Got a new motherboard for my P4/FX graphics system, it's an Asrock 775i65PE. This being a socket 775 board I put in a Pentium 4 HT 661 CPU C1 (3.60GHZ) which I have currently running at 4209.

Anyway here are some new scores with this hardware, I'm currently waiting for some D0 processors to show up that might overclock a bit more.

3dmark 2000 - 22806
3dmark 2001 - 17996
3dmark 2003 - 5628

Clearly the Quadro FX3000 is at it's limit in 2003, there's really no change there versus my 3.4 Northwood.

scores updated!

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with more voltage.

Anyway updated scores on my Asrock 775i65PE/P4/Quadro FX3000:
2000 - 24394
2001 - 19291
2003 - 6137

pete8475 wrote on 2025-08-01, 01:47:

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with mor […]
Show full quote

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with more voltage.

Anyway updated scores on my Asrock 775i65PE/P4/Quadro FX3000:
2000 - 24394
2001 - 19291
2003 - 6137

That looks like a solid 3rd on the 2000 😀

H3nrik V! wrote on 2025-08-01, 06:29:

pete8475 wrote on 2025-08-01, 01:47:

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with mor […]
Show full quote

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with more voltage.

Anyway updated scores on my Asrock 775i65PE/P4/Quadro FX3000:
2000 - 24394
2001 - 19291
2003 - 6137

That looks like a solid 3rd on the 2000 😀

Yeah I was surprised it increased that much versus the previous cpu. Now I want to get some conductive paint and try a voltage mod in this setup.

pete8475 wrote on 2025-08-01, 20:41:

H3nrik V! wrote on 2025-08-01, 06:29:

pete8475 wrote on 2025-08-01, 01:47:

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with mor […]
Show full quote

Got a D0 P4 651 in today and did a bit of overclocking. It seems stable so far at 4363mhz but I bet it could go higher with more voltage.

Anyway updated scores on my Asrock 775i65PE/P4/Quadro FX3000:
2000 - 24394
2001 - 19291
2003 - 6137

That looks like a solid 3rd on the 2000 😀

Yeah I was surprised it increased that much versus the previous cpu. Now I want to get some conductive paint and try a voltage mod in this setup.

Conductive paint? For voltage mod? Is that of the CPU or the GPU? It sounds like what people did to early Athlon socket A ..

H3nrik V! wrote on 2025-08-01, 23:20:

pete8475 wrote on 2025-08-01, 20:41:

H3nrik V! wrote on 2025-08-01, 06:29:

That looks like a solid 3rd on the 2000 😀

Yeah I was surprised it increased that much versus the previous cpu. Now I want to get some conductive paint and try a voltage mod in this setup.

Conductive paint? For voltage mod? Is that of the CPU or the GPU? It sounds like what people did to early Athlon socket A ..

This motherboard has no CPU voltage control, you can bridge various pads on the bottom of a 775 processor to get different voltages, fsb, etc. Conductive ink or paint seems to be the way most do it that I've seen.

I do have several much more "modern" 775 boards with PCI-E too, so I will put one of these processors in one of those and do some overclocking as well as benchmarking some silly video card like my GTX980TI at some point.

EDIT - Just to see if it would at least post with a 1066FSB in a Asus P5B I put one of the now three (yes it's silly I know) D0 SL9KE 651 P4 processors I have in there and bumped up the core voltage to 1.425V. It does in fact post at 4.55GHZ with a GTX970 and 4GB of DDR2, I have no spare SSD drives right now but if I can find a drive in the next couple of days I'll make a copy of the XP install from my i7 retro machine and see what that P4 scores with a 970. I can eventually try the 980 but I'll probably have to take that and the power supply out of the i7 and I don't want to do that at the moment.

Tried out XP on the P4 and interestingly 3dmark2003 scores a fair bit higher.

6559 vs 6137 on ME.

It's late today but on the weekend I'm going to do some testing with an X1950 Pro and then I'll swap out the board for my P5B so I can use a GTX970.

EDIT - I had to run 2003 on the X1950 Pro before calling it a night and it's just a hair faster than the FX3000. 15520 🤣

Ok so here's some 3dmark results with my overclocked 3.4 and a GTX970. The numbers are rather silly, if I can find the time this weekend I'll put the GTX980TI in there and see what happens.

3dmark2001 - 34268
3dmark2003 - 50100

2000 gives an error about needing at least 4MB of video memory.

The gains keep coming, but I'm going to bed now.

CPU at 4537 scores 52938 in 03.

This is probably my last test with the 970 in the P4, the max stable CPU speed I've achieved is 4759MHZ and it now scores 55231.

I'll try out the GTX980TI tonight.

pete8475 wrote on 2025-08-09, 20:28:

This is probably my last test with the 970 in the P4, the max stable CPU speed I've achieved is 4759MHZ and it now scores 55231.

I'll try out the GTX980TI tonight.

Looking forward to that. Maybe I should throw my own 980Ti from the daily driver in a Netburst rig ...

H3nrik V! wrote on 2025-08-09, 21:58:

pete8475 wrote on 2025-08-09, 20:28:

This is probably my last test with the 970 in the P4, the max stable CPU speed I've achieved is 4759MHZ and it now scores 55231.

I'll try out the GTX980TI tonight.

Looking forward to that. Maybe I should throw my own 980Ti from the daily driver in a Netburst rig ...

Testing done and the results are a bit underwhelming. If you have time please do try yours out too though!

P4 @ 4759 / GTX 980 TI
3dmark 2000 - same ram related error
3dmark 2001 - 36191
3dmark 2003 - 55641

EDIT: going to try this patch to see if it fixes 3dmark 2000. Running 3dMark 2000 on modern hardware

Patch works, 3dmark 2000 result is an undwerwhelming 19683.

Looks like you've run in to the CPU being the bottleneck ..

H3nrik V! wrote on 2025-08-10, 05:28:

Looks like you've run in to the CPU being the bottleneck ..

I guess so, it's just funny that the same CPU clocked about 500mhz lower with a much slower vid card scores 25% higher in my testing of 3dmark 2000 though. 🤣

Driver overhead is a bitch.

pete8475 wrote on 2025-08-10, 06:07:

H3nrik V! wrote on 2025-08-10, 05:28:

Looks like you've run in to the CPU being the bottleneck ..

I guess so, it's just funny that the same CPU clocked about 500mhz lower with a much slower vid card scores 25% higher in my testing of 3dmark 2000 though. 🤣

Probably feature emulation. Ie slow card does x in hardware, fast card has a software path run on the gpu to achieve the same result.

It's not uncommon.
As long as the hardware gives the correct result faster than the last optimised hardware, it doesn't matter.

Basically, you swap an "asic" for a few generations faster "cpu". Less efficient at doing that thing, but good enough for the few times it might be required.

myne wrote on 2025-08-10, 07:35:

Probably feature emulation. Ie slow card does x in hardware, fast card has a software path run on the gpu to achieve the same re […]
Show full quote

pete8475 wrote on 2025-08-10, 06:07:

H3nrik V! wrote on 2025-08-10, 05:28:

Looks like you've run in to the CPU being the bottleneck ..

I guess so, it's just funny that the same CPU clocked about 500mhz lower with a much slower vid card scores 25% higher in my testing of 3dmark 2000 though. 🤣

Probably feature emulation. Ie slow card does x in hardware, fast card has a software path run on the gpu to achieve the same result.

It's not uncommon.
As long as the hardware gives the correct result faster than the last optimised hardware, it doesn't matter.

Basically, you swap an "asic" for a few generations faster "cpu". Less efficient at doing that thing, but good enough for the few times it might be required.

Makes sense.

I'll test some other PCI-E Geforce cards I have later today, I know I have several 700 series cards I think a 460 around here somewhere.

EDIT - ran 2000 on the 460 and 770. I have a GTS250 too but that doesn't want to post in this board.
GTX460 - 19363
GTX770 - 19631

This is probably the end of my experiments with this stuff for now.