New advanced Benchmark - Page 2

Sagittaire · 19th April 2017, 16:17

Quote:

Originally Posted by NikosD

Hey,

I have made a few corrections to your "run.sh" file in order to be more consistent to what is running and what is being displayed.

For example, there is no "SSE3" instruction set tested, it's "SSSE3" and so on.

Also, the "All" setting is actually like "auto" so we don't need that.

I added also --no-asm option and I named it "No SIMD" in the final text, in order to see better the speedups of various SIMD sets compared to "No SIMD" setting

All of my changes are here:
http://txt.do/drlqm

yes, I have little bug for "all SIMD" bench.
I will make new script for choose the best x264 and x265 build between VS2017, GCC7 and ICC17

Sagittaire · 19th April 2017, 16:19

Quote:

Originally Posted by SquallMX

6700K Stock 4.0 Ghz

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i7-6700K      |  11.95  |   2.28  |    82  |   1.89  |   0.56  |    0.55  |   0.88  |   0.97  |   1.47  |   1.50  |   1.85  |    N/A  |

the most interessing result ... thx

you can (re)make benchmark at your max overclocking?

Quote:

Stock 4.0 Ghz

in fact stock is at 4.1 Ghz because turbo max for all core is 4.1 Ghz
You choose imposed 4.0 Ghz for all core or your i7-6700K is really in default stock frequency?

NikosD · 19th April 2017, 16:59

Quote:

Originally Posted by Sagittaire

yes, I have little bug for "all SIMD" bench.
I will make new script for choose the best x264 and x265 build between VS2017, GCC7 and ICC17

Don't forget to add the --no-asm option.

The "auto" is just fine for "all" SIMD sets.

Also, as I had told you in PM, the ffmpeg LAV video decoder is a lot faster in HEVC decoding than the script, probably due to more threads involved.

But using that script we can compare different CPUs to each other on the same script and not to absolutely fastest decoding.

SquallMX · 19th April 2017, 18:09

Quote:

Originally Posted by Sagittaire

the most interesting result ... thx

you can (re)make benchmark at your max overclocking?

in fact stock is at 4.1 Ghz because turbo max for all core is 4.1 Ghz
You choose imposed 4.0 Ghz for all core or your i7-6700K is really in default stock frequency?

i7-6700K stock clock for all cores at full is 4.0 GHz. Unfortunately is a DTR so no overclocking because of the cooling conditions

.

Motenai Yoda · 20th April 2017, 12:19

here my results.log

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i7  920 @3.8  |   7.90  |   1.14  |     49  |   0.96  |   0.37  |   0.37  |   0.60  |   0.65  |   0.96  |    N/A  |   N/A   |    N/A  |

RanmaCanada · 23rd April 2017, 17:40

Well it looks like according to these benches I need to upgrade to at least R1700.

Thank you for the comprehensive tables.

Fador · 25th April 2017, 08:29

Xeon E5-2699 v4 @ Stock

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Xeon  E5-2699 v4 |  31.04  |   6.79  |   135  |   4.95  |   1.72  |    1.73  |   2.66  |   2.88  |   4.21  |    4.22  |   4.87  |    N/A  |

Yanak · 27th April 2017, 15:58

Hello, CPU runs @4.2Ghz

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i7-3770K      |  10.35  |   1.46  |    72  |   1.22  |   0.47  |    0.47  |   0.76  |   0.81  |   1.22  |    1.23  |   N/A  |    N/A  |

shh · 30th April 2017, 09:51

2x Xeon E5-2699 v4 @2.20GHz, 128GB RAM, 2xThreads: 22/22 (HT: off)
https://ark.intel.com/de/products/91...Cache-2_20-GHz
2x22 full cores oviously are a little better than just one E5-2699v4 with Hyper-Threading on.

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  |
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|  2x Xeon E5-2699 v4 |  37.19  |   9.36  |   118   |   5.48  |   2.13  |   2.12  |   3.25  |   3.48  |   4.90  |   4.89  |   5.28  |    N/A  |

Selur · 3rd June 2017, 08:06

I overclocked my system to 4 GHz to see how this would affect the benchmark, and got some mixed results.

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Ryzen 7 1800X Eight-Core Processor          |  21.91  |   3.39  |    74  |   2.59  |   1.19  |    1.14  |   1.66  |   1.76  |   2.05  |    2.18  |   2.12  |    N/A  |4GHz (Mainboard default)
|  Ryzen 7 1800X Eight-Core Processor          |  22.14  |   3.43  |    73  |   2.52  |   1.21  |    1.15  |   1.64  |   1.73  |   2.22  |    2.24  |   2.32  |    N/A  |4GHz (only changed the cpu mult)
|  Ryzen 7 1800X Eight-Core Processor          |  22.35  |   3.41  |    74  |   2.59  |   1.14  |    1.14  |   1.56  |   1.75  |   2.17  |    2.32  |   2.37  |    N/A  | 4GHz (only changed the cpu mult) + MalwareBytes and Windows Defender disabled

I'm wondering why the LAVC benchmark is so low compared to the old unoverclocked result I posted before:

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Ryzen 7 1800X Eight-Core Processor          |  21.28  |   3.42  |   118  |   2.72  |   1.13  |    1.13  |   1.71  |   1.83  |   2.52  |    2.53  |   2.65  |    N/A  |@Stock speed

Not aware that I changed anything else on my system aside from updating Windows, since the last benchmark.

Cu Selur

Sagittaire · 3rd June 2017, 20:04

Hi Selur

AMD make new compilator optimized for Ryzen (linux only?)
http://developer.amd.com/tools-and-s...g-cc-compiler/

You can try to make compilation for x264 and x265, make test with this benchmark and post your result?

THX

jd17 · 29th July 2017, 15:32

If anyone is interested, I tested my i5-7500 as well, thanks for compiling the benchmark!

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i5-7500       |   9.21  |   1.89  |    64   |   1.55  |  0.45   |   0.45  |   0.72  |   0.78  |   1.19  |   1.20  |   1.52  |    N/A  |

sacd · 5th August 2017, 10:24

Interesting benchmark, measures all the right applications.
This is my new editing PC running the benchmark_auto from the first post dated April 14 2017 (CPU OC 4.5GHz):

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i9-7900X      |  32.97  |   6.24  |   157  |   4.70   |   1.57  |    1.58 |   2.50  |   2.74  |   4.07  |    3.98  |   4.82 |    N/A  |

drizzit · 6th August 2017, 17:07

My results with the auto bat, manually overclocked to 4.4GHz

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i7-4790K      |  11.70  |   2.10  |    78  |   1.74  |   0.49  |    0.49  |   0.82  |   0.89  |   1.39  |    1.40  |   1.75  |    N/A  |

Sagittaire · 10th August 2017, 15:34

you can say if you have real full CPU charge at 100% for x264 test with 8C/16T CPU and higher (if someone have 16C/32T).

thx

Clare · 10th August 2017, 21:45

I had to modify the script to run in Linux.

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
| Core i7-6700HQ      |  8.96   |   1.75  |    88   |   1.43  |   0.27  |    0.28 |   0.49  |   0.54  |   1.05  |    1.12 |   1.45  |    N/A  |

Here is the link to the modified script for Linux: https://gist.github.com/WyohKnott/72...10a4aea92af788
It depends on python-cpuinfo, and of course, you need to install ffmpeg, x264 and x265 from your distribution repository.

Balthazar2k4 · 22nd August 2017, 22:02

Here you go:
|------------------------------------------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
| CPU | x264 | x265 | LAVC | auto | MMX2 | SSE | SSE2 | SSE3 | SSE4 | AVX | AVX2 | All |
|------------------------------------------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
| Ryzen Threadripper 1950X 16-Core Processor | 37.59 | 6.50 | 136 | 4.65 | 2.02 | 2.00 | 2.95 | 3.10 | 3.89 | 4.10 | 4.26 | N/A |

Edit: x264 does NOT utilize all 32 threads. During that portion of the test I was seeing ~80% utilization.

kolak · 24th August 2017, 21:01

Quote:

Originally Posted by shh

2x Xeon E5-2699 v4 @2.20GHz, 128GB RAM, 2xThreads: 22/22 (HT: off)
https://ark.intel.com/de/products/91...Cache-2_20-GHz
2x22 full cores oviously are a little better than just one E5-2699v4 with Hyper-Threading on.

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  |
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|  2x Xeon E5-2699 v4 |  37.19  |   9.36  |   118   |   5.48  |   2.13  |   2.12  |   3.25  |   3.48  |   4.90  |   4.89  |   5.28  |    N/A  |

Far beyond reasonable scaling. Such a waste of processing power

This is only good for running many encodes at the same time or splitting long encodes into chunks.
Can you test it with 1 core, so we can see how all this cores are wasted

kolak · 24th August 2017, 21:05

Quote:

Originally Posted by sacd

Interesting benchmark, measures all the right applications.
This is my new editing PC running the benchmark_auto from the first post dated April 14 2017 (CPU OC 4.5GHz):

Code:

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
|     CPU             |   x264  |   x265  |   LAVC  |   auto  |   MMX2  |    SSE  |   SSE2  |   SSE3  |   SSE4  |    AVX  |   AVX2  |    All  | 
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------| 
|  Core i9-7900X      |  32.97  |   6.24  |   157  |   4.70   |   1.57  |    1.58 |   2.50  |   2.74  |   4.07  |    3.98  |   4.82 |    N/A  |

Just shows how good clock makes things fast.

bladerunner1982 · 27th August 2017, 20:06

|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
| CPU | x264 | x265 | LAVC | auto | MMX2 | SSE | SSE2 | SSE3 | SSE4 | AVX | AVX2 | All |
|---------------------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|---------|
| Pentium G4560 | 5.02 | 0.77 | 37 | 0.66 | 0.25 | 0.25 | 0.40 | 0.44 | 0.67 | N/A | N/A | N/A |

3rd June 2017, 20:04	#31 \| Link
Sagittaire Testeur de codecs Join Date: May 2003 Location: France Posts: 2,484	Hi Selur AMD make new compilator optimized for Ryzen (linux only?) http://developer.amd.com/tools-and-s...g-cc-compiler/ You can try to make compilation for x264 and x265, make test with this benchmark and post your result? THX __________________ Le Sagittaire ... ;-) 1- Ateme AVC or x264 2- VP7 or RV10 only for anime 3- XviD, DivX or WMV9

10th August 2017, 15:34	#35 \| Link
Sagittaire Testeur de codecs Join Date: May 2003 Location: France Posts: 2,484	you can say if you have real full CPU charge at 100% for x264 test with 8C/16T CPU and higher (if someone have 16C/32T). thx __________________ Le Sagittaire ... ;-) 1- Ateme AVC or x264 2- VP7 or RV10 only for anime 3- XviD, DivX or WMV9 Last edited by Sagittaire; 10th August 2017 at 21:27.

10th August 2017, 21:45	#36 \| Link
Clare Registered User Join Date: Apr 2016 Posts: 61	I had to modify the script to run in Linux. Code: \|---------------------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\| \| CPU \| x264 \| x265 \| LAVC \| auto \| MMX2 \| SSE \| SSE2 \| SSE3 \| SSE4 \| AVX \| AVX2 \| All \| \|---------------------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\| \| Core i7-6700HQ \| 8.96 \| 1.75 \| 88 \| 1.43 \| 0.27 \| 0.28 \| 0.49 \| 0.54 \| 1.05 \| 1.12 \| 1.45 \| N/A \| Here is the link to the modified script for Linux: https://gist.github.com/WyohKnott/72...10a4aea92af788 It depends on python-cpuinfo, and of course, you need to install ffmpeg, x264 and x265 from your distribution repository. __________________ still image formats comparison - video-formats-comparison Last edited by Clare; 11th August 2017 at 07:09.

22nd August 2017, 22:02	#37 \| Link
Balthazar2k4 Registered User Join Date: Mar 2009 Location: Here, There, & Everywhere Posts: 269	Here you go: \|------------------------------------------------------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\| \| CPU \| x264 \| x265 \| LAVC \| auto \| MMX2 \| SSE \| SSE2 \| SSE3 \| SSE4 \| AVX \| AVX2 \| All \| \|------------------------------------------------------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\| \| Ryzen Threadripper 1950X 16-Core Processor \| 37.59 \| 6.50 \| 136 \| 4.65 \| 2.02 \| 2.00 \| 2.95 \| 3.10 \| 3.89 \| 4.10 \| 4.26 \| N/A \| Edit: x264 does NOT utilize all 32 threads. During that portion of the test I was seeing ~80% utilization. Last edited by Balthazar2k4; 23rd August 2017 at 14:54.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

23rd April 2017, 17:40	#26 \| Link
RanmaCanada Registered User Join Date: May 2009 Posts: 331	Well it looks like according to these benches I need to upgrade to at least R1700. Thank you for the comprehensive tables.

27th August 2017, 20:06	#40 \| Link
bladerunner1982 Registered User Join Date: Apr 2003 Posts: 18	\|---------------------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\| \| CPU \| x264 \| x265 \| LAVC \| auto \| MMX2 \| SSE \| SSE2 \| SSE3 \| SSE4 \| AVX \| AVX2 \| All \| \|---------------------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\|---------\| \| Pentium G4560 \| 5.02 \| 0.77 \| 37 \| 0.66 \| 0.25 \| 0.25 \| 0.40 \| 0.44 \| 0.67 \| N/A \| N/A \| N/A \|