View Full Version : the need for speed... new P4's


Turd_Ferguson
23rd July 2003, 11:37
alright, it just took my system almost 8 hours to run DIF4U, DCCE4U, ReAuthorist, and then Scenarist on a 7.9GB NTSC DVD-9 to get it down to fit a 4.7GB DVD-5. of coarse CCE SP 2.50 took up the vast majority of that time. now when i built my system i did so that i could overclock the hell out of it, but that is not the case. this is what i got:

P4 2.5 w/100Mhz FSB OC'ed to 3.0 w/120Mhz FSB
Gigabyte GA-8IEXP w/intel 845E chipset AND onboard RAID ATA133
1x 512MB stick of Corsair XMS PC3200 low-latency
ATI AIW radeon 7500 64MB
1x 80GB WesternDigital HD w/8MB cache (on IDE channel)
2x 200GB Maxtor HD w/8MB cache (on dual-channel striped RAID array)
Pioneer A05
Enermax EG651P-VE 550W power supply

to keep my P4 from melting i built a rediculous custom water cooling system into my case. unfortunately, the 2.5 P4 gets unstable after 122 FSB, even with the CPU temp under 40C. i was thinking that a new 2.4C P4 with a 875/865 chipset mobo would smash this time, with the hyperthreading and dual-channel DDR400. of coarse i would OC the FSB to about 300Mhz and get 3.6Ghz out of it.

anybody know a better system that would cut encoding time significantly?

magomago
30th July 2003, 05:43
well that 800Mhz FSB would help anyways for encoding....and coupled with 865PE chipset...or even 875 ;) That would be faster but as for how much I wouldn't know. And then the other thing would be "Is it justified to buy a 2.4C-3.2C Which'll cost from 160-600$ depending what i get and then get a new 150-200 dollar motherboard)

I would find out to see if you know of a friend or someone else on the board and see how long an encode takes for them..then try to simulate that encode by using the same seetings and even DVD and see how long it'll take you and that will give you a gauge of whether that extra 10 minutes is potentially worth all that extra money


But have you checked to make sure it isn't your northbridge or southbridge overheating?

Also try loosening the timings on the ram becuase for the P4 Bandwith is more important than tight timings...see if you can raise the FSB that method. Also try to raise the voltages slightly on the ram (2.7v sounds like what you should try since you are aiming for a crazy overclock)

And in any situation a 2.53Ghz processor at 3Ghz isn't that bad of an overclock...thought I would point out that it could easily be acheived with a stock heatsink/fan.

But before you plunk down any more cash try those first...

dvd_maniac
13th September 2003, 12:35
REGION1 ntsc HERE. i AM GETTING AN AVERAGE 2.9 RT in CCE 2.66 with my setup. I use DVD2SVCD which pretty much uses the same programs except for re-authorist. But I think you can do a comparison with just the RealTime in CCE, so here is My system

2.8 P4 800FSB OC'ed to 3.2
1GB Samsung DDR400 Dual Channel (2 512's)
2 WD 80GB 7200rpm 8MB cache in raid (striped)
2 Seagate 120GB SATA 8MB cache in raid (striped)
"These 2 raids help in the speed department as I read the source from one and write to the other"
ATI AIW 9700 Pro
ASUS P4c800 Deluxe
DVR-A05 -r/w Recorder Primary master
NEC 1300A Dual Format Recorder Secondary Master

b00zed
5th October 2003, 08:57
The 2.5GHz P4 was the second fastest of the Northwood-A P4s, so it was never going to be very overclockable. The GA-8IEXP is validated to run 133MHz FSB Northwood-B processors, so running it at 120MHz isn't stressing it. If the PCI/AGP frequencies aren't locked (and Gigabyte isn't known for implimenting a PCI/AGP lock) then you could be getting stability issues from anything that's on the PCI, AGP and IDE buses.

You're very likely to have a lot of overclocking success with a Springdale/Canterwood board coupled with a 2.4GHz Northwood-C, since it's the slowest of these processors and has the most headroom. Most of them appear to be reaching 3.2GHz, so 3.6GHz isn't out of the question but I wouldn't take it for granted, though with good cooling it shouldn't be too much of a problem.

Solo
6th October 2003, 15:08
Hi there

I am now suspecting that a 2.4C overclocked to a 3 Ghz DOES NOT produce the same performance as a standard 3 Ghz CPU. I say this because I have a 2.4 @ 3.12 and the encoding figures (fps) are nowhere near what I have seen in reviews. For instance - I saw a couple of charts compaing the encoding results from a 2800+ Barton (which I had) to 3 Ghz (533fsb) P4 and the P4 totally outperformed the 2800+.
However the new 2.4C 800 fsb outperforms the 2800+ AMD Barton slightly. That I can confirm as the average fps (DivX 5.05 + AviSynth) are almost the same on both setups.
BUT the 2.4 @ 3 Ghz does not come anywhere near the figures I saw in the reviews ??? I worked out a P4 3Ghz should take about half hour for a 90 min movie, but it takes me about 68 minutes. This puzzles me.

3.1 - 3.6 is about the average overclock on the 2.4C - usually memory issues (using 5.4 or 3.2 dividers and Winbond CH5 chipsets) limit higher overclocks for example. But the Corsair XMS PC3200 should be using the BH5 chips so that should not be a problem. But try confirm that first.

The best two boards would be the Abit IC7 / IC7-G or the ultimate would be the MAX 3 (875 chipset). The other would be the ASUS P4P800 (865 chipset). But like I said both these (865/875) chipsets have a lot of issues with memory dividers and high FSB overclocking. A lot of people (including myself) are hitting a wall at 260 fsb. Just head over the the ASUS and Abit forums and you will see what I mean.

My system is Pentium 2.4 C @ 3.12 Ghz - Abit IC7 - Kingston HyperX KHX 3500 2 x 256 MB (@ 2.3.3.7) + Koolance watercooling. :devil:

Ramirez
7th October 2003, 02:37
Hey Solo ,have you tried this? >2/3,FSB-283/MEM-189 Timings: 2-2-5-2

I'd like to hear what kind of numbers you're getting using this combination. :)

P.S
You should disable all codec enhancements prior of starting any encoding benchmarks.
This is exactly what usually done in reviews.

b00zed
8th October 2003, 01:21
By rights a 2.4C @ 3.12GHz should be very close to, if not faster than, a 3.2C at standard speeds due to the increased FSB bandwidth. Relaxing the memory timings might of course have some negative effect, but the extra FSB clock should take care of that. Have you had any stability issues at that speed? (or have you run any processor-unfriendly benchmarks on it to test it?) Sometimes it may appear to be working fine until it's being really stressed... just a thought.

Solo
9th October 2003, 20:14
Originally posted by Ramirez
Hey Solo ,have you tried this? >2/3,FSB-283/MEM-189 Timings: 2-2-5-2

I'd like to hear what kind of numbers you're getting using this combination. :)



Hi Ramirez

Sorry took so long - I've been having problems accessing the forums here.

Unfortunately I am one of those few who ended up with the Winbond CH5 chipset and therefore hit a FSB wall of 260 (no matter what divider I use). Along with the 19 pages of people in this Abithttp://forum.abit-usa.com/showthread.php?threadid=13877 thread. Very very frustrating as the memory I need is the Corsair XMS (only uses BH5 chipset) but it is impossible to get over here in South Africa. Best my mem can do is 2-2-3-6 (well in correct order - different BIOS's).

Solo
10th October 2003, 13:04
Because this thread is already asking about overclocking and DivX encoding I'm sure you'd be interested in this indeapth article that xbitlabs did on Pentium 4's with memory timings and bandwidth.

The DivX ecoding page is here:
http://www.xbitlabs.com/articles/memory/display/p4-mem_11.html
Even though they still use Flask (yet again) the results are noteworthy.
The numbers for the MPEG4 encoding task look more true-to-life. At least, the CPU frequency is the decisive factor here to determine the overall system performance. Note also that the platforms with Pentium 4 2.4C overclocked to 3GHz and DDR500 memory are running faster than their analogs that use DDR400 with the diminishing coefficient, notwithstanding the more aggressive timings in the latter case.
Note that there is almost no difference between the following 3 when it comes to encoding:
Pentium 2.4c @ 3.3 Ghz (1100 FSB)DDR336(333)- 2.2.2.5 (divider 3:2) = 43.27 fps
Pentium 2.4c @ 3.3 Ghz (1100 FSB)DDR440(434)- 2.3.3.6 (divider 5:4) = 43.21 fps
Pentium 3.2c @ stock ( 800 FSB)DDR400(400)- 2.2.2.5 (1:1) = 42.97 fps

So the way I see it is get the cheaper 2.4C and OVERCLOCK to get same encoding performance as buying a more expensive 3.2 Ghz :devil:

Ramirez
10th October 2003, 21:35
Thanks for the invitation Solo, /Bookmarked. :)

I guess we're talking about some kind of weird BIOS limitation of the ABIT mobo as I know for a fact that many people out there can OC/3400 this proc using exactly the same setup as you're using (same mobo, same RAM) hell, one of my friends is running his system like that (no problems) I really dunno why it's not so in your case (2 variation of the mem chipset?, Diff BIOS revision?)

Solo
11th October 2003, 06:40
Cool. This link will explain some of the memory/Intel 865/875 compatibility issues. I recon your friends are lucky to have the older HyperX modules which did use the Winbond BH5 chipsets.
http://forum.abit-usa.com/showthread.php?s=&threadid=13877