TR Forums

Hello, I have a pentium EE 965 that gets ridiculously hot, even at stock settings being cooled by a 240mm AIO. It easily breaks into the mid 80c range and sometimes spikes above 90c. I believe the CPU die has poor thermal contact with the IHS, because I have used different coolers and tried reseating them multiple times and the results are usually similarly abysmal. I didn't think the solder under the IHS could go bad over time, but that's the only thing I can think of.

Can solder "go bad" or wear out? Should I delid the CPU?

If you try to delid a soldered CPU, you're most likely gonna kill it. Unless the entire solder puddle decoupled from the IHS, then it might delid pretty easy....

Anything can go bad. If you're tried multiple coolers, replaced the thermal paste, and checked the seating for good contact then I think you may be on the right track with your theory. As far as fixing it goes, you could always fry it up in a pan to reflow the solder! https://www.youtube.com/watch?v=Hf2KaYIR2A4

note: no promise this will work for you

How sure are you that the temperature reading is accurate? Is the CPU throttling?

just brew it! wrote:

How sure are you that the temperature reading is accurate? Is the CPU throttling?

I'm pretty sure the reading is accurate. I'm using Speedfan to measure temps, but I also checked in the BIOS and some other applications and they all seem to show similar CPU temps. AFAIK the CPU isn't throttling, but I'm not sure if these CPUs did thermal throttling, just thermal shutdowns (This CPU hasn't caused a thermal shutdown). This was before CPUs had dynamic clock speeds.

Has it been running at stock its entire life, or did it spend a lot of time overclocked? If it has been pushed hard for many years it may have sustained internal damage to the silicon itself which is raising power dissipation.

As others have noted, attempting to de-lid a CPU with a soldered IHS is a very risky undertaking.

just brew it! wrote:

Has it been running at stock its entire life, or did it spend a lot of time overclocked? If it has been pushed hard for many years it may have sustained internal damage to the silicon itself which is raising power dissipation.

As others have noted, attempting to de-lid a CPU with a soldered IHS is a very risky undertaking.

I don't know how this processor was treated by the previous owner, but I ran it at 4 GHz for a while because the temperature difference was minor, but it's just been so hot I turned it back to stock because it felt safer. Also I will have access to a hot air resolder work station, so hopefully that will make the procedure less risky. If not, I'll just put a spare Q9400 in the motherboard.

Any software reading of temperature is most likely going through one "Winbond" chip, so several software readings corroborating don't mean much. The only way to get a real reading is with a thermocouple.

cegras wrote:

Any software reading of temperature is most likely going through one "Winbond" chip, so several software readings corroborating don't mean much. The only way to get a real reading is with a thermocouple.

I'd have to delid the CPU to get the thermocouple in the right spot anyways.

If you have the appropriate gear, and it won't inconvenience you terribly if it dies, then yeah you might as well give it a shot. At the same time, if it's not super important hardware I'm not sure how much time I'd spend on it myself. Ark lists tcase as 68.6C, so like JBI said, if those temperatures are accurate it may be circling the drain already.

I have an AIO, and the radiator can get quite dusty over time. I'd blow out the radiator at least, and give the fans a good wipe.

Did you happen to update your BIOS recently? I had a P5Q Pro with a BIOS bug which caused it to misreport the temp about 20C too high. I knew it had to be some kind of error, and waited it out until the next revision, which magically fixed the problem.

The Egg wrote:

Did you happen to update your BIOS recently? I had a P5Q Pro with a BIOS bug which caused it to misreport the temp about 20C too high. I knew it had to be some kind of error, and waited it out until the next revision, which magically fixed the problem.

I updated the BIOS when I got the motherboard. It's the newest BIOS available for the board. That's very interesting because the motherboard I have is the Asus P5Q. I wonder if Asus fudged the numbers.

So we're talking about a Pentium 4 EE chip?

I see my time travel experiment has been a success!

Invest in Apple and Nvidia stock guys!

In my experience, I've found the temperatures reported by motherboards to be neither accurate nor precise, especially motherboards in that vintage. The only trustworthy/reliable information you can easily gather (without a thermocouple, as mentioned) is whether or not the CPU is throttling. There is a small utility you can download to check, and then you will know for sure. If it's not throttling, and you're not experiencing any crashes/shutdowns under load, then in my opinion you should just leave it alone and forget about it -- that is, unless you really want to break fix stuff

Takeshi7 wrote:

just brew it! wrote:

Has it been running at stock its entire life, or did it spend a lot of time overclocked? If it has been pushed hard for many years it may have sustained internal damage to the silicon itself which is raising power dissipation.

As others have noted, attempting to de-lid a CPU with a soldered IHS is a very risky undertaking.

I don't know how this processor was treated by the previous owner, but I ran it at 4 GHz for a while because the temperature difference was minor, but it's just been so hot I turned it back to stock because it felt safer. Also I will have access to a hot air resolder work station, so hopefully that will make the procedure less risky. If not, I'll just put a spare Q9400 in the motherboard.

if you have a spare Core 2 Quad I wouldn't hesitate to replace a P4. Should be night and day performance difference, and probably better thermals too.

Takeshi7 wrote:

The Egg wrote:

Did you happen to update your BIOS recently? I had a P5Q Pro with a BIOS bug which caused it to misreport the temp about 20C too high. I knew it had to be some kind of error, and waited it out until the next revision, which magically fixed the problem.

I updated the BIOS when I got the motherboard. It's the newest BIOS available for the board. That's very interesting because the motherboard I have is the Asus P5Q. I wonder if Asus fudged the numbers.

Hmm......is this a recent "new" build? If so, it could very well be the same bug I encountered, and you might try a few different BIOS revisions to see the results.

If it's an old build and you've been running the same config/BIOS for multiple years, it's something else. In either case, though, I question the accuracy of the temp reporting.

chuckula wrote:

So we're talking about a Pentium 4 EE chip?

I see my time travel experiment has been a success!

Invest in Apple and Nvidia stock guys!

It's a Pentium EE, not a Pentium 4 EE. So it's basically a dual core Pentium 4 EE.

The Egg wrote:

Hmm......is this a recent "new" build? If so, it could very well be the same bug I encountered, and you might try a few different BIOS revisions to see the results.

If it's an old build and you've been running the same config/BIOS for multiple years, it's something else. In either case, though, I question the accuracy of the temp reporting.

It is an old build, and i've been running this BIOS since I got it. Temps have always been awful, but it has always worked. It just makes me so nervous every time I see the temps so high.

You're going to get high temperatures from such a ridiculously inefficient chip.

If temperatures have always read that high, then I can't imagine those temperatures are accurate. A CPU run at 10-15 degrees hotter than its max tcase temperature almost certainly wouldn't have survived this long otherwise.

Takeshi7 wrote:

It is an old build, and i've been running this BIOS since I got it. Temps have always been awful, but it has always worked. It just makes me so nervous every time I see the temps so high.

Soooo......did it just recently begin running hotter, or having other problems? I'm confused as to why you're suddenly concerned 10 years later.

Takeshi7 wrote:

If not, I'll just put a spare Q9400 in the motherboard.

I try not to give people too much trouble about running older hardware, but why on earth would you still be using a Netburst chip when you have a drop-in C2Q just laying around?

techguy wrote:

Anything can go bad. If you're tried multiple coolers, replaced the thermal paste, and checked the seating for good contact then I think you may be on the right track with your theory. As far as fixing it goes, you could always fry it up in a pan to reflow the solder! https://www.youtube.com/watch?v=Hf2KaYIR2A4

note: no promise this will work for you

Cool video, but make sure to hit mute as soon as it starts, or some dude will talk French at you the whole time. I could barely stand it.

The Egg wrote:

Takeshi7 wrote:

It is an old build, and i've been running this BIOS since I got it. Temps have always been awful, but it has always worked. It just makes me so nervous every time I see the temps so high.

Soooo......did it just recently begin running hotter, or having other problems? I'm confused as to why you're suddenly concerned 10 years later.

Takeshi7 wrote:

If not, I'll just put a spare Q9400 in the motherboard.

I try not to give people too much trouble about running older hardware, but why on earth would you still be using a Netburst chip when you have a drop-in C2Q just laying around?

I've only had this build for about 9 months. I've always been sort of worried about the temps, but I figured I'd give it time to let the thermal paste cure or something.

Also I built this PC specifically to be a maxed out Netburst machine. The spare C2Q came with the motherboard.

Takeshi7 wrote:

Also I built this PC specifically to be a maxed out Netburst machine. The spare C2Q came with the motherboard.

Ahh, I see. A "because it is there" sort of exercise.

Takeshi7 wrote:

Also I built this PC specifically to be a maxed out Netburst machine. The spare C2Q came with the motherboard.

Man, I guess I suffered more during the NetBurp era than you did. I literally have a desk drawer full of Pentium Ds and late model Presshots, and I'm happy knowing that they rot silently, hurting no one. I wasn't allowed to talk about it at the time, but I remember clearly how terribly disappointed I was with the Prescott review kit. A Pentium 4 560, at 3.6GHz nominal speed (throttled constantly on the official HSF), and it was routinely outperformed by my Northwood 2.8C. Not by a little either, but by a lot.

Forge wrote:

Man, I guess I suffered more during the NetBurp era than you did. I literally have a desk drawer full of Pentium Ds and late model Presshots, and I'm happy knowing that they rot silently, hurting no one. I wasn't allowed to talk about it at the time, but I remember clearly how terribly disappointed I was with the Prescott review kit. A Pentium 4 560, at 3.6GHz nominal speed (throttled constantly on the official HSF), and it was routinely outperformed by my Northwood 2.8C. Not by a little either, but by a lot.

correct me if I'm wrong, but I don't think Netburst chips actually support Thermal throttling, just thermal shutdown when it gets too high. When I look at HWInfo64 the "Thermal throttling" section under the CPU isn't even there.

Takeshi7 wrote:

The Egg wrote:

Takeshi7 wrote:

It is an old build, and i've been running this BIOS since I got it. Temps have always been awful, but it has always worked. It just makes me so nervous every time I see the temps so high.

Soooo......did it just recently begin running hotter, or having other problems? I'm confused as to why you're suddenly concerned 10 years later.

Takeshi7 wrote:

If not, I'll just put a spare Q9400 in the motherboard.

I try not to give people too much trouble about running older hardware, but why on earth would you still be using a Netburst chip when you have a drop-in C2Q just laying around?

I've only had this build for about 9 months. I've always been sort of worried about the temps, but I figured I'd give it time to let the thermal paste cure or something.

Also I built this PC specifically to be a maxed out Netburst machine. The spare C2Q came with the motherboard.

Ah, I think I remember reading something about that. Not nearly old enough to be "cool" IMO, but to each their own.

Anyhow, yeah....you're likely seeing the same BIOS bug I encountered on the P5Q Pro, but which was apparently never fixed on the regular P5Q. You can mess around with older BIOS revisions if you want (maybe one will give you the correct number to ease your mind), but If you're running stock speeds with a decent HSF, I wouldn't worry about it too much. Doesn't sound like a critical system anyway.

The Egg wrote:

Ah, I think I remember reading something about that. Not nearly old enough to be "cool" IMO, but to each their own.

Anyhow, yeah....you're likely seeing the same BIOS bug I encountered on the P5Q Pro, but which was apparently never fixed on the regular P5Q. You can mess around with older BIOS revisions if you want (maybe one will give you the correct number to ease your mind), but If you're running stock speeds with a decent HSF, I wouldn't worry about it too much. Doesn't sound like a critical system anyway.

Thanks for telling me about your woes with the P5Q Pro. I think you're right. Nothing else seems to back up the temps being so high. This motherboard has an Asus EPU so I can see the CPU never draws much more than 90 watts.
And your tastes may differ, but what I think makes this system cool is that this old CPU has 4 threads and still feels modern (besides the power/heat) and completely usable today. It runs Windows 10 x64 from an SSD, has USB 3.0, and it can watch 1080p60 YouTube/Twitch videos, and I can even play Sea of Thieves on it.

I just want to give an update on this. My friend helped me delid the CPU with his hot air workstation. Once it was delidded we added a bit more solder and used a soldering iron to make sure the solder was nicely spread on both the CPU dies and the IHS. ( the IHS was harder because it has much more thermal mass) Then we relidded it and I put it in my PC. It now runs much cooler, Even under full load it is cooler than its old idle temp. Old idle temp was around 60C, load would go above 90C, After delidding/fixing solder, the new temp under load is 48ish (idle in the low 30s now). I will update this comment with a link to the Youtube video once I have it.

tl;dr: Yes the solder under the IHS can go bad, and it is possible to fix by delidding.