a few $ when they are configuring RAM for a new Avisynth machina...
*.mp4 guy
17th January 2007, 03:53
I said before that it was the timings, not the bandwidth of the ram that mattered for stuff like this, could you please test the ram with different timings and see how that effects the speed.
Adub
17th January 2007, 03:58
Have we established completely why Moitah is faster than Squid_80 and my test results, even thought the ram speed doesn't really matter. Also, It seems that my over clocked processor was performing about the same as stock, what the heck is going on?
Note: I do have Folding@Home running in the back ground. Maybe if I kill it it will improve my numbers, but I doubt it. It is a idle priority program.
Pookie
17th January 2007, 04:23
You're going to have to look at other variables, Merlin. Moitah's running Windows 2003; Squid_80 is running XP64. That alone is a huge difference. In your case, perhaps the L2 cache size is more important than raw clock speed when you hit these amazing numbers. Another possibility is that you've hit a ceiling of performance and the bottleneck is software related. Does the E6000 series throttle back during periods of high heat ? ffmpeg can turn a proc into a furnace.
Moitah
17th January 2007, 06:23
I said before that it was the timings, not the bandwidth of the ram that mattered for stuff like this, could you please test the ram with different timings and see how that effects the speed.
333 MHz, 4-4-4-12: 37.406s, 37.062s, 37.078s
333 MHz, 5-5-5-15: 37.531s, 37.297s, 37.375s
I tried to test 6-6-6-18 but it was too unstable. Shouldn't it have been more stable, if anything? I'm not really into overclocking, so I'm not sure.
*.mp4 guy
17th January 2007, 07:14
Some ram can be very particular about the timings, most ram modules work best with certain timings, but usually using more lax timings then the optimal ones doesn't result in instability. Anyway All I can surmise from your results is that c2ds don't benefit from tight timings like amd64's do, It looks like your fsb might be saturated, but at this point I'm just guessing.
Taurus
17th January 2007, 07:42
That would be considered not adhering to the GNU General Public License. You can't just throw the needed ones into a zip file. And most of all, I have too much respect for the plugin developers to even seem like I'm dissing them. It really ruffles their feathers.
Then there's the version of Windows, Avisynth, etc...
You're right, but for the sake of compatibility it would be nice to know the versions and/or date of release.
Never mind, it's a nice benchmark anyway. Thanks for your work and intelligence.
Audionut
17th January 2007, 09:27
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=1024 MB
256MB memory
bench: utime=181.703s
512MB memory
bench: utime=98.719s
960MB memory
bench: utime=89.906s
OC to 3.15GHz, with crap ram running at 702mhz 5-6-6-17 timings.
Using sse3 versions of removegrain and repair.
edit: using all the scripts etc from the first page. It looks as if you guys might be using different ones now.
Pookie
17th January 2007, 10:04
Those are very different results than the other 6600s. That's more than software version differences affecting your performance. Open the "Cpuinfo.txt" file that gets created and have a look at the first section for any anomalies. If something looks really off, like bus or core speeds, your bios settings might need adjustment.
*.mp4 guy
17th January 2007, 10:16
I think its becuase he didn't use multithreading.
Audionut
17th January 2007, 10:17
Here it is. minus a few lines to bring the post under 16000 characters long.
-------------------------
CPU-Z version 1.38
-------------------------
Processors Map
------------------------------------------------------------------------------------
Number of processors 1
Number of threads 2
Processor 0
-- Core 0
-- Thread 0
-- Core 1
-- Thread 0
Processors Information
------------------------------------------------------------------------------------
Processor 1 (ID = 0)
Number of cores 2
Number of threads 2 (max 2)
Name Intel Core 2 Duo E6600
Codename Conroe
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
Package Socket 775 LGA (platform ID = 0h)
CPUID 6.F.6
Extended CPUID 6.F
Core Stepping B2
Technology 65 nm
Core Speed 3150.1 MHz (9.0 x 350.0 MHz)
Rated Bus speed 1400.0 MHz
Stock frequency 2400 MHz
Instructions sets MMX, SSE, SSE2, SSE3, SSSE3, EM64T
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
FID/VID Control yes
FID range 6.0x - 9.0x
max VID 1.213V
Thread dumps
------------------------------------------------------------------------------------
CPU Thread 0
APIC ID 0
Topology Processor ID 0, Core ID 0, Thread ID 0
Type 01008001h
Max CPUID level 0000000Ah
Max CPUID ext. level 80000008h
CPU Thread 1
APIC ID 1
Topology Processor ID 0, Core ID 1, Thread ID 0
Type 01008001h
Max CPUID level 0000000Ah
Max CPUID ext. level 80000008h
Chipset
------------------------------------------------------------------------------
Northbridge Intel P965/G965 rev. C1
Southbridge Intel 82801HB (ICH8) rev. 02
Graphic Interface PCI-Express
PCI-E Link Width x16
PCI-E Max Link Width x16
Memory Type DDR2
Memory Size 1024 MBytes
Memory Frequency 350.0 MHz (1:1)
CAS# 5.0
RAS# to CAS# 6
RAS# Precharge 6
Cycle Time (tRAS) 17
Memory SPD
------------------------------------------------------------------------------
DIMM #1
General
Memory type DDR2
Manufacturer (ID) (0000000000000000)
Size 512 MBytes
Max bandwidth PC2-5300 (333 MHz)
Part number
Attributes
Number of banks 1
Data width 64 bits
Correction None
EPP no
Timings table
Frequency (MHz) 200 266 333
CAS# 3.0 4.0 5.0
RAS# to CAS# delay 3 4 5
RAS# Precharge 3 4 5
TRAS 9 12 15
TRC 12 16 20
DIMM #2
General
Memory type DDR2
Manufacturer (ID) (0000000000000000)
Size 512 MBytes
Max bandwidth PC2-5300 (333 MHz)
Part number
Attributes
Number of banks 1
Data width 64 bits
Correction None
EPP no
Timings table
Frequency (MHz) 200 266 333
CAS# 3.0 4.0 5.0
RAS# to CAS# delay 3 4 5
RAS# Precharge 3 4 5
TRAS 9 12 15
TRC 12 16 20
Dump Module #1
0 1 2 3 4 5 6 7 8 9 A B C D E F
00 80 08 08 0E 0A 60 40 00 05 30 45 00 82 08 00 00
10 0C 04 38 01 02 00 03 3D 50 50 60 3C 1E 3C 2D 80
20 20 28 10 18 3C 1E 1E 00 00 3C 69 80 18 22 00 00
30 00 00 00 00 00 00 00 00 00 00 00 00 00 00 12 73
40 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
50 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
60 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
70 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Dump Module #2
0 1 2 3 4 5 6 7 8 9 A B C D E F
00 80 08 08 0E 0A 60 40 00 05 30 45 00 82 08 00 00
10 0C 04 38 01 02 00 03 3D 50 50 60 3C 1E 3C 2D 80
20 20 28 10 18 3C 1E 1E 00 00 3C 69 80 18 22 00 00
30 00 00 00 00 00 00 00 00 00 00 00 00 00 00 12 73
40 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
50 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
60 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
70 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
DMI
------------------------------------------------------------------------------
DMI BIOS
--------
vendor Award Software International, Inc.
version F9
date 01/11/2007
DMI System Information
----------------------
manufacturer Gigabyte Technology Co., Ltd.
product 965P-DQ6
version unknown
serial unknown
UUID FFFFFFFF-FFFFFFFF-FFFFFFFF-FFFFFFFF
DMI Baseboard
-------------
vendor Gigabyte Technology Co., Ltd.
model 965P-DQ6
revision unknown
serial unknown
DMI System Enclosure
--------------------
manufacturer Gigabyte Technology Co., Ltd.
chassis type Desktop
chassis serial unknown
DMI Processor
-------------
manufacturer Intel
model Intel(R) Core(TM)2 CPU 660
clock speed 3150.0MHz
FSB speed 350.0MHz
multiplier 9.0x
DMI Memory Controller
---------------------
correction 8-bit parity
Max module size 1024MBytes
DMI Memory Module
-----------------
designation A0
DMI Memory Module
-----------------
designation A1
size 512MBytes (single bank)
DMI Memory Module
-----------------
designation A2
DMI Memory Module
-----------------
designation A3
size 512MBytes (single bank)
DMI Port Connector
------------------
designation PRIMARY IDE (internal)
connector On Board IDE
DMI Port Connector
------------------
designation SECONDARY IDE (internal)
connector On Board IDE
DMI Port Connector
------------------
designation FDD (internal)
port type 8251 FIFO Compatible
connector On Board Floppy
DMI Port Connector
------------------
designation COM1 (internal)
port type Serial Port 16450
connector 9 Pin Dual Inline (pin 10 cut)
connector DB-9 male
DMI Port Connector
------------------
designation COM2 (internal)
port type Serial Port 16450
connector 9 Pin Dual Inline (pin 10 cut)
connector DB-9 male
DMI Port Connector
------------------
designation LPT1 (internal)
port type Parallel Port ECP/EPP
connector DB-25 female
connector DB-25 female
DMI Port Connector
------------------
designation Keyboard (internal)
port type Keyboard Port
connector PS/2
DMI Port Connector
------------------
designation PS/2 Mouse (internal)
designation Detected (external)
port type Mouse Port
connector PS/2
connector PS/2
DMI Port Connector
------------------
designation USB (internal)
port type USB
connector Access Bus (USB)
DMI Port Connector
------------------
designation USB (internal)
port type USB
connector Access Bus (USB)
DMI Extension Slot
------------------
designation PCI
type PCI
width 32 bits
populated no
DMI Extension Slot
------------------
designation PCI
type PCI
width 32 bits
populated yes
DMI Extension Slot
------------------
designation PCI
type PCI
width 32 bits
populated no
DMI Extension Slot
------------------
designation PCI
type PCI
width 32 bits
populated no
DMI Physical Memory Array
-------------------------
location Motherboard
usage System Memory
correction None
max capacity 4096MBytes
max# of devices 4
DMI Memory Device
-----------------
designation A0
format DIMM
type unknown
DMI Memory Device
-----------------
designation A1
format DIMM
type unknown
total width 64bits
data width 64bits
size 512MBytes
DMI Memory Device
-----------------
designation A2
format DIMM
type unknown
DMI Memory Device
-----------------
designation A3
format DIMM
type unknown
total width 64bits
data width 64bits
size 512MBytes
Software
------------------------------------------------------------------------------
Windows Version Microsoft Windows XP Professional Service Pack 2 (Build 2600)
DirectX Version 9.0c
Resources
------------------------------------------------------------------------------
Port I/O Space, BA=0xCF8, size=0x8
Port I/O Space, BA=0xD000
Port I/O Space, BA=0xC000
Port I/O Space, BA=0xC400
Port I/O Space, BA=0xC800
Port I/O Space, BA=0xCC00
Port I/O Space, BA=0xF000
Port I/O Space, BA=0xFC00
Port I/O Space, BA=0x500
Port I/O Space, BA=0xD800
Port I/O Space, BA=0xDC00
Port I/O Space, BA=0xE000
Port I/O Space, BA=0xE400
Port I/O Space, BA=0xE800
Port I/O Space, BA=0xEC00
Port I/O Space, BA=0x7000
Port I/O Space, BA=0x9000
Port I/O Space, BA=0x9400
Port I/O Space, BA=0x9800
Port I/O Space, BA=0x9C00
Port I/O Space, BA=0xA000
Port I/O Space, BA=0xB000
Port I/O Space, BA=0x408, size=0x4
Port I/O Space, BA=0x480, size=0x40
Memory I/O Space, BA=0xFB105000
Memory I/O Space, BA=0xFB100000
Memory I/O Space, BA=0xFB104000
Memory I/O Space, BA=0xFB106000
Memory I/O Space, BA=0xF4000000
Memory I/O Space, BA=0xE0000000
Memory I/O Space, BA=0xF5000000
Memory I/O Space, BA=0xFB000000
Memory I/O Space, BA=0xFA000000
Memory I/O Space, BA=0xF0000000, size=0x1000
Memory I/O Space, BA=0xFEE00000, size=0x1000
SMBus I/O Space, BA=0x500
PCI Express I/O Space, BA=0xF0000000, size=0x1000
PCI I/O space, BA=0xCF8, size=0x8
It was only using one cpu/thread while running the test.
Audionut
17th January 2007, 10:25
Here's the results using setmtmode(2,0)
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=1024 MB
256MB memory
bench: utime=129.687s
512MB memory
bench: utime=67.109s
960MB memory
bench: utime=55.937s
squid_80
17th January 2007, 10:30
I only used multithreading in the posts where I said so. I made this post (http://forum.doom9.org/showthread.php?p=933579#post933579) to spell that out pretty clearly.
Try winding the FSB back to stock speed and set the memory timings back to auto and see what happens.
Audionut
17th January 2007, 10:34
Using 256MB of memory, my overclocked cpu is quite a bit faster.
However when using more memory, my cheap shit house memory shows just how cheap and crap it is.
edit: I can't reboot yet, but will try a few things on the weekend.
BigDid
17th January 2007, 18:21
Hi, some updates
I re-arrangered my previous post: http://forum.doom9.org/showthread.php?p=937879#post937879 to have all the infos grouped, borrowed the good idea from Moitah
@Pookie, do you have the version number for the fluxsmooth.rar you have uploaded?
Now that it is working and to avoid the the scattered infos I will try to update my previous post only if some changes.
One could be using the 1.4.13 mvtools instead of the 1.3.00 if quicker.
Another would be finding why I (and a few others) are having better results with more ram, even more than half of the total installed; atm I think about the old memory manager I use (optimize memory) but I could be wrong.
If/when I change my gear to go on the darkside twice (Intel C2d) :D I will try to make another summary.
Did
Pookie
17th January 2007, 20:37
It has no version info associated with it.
Created:Tuesday, August 23, 2005, 9:44:24 PM
72.0 KB (73,728 bytes)
I think your performance results are consistent with your processor and ram. I've got a similar machine at work and I'll do a comparison.
Taurus
17th January 2007, 21:08
I think it's useless comparing speed as long as not everybodies using the same setup. Look at my post one page back.
By just changin around some plugins I've got a performance boost:
-------------------------------------------------------
Before optimizing:
-------------------------------------------------------
Specification AMD Athlon(tm) XP 2200+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 256 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=512 MB 133Mhz
-------------------------------------------------------
with new removegrain.dll:
-------------------------------------------------------
128MB memory
bench: utime=626.971s
256MB memory
bench: utime=0.000s
448MB memory
bench: utime=241.718s
--------------------------------------------------------
with older removegrain.dll
--------------------------------------------------------
128MB memory
bench: utime=644.537s
256MB memory
bench: utime=261.616s
448MB memory
bench: utime=252.723s
--------------------------------------------------------
after optimizing:
--------------------------------------------------------
128MB memory
bench: utime=425.972s
256MB memory
bench: utime=230.962s
448MB memory
bench: utime=125.360s
--------------------------------------------------------
Of course: Same script/same environment.
Don't forget to multiply the numbers x2 :D
Pookie
17th January 2007, 21:33
What does "after optimizing" mean ?
Taurus
17th January 2007, 21:43
What does "after optimizing" mean ?
just fiddling around with the avisynth plugins......
Trial and error principle.
BigDid
17th January 2007, 22:06
just fiddling around with the avisynth plugins......
Trial and error principle.
Hi Taurus and happy new year,
Would you mind giving your plugins versions and/or your other optimizations?
I changed the mvtools from 1.3.0 to 1.4.13 and got around 10% less time, will update my summary post with complete data later.
Did
Pookie
18th January 2007, 00:44
Specification AMD Athlon(tm) 64 Processor 3000+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed= 3072MB
768MB memory
bench: utime=107.595s
1536MB memory
bench: utime=77.892s
3008MB memory
bench: utime=91.692s
---------------------------
Using Bigdid's memory amounts:
256test
bench: utime=389.350s
512test
bench: utime=309.475s
960test
bench: utime=80.266s
----------------------------------------------------
medianblur.dll Version: 0,8,4,1
mt_masktools.dll Version: 2,0,30,0
TTempSmooth.dll Version: 0,9,4,0
mvtools.dll Version: 1,6,2,0
The rest of the Dlls won't return version info
If you want to check dll versions, download
http://helpdesk.kixtart.org/Download/Utils/GetVers.exe
At the command prompt, type in the following
for %a in (*.dll) do getvers %a >>dllvers.txt
Adub
18th January 2007, 05:33
medianblur=0.8.4.1
mt_masktools=2.30
ttempsmooth=0.9.4
mvtools=1.6.2.0
No multithreading.
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=2048 MB
512MB memory
bench: utime=158.766s
1024MB memory
bench: utime=43.750s
1984MB memory
bench: utime=43.532s
Multithreading:
setMTmode(2)
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=2048 MB
512MB memory
bench: utime=69.328s
1024MB memory
bench: utime=21.500s
1984MB memory
bench: utime=21.750s
Moitah
18th January 2007, 06:49
mvtools=1.6.2.0
That explains it... your result is quite good then.
Pookie
18th January 2007, 07:28
Amazing performance, but remember, it wouldn't work on a "real" video file because of the MVtools/MT incompatibility issue. Doesn't change the results - if you got the dough, the E6600 is the way to go....
Also mentioned in another thread - the E6300 and E6400 are going to get 4MB L2 cache as a standard beginning in Q2 2007
BigDid
18th January 2007, 19:10
Regarding MVT 1.6.2, comparison updated from here: http://forum.doom9.org/showthread.php?p=937879#post937879
Complete test with 30 frames (trim 100,129) and MVTools 1.4.13.0 :
256MB memory
bench: utime=271.516s (MVT 1.3.0.0=301.625s, MVT 1.6.2=321.547s )
512MB memory
bench: utime=209.062s (MVT 1.3.0.0=232.828s, MVT 1.6.2=255.329s )
960MB memory
bench: utime=153.640s (MVT 1.3.0.0=160.734s, MVT 1.6.2=102.390s)
For an Atlon monocore, MVT 1.4.13 is always better than MVT 1.3.0.0; and is better than MVT 1.6.2 excepted for 960Mb
Anybody could confirm for another config and/or more RAM?
Did
Taurus
18th January 2007, 21:15
Regarding MVT 1.6.2, comparison updated from here: http://forum.doom9.org/showthread.php?p=937879#post937879
Complete test with 30 frames (trim 100,129) and MVTools 1.4.13.0 :
256MB memory
bench: utime=271.516s (MVT 1.3.0.0=301.625s, MVT 1.6.2=321.547s )
512MB memory
bench: utime=209.062s (MVT 1.3.0.0=232.828s, MVT 1.6.2=255.329s )
960MB memory
bench: utime=153.640s (MVT 1.3.0.0=160.734s, MVT 1.6.2=102.390s)
For an Atlon monocore, MVT 1.4.13 is always better than MVT 1.3.0.0; and is better than MVT 1.6.2 excepted for 960Mb
Anybody could confirm for another config and/or more RAM?
Did
That's exactly what I'v got too:
----------------------------------------
128MB memory
bench: utime=425.552s
256MB memory
bench: utime=229.490s
448MB memory
bench: utime=191.626s Mvtools 1.4.13.0
----------------------------------------
128MB memory
bench: utime=464.598s
256MB memory
bench: utime=254.927s
448MB memory
bench: utime=137.037s MVtools 1.6.2.0
----------------------------------------
I guess we all should stay at this plugins for benchmark purposes:
Fluxsmooth.dll Vers.: unknown version build date: Thursday, 9. November 2006 Size: 72,0 KB (73.728 Bytes)
Mt_masktools.dll Vers.: 2.0.30.0
Medianblur.dll Vers.: 0.8.4.1
Mvtools.dll Vers.: 1.6.2.0
Removegrain.dll Vers.: unknown version build date: Sunday, 1. Mai 2005 Size: 23,0 KB (23.552 Bytes) from authors Website (0.9)
Repair.dll Vers.: unknown version build date: Sunday, 1. Mai 2005 Size: 21,5 KB (22.016 Bytes) from authors Website (0.9)
TTempSmooth.dll Vers.: 0.9.4.0
zilog jones
18th January 2007, 22:02
Jeremy Duncan already tried to start off a more controlled approach to this test using specific plugin versions in this thread (http://forum.doom9.org/showthread.php?t=120819).
BigDid
18th January 2007, 22:39
Jeremy Duncan already tried to start off a more controlled approach...
Hi,
I will let the "controlled approach" to it's author.
...
I guess we all should stay at this plugins for benchmark purposes:
...
Mvtools.dll Vers.: 1.6.2.0
Precisely this is the point aka in real time what will be the quickest settings/plugins? I'm not sure 1.6.2 is the answer.
Until now I had a setmemorymax.avsi set to 192 but that was for avisynth 2.56; what for 2.57? no setmemorymax needed cause no more memory leak?
If this is true maybe Pookie could give a modified runme.bat (adding a no fixed mem size, and/or returning optimal mem size used internally) to verify this particular point?
I still have a long way to go to master avisynth so that's why this thread is interesting :)
Did
zilog jones
18th January 2007, 23:15
I will let the "controlled approach" to it's author.
I was just saying, as a response to Taurus suggesting we use specific versions of the plugins that Jeremy already suggested the exact same thing (and put it in a less cluttered new thread). If we did this along with not changing the order of loading the plugins and whatever else we'd have a chance of getting more comparable results.
Pookie
19th January 2007, 03:08
It's all good... I sorta knew this thread was going to splinter into a bunch of different directions once we uncapped the 128MB memory limits. As far as "Dunc's" branch is concerned - I think the train has already left the station on this issue. Besides, if you like Peter Sellers movies, you can't help but be entertained.
Remember, the original goal was to determine the fastest processor if one was going to buy/build a new machine today. Done.
Changing file versions of the Avisynth plugin dlls is interesting, but again, there's no way to standardize the other O/S based dlls. We're all running different versions of Windows, including different service packs. Then there's the issue of Mvtools 1.62 vs 1.4x.
You're trading speed for quality on that one, especially during fade-ins/outs. Is it worth the 10% speedup/penalty? Depends on you and your source file.
Regarding 2.56 vs 2.57 - I don't know. Notice Moitah's posting regarding having to switch to Tritical's 2.57 build from the official because of a crash he was experiencing at total RAM-64MB (the third test). Then again, he's running Windows 2003 server, so who knows how that is affecting RAM allocation with Avisynth. I used those percentages for the test due to an old posting from Ian B stating that the largest setmemorymax statement is Total Ram -16MB, and that using a number higher than total RAM would result in pagefile swapping. Perhaps that forumula changed in the latest version. Still, 50% of total RAM seems like a stable and performance optimized value to use, and that works for both 2.56 and 2.57 builds.
Pretty head-spinning - too many variables all around :D
zilog jones
19th January 2007, 23:27
Does SetMemoryMax take into account memory already used by other running (non-AviSynth) processes? If it doesn't (which I assume not), this could vary what the optimal settings will be greatly for people depending on what else they're running. For example, right now I'm just running Opera, iTunes and some other junk, and about 500MB is already being used :(
Pookie
20th January 2007, 02:58
You can download http://www.stereopsis.com/eatmem/eatmem.zip
Run it in another Command Prompt Window
Syntax : Eatmem 256 (or whatever amount you want to allocate). Wait until it has allocated the amount you have specified on the command line (it'll tell you), then run the Avisynth test. See if the results differ from your original test.
Taurus
20th January 2007, 12:54
Pretty head-spinning - too many variables all around :D
You never will get exact reasults as Pookie pointed out in his post.
There are to many variables to take into account.
But for a rough guideline on processor performance it's almost clear to see which ones on top.
So this is all information really needed.
Still it is interesting to see different plugins behavior on memory management.
My own tests clearly shows the script was getting slower as memory gets saturated. Page file swapping occurs at the -64 entry. Part of the OS and background tasks got swapped.
Whole system became sluggish.
So the only logical conclusion is to make sure you never allocate to much Setmemorymax=? to your scrypt.
Just watch taskmanager how much memory is already in use.
Maybe kill some unneccesary background tasks.
Kill networking, because if there's heavy disc traffic, no benchmark will show correct numbers.
Examples:
With almost all Background tasks and some services disabled:
------------------------------------
128MB memory
bench: utime=420.174s
256MB memory
bench: utime=227.607s
448MB memory
bench: utime=117.639s
------------------------------------
Normal windows environment:
------------------------------------
128MB memory
bench: utime=425.552s
256MB memory
bench: utime=229.490s
448MB memory
bench: utime=191.626s
-----------------------------------
Draw your own conclusions.
-----------------------------------
I've got the information I wanted from this thread.
Now I know which processor to buy if I can afford it:p
Thanks again Pookie for starting this.
mooms
21st January 2007, 02:24
wohoo i have the fastest processor in history !
Specification AMD Athlon(tm) 64 Processor 3500+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=1024 MB
256MB memory
bench: utime=0.828s
512MB memory
bench: utime=1.062s
960MB memory
bench: utime=0.625s
what am i doing wrong ?
Audionut
21st January 2007, 02:45
what am i doing wrong ?
Have you got all the required plugins?
Ensure that you are using the correct file path in the test.avs.
Taurus
21st January 2007, 18:16
what am i doing wrong ?
Is your test.avs playing in any mediaplayer? If not: look @ Audionut's post.
Audionut
22nd January 2007, 00:36
A little update. Bus speed 400mhz. Cpu 8x multi 3.2Ghz. Mem at 400mhz 5-6-6-18.
Specification Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 4096 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=1024 MB
256MB memory
bench: utime=131.250s
512MB memory
bench: utime=65.453s
960MB memory
bench: utime=48.937s
A fraction slower at 256mb, but about 7 secs faster at 960mb.
BigDid
23rd January 2007, 17:34
Same config as my summary post: http://forum.doom9.org/showthread.php?p=937879#post937879
Was:
with 15 frames (trim 100,114) and MVTools 1.4.13.0 :
256MB memory
bench: utime=85.641s
512MB memory
bench: utime=41.656s
960MB memory
bench: utime=49.875s
Same config + thunderbird+Firefox+trays programs+eatmem 256 (539Mb free):
256MB memory
bench: utime=129.594s
512MB memory
bench: utime=102.812s
960MB memory
bench: utime=235.047s
Same config + thunderbird+Firefox+trays programs+eatmem 256 + mem manager tuned up:
256MB memory
bench: utime=143.562s
512MB memory
bench: utime=59.000s
960MB memory
bench: utime=190.875s
Did
Pookie
23rd January 2007, 21:56
Wow, interesting info, BigDid. There are a few freeware apps out there which will allow you auto close a list of running apps and services which you specify. They are mostly intended to keep CPU utilization down for DVD burning, but they can certainly work as part of the transcoding procedure for Avisynth. FSautostart is a good one. You can tell it to stop all unecessary apps and svcs, run an avisynth script, then restart the same apps and svcs after the script is complete.
http://www.wspilots.com/index.php?page=4&action=file&file_id=944
http://www.softpedia.com/get/Tweak/System-Tweak/FSAutoStart.shtml
BigDid
23rd January 2007, 23:54
Wow, interesting info, BigDid. There are a few freeware apps out there which will allow you auto close a list of running apps and services which you specify.
Hi Pookie,
Thanks. I'll redo the test with an eatmem 384 or 512 to get less than 512 free mem.
... FSautostart is a good one. You can tell it to stop all unecessary apps and svcs, run an avisynth script, then restart the same apps and svcs after the script is complete.
http://www.wspilots.com/index.php?page=4&action=file&file_id=944
http://www.softpedia.com/get/Tweak/System-Tweak/FSAutoStart.shtml
Until now I was more concerned by free mem or allocated mem between avisynth and the encoder+codec; now I will also take tray programs into account :thanks:
Did
Edit: Have tuned the mem manager and updated the above results.
mooms
27th January 2007, 22:17
Is your test.avs playing in any mediaplayer? If not: look @ Audionut's post.
http://img297.imageshack.us/img297/774/sanstitrepm6.png
which plugin is causing that message ?
foxyshadis
27th January 2007, 22:45
A too-old version of mvtools.
mooms
28th January 2007, 05:31
A too-old version of mvtools.
i have 0.9.9.1 and i think is the latest version isn't it ?
Pookie
28th January 2007, 05:59
http://avisynth.org.ru/mvtools/mvtools.html
mooms
28th January 2007, 18:15
thanks Pookie that do the trick!
Specification AMD Athlon(tm) 64 Processor 3500+
L1 Data cache 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 512 KBytes, 16-way set associative, 64-byte line size
Total Memory Installed=1024 MB
256MB memory
bench: utime=292.438s
512MB memory
bench: utime=156.484s
960MB memory
bench: utime=95.578s
....with kaspersky emule sygate firefox and some others programs running in the background
i will test with no other proggies later
BigDid
3rd February 2007, 05:23
New config E4300 Core2duo + Asrock 775 DualVSTA + 2*512 DDR 400 RAM (same as old config). No memory manager
Specification Intel(R) Core(TM)2 CPU 4300 @ 1.80GHz
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
Core Stepping L2
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
Total Memory Installed= 1024MB
Same avisynth config as in this post http://forum.doom9.org/showthread.php?p=937879#post937879 except avisynth 2.57.3 from TSP http://www.tsp.person.dk/avisynthMT257.3.zip and MT 0.5 http://www.avisynth.org/tsp/MT_05.zip
With 15 frames (trim 100,114) and trays programs:
E4300@1800Mhz
256MB memory
bench: utime=29.187s
512MB memory
bench: utime=25.484s
960MB memory
bench: utime=25.641s
E4300@2400Mhz
256MB memory
bench: utime=20.031s
512MB memory
bench: utime=19.953s
960MB memory
bench: utime=19.843s
E4300@2400Mhz + SetMt(2)
256MB memory
bench: utime=9.890s
512MB memory
bench: utime=9.500s
960MB memory
bench: utime=9.640s
With 30 frames (trim 100,129) and trays programs
E4300@2400Mhz + SetMt(2)
256MB memory
bench: utime=19.906s
512MB memory
bench: utime=19.953s
960MB memory
bench: utime=19.688s
Did
Edit: to get more info on the config, see this thread: http://forum.doom9.org/showthread.php?goto=newpost&t=120509
Pookie
3rd February 2007, 06:11
You're beating the 6600s ?:confused:
BigDid
3rd February 2007, 13:07
You're beating the 6600s ?:confused:
It seems so :D
Keep in mind this is not my final XP as I had to install a fresh one. I actually have the SP2, the MS corrections and updates, my MB and ATI drivers and dx9, Virus-scan8, the .Net 1.1,the RM clock utility, Supercopier, netmeter and that's all.
Also the first thing I tuned was the memory: see the hardware thread. I believe the 1t and mem@200 gains a lot.
Following is a partial dump of the cpuzinfo report:
CPU-Z version 1.38
-------------------------
Processors Map
------------------------------------------------------------------------------------
Number of processors 1
Number of threads 2
Processors Information
------------------------------------------------------------------------------------
Processor 1 (ID = 0)
Number of cores 2
Number of threads 2 (max 2)
Name Intel Core 2 Duo E4300
Codename Conroe
Specification Intel(R) Core(TM)2 CPU 4300 @ 1.80GHz
Package Socket 775 LGA (platform ID = 0h)
CPUID 6.F.2
Extended CPUID 6.F
Core Stepping L2
Technology 65 nm
Core Speed 1596.9 MHz (6.0 x 266.1 MHz)
Rated Bus speed 1064.6 MHz
Stock frequency 1800 MHz
Instructions sets MMX, SSE, SSE2, SSE3, SSSE3, EM64T
L1 Data cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L1 Instruction cache 2 x 32 KBytes, 8-way set associative, 64-byte line size
L2 cache 2048 KBytes, 8-way set associative, 64-byte line size
FID/VID Control yes
FID range 6.0x - 9.0x
max VID 1.213V
Chipset
------------------------------------------------------------------------------
Northbridge VIA PT880 Pro rev. 00
Southbridge VIA VT8237A rev. 00
Graphic Interface AGP
AGP Revision 3.0
AGP Transfer Rate 8x
AGP SBA supported, enabled
Memory Type DDR
Memory Size 1024 MBytes
Memory Frequency 199.6 MHz (4:3)
DRAM Interleave 4-way
CAS# 3.0
RAS# to CAS# 3
RAS# Precharge 3
Cycle Time (tRAS) 8
Command Rate 1T
Memory SPD
------------------------------------------------------------------------------
DIMM #1
General
Memory type DDR
Manufacturer (ID) Corsair (7F7F9E0000000000)
Size 512 MBytes
Max bandwidth PC3200 (200 MHz)
Part number CMX512-3200C2
Attributes
Number of banks 2
Data width 64 bits
Correction None
Registered no
Buffered no
EPP no
Timings table
Frequency (MHz) 200
CAS# 3.0
RAS# to CAS# delay 3
RAS# Precharge 3
TRAS 8
DIMM #2 (same as dimm1)
Monitoring
------------------------------------------------------------------------------
Sensor Chip Winbond W83697HF
PCI Device List
------------------------------------------------------------------------------
Host Bridge
bus 0 (0x00), device 0 (0x00), function 0 (0x00)
Common header
Vendor ID 0x1106
Model ID 0x0308
Revision ID 0x00
PI 0x00
SubClass 0x00
BaseClass 0x06
Cache Line 0x08
Latency 0x08
Header 0x80
PCI header
Address 0 (memory) 0xD0000000
Subvendor ID 0x1849
Subsystem ID 0x0308
Int. Line 0x00
Int. Pin 0x00
Capabilities
AGP Capability
Version 3.5
Status enabled
Transfer rate 8x (max 8x)
Queue lenght 2 (max 32)
Dump
00 01 02 03 04 05 06 07 08 09 0A 0B
02 50 35 00 1B 0A 00 1F 02 03 00 00
Power Management Capability
Version 1.1
Dump
00 01 02 03 04 05 06 07
01 00 02 00 00 00 00 00
DMI
------------------------------------------------------------------------------
DMI BIOS
--------
vendor American Megatrends Inc.
version P2.10
date 10/27/2006
DMI System Information
----------------------
manufacturer To Be Filled By O.E.M.
product 775Dual-VSTA
version To Be Filled By O.E.M.
serial To Be Filled By O.E.M.
UUID 00020003-00040005-00060007-00080009
DMI Baseboard
-------------
vendor unknown
model 775Dual-VSTA
revision unknown
serial unknown
DMI System Enclosure
--------------------
manufacturer To Be Filled By O.E.M.
chassis type Desktop
chassis serial To Be Filled By O.E.M.
DMI Processor
-------------
manufacturer Intel
model Intel(R) Core(TM)2 CPU 4300 @ 1.80GHz
clock speed 2400.0MHz
FSB speed 200.0MHz
multiplier 12.0x
DMI Memory Controller
---------------------
correction 64-bit ECC
Max module size 2048MBytes
DMI Memory Module
-----------------
designation DIMM0
size 512MBytes (double bank)
DMI Memory Module
-----------------
designation DIMM1
size 512MBytes (double bank)
DMI Extension Slot
------------------
designation PCI1
type PCI
width 32 bits
populated no
DMI Extension Slot
------------------
designation PCI2
type PCI
width 32 bits
populated no
DMI Extension Slot
------------------
designation PCI3
type PCI
width 32 bits
populated no
DMI Extension Slot
------------------
designation PCI4
type PCI
width 32 bits
populated no
DMI Extension Slot
------------------
designation PCIE1
type A5
populated no
DMI Physical Memory Array
-------------------------
location Motherboard
usage System Memory
correction None
max capacity 4096MBytes
max# of devices 2
DMI Memory Device
-----------------
designation DIMM0
format DIMM
type SDRAM
total width 64bits
data width 64bits
size 512MByte
DMI Memory Device
-----------------
designation DIMM1
format DIMM
type SDRAM
total width 64bits
data width 64bits
size 512MBytes
Software
------------------------------------------------------------------------------
Windows Version Microsoft Windows XP Professional Service Pack 2 (Build 2600)
DirectX Version 9.0c
Did
BigDid
3rd February 2007, 19:45
Update!
1 change but big effect: changed net framework 1.1 (quicker) to net framework 2.0 (slower)
same config as post 155 above
With 30 frames (trim 100,129) and trays programs
.net 2.0
256MB memory
bench: utime=71.719s
512MB memory
bench: utime=39.657s
960MB memory
bench: utime=21.313s
.net 3.0
256MB memory
bench: utime=67.984s
512MB memory
bench: utime=33.562s
960MB memory
bench: utime=23.110s
Same in windows safe mode:
256MB memory
bench: utime=20.704s
512MB memory
bench: utime=20.422s
960MB memory
bench: utime=20.453s
Did
Terka
4th February 2007, 11:10
so which framework is faster? im not sure what results from your posts compare.
Pookie
4th February 2007, 11:50
BigDid, I suspect the transcoding is failing for you in Safe Mode. Might be interesting to note the size of the transcoded output file in normal and safe modes - betcha the Safe Mode one is small, maybe even 0 bytes ?
Aside from that, pretty nice to have a dual core that runs on DDR1 RAM. I've got tons of the stuff myself.
BigDid
4th February 2007, 18:32
so which framework is faster? im not sure what results from your posts compare.
Hi,
1.1 is faster than 2.0
Will install the 3.0 (2.0 compatible) and report if better.
If not 2 solutions: encode in safe mode or use the utility Pookie pointed out: FSautostart http://forum.doom9.org/showthread.php?p=943039#post943039
Edit: @Pookie
The files test.avi (258Mb) and out.avi (7.42Mb) have same size in normal or safemode for 15 frames.
Aside from that, pretty nice to have a dual core that runs on DDR1 RAM. I've got tons of the stuff myself. with (it seems) no real speed penalty for common use and an AGP slot.
Did
Adub
4th February 2007, 20:59
Oh, on my results, I was using .net version 3.0. In case that helps.
BigDid
4th February 2007, 21:25
Oh, on my results, I was using .net version 3.0. In case that helps.
Hi,
I updated the previous post http://forum.doom9.org/showthread.php?p=948720#post948720
Seems more or less same results (differences due to progs running or not, test run at Win startutp or not etc...)
I'll go for .net 3.0 or maybe not will see for encoding tests not just the avisynth script...
Did
Edit: Back to .net 2.0, a few percent better for encodes; once .net 3.0 is installed, you should have net 2.0 and below net 3.0
-> uninstall net 3.0 only and you will have net 2.0 still installed
Pookie
4th February 2007, 22:15
30 frame test = 15MB Resulting File
15 frame test = 7.5MB Resulting File
How would .NET versions affect the results of the test ?
E4300 selling for $180. Not bad ;)
BigDid
4th February 2007, 22:58
30 frame test = 15MB Resulting File
15 frame test = 7.5MB Resulting File
Hi,
It was for 15 frames, I edited the previous post.
How would .NET versions affect the results of the test ?
Cause it was the only change I made (1.1 to 2.0) and there is a process called "FrameworkService.exe" (and maybe others related I am not aware) running in background
Did
Pookie
4th February 2007, 23:02
Gotcha...
vBulletin® v3.8.5, Copyright ©2000-2012, Jelsoft Enterprises Ltd.