Prescott cores were nicknamed 'Pres-hot' for good reason 😀 . I much prefer Northwoods in a socket 478 system- essentially the same performance per clock but with less heat.

That said, S478 motherboards were manufactured during the height of the capacitor plague. If your system is so unstable that you can't even install an OS, my first instinct would be to re-cap it.

The other system I have is Northwood and I think that is what I used back in the era.

Regarding capcitors - I was thinking the same thing. The capacitors aren't boiling electrolyte but that doesn't mean they are good. I guess I'll have to save this project for a point in time when I can justify spending that much time on a project like this.

I had a socket 939 board with similar problems and it was due to bad VRM capacitors. After full recap everything was fine again. How do your caps look like? Given the time period I'd probably try a recap anyway, even if caps look fine. I haven't had a board from this time period that didn't have any cap issues at all to be honest.

here is a shot of the VRM before I installed the board

can bad caps lead to high CPU temps?

55C and even 62C (with a crappy fan) at idle is normal for a Prescott P4 CPU. It does not necessarily mean that anything is bad. Especially with 2 different CPUs - the chances they are both defective is rather low. BSODs that won't even let the Windows installer to complete are either due to bad RAM, a (very) faulty GPU or (most likely) a failing motherboard.

What is the critical temperature for the Prescott parks? Up until recently I used to be able to find most of this on Intel's Ark, but that seems to be gone now. Retrospectively I believe I only ever personally owned the Northwood parts back then, I never purchased or personally used Prescott parts that I can recall. So even the recollection I have is not very useful to be honest.

That is a QDI motherboard I believe, so most likely Panasonic caps with the "T" vent on top. I have recapped many P4 motherboards at this point, and even at this age I have found Panasonic caps from that time frame to still be mostly in spec. So you should visually check what brand the caps are. Many 845 based motherboards needed BIOS updates to properly support Prescott cored P4 CPUs and even then 800MHz FSB CPUs were not really meant for the 845 chipset. I suggest you replace that CPU with a 400/533MHz FSB Northwood Pentium 4 and see how everything comes around......

Yes it is definitely a QDI motherboard. Box says one thing, motherboard says QDI. I had only grabbed the Prescott part because it was cheap. Prices for northwood parts are all over the place, so I will just order a cheap one for now. I have ordered a CPU with stepping SL6WJ that I can try. This is apparently 800 FSB but costs next to nothing, I'll see if the system is anymore stable before trying anything else.

One detail I forgot - once I installed the second CPU I can actually "Save & Exit" in the BIOS without any weird behavior.

386DX40 wrote on 2025-08-25, 16:34:

That is a QDI motherboard I believe, so most likely Panasonic caps with the "T" vent on top. I have recapped many P4 motherboards at this point, and even at this age I have found Panasonic caps from that time frame to still be mostly in spec. So you should visually check what brand the caps are. Many 845 based motherboards needed BIOS updates to properly support Prescott cored P4 CPUs and even then 800MHz FSB CPUs were not really meant for the 845 chipset. I suggest you replace that CPU with a 400/533MHz FSB Northwood Pentium 4 and see how everything comes around......

It's surprising that the Prescott CPU is recognized on some i845 motherboards...
In my case, only Pentium 4 Willamette (400MHz FSB) works without issues on my Chaintech 9BID0-C100A (i845D) motherboard. Pentium 4 Northwood (800MHz FSB) is also works at half CPU frequency due to system bus limitation at 400MHz. Pentium 4 Prescott is not supported at all, latest BIOS doesn't know these processors.

Pentium 4 3.0 HT (SL7PM) has a TDP of 89W. While not the most power hungry P4, it's not the most efficient either.

For lower power consumption and lower temps, you want earlier Northwood processors (54W to 68W) or late model D0 Cedar Mill processors (65W). The latter would require an LGA 775 board.

I'll join the others in saying that I'm surprised this motherboard even POSTed with a 3.0/1024/800.

My only 845 P4 board is an Asus P4B 266 that came with a 1.7/256/400 CPU. It was badly in need of being recapped, so I did, and after that it worked perfectly - until a I tried a faster CPU (IIRC I tried both a 2.2/512/400 and a 2.0). Any CPU I tried that was faster than the 1.7 would eventually lead to it locking up while installing an OS, or shortly after reaching the desktop if you had installed the OS with the 1.7 CPU. It seemed that something on the board was overheating, but I haven't any idea what. With the 1.7 GHz CPU it never has any issues.

Just to add to this, here is an old discussion on the [H]ardForum from one of the main motherboard reviewers back when [H]ardOCP still existed: https://hardforum.com/threads/845-vs-865-vs-8 … post-1027752595

well that is fun . This is the manual page I was basing my purchasing decisions off. I guess whoever wrote this was very optimistic about what "support" meant

As others have said, expect high temps with P4. But (I believe) it will run at full speed idling in BIOS, so it will get hotter faster than in Windows.

One reason I greatly prefer Pentium 3 if I don't need the extra speed is the much lower TDP.

Regarding the thermals, first thing to do is identify where the problem is. Yes, the Prescott is a serious hothead by early 2000s standard - but utterly unremarkable when compared to CPUs today. The 89W of that 3.0HT is less than many Core i5s today, let alone the bigger monsters. Why isn't that such an issue? Because heatsinks, fans and airflow have been massively improved over the past years.

Idle CPU temp of >60C is high. Question is what temperature the air in the case is. What are the other temps (i.e. case temperature)? And what's the temperature of the air surrounding the case?

My son was complaining about similarly high temps on his Ryzen build - the CPU was hitting 95C and then thermally throttling. He was blaming the (stock) heatsink+cooler. That's definitely not the best one out there, but his case temps were sky-high, almost 70C. He'd built this in an old 2000-era case of mine as his money ran out. The only fan was in the PSU. We added an 80mm outtake fan (the only option available), case temps dropped by 10C and CPU temps als immediately dropped to max 85C and no thermal throttling. Still not great, but a very clear indication the biggest problem was case airflow, not CPU HSF, let alone thermal compound. He's getting a different case once he's saved enough next month.

You probably don't want to put this system in a fancy new case, but the same logic applies: it woudn't surprise me if your primary issue is case airflow and - if the holes are there - adding a case fan or two might do more for thermals than messing around with the contents of the case.

That doesn't mean that you can't also have a problem with overheating VRMs, but there as well the cooling efficiency is directly related to temperature difference between hot component and air around it.

hydrogen18 wrote on 2025-08-25, 22:54:

well that is fun . This is the manual page I was basing my purchasing decisions off. I guess whoever wrote this was very optimis […]
Show full quote

well that is fun . This is the manual page I was basing my purchasing decisions off. I guess whoever wrote this was very optimistic about what "support" meant

As I presume, 533 MHz FSB CPU are highest oficially suported by i845 chipsets (all variants). So 800 MHz variant is overclocked (as manual stated too) and thus oced even AGP/PCI bus? Second thing: all Prescott cores P4 are too hot CPU´s, so better stay with much cooler Northwood, which have same or almost same performance as Prescott, but has much less cooling demand.

The case is an early ATX case. I have ran it with the case side off in all cases. Ambient temperature is 24 C. There is really no way airflow is an issue, the fan I am using at present moves a minimum of 1200 CFM in the worst case.

If the 800 MHZ bus is overclocking the AGP bus, that probably explains many issues. But when I manually lowered the FSB to the minimum value in the BIOS, there was no improvement. So I don't think that is the problem here.

My actual experience with motherboards and CPUs at this time period was oddly enough the chipsets from VIA and the original Northwoods without HT. I effectively quit building systems at that time as I had no further personal reason to upgrade & assembling PCs was not really that profitable for me

So this has been quite the learning experience for me. I've realized a couple things here

1. whoever wrote this manual, was out of their mind. Intel i845 was never meant for these Prescott 800 MHz chips
2. the box says this is an AGP 8x board. Intel i845 officially supports AGP 4x. Maybe they actually got it to run at 8x? Manual for it says 4x in the picture above
3. testing and troubleshooting hardware is really hard when there are no more known good components. Effectively ever component is suspect and there are no longer any NOS motherboards, etc. to use as a comparison

The other system I have working is Northwood,533 MHz FSB at 2.53 GHz clock. Non-HT cpu. AGP 4x slot is working fine. No issues I've ran into, capacitors actually look perfect as well on the board. I don't think there is any reason to actually get this system running at this time. The only way I could make this faster than that system would be to use Pentium 4 3.06 HT which is Northwood with Hyperthreading at 533 MHz FSB. Only stepping I can find on eBay is SL6S5. It's $100 to get it to me, no way I am spending that much on this.

I will throw in the SL6WJ chip when it arrives as it was extremely cheap delivered to me. If it works or if I can tweak FSB to make this stable I'll proceed with it. If not the components are going to go back in the box for now. It can be a backup board