AMD X2 3800+ temps

[GeezerMan]

Originally posted by: MarcVenice
1.475vcore isn't that low. I can imagine it hits 60celsius with a stock HSF.

That's right. That is not a low voltage

 

[bryanW1995]

Originally posted by: Yoshi911
I hav'nt read all the posts, but you should try re-seating the heatsink. I just mounted a new 165 opty and was getting load temps of 80C after overclocking it to 1.475 volts, WAY TO MUCH. so I applied a little more Arctic silver 5, re spread it, and reseated the heatsink. Now im getting load temps of 60, way less (but still to much) i think i need to lap my heatsink.

that is the right temp for an overclocked opty. I had 57c with a zalman on mine running seti@home with an operton 180 @ stock.

 

[JC]

I'm running this X2-3800 at 2.5GHz, 1.45v, 4-pipe AMD HSF w/AS5. Runs around 60C fully loaded (which is 24/7, I run Rosetta@Home).

This is me, not worried about the temp

 

[geokilla]

I just thought of something. If Marc is getting 60C with a stock HSF and 1.475 vcore, shouldn't I be getting something like 50C since the chip is undervolted to 1.25V at 2.1Ghz? My temps seem to be way too high. Then there's JC getting around 60C at 1.45V. I understand that he's using AS5 so that will lower the temps quite a bit. I seriously think that I should reseat the HSF.

Also, is the Artic Cooling Freezer 64 Pro the best bang for buck HSF? If not, do you have a suggestion for a better aftermarket HSF that's around $30 from Canada Computers?

BTW, Running F@H 80% load, I get around 53C, while PC Probe tells me around 47C.

 

[NesuD]

core temp can be decieving. You need to be looking at the Tcase Temp rather than the core temps. Core temps will spike up very high at times. AMD specs their cpu by the tcase temps. As best as i can determine Tcase is measured roughly where the die meets the heat spreader. measuring at that point will give a pretty fair indication of how well you are dissipating the heat. The core temps are measured deeper in the die core so you will always get much higher temp readings at that point. If you are using speedfan to measure usually temp 1 is the tcase measurement. CoreTemp will tell you what the Tcase temp rating is for your cpu.

 

[geokilla]

Can I read the tcase temps from ASUS PC Probe? Core Temp doesn't show the TCaseMax for my chip for some reason. I'm using Core Temp version 0.95 to check my current temps. In SpeedFan, Temp 1 states 45C, with Temp 2 at 37C and Temp 3 at 30C. Then under HD0, I get another Temp1 at 40C and Core at 43C. In PC Probe, I'm getting the same temp as Temp 1, which states 45C.

Also, any advice on the aftermarket best bang for buck HSF?

 

[Bill Kunert]

Also, any advice on the aftermarket best bang for buck HSF?[/quote]

I'm happy with my Arctic cooling Freezer 64 Pro. See my post early in this thread(post #13).

 

[geokilla]

I've come to the conclusion that Temp 1 from SpeedFan is the same as the CPU temp from PC Probe. Assuming that they read the temps from the Tcase Temp, it seems that I have around a 5C headroom. I did some googling, and it said that the Tcase max Temp for my 3800+ is from 55C to 72C. How do I know what the max temp of my chip is? Core Temp doesn't show the TCaseMax for some reason.

@Bill, is that with the MX1 paste or with AS5? I'm planning on getting the Arctic cooling Freezer 64 Pro if my parents allow me, and I need to determine whether it's hard to install and whether I should stick with the MX1 or buy myself some AS5.

 

[GeezerMan]

Originally posted by: geokilla
I've come to the conclusion that Temp 1 from SpeedFan is the same as the CPU temp from PC Probe. Assuming that they read the temps from the Tcase Temp, it seems that I have around a 5C headroom. I did some googling, and it said that the Tcase max Temp for my 3800+ is from 55C to 72C. How do I know what the max temp of my chip is? Core Temp doesn't show the TCaseMax for some reason.

@Bill, is that with the MX1 paste or with AS5? I'm planning on getting the Arctic cooling Freezer 64 Pro if my parents allow me, and I need to determine whether it's hard to install and whether I should stick with the MX1 or buy myself some AS5.

The MX1 paste is pretty good, real close to AS5. Here is the Tcase max utility. The AC64 is pretty easy to install. I have used the 64 pro myself, but have gotten a little better temps out of the AMD heatpipe HSF, especially if you want to short out the thermistor on the fan to allow even greater fan speed, like during stress testing.

Tcase max Utilty

 

[geokilla]

Thanks Geezer, but unfortunately it can't find the Tcase max because I have the F2 stepping (AM2). I did find this, but I don't get it. I did find this though, Thermal Data from AMD's site. If anyone understands it, can you tell me what the Tcase max is for my CPU? The OPN is ADA3800CU B0X.

I also found this My CPU specs from AMD

 

[Bill Kunert]

@Bill, is that with the MX1 paste or with AS5? I'm planning on getting the Arctic cooling Freezer 64 Pro if my parents allow me, and I need to determine whether it's hard to install and whether I should stick with the MX1 or buy myself some AS5.[/quote]

That's with AS5. I had to lap the heatsink and the heat spreader because both were concave. The heat sink is not difficult to install. Hook the latches over the pins on the mount and flip the lever.

 

[geokilla]

Does anyone have any idea what my Tcase Max is? I can't seem to figure it out, but I think it's around the 55C area.

Thanks Bill. Judging from your post, it seems that even a noob (me) can install it with little difficulty.

Edit: Can someone correct me if I'm wrong on the following:

PC Probe and Temp 1 from SpeedFan reads the temperatures from the TCase.

 

[geokilla]

So I guess I'm fine with the Tcase Max being around 55C?

 

[Bill Kunert]

Max TCase for the X2-3800 Toledo is shown as 67C in the AMD64 MaxTCase program.

 

[geokilla]

I have the X2 3800 Windsor one, so the AMD64 MaxTCase program can't read it. I have to rely on guessing I guess.

Is the Scythe Katana 2 better than the Artic Cooling Freezer 64 Pro? The reason I ask this is because the price difference between the two coolers are $6, and the Katana one looks better. It also supports AM2 sockets.

Arctic Cooling Freezer 64 Pro.

Scythe KATANA 2.

Here is a review for the Scythe Katana: Anandtech's Review on Scythe Katana. It seems that the Scythe is a pretty good bang for buck cooler as well.

 

[Rhoxed]

My 3800+ used to reach temps of about 57oC load at 2.5ghz 1.45vcore so i bought a zalman9700nt cooler, temps wont reach over 48oC even at 2.7 (cant get past 2.7 stable though)

the cooler gave me an easy 10-15 degree difference.

 

[geokilla]

@Rhoxed, I don't think I can get a Zalman because they are too expensive. My parents were annoyed and mad when I said I want to get a new HSF for around $30 CAD.

 

[f4phantom2500]

mrk6 is right; touch it. how hot does the heatsink feel? if it's not even warm then the heatsinks not seated right or is defective or something, or the temp readings are incorrect. if it is hot then you probably just have a really really hot chip (although i doubt that's the case, at 1.25v it shouldn't be hot at all). if that's the case, then yeah you need a new heatsink.

 

[geokilla]

Ok, so it's near idle right now. I touched the heatsink (top and right side) and it feels just a bit warm. It's definitly not hot, but it feels like even my monitor is hotter than the heatsink. I'll put the CPU under some load, and then touch it again.

Edit: Ok, 17 min of Orthos Small FFTs, the HSF feels warm at the top, but at the side it feels cool, just like before. The difference between idle and load was not too big. It just feels a bit warmer, but not much.

Seems like the HSF isn't properly installed afterall.

 

[geokilla]

Um, no suggestions?

Anyways, my CPU idles at 32C with 1.25V. This reading was from the BIOS. And I'm pretty sure AMD takes there temps from the Tcase and not hte core die that Core Temp reads.

 

[geokilla]

Anyone please?

 

[MarcVenice]

Erm what suggestion do you need ? what temps do you hit when you stress the cpu with smal fft's ? For example, with the stock hsf, at 2ghz, stock voltage, 1.35vcore, small fft's would heat up my cpu 0 to high 50's celsius. Idle would be around 40ish.

Right now, with my new arctic freezer pro, I'm idling around 35 celsius, at 2.6ghz, stock vcore, and I'm topping out around 55 celsius using small fft's for 30 minutes, I doubt it will go any higher. Using a x2 3800+ btw, toledo core, I think this puppy is gonna run at 2.7ghz or even higher no problemo The heatsink feels slightly warm to the touch, I'm guessing around 35-40ish celsius, slightly heated up water kind of temperature. I've also got a casefan directly pulling the hot air out, might wanna consider doing the same ...

 

[GeezerMan]

Originally posted by: MarcVenice
Erm what suggestion do you need ? what temps do you hit when you stress the cpu with smal fft's ? For example, with the stock hsf, at 2ghz, stock voltage, 1.35vcore, small fft's would heat up my cpu 0 to high 50's celsius. Idle would be around 40ish.

Right now, with my new arctic freezer pro, I'm idling around 35 celsius, at 2.6ghz, stock vcore, and I'm topping out around 55 celsius using small fft's for 30 minutes, I doubt it will go any higher. Using a x2 3800+ btw, toledo core, I think this puppy is gonna run at 2.7ghz or even higher no problemo The heatsink feels slightly warm to the touch, I'm guessing around 35-40ish celsius, slightly heated up water kind of temperature. I've also got a casefan directly pulling the hot air out, might wanna consider doing the same ...

That's my experience as well with the Toledo core, both a Opteron 165 and the 3800 X2, around 2.6 GHz, with the better AMD heatpipe HSF, temps in the mid 50s under heavy test load. Some software reports the core, some reports the CPU case temp. Core is always hotter than case.

 

[MarcVenice]

CPU rightmark is what im using, cpu 0 shows it idling at 35celsius, core 1 would be idling at 24 celsius, obviously near inpossible at ambient temps of 19-20 celsius, so thats the case temp. Funny thing is, when I run orthos small fft's, the case temp becomes 50ish celsius, and like I said the core will be at 55 celsius, so the discrepancy becomes smaller.

2.6ghz suits my needs just fine, okay, it won't beat a core 2 duo at 3-3.2ghz, but it's nearly as fast as a stock e6600, which would have costed me 2.5 times the money. It's not gonna be the limiter in most if any games, so game on !

 

[geokilla]

Umm, the suggestion I need is about the stock heatsink not feeling warm and idle, and just a bit warm under load. Even my monitor is warmer than the heatsink when it's under load.

PC Probe gives me 50C when I run Orthos Small Ftts for 20min, while Core Temp states 54C. This is with stock hsf at 2.1Ghz, 1.25V.

I've got an exhaust fan and a side intake fan.

Also, the AMD thermal data sheet that was linked in my above posts state that AMD reads their temps from the Tcase and not from the core. So now I trust PC Probe since it reads the TCase one.

I also get that huge temperature difference when my CPU is at idle, but under load, like what Mark said, the disrepancy becomes smaller.

Reason I was asking for suggestions is because I'm not sure what I should do now. Temps seem way to high even for an undervolted chip.

Thanks.