The heat sink on the right in this photo is for AM3+; in an effort to make is work with my 939 and the original flexy mounting bracket, I swapped the locking mechanism from a stock 939 heat sink. Then I tried it, and I discovered that it is too wide for the 939 mounting bracket, which keeps it from making contact with the CPU.

Ok, i decided to work on the machine again after taking a significant break and listening to a lot of music (Rings of Saturn and After the burial are good tho) just to calm my nerves a bit. I decided to try a couple of things today, using the pea method, unscrewing the bracket and using the heatsink alone. I had to compress the images to 1024x768 via paint.net because I borrowed my dad's iPhone and they were too big for vogons, but here's pics of the cpu, heatsink and temperature readings

Just as a sidenote I am still using the same arctic mx-4 thermal paste. I do not know if 50-60 C is normal for idle temps (especially when using the heatsink with and without the bracket) but at least it's better than the 80-131 C that I had previously, the pea method seems to be working decently for now. It seems to be staying at a stable 60-61 C at the BIOS. Maybe im just so used to modern intel temps (those from core i3, i5 and i7) that I find AMD temps to be abnormal for some reason. Thanks guys 😀

An apples to apples comparison isn't possible but; my 939 system with an Opteron 180 overclocked to 2.7 GHz with the stock copper/heatpipe 6000+ cooler, Noctua thermal grease in a case with no side panel installed: idle 26°, 33° right after running 3d Mark 03. That's using Gigabyte's Easy Tune 5 for temperature monitoring in Windows - rebooting to check in the CMOS settings shows 30° CPU with a system temperature of 34°.

OK, that's quite a difference in temperatures here between what you started with and what you're dealing with now. So I'm still having a hard time telling if the motherboard sensor(s) could be off.

Do as I suggested in my previous post and try different software like HWInfo, OCCT, and Speedfan to see if they report those temperatures too, or if there are additional sensors for the CPU. It could be that the temperature shown in BIOS is using a flaky/dud/non-existent sensor. I can't remember on which motherboard I had this issue, but it was showing my CPU as -127C... which is rather amusing.

And if the BIOS temperatures turn out to be accurate... then yes, even 50C is still way too much for a socket 939 CPU.
As a more fair comparison, I have several 3200+ and 3500+CPU in various different motherboards, all running stock heatsinks and fans. On my OC'ed 3200+ (2-->2.5 GHz on stock 1.44V core voltage), I get about 34-35C with 27C ambient room temperature and top at 50C after hours of running a synthetic benchmark to keep the CPU at 100% all the time. With regular "100%" CPU load from games, the CPU won't break 45-46C. And again, that's on a stock heatsink with 70 mm fan running @ about 2000-2500 RPM. On my other 939 systems, I get low-mid 40's Centigrade under same load and same ambient temperatures. If ambient room temperature is 5C lower, I get an equivalent drop in CPU temperature on all machines... so I can attest to the accuracy of the sensors of those machines.

In regards to thermal compound: pea-sized method is good for testing if you get good spread on the heatsink. But I personally advise to cover entire surface area of the CPU IHS with a really thin layer thermal compound - this way you can guarantee that all of the CPU's IHS comes into contact with the surface area of the cooler for better heat transfer. For CPUs with IHS, I find the pea-sized method to not be enough to cover the entire IHS - especially around the corners.

I'd bet there's something wrong with the cooler mounting mechanism. 61C is not normal idle temps for a 3700+. Sometimes the lever that secures the cooler to the CPU goes bad, it either bends or stretches, and will fail to seat correctly or apply sufficient mounting pressure. It's also possible (like other users stated) that the cooler is not original to the motherboard and is not applying sufficient mounting pressure to the chip. Try sourcing another cooler, preferably one from another 754/939 pc, and install that. If you don't have another one handy, order a cheap socket am2/am3 cooler that fits on the stock mounting bracket. I Highly doubt you have an issue with the thermal compound.

momaka wrote on 2024-03-08, 09:09:

I find the pea-sized method to not be enough to cover the entire IHS - especially around the corners.

I've been using the pea method since the 90s. I was a system builder and have built several hundred PCs since then. And have always had perfect coverage when removing the heatsink afterwards.

I do agree with you for 13th/14th gen Intel Core. Those rectangular ratio IHS need a little help in the corners.

Socket3 wrote on 2024-03-08, 21:20:

It's also possible (like other users stated) that the cooler is not original to the motherboard and is not applying sufficient mounting pressure to the chip.

I Highly doubt you have an issue with the thermal compound.

If the heatsink doesn't have a Compaq part number sticker on it, I would be suspect. Agree about the thermal compound.

the cooler is not clamped down to the cpu, if we look at the picture OP post, the cooler basically just sitting on the cpu and using it own weight as the pressure on the cpu so 50*c idle is pretty decent with that kind of mounting pressure 😁

Looks like a s.754 or A h/s to me.
That is not the right one and I can see a missing mounting mechanism.

nhattu1986 wrote on 2024-03-09, 01:48:

the cooler is not clamped down to the cpu, if we look at the picture OP post,

Yes, if you read the last page, we asked the O/P to do this to hopefully rule out incompatibility between cooler and mounting bracket.

nhattu1986 wrote on 2024-03-09, 01:48:

the cooler basically just sitting on the cpu and using it own weight as the pressure on the cpu so 50*c idle is pretty decent with that kind of mounting pressure 😁

Socket3 wrote on 2024-03-08, 21:20:

It's also possible (like other users stated) that the cooler is not original to the motherboard and is not applying sufficient mounting pressure to the chip

You don't need a ton of mounting pressure to keep the temperatures low. Just some initial pressure to "smoosh" the thermal compound the first time. I do this by slightly turning the cooler in both directions while pressing down on it to compress the thermal compound. This gets rid of the air bubbles and couples the cooler nicely to the CPU. Once this is done, you can actually forgo using a clamp... again, so long as the motherboard is sitting in a horizontal position.
I ran a system like this for a long time... ironically also a socket 939 system. The temperatures were fine and not any higher than with clamp installed.

That being said, I absolutely DO NOT recommend doing this with exposed-core CPUs/GPUs - at least not long-term. And in the case of socket A CPUs, not using the clamp = dead CPU.

As for socket 754, 939, and AM2/3, I often see cooler clams that over-apply too much pressure, especially some stock and OEM ones. I have quite a few motherboards that have become bent like a bow due to too much pressure from the cooler. The only ones worse are Intel pushpin coolers for LGA 775/115x - those I *HATE* with a passion. Have seen one too many motherboards with physically cracked BGA under the CPU socket due to the crappy pushpin coolers warping the board like crazy over time.

momaka wrote on 2024-03-10, 16:23:

As for socket 754, 939, and AM2/3, I often see cooler clams that over-apply too much pressure, especially some stock and OEM ones. I have quite a few motherboards that have become bent like a bow due to too much pressure from the cooler. The only ones worse are Intel pushpin coolers for LGA 775/115x - those I *HATE* with a passion. Have seen one too many motherboards with physically cracked BGA under the CPU socket due to the crappy pushpin coolers warping the board like crazy over time.

Absolutely. Take some budget boards, they give you the impression of a likely very quick demise.
I have some Asrock (8.9-in x 6.7-in, 22.6 cm x 17.0 cm) that, if you use a copper core pushpin cooler (the ones that protrude a bit), will bow like about to break in two. This is why I started testing stuff with heat pipes coolers that allow a clamp; the clamp can be set to allow different traction (force).

Socket A are a good example of pain with bad positioned coolers.

Basically OP's issue is a cooler missing. At least it makes it easy to solve. 😀

kingcake wrote on 2024-03-08, 22:57:

Socket3 wrote on 2024-03-08, 21:20:

It's also possible (like other users stated) that the cooler is not original to the motherboard and is not applying sufficient mounting pressure to the chip.

I Highly doubt you have an issue with the thermal compound.

If the heatsink doesn't have a Compaq part number sticker on it, I would be suspect. Agree about the thermal compound.

I think the last time I looked at the heatsink it indeed doesnt have a part number/sticker on it, so im guessing its probably a bad heatsink then. What kind of heatsinks for 939 do you guys recommend?

nhattu1986 wrote on 2024-03-09, 01:48:

the cooler is not clamped down to the cpu, if we look at the picture OP post, the cooler basically just sitting on the cpu and using it own weight as the pressure on the cpu so 50*c idle is pretty decent with that kind of mounting pressure 😁

I forgot to take a picture of the cooler being clamped after I screwed the bracket back in. That was just a picture i took testing the cpu and cooler for temps, and I didnt have time to take another picture with the bracket fully in the motherboard. Apologies

RoozerXC wrote on 2024-03-11, 01:52:

I think the last time I looked at the heatsink it indeed doesnt have a part number/sticker on it, so im guessing its probably a bad heatsink then.

None of the OEM heatsinks I've ran across had any part numbers on them.
The one you have looks identical to one I got with an Athlon 3500+ from a Compaq -OEM'ed MSI motherboard.
I ran a 4000+ with that heatsink and the temperatures were exactly the same as with the 3500+.
Socket 939 CPUs are actually pretty true to their stated TDP. For most, that's 67W IIRC. Only the X2 were 89W, with the FX-57 being 105W and FX-60 110W... well, supposedly. I personally never saw my FX-57 go that high, even with benchmarks.

RoozerXC wrote on 2024-03-11, 01:52:

What kind of heatsinks for 939 do you guys recommend?

Anything with 2 or more heatpipes and a large fan would be nice.
I bought a China no-name cooler with 2 heatpipes and 90 mm fan, and it keeps my 89W socket 754 3400+ about 10C lower than the stock cooler while also being much more quiet (no fan control from the motherboard, so fan is at 100% all the time... but it's a quiet fan.)

Nexxen wrote on 2024-03-10, 16:40:

Absolutely. Take some budget boards, they give you the impression of a likely very quick demise.
I have some Asrock (8.9-in x 6.7-in, 22.6 cm x 17.0 cm) that, if you use a copper core pushpin cooler (the ones that protrude a bit), will bow like about to break in two. This is why I started testing stuff with heat pipes coolers that allow a clamp; the clamp can be set to allow different traction (force).

Yup - just loosen or tighten the screws to get the desired optimal clamping force (not too loose, not too tight)... and all of that without warping the board.

That said, I still like the stock "full-height" Intel heatsinks too much (especially the ones with the copper insert) to just chuck them away. So what I do is I remove the push pins and assemble my own rear motherboard clamp and mounting mechanism. I used to work in a console repair shop back in the Xbox 360 and PS3 days, so I have a ton of metal clamps from Xbox 360's and PS3s. The ones from PS3s can be easily modified for diagonal retention on s775 or normal retention on s939... though I modify the clamps and brackets for socket 754 and 939 boards in a different way, using the original hardware and just by adding a shim under the CPU clamp and longer bolts between top and bottom part to account for the shim (and those pesky coolers that have really tight/rigid clamps that bent the heck out of boards.)

Nexxen wrote on 2024-03-10, 16:40:

Socket A are a good example of pain with bad positioned coolers.

Yeah, socket A (and 370) are not very user-friendly or mod-friendly... not without some more serious mechanical understanding / abilities.
But one thing I do give credit to its design is that the pressure for the heatsink IS actually supplied by a real metal clamp and the clamping force is NOT passed onto the motherboard to make it warp. With some many socket 754, 939, and AM2/AM3 stock coolers, the metal "clamp" in the heatsink is not actually a clamp, but a rigid piece of metal that doesn't bend at all. Instead, it transfers all of the travel from the retention lock down onto the back plate / bracket on the motherboard, which in turn transfers most of that travel onto the motherboard's PCB. The result is often a severely warped motherboard. It's a good way to kill a motherboard over time.

RoozerXC wrote on 2024-03-11, 01:52:

kingcake wrote on 2024-03-08, 22:57:

Socket3 wrote on 2024-03-08, 21:20:

It's also possible (like other users stated) that the cooler is not original to the motherboard and is not applying sufficient mounting pressure to the chip.

I Highly doubt you have an issue with the thermal compound.

If the heatsink doesn't have a Compaq part number sticker on it, I would be suspect. Agree about the thermal compound.

I think the last time I looked at the heatsink it indeed doesnt have a part number/sticker on it, so im guessing its probably a bad heatsink then. What kind of heatsinks for 939 do you guys recommend?

My experience with Socket 754/939 coolers has been that the stock ones are fine if you are not overclocking the CPU. I still don't understand why your temperatures are so high - unless there is something wrong with the clamp or the fan on the heatsink. This one is either very similar, or identical to what I'm using on my overclocked Opteron, so it should be more than adequate for your needs.

https://www.ebay.com/itm/165923344456?itmmeta … %3ABFBM0sqL5MVj

Repo Man11 wrote on 2024-03-11, 16:09:

RoozerXC wrote on 2024-03-11, 01:52:

kingcake wrote on 2024-03-08, 22:57:

If the heatsink doesn't have a Compaq part number sticker on it, I would be suspect. Agree about the thermal compound.

I think the last time I looked at the heatsink it indeed doesnt have a part number/sticker on it, so im guessing its probably a bad heatsink then. What kind of heatsinks for 939 do you guys recommend?

My experience with Socket 754/939 coolers has been that the stock ones are fine if you are not overclocking the CPU. I still don't understand why your temperatures are so high - unless there is something wrong with the clamp or the fan on the heatsink. This one is either very similar, or identical to what I'm using on my overclocked Opteron, so it should be more than adequate for your needs.

https://www.ebay.com/itm/165923344456?itmmeta … %3ABFBM0sqL5MVj

Thank you so much. I'll try it when i get it delivered at some point and see if the temps get any better. I'm still using the pea method for now and im considering adding more fresh paste onto the cpu as momaka said on their post earlier

I never overclocked this CPU so it could be something else entirely, but for now ill see if changing the heatsink helps anything.