I checked Ebay for Socket 478 Gallatin P4 EE CPUs and at first I diddnt find any but then I looked some more and found some! 😀

These are P4 EE 3.2 GHz, they are just mislabeled. There are 4 of them and the price is fair.

http://www.ebay.co.uk/itm/Working-Intel-Penti … 6EAAOSwpDdVcmuR

I'm pretty convinced that the Prescott is better than the Northwood for decoding H.264 video, at least in VLC player anyway. It matters because those videos are marginal on a P4. The Prescott can manage to play things smoothly that will stutter on a Northwood. Maybe it's the cache, but I think it's more likely that the reason is the player/codec/whatever is programmed to benefit from SSE3.

What GPU U were using ?
Northwood is around 100MHz faster at the same clock speed (ie. 3,2GHz Prescott = 3,1GHz Northwood or 3,4GHz Northwood = 3,5GHz Prescott).
If U have enough difference in clock speed, Prescott will be faster.
Also, I don't think SSE3 is THAT important.

PS. How fast a Prescott CPU would have to be, to beat this scores ?
Cinebench 2003 : LINK
Cinebench R11.5 : LINK

agent_x007 wrote:

What GPU U were using ? Northwood is around 100MHz faster at the same clock speed (ie. 3,2GHz Prescott = 3,1GHz Northwood or 3,4 […]
Show full quote

I'm pretty convinced that the Prescott is better than the Northwood for decoding H.264 video, at least in VLC player anyway. It matters because those videos are marginal on a P4. The Prescott can manage to play things smoothly that will stutter on a Northwood. Maybe it's the cache, but I think it's more likely that the reason is the player/codec/whatever is programmed to benefit from SSE3.

What GPU U were using ?
Northwood is around 100MHz faster at the same clock speed (ie. 3,2GHz Prescott = 3,1GHz Northwood or 3,4GHz Northwood = 3,5GHz Prescott).
If U have enough difference in clock speed, Prescott will be faster.
Also, I don't think SSE3 is THAT important.

PS. How fast a Prescott CPU would have to be, to beat this scores ?
Cinebench 2003 : LINK
Cinebench R11.5 : LINK

Press articles from 2004 pretty widely agreed that the Northwood was faster than Prescott at equal clocks, but they were using software from 2004. I question whether those results would be matched when using newer software. I had difficulty trying to find any newer comparisons. The press stopped caring about these CPUs as soon as something newer appeared on retail shelves.

The GPU in the machine currently is a Geforce 7600GS 256MB AGP. It doesn't have any meaningful H.264 acceleration, so playing back those videos is CPU intensive. H.264 acceleration isn't easy to come by on AGP, which is an unfortunate irony since those are the machines that would benefit the most from it.

I've never used Cinebench, but I'm not surprised if Northwood is better at it. My observation though is that Prescott has been performing better at H.264, and this is a major consideration for me.
I'd love to try a Gallatin but I haven't seen any that were cheap enough for me to be willing to buy them.
I didn't realize until the past day that Gallatins were available in LGA775 - and they seem to be cheaper than the mPGA478 versions. I had thought LGA775 P4 boards were condemned to Prescott only. If I had a 775 board that I didn't hate, I'd probably look for one of those chips.

I don't have directly equal Prescott and Northwood CPUs - they're all at slightly different combinations of clock speed, FSB, and HT support. But the pattern I was sensing from the chips I have was that hyperthreading helped a lot, and so did being a Prescott (whether that's because of SSE3 or some other reason). The Prescott advantage was particularly suggested by the comparison of a 2.6/800 Hyperthreaded Northwood vs a 2.8/533 non-HT Prescott - the latter doing measurably better, I felt it was too much to be explained by the slight difference in clock.

I read the description of Cinebench on their web site, and it sounds like it's a test of unaccelerated CPU based high quality 3D rendering performance, like for somebody running 3ds max (or whatever people use now). Is it relevant to 3D games or anything? Otherwise it seems like it doesn't matter.

shamino wrote:

Press articles from 2004 pretty widely agreed that the Northwood was faster than Prescott at equal clocks, but they were using s […]
Show full quote

Press articles from 2004 pretty widely agreed that the Northwood was faster than Prescott at equal clocks, but they were using software from 2004. I question whether those results would be matched when using newer software. I had difficulty trying to find any newer comparisons. The press stopped caring about these CPUs as soon as something newer appeared on retail shelves.

The GPU in the machine currently is a Geforce 7600GS 256MB AGP. It doesn't have any meaningful H.264 acceleration, so playing back those videos is CPU intensive. H.264 acceleration isn't easy to come by on AGP, which is an unfortunate irony since those are the machines that would benefit the most from it.

I've never used Cinebench, but I'm not surprised if Northwood is better at it. My observation though is that Prescott has been performing better at H.264, and this is a major consideration for me.
I'd love to try a Gallatin but I haven't seen any that were cheap enough for me to be willing to buy them.
I didn't realize until the past day that Gallatins were available in LGA775 - and they seem to be cheaper than the mPGA478 versions. I had thought LGA775 P4 boards were condemned to Prescott only. If I had a 775 board that I didn't hate, I'd probably look for one of those chips.

I don't have directly equal Prescott and Northwood CPUs - they're all at slightly different combinations of clock speed, FSB, and HT support. But the pattern I was sensing from the chips I have was that hyperthreading helped a lot, and so did being a Prescott (whether that's because of SSE3 or some other reason). The Prescott advantage was particularly suggested by the comparison of a 2.6/800 Hyperthreaded Northwood vs a 2.8/533 non-HT Prescott - the latter doing measurably better, I felt it was too much to be explained by the slight difference in clock.

I read the description of Cinebench on their web site, and it sounds like it's a test of unaccelerated CPU based high quality 3D rendering performance, like for somebody running 3ds max (or whatever people use now). Is it relevant to 3D games or anything? Otherwise it seems like it doesn't matter.

Only benchmark I have found that favors Prescotts is 3DMark06, I don't have enough CPUs to do a good comparison but it seems to be about 600Mhz faster that a Northwood has to be for it beat a Prescott 3DMark06 score and about 200Mhz with a Gallatin so I don't think its just the cache making Prescotts score better

I thing that's an interesting topic, hopefully I can do some testing in this regard in the near future 😀

I have a 6600 GT AGP but hopefully a faster X850 XT soon, that should make it easier to compare. I also have a HD 3650 AGP but had nothing but issues when using it with a XP 3200+ system, so hopefully the X850 XT works better.

stuvize wrote:

Only benchmark I have found that favors Prescotts is 3DMark06, I don't have enough CPUs to do a good comparison but it seems to be about 600Mhz faster that a Northwood has to be for it beat a Prescott 3DMark06 score and about 200Mhz with a Gallatin so I don't think its just the cache making Prescotts score better

Well I did run 3DMark06 on Gallatin @ 4GHz, but it was with "OP" GPU : LINK - not sure how it affects CPU score (driver overhead). Don't compare it to older results, because I used Windows 7 (not XP).

533MHz FSB is a killer with 512kB L2 cache.
Slow FSB = Slow data rate between RAM and CPU.
Combined this with small L2, and U can get a performance drop higher than clock difference.

Basic x264 hardware acceleration on AGP will give HD 2600 XT (better get DDR3 version), I recommend it if U want to play games on this computer.
Don't know if it will be better suited for Athlon XP rig, but the most important part of this are the drivers ("AGP fix" ones to be exact). Other than that, u may be forced to install them from Device Menager (ie. "Driver update", with manual choosing of the right GPU model from the list of devices).

As for games on 130nm P4EE CPU's : LINK 😉

I bought a few late AGP cards a few months ago. The best one was an HD2600XT, but I haven't tried it out much yet. Anything faster than that gets pretty expensive.
I'm half heartedly experimenting with different P4 CPUs and video cards, and feel like maybe I should post some test results in a new thread, but I need to be more organized if I'm really going to do that.

If the OP is still watching, I suggest not letting your motherboard be dictated by the 3GHz Prescott mPGA478 chip that you found. If you decide you'd rather have LGA775 then go for it - an equivalent (or faster) 775 chip would be cheap. However, 775 also means a different HSF, and usually different RAM and video card type, so depending what components you already have on hand, changing platforms could start to add up to real money.