[Maverick0984]

Brand new to the forums, hopefully I'm accepted graciously

Alright well, after a couple days of running the PC I have a better understanding of where my CPU is going to be sitting at idle and on load in terms of temperature.

I am also using a Zalman 9500 heat sink.

A can have basically 3 basepoints, using CoreTemp (because I like the taskbar temp readings). After leaving it sit w/ Windows running with me doing absolutely nothing, meaning the monitor goes into power save and all that, I can check the logs and see that my core temperatures got as low as 35C. Doing low-moderate tasks like browsing and what have you, it's anywhere from 41C-44C. If I put it on full load, like running two instances of prime95 it can get to high 60's C or programs like TAT I even saw 71C, although it is usually a couple degrees higher than CoreTemp. That means, at dead idle I am at 35C, and full load, probably about 68-70C. That seems like a very large increase from what I've read online. Most are saying expected increase is about ~15C from idle to load.

Would I consider my 35C idle? or 44C idle? I assume 35C, which means it nearly doubles under load. I wouldn't think ambient temperatures or anything like that could cause something like that. Mind you, it is OC'd to 3.4ghz and it is pretty stable, so it's not at stock speeds. I'm at work right now but on an MSI board the Vcore settings are done by # over stock, and I am .0250v over stock. Should I try reseating my Zalman 9500? It can be a pretty big ordeal to do something like that, and I'm pretty sure I did it correct, I'm just questioning myself because of the increase in temps.

I can still kind of rotate the heat sink ever so slightly if I give it enough pressure in one direction, but it is on pretty tight and I wouldn't want to make it any tighter. This is the first computer I've built so I'm open to suggestions. However, I did do my research on all of this and understand the importance on quantity of thermal paste (small) and the type of contact, as well as copper > aluminum in the matter. I am a computer engineer though, most of that is pretty obvious to me. Thanks for the continued help. I'd really like to get my temps down, maybe push the OC a bit more because of it.

It is a relatively new install so the Arctic Silver might not be fully "broken in" but I can't imagine it changing anything enough to make me satisfied with it.

Other details include: MSI P6N Diamond 680i Mobo, Geforce 8800 GTX, and it's in a Lian Li v1000B case that contains 3 120mm fans. One pulls out air directly from the heat sink real nicely, the other is on the side right next to the GPU, and the 3rd is in the front bottom, not really close in relation to the CPU (more for hdd cooling).

Thanks for any possible insight The temps make me nervous. I've read all the specs online I can find, but nothing really talks about the large swing I have in temps. Not sure if it could mean the heat sink has a bad connection or what.

[orion456]

Welcome to the Forums!

You are reading core temperatures and yours are around usual readings when overclocking to 3.4ghz.

Get speedfan and read the surface temps. You don't want to exceed 60c or so on the surface. Normally surface temps are about 15-20c less than core so at 71c, you are around 55c on the surface, which is good. Your HS is mounted properly and your temps will improve a bit after a while.

What's the room temperature?

[Maverick0984]

Room temp isn't anything too cool or warm, the past week it has been about 75F in Chicago here, so about 24C. No need for the AC, but it warms up this week!

So when people quote their temps, are they quoting surface temps then or core temps? I had installed Speedfan, but it lists core temps, and after 50C the icon turns into a little flame

The other temperature readings are Temp1,Temp2, and Temp3. Temp1 was like 22C and Temp3 was like -140C (doesn't work on my mobo or something), and Temp2 was 40-50ish, I can't really remember. Like I said, I'm at work and can't check the numbers, but I will as soon as I get home.

[Maverick0984]

So which temp do I believe then? Speedfan gives me temps like the following:

Temp1: 28C
Temp2: 35C
Temp3: -128C (unreliable obviously)
Core0: 44C
Core1: 45C

These are at idle in Win XP 32bit, is there anyway to know what those mean?

[treatmentx]

Core0/Core1 is what you look at. Usually (not always) Temp1 is the surface temp of the cpu and Temp2 is the chipset/Motherboard temp.

[orion456]

Originally Posted by treatmentx

Core0/Core1 is what you look at. Usually (not always) Temp1 is the surface temp of the cpu and Temp2 is the chipset/Motherboard temp.

I'm thinking temp2 is surface of the CPU, but its easy enough to tell, just give it some juice and one of them will rise immediate...that's the CPU.

Unfortunately, people are giving both core and surface temps and not stating which is which. That's happening because core temps have only recently become available.

Surface less than 60c, core less than 100c, are usually the quoted numbers.

The flame icons can be changed using configuration/temperatures and clicking on the temperature you want to change. I usually use 55c and 80c for the surface and core.

[Maverick0984]

Ok, well. Giving it some gas, raises Temp1 to over 40C so that is the surface temp.

I can't help but thinking you two are contradicting each other though?

[treatmentx]

We were. I still think chip surface is Temp1, but I've read some posts where that may not be the case for everyone. The reason why I believe it is my chip surface is, and as you have experienced, is because it increases as I load orthos (gas, as you said). Northbridge does increase in temp, but not nearly as much as the chip, when you're doing something like Small FFT's in Orthos.

I believe chipset temp (northbridge temp) is Temp2 because when I up my voltage for my northbridge, Temp2 goes up. Seems logical enough to me.

However, I have seen it where people's motherboards are different and have swapped it around etc etc, which could lead to why sometimes different people believe in different representation of Temp1, Temp2, Temp3, etc.

[Maverick0984]

Originally Posted by treatmentx

We were. I still think chip surface is Temp1, but I've read some posts where that may not be the case for everyone. The reason why I believe it is my chip surface is, and as you have experienced, is because it increases as I load orthos (gas, as you said). Northbridge does increase in temp, but not nearly as much as the chip, when you're doing something like Small FFT's in Orthos.

I believe chipset temp (northbridge temp) is Temp2 because when I up my voltage for my northbridge, Temp2 goes up. Seems logical enough to me.

However, I have seen it where people's motherboards are different and have swapped it around etc etc, which could lead to why sometimes different people believe in different representation of Temp1, Temp2, Temp3, etc.

Well, sorry, I guess I should have clarified. I'm not really asking what they are specifically. I do agree for me Temp1 is my surface temp, but that varies from board to board.

What I was curious about was when people talk about their "temps on load" or "temps on idle" they are always quoting numbers well below my core temps, and basically corresponding with my surface temps. Like if someone says they are idle at 25C, I can hit that for a surface temp, but my core temps are nowhere near that. The way I am understanding it from Orion456, is that that is OK and that I'm doing just fine. Prior to him saying that, I was comparing my core temps to peoples quoted 25C and I thought I was running insanely hot.

Now, I thought you were talking about that, and saying that I should be worried about core temps and comparing those to other's quoted temps.

Does that make any sense?

Basically, do you agree with Orion456, that a mid-20C surface temp is good, and therefore a low to mid-40C core temp is still good? These being on idle. Since my load core temps reach nearly 70C when OC'd. Seeing others having overclocked this chip well passed what I was at with literally 15-20C differences in temp really scared me. Again, since all they are listed is a generic temp. (usually, I'm sure alot of people are specific)

I was also worried because programs like Speedfan had little icons that displayed flame once core temps surpassed 50C. Yes you can change that, but I trusted them that it was there for a reason.

Do I have cause to still be worried? Or am I to assume most quote surface temps when they omit the source, and that core temps can really reach 100C like Orion claims.

No offense to Orion but the more sources of info that collaborate something, the better I can sleep .

I'm really just nervous since this is my first build and like those more experienced reassuring me .

I guess I could up my NB voltage to see if that corresponds with Temp2 just to know for the future.

Thanks!

[treatmentx]

I apologize, now that I re-read, I see what you mean.

I have ran into the dilemma myself, and I've looked over this thread.
http://www.ocforums.com/showthread.php?t=179028

My take is this. EVEN when people are quoting, for a reason I don't understand, there are even those people who choose to lie about it, as if attaining a better cooling solution. *shrug* i just know for a fact it happens, i've seen a contradictory screenshot before, then the screenshot taken down and replaced.

I think the better way to answer is "is it too hot?" is to follow several steps

1. Air cooling or Water cooling or Extreme? That will put you in different class/category.
2. Get a measurement of your room temp
3. Get a measurement of your case temp
4. Note the voltage that is used in bios,
5. Open CPUZ, keep an eye on the fsb speed and multiplier (C1E/Speedstep etc)
6. Open a voltage monitor, something like PC Probe or something that came with your system to monitor vcore.
5. Do a idle temp using CoreTemp and SpeedFan (core0/core1)
6. Then do a load temp using CoreTemp and SpeedFan (watch your vcore), and write down the vcore as well as the temps
7. Compare your idle/load temps with ambient temps, how far is it off? It could be my imagination, but I feel like watercooling is MORE influenced by ambient temps.
8. Now change your vcore and repeat the process.

There's a guide somewhere, but once you start graphing out the data you've achieved, it should be a log-graph reaching a plataeu. If the plataeu starts real early, say in the 1.35vcore range, then I'd be suspicious of poor cooling (either remount or mind the ambient temps, etc).

Try that, maybe we can help you more after we get more data?

[Maverick0984]

I'll try all that when I get home from work. One thing I did notice, my temps increased enough to notice while increasing the FSB and not moving vcore yet. Which disappointed me, because I could get pretty far up from stock w/o increasing the vcore voltage.

I should also note, that looking at CoreTemp and you probably already know this, my Tjunction is at 85C meaning the CPU will shut it self off at 85C, so I would never come close to the 100C Orion456 mentioned.

Also, I'm on air cooling. Water cooling scares me.

[nikhsub1]

Originally Posted by orion456

Get speedfan and read the surface temps. You don't want to exceed 60c or so on the surface.

First, no program can read surface temps (known as Tcase). The only way to get a Tcase value is to GROOVE the IHS and put a thermocouple (thermometer) in the groove and read from it. There is no sensor by default on the top of the IHS. Second, what are you using to generate load temps? If you are use TAT to generate load please be advised that NOTHING ON EARTH can get a C2D chip hotter. TAT is what is known internally as a 'power virus' and will put the chip some 20% OVER Intel's rated TDP. You will never encounter a load like TAT out in the real world. TAT is what I use for WB test however as I need the high heat load. FWIW TAT will get your C2D about 15C hotter than Orthos.

[jason4207]

Originally Posted by nikhsub1

First, no program can read surface temps (known as Tcase). The only way to get a Tcase value is to GROOVE the IHS and put a thermocouple (thermometer) in the groove and read from it. There is no sensor by default on the top of the IHS. Second, what are you using to generate load temps? If you are use TAT to generate load please be advised that NOTHING ON EARTH can get a C2D chip hotter. TAT is what is known internally as a 'power virus' and will put the chip some 20% OVER Intel's rated TDP. You will never encounter a load like TAT out in the real world. TAT is what I use for WB test however as I need the high heat load. FWIW TAT will get your C2D about 15C hotter than Orthos.

I built a e6300/P965 system last December for a friend and used TAT to test for stability. I ended up w/ an undervolted 3GHz setup w/ stock cooling. Pretty nice, but it sounds like I could have gone further had I just used Orthos. I had both cores hovering between 60 and 65 (iirc) running TAT full blast. Oh well...at least I know he's got a good stable setup.

[Maverick0984]

Originally Posted by nikhsub1

First, no program can read surface temps (known as Tcase). The only way to get a Tcase value is to GROOVE the IHS and put a thermocouple (thermometer) in the groove and read from it. There is no sensor by default on the top of the IHS. Second, what are you using to generate load temps? If you are use TAT to generate load please be advised that NOTHING ON EARTH can get a C2D chip hotter. TAT is what is known internally as a 'power virus' and will put the chip some 20% OVER Intel's rated TDP. You will never encounter a load like TAT out in the real world. TAT is what I use for WB test however as I need the high heat load. FWIW TAT will get your C2D about 15C hotter than Orthos.

Ya, I noticed that. TAT got my CPU a good 15C over Orthos.

Well, I'm thinking one of my reasons for a higher temp is the MSI GeForce 8800 GTX HD-OC just a few inches from the CPU, probably increasing ambient temps in the area quite a bit.

[Maverick0984]

Alright, I took some measurements. Mind you I tried my best to "average" the temps I copied down, but they obviously fluctuated on each side of the numbers below. I should mention again, that I do have a 8800 GTX which does tend to get pretty warm which probably doesn't help my CPU cooling. It's a Lian Li case with 3 120mm fans as well, so I would think it is doing a pretty darn good job for air cooling. While doing these tests, my bios posted temps of anywhere from 29C-32C for the CPU, and a 35C System Temp. In windows, my numbers were considerably higher, and it is kind of dissapointing.

edit: Sorry for the delay, I typed up some code and threw it on my server

http://brandonmovielist.dyndns.org/specs.htm

That contains what I tested at and my stock settings, and current settings. Any suggestions?

[Maverick0984]

since yesterday when those temps were taken, my idle temp at current settings has dropped to 42C, perhaps the AS5 settling in, even though alot more time will be needed.

Would it be worth purchasing the Thermalright Ultra 120 Extreme ? I have a Zalman CNPS9500 in there right now.