With the help of some of the posts about lm_sensors and gkrellm on this forum, I just got my sensors working and displayed in gkrellm. Now I need to figure out what I am looking at.

Gkrellm indicates that temp1 is 41.0C and temp2 is 59.5C, with occasional spikes to 60.0C. What do temp1 and temp2 measure, and what are the optimal values I am looking for? I am running an AMD Athlon XP 2600+ at its normal speed - no overclocking. I am also running the distributed.net client, so my CPU load stays at about 99 - 100%.

It sounds like temp1 is your motherboard temp and temp2 is your CPU.

My 2.8 GHz P4 (Northwood/Socket 478/Hyperthreading) peaks at around 50C, with its maximum rated temperature at 75C. Under full load, my motherboard is around 38C, which is comparable to your temperature.

Okay, sounds like I might want to add another fan, but I can hold off on the water cooling.

I need to check into what the maximum rated temperature for my CPU is.

I got concerned about CPU temperature in the last couple of days, because it has been so hot here. My desktop rebooted a couple of times, and I thought it was because of overheating. Looks like it was actually atributable to a powerline ethernet adapter going bad and sending spikes through the outlet I have my machine plugged in to.

hussar

those are not good temperatures for that processor.

I have an amd 2500 overclocked to 3200+ speeds, and my front side bus is 400 and not 333 like your's and I'm in the 35 deg celsius range.

That's just one of many AMD's that I push to the limits.

Your numbers frighten me to be honest.

If you cpu load is really going to be 99.9% I reccommend the following on amd procs

1. very important: power supply fan minimal 2000 rpm 80" diam in top of tower, at the outlet of the case. none of these push pull designs. psu should also have side intake preferably , and not a bottom intake.

2. arctic silver 10% comound on the cpu, remember more is not better. clean the die to shiny before using.

3. you need a good heat sink. dont just run out and get any old heat sink. and dont just buy a thermaltake or a mad dog you see in a computer store. i recommend you go to toms hardware guide and get a little eduction there. they have benchmarks there for you to compare and overcome many of these manufacturer's claims.

4. amd actually wants a fan at the back of the case nearest to the HSF (heat sink fan) and I agree this truly does pull the air up to the proc area, which in turn is near the power supply.

5. about a 120 mm diam fan at 3000 rpm in bottom of the case is usually fine.

The above is good for a typical home use pc with 1 hard drive and a 300 watt psu I find.

AMD's always run hotter than pentiums as a rule of thumb, they use a higher vcore as a generalization

If you have more hard drives look into hdparm to put them to sleep if need be. Or hard drive coolers that cycle on when the hard drive surface temperatures get to 105 deg F.

The biggies in general is get the heat out of the PSU (power supply) is a very big factor.

Hard drives run hotter over time and are a big factor to heat.

If your not overclocked, and if your memory is about 80 deg F or so, then your northbridge and southbridge are probably fine.

Another thing you may want to do is check in your bios that your not above 1.7 volts at all.

I find these amd's run fine 1.60 without being overclocked. Your mileage may vary, and you may actually need a higher vcore as your PSU may be crappy and there is no way for me to tell without being there

Many times for whatever reason on gigabyte mobo's I've seen AMD's get loaded by the bios with as much as 1.8 volts

Those high volt temps can lead to that really bad/high "die" temperature your reporting to us.

Hope this helps.

I'm a PC repair tech by trade and I own my own PC business,...

Your temps are a little high but nothing to worry about

Its prob because of your recent higher ambient temps... It couldn't hurt to add another fan to the case. For 99% load I really gotta say thats not bad.

also it depends on your bios type and Rev. as to what the temp sensors will acctually display...

Just for comparison sake, I've seen (on more than one occasion) customers factory heatsinks get so clogged with dust and pet fur that the fan can't get any air to the heatsink !!! even then the temps I've seen on these machines have been upwards of 90C before the PC starts acting real screwed-up...

Hope this helps

I have an AMD 2400 overclocked to a 2.2Ghz(not a whole lot but hey...)
CPU is 37C right now and is normaly between 35-40C
MB is 32C right now and is normaly between 30-35C
Im using heatpipe cooling and its the coolest and quietest of any other cooling ive used. Had one fan that sounded like a dustbuster and still didnt cool as good as the heatpipe does.

Hussar, just thought I'd give you a little more info on your processor. It's a model 6 processor. The max temp rating for these are 85 Deg C and most likely your motherboard honors the "loosely" defined specification that AMD issued.

For the most part your pc will probably shutdown at 75 deg C and there is most likely nothing that can be done for that. There are barely any manufacturer's (mainly DFI only really on a large scale) making boards where you can even adjust this 'high limit'.

Your spec's can be on page 35 of 106found at this link from AMD's web-site (it is in english):

http://www.amd.com/us-en/assets/cont...docs/24309.pdf

You will also note the vcore of 1.65 on that data sheet as I pointed out, be sure to check out your mobo on that.

All I can tell you Hussar, is that even Intel makes pentium 4's to cut back on cpu cycles starting at 65deg C a noticeable performance drop is only a few points away for you.

These AMD's have a very noticeable performance drop once you hit ~ 120 just go to tomshardware or hard oc for benchmarks.

Tho it is a handy tool, the best way to measure your temperature is not thru gkrellm to be honest, but it is to use the temp readings in the motherboard. In there you can establish what is referred to as your "baseline" reference point. That is, your maximum full load 100% cpu load temperature given your room ambient.

When you are in your bios, your cpu is instructed to be at 100% load. That is on purpose and is a standard to allow technicians to fine tune. You are also supposed to allow for 5+/- deg F on your readings in the mobo as tolerance believe it or not. Then you're supposed to bounce into linux and load up the proc, and set the gkrellm. You will notice in configure there are adjustments to calibrate. That's because it does not know every mobo's setting in the world.

PS...Ya those heatpipes own Anarchy!

Thanks for your responses.

I am using the standard AMD fan and heat sink that came with the CPU. I also have one of those intake fan sets that mounts in a 5 inch bay at the front of the computer. I have it up high so that it collects a minimum of dust bunnies and channels the air almost directly at the CPU, and I vacuum-clean it every now and again. I also have an 80mm fan mounted at the back about level with the CPU to pull air through. My power supply pulls its air in from inside the case and exhausts out the back above the 80mm fan I just mentioned.

What I might try is adding another couple of fans to the case. I think there is still an open mounting area at the front, and I have an 80mm fan here from an old machine. I also have one of those exhaust fans that mounts in an open expansion card slot at the back. What that might do is take some of the heat from the graphics card and the hard drive and pull it out the bottom rather than up past the CPU.

Also check to make sure that the cpu temperature you are getting is not a "calculated" temperature. I have an Abit KV-85 and a Sempron 3400+ and no matter how long the pc is running it's always reading 50C-52C (when it's a cold boot or the pc has been running under load for an hour). This is proven false because I have an Aspire X-QPack with a thermal probe and the heatsink temperature never rises above 34C.

I have quite a few AMD CPUs from the last few years (Durons, Athlons, Semprons) and all their temps are between 33-46 C.

Aside from adding a new fan, maybe you should look into a better HSF for the CPU?

That's probably a good idea, too. A new HSF might be a bit quieter also, and that would be an added benefit. Any recommendations?

If you want a great cooler that also won't break the budget, look at this one:

MASSCOOL 5F394B1L3G Full Copper Heat Sink CPU Cooler

I currently have that one in my Athlon XP server. It's all copper, nice and heavy, and has an 80mm fan. The fan it comes with does the trick, but you could replace it with an even better one (like a Panaflo) if you wanted.

Also, go with some Arctic Silver 5, or Arctic Silver Ceramique. About $5-6 a tube plus shipping, but it really does make a difference.

I've bought the cheaper coolers before from Newegg, and I can say I've always been less than satisfied with them. I got a $30 Thermaltake Volcano cooler on my OC'd socket A duron and it lowered the load temps by ~15 C. Not to mention it comes with two rheostats, and an auto temp controller.

the cpu heat sink i have is the aerocool ht-102 its pretty big(ht stands for hightower) but it is really quiet and cools very well.

Hussar,

Heat sink fans to me are very "personal" and I like to fit the right one for the job. One size surely dont fit all.

Here is a link for updated performance HSF's.

Where you can validate your decision what to do.

http://www.tomshardware.com/2001/05/21/can/index.html

Remember,your maximum thermal wattage output is already shown on the AMD link that I gave you.

On this guide you get the data you need to quickly size up whatever HSF you want and it's noise level. Just size up the one you need, and spend no more, no less, and keep sanity with low noise in your house, office etc.