VOGONS


Bad socket 370 CPU?

Topic actions

Reply 20 of 49, by Grem Five

User metadata
Rank Member
Rank
Member
red_avatar wrote on 2023-03-07, 21:13:

The motherboard is perfectly fine - it has really high quality caps which is why I bought it in the first place (that and because it was a high end mobo at the time so has lots of bells & wistles).

Probably has the same caps my TUSL2-C had and it had to have some leaky caps replaced (leaky current wise not physically leaky). They may be high quality but they are 20+ years old.

If you cant test that CPU in another board and another one is on the way, why not wait until the other one arrives before asking for a refund?

Reply 21 of 49, by smtkr

User metadata
Rank Member
Rank
Member
TrashPanda wrote on 2023-03-08, 00:48:
Imperious wrote on 2023-03-08, 00:24:

Pentium 3 socket 370 systems also don't have the PSU issues that plagued the early Athlon setups, their wattage requirements are really no different than what a Pentium II would require, heck you could run a P3 setup off a 350watt PSU even with a weak 5v rail ~18-20 Amps. (The GPU might not like it but the core setup wouldn't care a bit)

Most Pentium 3 systems had 250W power supplies back in the day. I saw some even smaller. And that makes sense--600Es were 16W parts and the very common (in 2000) 800EBs were 21W parts. Vogons people have a tendency to worry about 5V rails too much.

Reply 22 of 49, by TrashPanda

User metadata
Rank l33t
Rank
l33t
smtkr wrote on 2023-03-08, 03:51:
TrashPanda wrote on 2023-03-08, 00:48:
Imperious wrote on 2023-03-08, 00:24:

Pentium 3 socket 370 systems also don't have the PSU issues that plagued the early Athlon setups, their wattage requirements are really no different than what a Pentium II would require, heck you could run a P3 setup off a 350watt PSU even with a weak 5v rail ~18-20 Amps. (The GPU might not like it but the core setup wouldn't care a bit)

Most Pentium 3 systems had 250W power supplies back in the day. I saw some even smaller. And that makes sense--600Es were 16W parts and the very common (in 2000) 800EBs were 21W parts. Vogons people have a tendency to worry about 5V rails too much.

Only time I worry about it is with early Socket A boards that don't have the P4 12v plug and late model Tualatin setups, they do require stronger 5v rails (~25Amps+) but if the P4 plug is there then the PSU doesn't matter a whole lot.

Last edited by TrashPanda on 2023-03-08, 04:05. Edited 1 time in total.

Reply 23 of 49, by TrashPanda

User metadata
Rank l33t
Rank
l33t

ok so the forum is being fucky with quotes for some reason

cyclone3d wrote on 2023-03-08, 01:04:

I still think it is the PSU or capacitors on the motherboard.

Only way I see it being thermal shutdown is if the heatsink isn't on the CPU flatly.

I would agree with it being a misaligned heatsink or bad VRM caps, even if they are not leaking they may have still dried out and cant hold a stable capacitance under load.

Reply 24 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie
TrashPanda wrote on 2023-03-08, 04:04:

ok so the forum is being fucky with quotes for some reason

cyclone3d wrote on 2023-03-08, 01:04:

I still think it is the PSU or capacitors on the motherboard.

Only way I see it being thermal shutdown is if the heatsink isn't on the CPU flatly.

I would agree with it being a misaligned heatsink or bad VRM caps, even if they are not leaking they may have still dried out and cant hold a stable capacitance under load.

It may be the caps but it seems unlikely to me. All the caps look perfectly in order visually but since the 800Mhz system was also solid as a rock with ZERO lock ups ... it's highly unlikely that a CPU that is slightly faster would suddenly make it instantly crash upon loading Windows and also always in the exact same spot (right after it has fully loaded Windows - the moment the mouse should let you click on things). The heatsink feels cool to me and the temperature (while it may not be entire reliable) always shows around 32-33°C. When the PC is just turned on it shows 23°C and rising so it's accurate enough in that way).

I should have the new CPU in a few days but to make sure, I'll lower the multiplier to 6x to basically have the identical CPU as I had before. If it STILL crashes, then I'm sure it's NOT the mobo.

As for the PSU: as others said, the P3 is pretty power efficient and isn't like the P4. I've used the same PSU in a Pentium II system without issues.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 25 of 49, by rasz_pl

User metadata
Rank l33t
Rank
l33t
smtkr wrote on 2023-03-07, 23:40:

Also, what's the S-code on the 1Ghz CPU you have that's giving you problems?

Hmm, we know Intel struggled in early 2000 alter losing 1GHz race to AMD. They tried to reclaim the lead by shipping heavily overclocked 1.13-GHz Pentium III just to be forced to recall all of them https://www.zdnet.com/article/intel-recalls-1 … z-pentium-iiis/. But afaik 1GHz didnt have any problems, other than being a fake paper launch https://www.zdnet.com/article/its-official-am … rst-5000096067/

Imperious wrote on 2023-03-08, 00:24:

Your cpu will be running off the 5v rail from the PSU. I'm amazed no one has yet mentioned that modern PSU's do not provide many Amps on the 5v rail as
everything after P4 and Athlon 64 uses the 12v rail.

1GHz P3 is <30W, not a problem yet.

red_avatar wrote on 2023-03-08, 08:11:

All the caps look perfectly in order visually

sadly that doesnt really matter. You can spot leaky caps, you can spot swollen caps that boiled, but you wont see a thing if they just slowly dried

red_avatar wrote on 2023-03-08, 08:11:

but since the 800Mhz system was also solid as a rock with ZERO lock ups ... it's highly unlikely that a CPU that is slightly faster would suddenly make it instantly crash upon loading Windows and also always in the exact same spot (right after it has fully loaded Windows - the moment the mouse should let you click on things).

set fsb to 66/100 to verify

red_avatar wrote on 2023-03-08, 08:11:

The heatsink feels cool to me

this is bad, should be warm. Afair radiators from that time have a lip, if you mount them in reverse one side of radiator will sit on the socket away from CPU core. Socket A = smoke, Intel = thermal shutdown.

red_avatar wrote on 2023-03-08, 08:11:

I'll lower the multiplier to 6x to basically have the identical CPU as I had before. If it STILL crashes, then I'm sure it's NOT the mobo.

did P3s have unlocked lower multipliers?

Open Source AT&T Globalyst/NCR/FIC 486-GAC-2 proprietary Cache Module reproduction

Reply 26 of 49, by H3nrik V!

User metadata
Rank Oldbie
Rank
Oldbie
rasz_pl wrote on 2023-03-08, 11:47:
red_avatar wrote on 2023-03-08, 08:11:

I'll lower the multiplier to 6x to basically have the identical CPU as I had before. If it STILL crashes, then I'm sure it's NOT the mobo.

did P3s have unlocked lower multipliers?

They did not, they are hard-locked. Only way is lower FSB like you suggest.

Regarding power - yes, the 1GHz may have similar TDP to the 800, but I'd be highly surprised if it didn't have some pretty heavy current surges under high load, that must be a lot larger than the the 800 MHz.

Yes, 1 GHz is "only" 25% more than 800 MHz, but voltage is higher as well, IIRC. This could give like 50% higher spikes in current draw, give or take a bit.

I'd wait for the new CPU to arrive, and test that, before asking for a refund. If the new one also crashes, VRM capacitors would be my first choice ...

Please use the "quote" option if asking questions to what I write - it will really up the chances of me noticing 😀

Reply 27 of 49, by Imperious

User metadata
Rank Oldbie
Rank
Oldbie

It was really just an idea I was putting forward. Can't You turn off the thermal protection in the bios, or is that for later architectures?

Actually I bought a Socket 370 motherboard last year and it is highly unstable causing reboots, could be capacitors even though they all look perfect.
The thing with electrolytic caps is it's normally when they are cold that they cause problems, not when warmed up, but I guess when
the power supply and motherboard is under load then that would cause problems as well.

Atari 2600, TI994a, Vic20, c64, ZX Spectrum 128, Amstrad CPC464, Atari 65XE, Commodore Plus/4, Amiga 500
PC's from XT 8088, 486, Pentium MMX, K6, Athlon, P3, P4, 775, to current Ryzen 5600x.

Reply 28 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie

It's not thermal runaway from what I can tell.

I used an infrared measuring device on the heatsink and it showed roughly 30-35°C. When I touch it during a benchmark it feels around 35°C. This is also what the BIOS says.

Here's the results of my tests with 3DMark 2001 SE (default benchmark):

1Ghz: it does all the tests and then crashes right after
750Ghz (FSB 100): it does all the tests and then crashes right after
500Ghz (FSB 66): it does all the tests and then crashes right after

--> then I swapped to my 800Mhz CPU:
it does all the tests and NO crashing.

Sure the voltage is higher but it's what, 0.05V higher? It says 1.75 for the 1Ghz and 1.7 for my 800Mhz. I could even raise it to 1.75 on my 800Mhz and see what happens but I bet it will run just fine.

Also: the heatsink is on correctly - originally the metal lever on top was oriented the wrong way (the lowest point is supposed to be directly above the CPU die and it was the other way round) so I corrected this.

With the current high speed fan, the CPU really doesn't get warmer than say 35-40°C tops.

There's 21 caps on this motherboard (I counted them) so I'm not very anxious to replace them when the 800Mhz CPU works fine. I still feel it's unlikely to be either the mobo or PSU since the PC is rock solid on the 800Mhz. I mean, I used the 800Mhz to 7ZIP multiple Gbs of CD images at ones (I'm making 1:1 copies using Alcohol 120% to run off 128GB USB sticks - way faster load times and no CD drive noise!) and it worked fine even though this would mean the CPU is maxed out for ages at a time and it did it just fine.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 29 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie
Imperious wrote on 2023-03-08, 13:25:
It was really just an idea I was putting forward. Can't You turn off the thermal protection in the bios, or is that for later ar […]
Show full quote

It was really just an idea I was putting forward. Can't You turn off the thermal protection in the bios, or is that for later architectures?

Actually I bought a Socket 370 motherboard last year and it is highly unstable causing reboots, could be capacitors even though they all look perfect.
The thing with electrolytic caps is it's normally when they are cold that they cause problems, not when warmed up, but I guess when
the power supply and motherboard is under load then that would cause problems as well.

You can't - Pentium IIIs didn't even get that hot so it was probably simply not needed.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 30 of 49, by smtkr

User metadata
Rank Member
Rank
Member
rasz_pl wrote on 2023-03-08, 11:47:
smtkr wrote on 2023-03-07, 23:40:

Also, what's the S-code on the 1Ghz CPU you have that's giving you problems?

Hmm, we know Intel struggled in early 2000 alter losing 1GHz race to AMD. They tried to reclaim the lead by shipping heavily overclocked 1.13-GHz Pentium III just to be forced to recall all of them https://www.zdnet.com/article/intel-recalls-1 … z-pentium-iiis/. But afaik 1GHz didnt have any problems, other than being a fake paper launch https://www.zdnet.com/article/its-official-am … rst-5000096067/

I wasn't implying that OP had a notorious model. I'm just genuinely curious which model it is.

Reply 31 of 49, by Imperious

User metadata
Rank Oldbie
Rank
Oldbie

After all that testing definitely try the cpu first, hopefully it's just that.

Atari 2600, TI994a, Vic20, c64, ZX Spectrum 128, Amstrad CPC464, Atari 65XE, Commodore Plus/4, Amiga 500
PC's from XT 8088, 486, Pentium MMX, K6, Athlon, P3, P4, 775, to current Ryzen 5600x.

Reply 32 of 49, by cyclone3d

User metadata
Rank l33t++
Rank
l33t++

Yeah, with that testing it sounds like the CPU though it is odd that it crashes right after the test no matter the speed the CPU is set to.

Yamaha modified setupds and drivers
Yamaha XG repository
YMF7x4 Guide
Aopen AW744L II SB-LINK

Reply 33 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie
cyclone3d wrote on 2023-03-09, 02:50:

Yeah, with that testing it sounds like the CPU though it is odd that it crashes right after the test no matter the speed the CPU is set to.

If I were to take a guess, it crashes after Windows loads the desktop. The benchmark is full screen (at a different resolution) and then when it drops out to the desktop, it takes about the same amount of time to crash as it does when Windows boots up. The weird thing is, if it boots up without crashing, it boots up fine and I can leave it at the desktop for 15 minutes without crashing.

Also, the CPU code is SL52R for both. These seem to be the most common ones, probably because some OEM brand like Dell used them in their PCs. There's also the SL5QJ which I can get so if the second CPU doesn't work, I'll try that one and return the other ones. I found a topic on Vogons from VooDooMan who said the SL5QJ is easiest to OC which to me means it's also likely to be the least "sensitive" to dying.

I believe the 800Mhz is the SL4CD if I recall correctly.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 34 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie

So ... I decided to check another way to see if my motherboard is to blame or not. I took my 800mhz CPU, changed the voltage to 1,75V and overclocked it to 1Ghz. And ... surprise surprise, it's stable as a rock. I've done 4x 3D Mark 99 benchmarks so far without a single crash. Oddly enough the motherboard made me underclock my RAM to 100mhz so the memory is running a little slower.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 35 of 49, by H3nrik V!

User metadata
Rank Oldbie
Rank
Oldbie

SL5QJ is Coppermine-T, so you should check compatibility with your motherboard if going down that road ...

Please use the "quote" option if asking questions to what I write - it will really up the chances of me noticing 😀

Reply 36 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie
H3nrik V! wrote on 2023-03-09, 18:02:

SL5QJ is Coppermine-T, so you should check compatibility with your motherboard if going down that road ...

Yeah I read about that but found several articles & posts (also on Vogons) that basically said any motherboard that supports Coppermine would support Coppermine-T. I'll wait till the other CPU arrives first but because the SL5QJ uses less power and heat, it's the superior CPU to try and OC.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 37 of 49, by red_avatar

User metadata
Rank Oldbie
Rank
Oldbie

Another update: my second 1Ghz CPU (exact same model) arrived and ... zero issues so far. I've done a 3D Mark 99 bench and no crashes. I'll be installing Windows XP tonight and will be swapping my Voodoo 3 over with a GeForce 4 Ti 4200. I forgot I already had a Voodoo 3 3000 in my Pentium II 333 so there's too much overlap there - I might as well go for games up to 2002 with this system.

Retro game fanatic.
IBM PS1 386SX25 - 4MB
IBM Aptiva 486SX33 - 8MB - 2GB CF - SB16
IBM PC350 P233MMX - 64MB - 32GB SSD - AWE64 - Voodoo2
PIII600 - 320MB - 480GB SSD - SB Live! - GF4 Ti 4200
i5-2500k - 3GB - SB Audigy 2 - HD 4870

Reply 38 of 49, by H3nrik V!

User metadata
Rank Oldbie
Rank
Oldbie

Interesting. That proves that something's wrong with the first one. Puzzles me though, how it can run so much, and still be defective

Please use the "quote" option if asking questions to what I write - it will really up the chances of me noticing 😀

Reply 39 of 49, by bloodem

User metadata
Rank Oldbie
Rank
Oldbie
H3nrik V! wrote on 2023-03-12, 12:52:

Puzzles me though, how it can run so much, and still be defective

This is nothing! In 2001, a friend of mine had a 700 MHz Slot 1 P3 Coppermine that died in a very unusual way: basically everything was working, except for multimedia content (games, video & music files).
Accessing any game resulted in artifacts followed by a total system freeze. The behavior was similar for video files (random colored patterns without the OS actually freezing) and a screeching sound that could wake the dead (same for MP3 files -> screeching sound).
As you could imagine, nobody thought it was the CPU (not even the guys at the computer store), but after ruling out everything else... it became clear that the CPU was without a doubt the actual culprit (and was replaced by the computer store under warranty). We always assumed that the issue was probably related to some specific multimedia instructions (maybe the MMX circuitry?)

1 x PLCC-68 / 2 x PGA132 / 5 x Skt 3 / 9 x Skt 7 / 12 x SS7 / 1 x Skt 8 / 14 x Slot 1 / 5 x Slot A
5 x Skt 370 / 8 x Skt A / 2 x Skt 478 / 2 x Skt 754 / 3 x Skt 939 / 7 x LGA775 / 1 x LGA1155
Current PC: Ryzen 7 5800X3D
Backup PC: Core i7 7700k