rasz_pl wrote on 2022-12-11, 05:47:

as I said you most likely fried clockgen with P4 😮, or it got so hot somehow one of the pins responsible for serial communication (54 55) disconnected from the track

It certainly seems that way. This motherboard isn’t anything I care about though so I’m not going to attempt any kind of repair on it.

So just for fun I put the Celeron 1.6 in my MSI G31 board to see what would happen. There are some basic overclocking options so I'll see what it can do over the next couple of hours. First I am doing the same benchmark runs with the GTX770 to see how it compares to the P35 board.

GTX770 - Celeron D 420 @ 1.60 - P35
3dmark01 - 20823
3dmark03 - 29935
3dmark05 - 7384
Aquamark3 - 65380
UT2003 1280x960
Flyby - 210.124344
Botmatch - 66.980415

GTX770 - Celeron D 420 @ 1.60 - G31
3dmark01 - 21365
3dmark03 - 31053
3dmark05 - 7822
Aquamark3 - 66782
UT2003 1280x960
Flyby - 215.984421
Botmatch - 68.840500

GTX770 - Celeron D 420 @ 1.92 - G31
3dmark01 - 25367
3dmark03 - 35816
3dmark05 - 9090
Aquamark3 - 78725
UT2003 1280x960
Flyby - 254.839401
Botmatch - 81.351585

GTX770 - Celeron D 420 @ 2.13 - G31
3dmark01 - 27056
3dmark03 - 39390
3dmark05 - 9806
Aquamark3 - 82958
UT2003 1280x960
Flyby - 271.775116
Botmatch - 86.908012

GTX770 - Celeron D 420 @ 2.40 - G31
3dmark01 - 29335
3dmark03 - 42179
3dmark05 - 10904
Aquamark3 - 90877
UT2003 1280x960
Flyby - 298.411469
Botmatch - 95.235298

I like these performance tests.
Some time ago I bought some processors and among them I have a Pentium 4 670 at 3.8ghz.

It is quite expensive today as they have said before. Now I sure wouldn't buy it.
In fact, I did some tests a while ago in 3dmark that gave this result:

3DMARK 99:

20576 3DMarks
51672 CPU 3DMarks

PC:

- Pentium 4 670 3.8ghz
- 512mb ram ddr
-Nvidia 6600GT
- 80gb hdd
-windows 98se

It's a beast but it's also too hot and I replaced it with a pentium 4 3.0ghz.

Maybe that data can be useful to someone.
Greetings.

This Celeron is overclocking very well so far! 🤣

GTX770 - Celeron D 420 @ 2.67 - G31
3dmark01 - 31451
3dmark03 - 45488
3dmark05 - 11852
Aquamark3 - 96536
UT2003 1280x960
Flyby - 314.770233
Botmatch - 101.823257

looks like 1/3 IPC boost over P4, I expected better 😀 Celeron 420 should overclock all the way to 3GHz no sweat, 3.2GHz is the limit on normal air cooling.

rasz_pl wrote on 2022-12-12, 00:42:

looks like 1/3 IPC boost over P4, I expected better 😀 Celeron 420 should overclock all the way to 3GHz no sweat, 3.2GHz is the limit on normal air cooling.

2.71GHZ might be about the limit on this CPU/MB combo. It doesn't post at 2.8GHZ.

give it 1.45volt

rasz_pl wrote on 2022-12-12, 01:28:

give it 1.45volt

There are no voltage options in this motherboards bios.

there might be a way masking cpu pins 😀

rasz_pl wrote on 2022-12-12, 02:01:

there might be a way masking cpu pins 😀

Yeah I did some quick googling and it seems you can bridge a few pads to get 1.4V pretty easily with a conductive ink pen. I don't have one and the cheapest one I can find is about $15.00 online, which is more than the 420 CPU cost. 🤣

This entire thing was just a fun little time killer to mess with but this isn't a system I'm going to use for anything so I'm not going to continue with it any further. I do all my XP era gaming on my GTX970/Xeon rig anyway.

Yes this stupid thread is back. Why? Well this stupid man made a stupid purchase. I got a Celeron D 450 for $5.00 cash on kijiji.

Going to play around with it a bit tonight and post some overclocking and benchmarks in a bit. Not going to bother with anything except the GTX770 that's already in the machine.

EDIT - first round of testing completed at ~2860mhz

GTX770 - Celeron D 450 @ 2.86 - G31
3dmark01 - 31502
3dmark03 - 46263
3dmark05 - 11790
Aquamark3 - 97044
UT2003 1280x960
Flyby - 328.599579
Botmatch - 104.189194

ROUND 2:

GTX770 - Celeron D 450 @ 2.97 - G31
3dmark01 - 31586
3dmark03 - 47069
3dmark05 - 12020
Aquamark3 - 97525
UT2003 1280x960
Flyby - 330.017334
Botmatch - 104.673988

That's probably it for me tonight, I need to have dinner and get to sleep.

I'll see if it'll do anything above 3ghz reliably tomorrow. It's already way faster than the previous p4 or 420 celeron though.

EDIT - couldn't resist one last round of testing. The ram was severely underclocked on the previous set of tests so now it's running at essentially it's normal speed.

GTX770 - Celeron D 450 @ 2.97 - G31 - 810mhz mem
3dmark01 - 32594
3dmark03 - 47781
3dmark05 - 12499
Aquamark3 - 99775
UT2003 1280x960
Flyby - 340.840546
Botmatch - 108.210584

This is probably the last set of benchmarks. Overclocking any higher will result in a post but bluescreens during any heavy CPU usage.

GTX770 - Celeron D 450 @ 3.047 - G31 - 829.6mhz mem
3dmark01 - 33678
3dmark03 - 48599
3dmark05 - 12662
Aquamark3 - 103364
UT2003 1280x960
Flyby - 349.464813
Botmatch - 111.050087

If a motherboard with CPU voltage adjustment falls into my lap I'll revisit this but I am not ordering anything on ebay and putting a limit of $10-20 on any further kijiji/craigslist spending for this silly project.

So I came across an Asus P5B with bent pins last night and gave a friend a coffee for it.

Straightened the pins today and it works, so I'm back to doing silly things with this Celeron and P4. I've upped the Celeron voltage to 1.36 and gotten a bit more out of it. Once I reach the limit on the Celeron the P4-661 is going back in!

GTX770 - Celeron D 450 @ 3.162 - Asus P5B - ~718mhz mem 1.36V
3dmark01 - 34106
3dmark03 - 49483
3dmark05 - 13190
Aquamark3 - 107325
UT2003 1280x960
Flyby - 359.355804
Botmatch - 113.402908

Running some tests at 3.24 then I'm taking a break for dinner, might come back later tonight.

GTX770 - Celeron D 450 @ 3.24 - Asus P5B - ~736mhz mem ~1.39V
3dmark01 - 35141
3dmark03 - 51236
3dmark05 - 13519
Aquamark3 - 109475
UT2003 1280x960
Flyby - 366.925018
Botmatch - 116.577873

So 3288mhz seems to be about it for this CPU. Anything higher and it bluescreens or doesn't post. Kind of neat to get over a 1000mhz overclock out of it though.

GTX770 - Celeron D 450 @ ~3.29 - Asus P5B - ~743mhz mem ~1.39V (set to 1.425 in bios)
3dmark01 - 35740
3dmark03 - 50713
3dmark05 - 13616
Aquamark3 - 109729
UT2003 1280x960
Flyby - 370.641785
Botmatch - 117.543488

Breaking out the big guns to see what I can do with this P4 on this Asus board.

I was using this Big Typhoon cooler on an Athlon 64X2 but that board has some bad caps and has been retired for now.

Hey, I just noticed this thread. This might sound silly, but did you try to mess with the Floppy Controller and/or related floppy settings in the BIOS for that board that wouldn't overclock?

I have an ASUS Nforce4 s939 board from 2005 that I couldn't figure out why it didn't want to cooperate when I increased the Clock frequency. The increased frequency did not take effect for the CPU, RAM and HT, even though the BIOS saved the value. CPU-Z showed no difference.
Then I found an old thread somewhere where someone said they had the same issue and eventually found out they had to set the A: drive as a 1.44MB 3.5' floppy, even though they had no floppy installed. Sure enough, I didn't have a floppy installed, there was no space for it on the (modern) case. Followed the steps, WHAM, it works! Hopefully that was the issue all along in your case as well!

Garrett W wrote on 2023-01-29, 10:25:

Hey, I just noticed this thread. This might sound silly, but did you try to mess with the Floppy Controller and/or related floppy settings in the BIOS for that board that wouldn't overclock?

I have an ASUS Nforce4 s939 board from 2005 that I couldn't figure out why it didn't want to cooperate when I increased the Clock frequency. The increased frequency did not take effect for the CPU, RAM and HT, even though the BIOS saved the value. CPU-Z showed no difference.
Then I found an old thread somewhere where someone said they had the same issue and eventually found out they had to set the A: drive as a 1.44MB 3.5' floppy, even though they had no floppy installed. Sure enough, I didn't have a floppy installed, there was no space for it on the (modern) case. Followed the steps, WHAM, it works! Hopefully that was the issue all along in your case as well!

I think it was set to off, but this is something I'll check into for sure!

Well I seem to have reached the end of the road with this P4 661 CPU. 4.77 seems to be limit, going even 1mhz higher on the bus speed and it does not post.

GTX770 - Pentium 4 @ 4.77 - Asus P5B
3dmark01 - 36778
3dmark03 - 54442
3dmark05 - n/a - bluescreens
Aquamark3 - 101942
UT2003 1280x960
Flyby - 342.666321
Botmatch - 106.775185

pete8475 wrote on 2023-01-29, 16:45:

Well I seem to have reached the end of the road with this P4 661 CPU. 4.77 seems to be limit, going even 1mhz higher on the bus […]
Show full quote

Well I seem to have reached the end of the road with this P4 661 CPU. 4.77 seems to be limit, going even 1mhz higher on the bus speed and it does not post.

GTX770 - Pentium 4 @ 4.77 - Asus P5B
3dmark01 - 36778
3dmark03 - 54442
3dmark05 -
Aquamark3 -
UT2003 1280x960
Flyby - 342.666321
Botmatch - 106.775185

Still a good bit more than a GHz overclock on a CPU from an architecture that newer released past 3.8 ... That's pretty neat