How can the TYPE of Mobo effect CPU Temps?? Why are KT333 mobos temps so HIGH?

[Se7en01]

OK, I have tested 4 different mobos in the past week. They are as follows, MSI KT3 Ultra, Soyo Dragon Lite, Gigabyte 7VRXP, and Epox, 8K3A+. I am using ALL of the same components, (CPU, HS/Fan, Memory, OS, etc.) on all to test speed and stability. The thing that has puzzled me is this:

On the MSI board, the CPU temps stayed around 46-48c, However on the other 3 boards, the CPU temps are ranging from 49-59c depending on the load. Now, I am no computer genious, but isn't that kinda high? The Epox board even hit 65c at one time. I am using a Volcano 6cu+ HS fan and have tried the coolmaster that came with the Retail CPU but the retail HS/Fan was even worse. BTW, the CPU is an 1800xp.

I remember not too long ago when I was running KT266A and before that KT133A chipsets and my temps never exceeded 40c.

Is this something that is wrong with the KT333 chipset or does this mean I will have to start using a better HS/Fan as the chipsets get faster?

Anyone else notice this??

And YES, I do have the HS/Fan mounted correctly and am using Artic Silver II. I have mounted the HS/Fan the same way I ALWAYS have in the past on all the boards.

All of this is done W/O any overclocking or change to the voltages.

 

[Buz2b]

Thank you for your post. This poses some interesting questions and I LOVE a good debate. J/K Off the top of my nearly bald head I think you would need to look at the performance of each chipset.

does this mean I will have to start using a better HS/Fan as the chipsets get faster?

As we all have come to accept, higher performance usually equals higher temps. No one said it was ONLY the cpu that got hot! Second, look at the BIOS settings. I am betting that you made sure that the BIOS settings were very similar but what is apples to one board will be oranges to another. Different hardware can react in a different manner on (I'm sorry) different MB's also. The days of "it all works the same" are over. What one MB can accept, causes another to crash. Sometimes it really suxx. My only advice would be to make damned sure your "variables" are covered; HSF, thermal paste, RAM amount and installation, video card, system fans, and on and on and on, etc. If you still have some higher temps, deal with them as best you (and some help from your fellow AT'ers) can do.
On the other side of the coin, you could always start a research project on "chipsets and thier relative effects on system and cpu temperatures" and maybe sell the article/research to a "popular" website or magazine.

 

[mechBgon]

Different calibration and/or different approaches to reading the CPU temperature. I have a couple of boards that read about 10C apart for the same CPU/heatsink/fan combo, and it's just the way they're calibrated.

The EPoX 8K3A+ is the real deal, it reads the actual on-core temperature sensor, and that's probably why it gives realistic readings in the 60C area, instead of artificially-low ones like you've been accustomed to. If the actual on-core temperature exceeds 75C, then you're heading for trouble.

Think of it this way. You preheat your oven to 350F. You get four oven thermometers and put one inside, one on the top, stick one to the door and put another one on top after bending the needle to try to make it read right. They'll give you four different readings, but the oven is at 350F no matter what the four thermometers say. The EPoX board is the first one to put the thermometer inside the oven, so to speak. I want one.

 

[Buz2b]

Better yet..... What he (mechBgon) said, along with what I said and you have your answer; in one form or another. Not a "cop out" either; there are no "standards" that demand what a MB builder must do and not do. Or, as is important to this thread, what they must report or not report. "Apples to Oranges". Ain't high performace fun!?!?

 

[Se7en01]

Wow, great replies guys, thanks. What you both said makes perfoect since and I understand what your saying. Higher performance must equal higher temps. I'm just wondering why the HS/Fan that actually CAME with the CPU in the retail package was so insufficient. (or not, it was cooling @ 60c.) The volcano kept it 6-8 degrees cooler. If this IS the case and Higher performing mobos are gonna cause higher CPU temps, then why hasn't this been more of an issue on forums and reviews? I noticed it b/c I am an experienced builder/tester. However, what about the guy that doesn't really notice it. It seems to me that if the newer mobos are going to cause the CPUs to perform hotter, then there would be more info to the consumer regarding this issue. On the other hand, like "Mechbgon" said, maybe the temps I am seeing now are the TRUE temps that the CPU has been running at all along, even dating back to the KT133 boards. I haven't checked AMD's site yet, (I will in a few) but is there anywhere on the site that tells what a SAFE operating range is? I don't want to burn up my CPU just because I upgraded my mobo and the older HS/Fan that AMD not only recommends but provides with thier CPU's is not sufficient.

When I first noticed the higher temps when testing the Soyo board I called Soyo "tech support" and asked about this and the tech told me that it was ok as long as it stayed below 55c. ???? Hmmm, well, using the Stock "Coolmaaster" HS/Fan that came with the CPU on the Epox and Soyo board, I couldn't get it below 60c. Using the Volcano it is staying between 46-54c.

With that being said, and tested. If 55c and below is desirable, then why hasn't AMD, and all the mobo manufacturers delt with the issue of needing MORE cooling for the CPU than what we have been accustomed too?

I understand that AMD ships a CPU/HS/Fan in the retail package not knowing what mobo it will go on, and I realize that the 1800xp came out BEFORE the KT333 boards, however, it seems to me that they AND the mobo manufacturers would recommend better cooling and address the issue by providing more info on the subject.

I am also surprised that in all the reviews I have read on These KT333 boards, none of them mention higher CPU temps. I know I am not the only one who has noticed this as I have talked to many people experiencing the same thing. It is an issue, so why don't the people who write these reviews address it?

Anyway, bottom line is this. I don't want to buy a NEW CPU right now so I would love to know if there temps I am getting are OK or if I need to buy a Bigger and Better HS?Fan. And if the answer is the latter, then why haven't the people who make the hardware let me know this??

 

[mechBgon]

I believe AMD recommends a maximum actual core temperature no higher than 90C, and I would try to keep it below an actual core temperature of 70C to 75C, as I said. I don't think KT333 has anything to do with it at all, since it is derived from KT266A, which was on the market when the AthlonXP was introduced if I recall correctly.

You want to know if 55C is a safe operating temperature for an XP1800+. Because boards are calibrated differently and you didn't say which board this reading is from, I'm going to generalize and say it's fine. Are you having any problems that would seem to indicate otherwise?

I guess I'll add a comparison in case it helps you out. At work we have a new KT333-based system with an Asus A7V333-R and an AthlonXP 1900+ using the retail heatsink/fan unit. If the CPU temperature drops below a socket-thermistor reading of 60C, the motherboard will actually drop the CPU fan speed in order to hold 60C readings. I'm too tired to give a long explaination of Asus' temperature-reading schemes and stuff, but suffice it to say that the system runs great. If you want to delve into a detailed explaination of what makes an A7V333-R tick, read here: thread with links to my experiments on my own personal A7V333-R. Good night, hope that helped you out some

 

[Buz2b]

Don't fixate on one line that suggested "higher performance = higher temps". That can be true but you need to look into all areas, like BIOS settings, hardware involved, MB used, etc to get a better overall understanding as to why your temps are a tad higher. Actually the on the XP line of processors they brought down the voltage a notch to help combat the temps (well, and the new core didn't require the higher voltage). In theory, that could translate to a slightly lower temperature. However, with chipsets and RAM getting higher performance, this is offset by their slightly increased temps.
As to the "optimal" operating temperature for an AMD cpu, again it is going to vary. The chip itself could be slightly different from one batch to another (although that shouldn't be much of a vairable), the MB could be a variable and even the installation could cause a difference (not enough thermal paste, wrong type, HS not centered, etc, etc). 55 C is not a terrible thing, although I like to keep mine below 50 C. Getting above 60 C would make me a bit nervous. However, I'm sorry but this will depend on your individual hardware. If your system gets unstable at anything above 60 C, then that is your "optimal" level. Mine will continue to function into the high 60's (found out once by the previously mentioned improper HSF installation ) The bottom line is what mechBgon says, AMD doesn't have a recommended "normal" operating zone that they recommend; only the Failure point of 90C. I can assure you that your cpu would cease to operate with any stability before that though.
By the way, IMHO you can get a better HSF for the same or less than you are paying for the volcano. The AX-7 is on sale on one site for about $23.00. If you want lower temps you might start there. You didn't specifically mention but were the cases all the exact same model for these builds? Same location and speed case fans?

 

[Mikewarrior2]

the other thing to consider is your "system temps" on the various systems you tested out.

System temps have a direct corellation to CPU temps. So a 35C case temp situation can't be compared to a 20C case temp situation, or any degree in between, etc.

Also, this is the negative about being brainswashed by socket-thermistor's and their "artificially" low readings. Everyone gave Epox heat for the supposed "wrong" and "too high" readings, just like htey did to Asus. Well, guess it turns out everyone else was just way too low. No one got mad at the artificially low reaidngs though .


Mike

 

[Se7en01]

Hmmm, even more to think about now. I will, however, emphisize that in all my tests, EVERYTHING was the same except the mobo. I used the same case, fans, cpu, video, sound, nic, bios settings, same desk in the same room, etc. etc.. The only thing different was the mobo. And, with the exception of the MSI board, all the KT333 boards had higher temps then any KT266A or KT133 board in the past. Maybe it doesn't have anything to do with the KT333 chipset, but somethng is causing higher temps on KT333 boards. Either that, or the temp readings are more TRUE then they were before. Anyway, Like Buz, I like to keep my temps below 50c and I understand that as long as the system is stable it should be OK. However, I wonder what these higher temps are doing to the LONGIVITY of my CPU. Sure it might run stable at 60c, but how much is running at that temp shortening the life cycle of the CPU?

I'm gonna do a lot more research into this. And Buz thanks for the recommendation on the HS/Fan. I might give it a try. However, it still kinda rubs me the wrong way to think that:

If the newer mobos are going to cause higher CPU temps and IF these higher temps cause either instability OR shorter CPU life, then that means I will need to buy a newer more advanced HS/Fan everytime I upgrade my mobo. And even though AMD has their 90c failure benchmark, it would be nice to know (If possible) what the best temp range is for (Longest life).

I have never been much of an overclocker, however, am I right in assuming that the reason why AMD will not support overclocking and why they supposedly will not honor a warranty if a CPU has been overclocked is due to the fact that overclocking raises the CPU temp and IN TURN shortens the life of the CPU. I mean, that is why there are SO MANY after-market HS/Fan combos out there.

Wait a minute!!! I think I just stumbled across a conspiracy...... Check this:

If AMD or Intel WERE to give an optimal Temp range for stability and longivity, then they might put all the HS/Fan people out of business right? Lets say that a safe rage is 30c-80c. Well, even the stock retail HS/Fan will handle that. However, if somewhere along the line SOMEONE made us believe that anything over 50c is reaching unsafe, WELL, then, HS/Fan business just became lucrative!

Well, its just a thought. I know I am probably making too much a big deal out of this. On the other hand, It still kinda irks me that there isnt more of a standard or benchmark for knowing whats safe and optimal for a CPU that I MIGHT and probably WILL use on several diferent mobos.

IMHO there is WAY TOO MUCH "opinions" in this community, (Computers in general, not just anands). What I mean is this.

You ask five people what a safe range for overclocking a Video card is and you get 5 different answers. You ask 6 people if you should install driver "A" before driver "B" and you get 3 differrent answers. Not that I don't value people opinions which are usually based on thier own experiences and training. However, I get tired of all the "UnKnowns" and so-called "PROS" who consisitently disagree on so many issues and the people who are less experienced and less trained not only get confused but end up damaging their hardware.

There needs to be more "Absolutes" and "Standards" in the industry. More info to the consumer.

The whole thing boils down to competition and $$$$$. Thats the sad state of our economy. Oh no, I'm off on a tangent now. Sorry.

I guess the bottom line is: I just want the hardware companies to be MORE responsible for providing ME the consumer information about what works and doesnt work. What I need for part "A" to make sure it works with part "B". And if I use part "A", will it hurt part "C".

All you have to do is look at all of the "incompatibility" issueswe have seen over the years to see that these hardware companies do not CARE about US the consumer, all they care about is being the first to produce the new "HARDWARE" regardless of whether it works witht he OLD hardware.

I think it is about time that the hardware companies do a better job of ensuring the consumer is more informed. Hell, they would be helping themselves in the long run, as the RMA's would go down dramatically.

 

[Mikewarrior2]

Se7en,

That's why I've always said there is no general safe-temp "range" for socket-a motherboards, rather jsut a set of "reasonable" range of temps per motherboard. Every socket-a motherboard reads temps differenlty, as to the reasons MechBGon posted above. Different scoket-thermistor location, even a 1/4inch off, changes the readings significnatly.

the problem with socket-thermistors again are the variability in motherboard readings, and the gross inaccuracy. How is AMD able to provide a set "max temp" when that max temp is 60C with motherboard a and 40C with motherboard B?

Also, this is why Intel has on-board thermal protection (something AMD should have). Without the need to depend on a motherboard or a motherboard's "temp reading problems", the CPU has the ability to protect itself from destruction.

Amd relies on the motherboard manufacturer for this, and that is plain wrong in my book. YOu can never depend on a manufacturer .

There is no HSF industry conspiracy. If anything, socket-thermistor's hurt hte HSF industry because the readings are compressed. Now that we are seeing the 8k3a temp readings, and seeing spreads of 30C between good water cooling and bad Air cooling, the advantages of higher priced HSF units is much more readily seen.


Mike

 

[mechBgon]

Either that, or the temp readings are more TRUE then they were before.

Bingo. Or perhaps I should qualify that and say, some of them are more true, while other companies are still spoon-feeding us artificially-low readings.

 

[Se7en01]

Originally posted by: Mikewarrior2
Se7en,

That's why I've always said there is no general safe-temp "range" for socket-a motherboards, rather jsut a set of "reasonable" range of temps per motherboard. Every socket-a motherboard reads temps differenlty,

Mike,

OK, I agree with you and I do wish there were a more consisitent way of measuring temps on socket A. However, I still have to get back to my original concern and maybe you can help me to make a choice or at least feel better about my current situation. Here is the deal:

*Note: System specs were the SAME on all builds. HS/Fan was Thelmaltake Volcano 6u + and I installed and re-installed the HS/Fan 3 times on each with Artic Silver to make sure of constitent install.

With all of my KT266/A boards temps ranged 35-40c. Back in my KT133A days temps usually ranged in the 30-40c range.

Now, with these KT333 boards I am seeing 50-60c and EVERYTHING is the same as with the KT266A.

So, my questions are:

1. Should I just accept this as normal and within range and forget it?

2. If so, WHY are the temps higher? (or are they just more accurate than b4 like MechBgon suggested?)

3. Is 50-60c going to be OK in the long run (Longivity) or should I go out and buy a Better HS/Fan even though the one I am using is on the AMD "recommended" list for my CPU.

4. What temps do "equivilently clocked" P4 CPUs run at? And is there some kind of standard SAFE operating range for them?

 

[Vette73]

This is the reason that board makers need to use the temp. sensor already built into the Athlon XP's. Its already there they just need to program the board to look at the on-chip sensor and not the one under the CPU.

The P4 temps should all be about the same as I think all the currrent boards use the on-chip sensor. I don't know why they don't use AMD's. It should lower the cost of the board and add the ability to turn of the computer if it gets to hot.

 

[Buz2b]

So, my questions are:
1. Should I just accept this as normal and within range and forget it?

2. If so, WHY are the temps higher? (or are they just more accurate than b4 like MechBgon suggested?)

3. Is 50-60c going to be OK in the long run (Longivity) or should I go out and buy a Better HS/Fan even though the one I am using is on the AMD "recommended" list for my CPU.

4. What temps do "equivilently clocked" P4 CPUs run at? And is there some kind of standard SAFE operating range for them?

1) As long as you CYA and all things are covered that we've discussed, you could indeed.
2) Already been answered to the best of our combined abilities.
3) 50-60 C should be fine. The question becomes stability. At the point of instability, you will know the CPU is in distress. Running a chip at 50-60 C vs 40-50 may shorten the life of the chip but I doubt it will be a significant amount; if at all. Speculating here, I could say that all things being equal, you could theorize that you might shorten the life of the CPU by 10-25%. So what?? How long would a CPU last if EVERYTHING were PERFECT?? No one knows but you could again speculate that it might last 5-10 years at minimum. So, even if you knocked 25% off of that, who is going to keep a computer longer than 4-7 years?? Personally, I would try to lower it to the 40-50 C range as I personally (sorry, but that DOES come into play sometimes) would not want to run a CPU an additional 10 C above the "accepted" norms for good cooling. Call me "anal"; many do. :Q
4) "Equivilently clocked" P4 CPU's (in case you haven't bothered to look) run significantly lower temps. In addition they have thermal protection built in that protects the CPU from overheating damage. They don't need a "Safe operating temp" built in. They work until they get too hot, then they throttle down and eventually shut down. Is it a perfect system? Probably not. But until MB builders decide to use the new AMD system of protection, Intel's is better than none.

I think we have dragged this out way beyond it's meaningful life. I'm not sure you are going to get any better answers than you already have. But, who knows? I think you need to make a choice at this point; decide that there is no "standard" with the AMD MB/CPU combos for temps and learn to deal with it in as professional manner as you can, OR, you can start building Intel systems that have lower temps and somewhat better support for their CPU's--like the Intel versions. I guess another option would be to build systems with the "lower of the class" of AMD system also.
Don't get me wrong, I agree that it would be a heck of a lot better if the companies like AMD, ASUS, MSI Epox, etc, etc would get together and come up with some "standards", but I think you are "wishing on a star" for that. That is like asking MS and Apple to standardize their file systems and software compatibility. Ain't gonna happen; no way, no how! Too much of the almighty dollar at stake there. Money talks, BS walks is the unfortunate way that business has to be done today. Hopefully, all your questions (whether in my post or anothers) have been answered well enough.

 

[Se7en01]

Buzz,

Thanks again for the input! Sorry for dragging it out like I did but it's in my personality to milk things to death. Anyway, I agree with all you said and although I am not satisfied with the higher temps, I guess I will just accept it and hope that AMD improves on things in the past.

Side NOTE*** I did talk to AMD tech support today and the tech I talked to told me that the TEmp sensor that is "INSIDE" the AMD CPU is NOT a temperature indicator, it is just a fail safe that will alert you if the 90c ceiling is met. It will not and cannot give you accurate CPU temps even if the mobos could read it. It is there simply to tell you if your CPU is too hot.

BTW Buzz, I love your final quote in your sig. I am speachless at the though of our high courts changing something so important to our country's history, not to mention our source of strength. I am a teacher and it will be hard to NOT say the pledge b4 class each day. Hell, I might do it anyway. What will the do, put me in jail?

 

[Mikewarrior2]

Se7en01,

That tech support person was wrong . The Internal diode (tested at Procooling.com, Voidyouwarranty.net) of the XP/MP chip is perfectly function as a temp reader. Looks like this guy didnt' know what he was talking about (gasp, the horror and shock, a tech support person that doesnt' know their stuff? ).



Mike

P.S. ProCooling link from Phaestus. An excellent read. Modified asus mb reading temps from the int. diode.

P.S.2. The internal diode is there for a reason. To artificially make it unreadable is just plain dumb. Not using it further exacerbates the problem of motherboards artificially reading too low.

 

[Se7en01]

Ya, its pretty sad that the people getting paid to give advice by a company that large would give false info.

Mike, Just out of curiousity, what HS/Fan are you using on your CPU and what kinda temps do you get?

 

[Mikewarrior2]

On my p4, relatively about 50/32 cpu/case temps right now, retail hsf, AS3.

my athlon xp(8kha+ socket-thermistor), 48/32 cpu/case. Figure the 8kha+ is probably 8-12C below actual die temp when under load .



Mike