52.16 heat issue

[Beep99]

These new drivers-Forceware-when installed on my system report as much as a 15 degree Celcius increase IDLE then the older 45.23's did.

(it only shows up on FX cards in the driver settings list under or so I'm told by others)

My question is this......has it been solved as to why there is such a high discrepancy with a simple change from one set to the next? Is this a bug in the new drivers or were the older ones the ones that reported the wrong temps?

No one seems to know anyplace I surf to find out.

Thanks!!! :whistle: :x :ninja:

[DirtyLarry Veteran]

okay first off i want to say this is not a defintive answer... just trying to "troubleshoot" per say and i am curious as to what others have to think about it...

according to my info in my drivers settings my gpu is 59 degrees...

when i check the temperature of my cpu in my BIOS it is 51 degrees...

so, as a result of my cpu's temperature in my BIOS, currently my gpu is 8 degrees hotter than my cpu. if indeed these drivers are wrong and show a 15 degree higher temperature for gpu's, than my gpu would then in theory be 7 degrees cooler than my cpu and would be in reality 44 degrees.

so i think the question is this? what generally has the higer temperature? the cpu or the gpu.

my hunch is the cpu, but i defintely could be wrong, as i dont overclock so i never pay much attention to temperature at all.

so if the cpu is indeed always hotter than the gpu, this leads me to believe the drivers have a bug.

and if the gpu is normally hotter than the cpu, these leads me to believe the drivers are correct.

anyone know which is usually hotter in temperature the gpu or the cpu?

i am thinking that experienced overclockers could answer this question easily... and as a result we can come to our own well grounded conclusion...

[DirtyLarry Veteran]

okay so nobody on neowin knows if a graphics card is a higher temperature than a cpu normally? sorry if this is such a nOOb question but i am curious... thanks to anyone that answers...

[Beep99]

they cant larry ol chum.....they're all too busy concocting up thier next nvidia-bashing .

Seriously.....you do raise an interesting scenario.......I can only just say this.....while overclocking my last graphics card....a 32 meg GeForce 2 ....I was able to get it to max out and depending on room temp it would go as high as 55 -- 56 degrees C before it would crash.I use to use SmartDoctor from Asus that was made to monitor the card plus overclock it.

Now with this newer FX card I have...the Smartdoctor is a sham because it only overclocks the card/gpu so much and is limited in that regard.....plus.......at the same time....and I might add I am still waiting for Asus to do..is make an upgrade the latest version that doesnt do one thing that the older Smartdoctor did and that is monitor gpu and card temps.......What puzzles me is why cant they?.....what's the deal here?........I use Nvidia drivers on the card and Lo and Behold it has ""Temperature Setting"" on it.......so if a graphics driver maker can have it on thier drivers....why cant the maker on the card have it on thier monitoring proggy they've had all along?........I really think we are not being told the whole truth here on this matter and this latest discrepency temps we see on a simple switch of graphic drivers is an indication of it I should think.

[afecelis]

I've had the same problem with the latest drivers so I reverted to 45.23 but now I'm getting the "Nvidia system sentinel" warning and my gpu is slowering its power down. This started happening since I installed the 52.16...damn buggy drivers! And they also increased my temperature in 10-15 degrees celsius. Ah! one more thing, the cpu's temperature is supposed to be higher than the gpu's.

here's a pic of the warning:

http://www.theshower.nl/cgi-bin/genesis/we...is/sentinel.gif

[DirtyLarry Veteran]

they cant larry ol chum.....they're all too busy concocting up thier next nvidia-bashing .

Seriously.....you do raise an interesting scenario.......I can only just say this.....while overclocking my last graphics card....a 32 meg GeForce 2 ....I was able to get it to max out and depending on room temp it would go as high as 55 -- 56 degrees C before it would crash.I use to use SmartDoctor from Asus that was made to monitor the card plus overclock it.

Now with this newer FX card I have...the Smartdoctor is a sham because it only overclocks the card/gpu so much and is limited in that regard.....plus.......at the same time....and I might add I am still waiting for Asus to do..is make an upgrade the latest version that doesnt do one thing that the older Smartdoctor did and that is monitor gpu and card temps.......What puzzles me is why cant they?.....what's the deal here?........I use Nvidia drivers on the card and Lo and Behold it has ""Temperature Setting"" on it.......so if a graphics driver maker can have it on thier drivers....why cant the maker on the card have it on thier monitoring proggy they've had all along?........I really think we are not being told the whole truth here on this matter and this latest discrepency temps we see on a simple switch of graphic drivers is an indication of it I should think.

hmmm also very interesting points you are bringing up as well..i just dont know really.... as far as your reference to your smart doctor i think i might have a solution for you for monitoring everything else... nvidia just released on monday an app called NVIDIA system utility for any motherboards that use nforce2 or nforce3, which im not sure if you do or not.

heres the link...

http://www.nvidia.com/object/sysutility.html

maybe this sytem utility thing will work for you, as here is a brief section of the write up...

NVIDIA? System Utility?is the ultimate utility for accessing, monitoring, and adjusting your system. Whether you want to monitor temperature or voltages, overclock your system for ultimate performance, or underclock for quiet operation, the System Utility enables total system control with user-friendly control panels.? With the profile manager you can setup and save individual profiles such as an ultra-fast profile for gaming, a quiet profile for watching DVDs, or a standard profile for office productivity.

However it only fully supports some motherboards right now, and I installed it on my system although my motherboard was not fully supported and I did not see temeprature settings anywhere, maybe becasue my motherboard was not fully supported....

Well i guess i say good luck to all of us. Although these drivers have been working well for me ona whole this one little temperature thing does not sit well with me at all, and although I have never had a problem with my card to date nvidia can be doing way, way better in my book...

[Beep99]

well......it's been a while still I started this post....and NO PLACE yet have I heard of anyone revealing or at least finding any more info on the "Temperture Settings" degree discrepency that a simple changeover from the previous 45.23 Det's to the new Forceware 52.16's gives in reporting the temperature state of your GPU.

As a side note.....I also have installed the latest 3.13 Nvidia system drivers and so far they run flawless but then again as you install them on your system it also installs the 52.16's all over again......at least it did here for me.It made no mention of changing video drivers and I've never seen where installing system drivers install new video drivers at the same time as well.System drivers are ...well....system drivers....right? First time for me that I've ever seen where installing system drivers also installs video drivers at the same time too.

I've rolled back to the older 45.23's since then and all is fine.

Just makes you wonder .

[sandman45654]

Thanks for the post Beep99. I was wondering what was dragging my fresh install of WinS2k3 down. I can't confirn the temp problem (no sensor) but there is definitely something screwy about the 52.16 drivers. Installed the 45.23 and everything is back to normal.

[arcaine25]

Here is a reply I got directly from Nvidia about the "heat issue" that I was having on my FX 5900 Ultra.

"Thanks for the feedback, but it's actually not a bug or problem. The newer drivers did not change the clock speeds at all -- they merely fixed a long-standing bug in how they used various GPU features. In some scenarios it might lead to increased GPU temperature due to the change in workload, but the temperatures still are well within the the operating tolerances NVIDIA designed for the GPU product."

As you see it is not a problem, but it is actually due to the fact that they "unlocked" the GPU's potential.

I never had any quirky or screwy problems after installing the WHQL drivers, except I though the heat thing was bad, but according to Nvidia, it is not bad, and within range.

I am satisfied with Nvidias answer, and I trust what they are saying. I have always trusted Nvidia products from the begining and I don't plan on stopping now :-)

[Premgenius]

NVIDIA rocks but how do you mesure the GPU temp cos i have a P3 with not sesors on the mobo and a NVIDIA GeForce4 TI 4600?

[sinatosk]

makes me wonder... if this is what caused the death of my GeForce 4 TI4200

[DirtyLarry Veteran]

man that would suck if thats what killed your ti4200, but from aracaine25's response i get the feeling maybe this is all legit with the temperature, makes sense to me if its doing more with the new drivers it would therefore be a higher temperature... wow i thought this thread was dead and done about a week ago... still it sucks sintosk about your card sorry to hear it...

[Beep99]

Well as far as I am concerned ...that answer that Nvidia gave isnt acceptable at all.Especially in the case where either the temps it was giving before were accurate or they werent...it's either one or the other.So according to Nvidia now with this Forceware set the temps it registers are well within the range....well just what is the RANGE????? A difference of temp settings that change from 42 up to 57 with nothing more than a simple change in a driver set isnt what I would call acceptable.Either the old drivers were inaccurate all along or the newer ones are totally inaccurate from the word GO.

Now which is which?

I own and use an ASUS 5900 FX card and currently ASUS will not or has not provided an update to thier SMARTDOCTOR series of video card monitor utility because they claim there is some sort of change to the architecture within the newer FX card that doesnt seem to work properly and given this latest answer that we see a few posts up that was so kindly provided by someone who has the time to pass it along to all of us who may not have albeit real life getting in the way....smile.....I'd say that we have a problem here that somehow is getting swept under the rug by not only the card manufacturer and developer but 3rd party distributor as well.

Until Nvidia has the decency to announce publicly that the temps on the newer cards are wrong.....or the temps given to older card users by those that used thier older reference driver sets were wrong....then we do indeed have a problem and everyone who basically uses thier cards has one too because although they claim by overclocking thier products voids the warranty.....don't they void thier own warranty by being dishonest to provide the false means to achieve them to begin with?

ALL I WANT TO KNOW IS THIS......WHY DID MY BRAND NEW ASUS FX 5900 CARD I HAVE HAD FOR just over 2 MONTHS suddenly BEGIN TO REGISTER A 15 DEGREE CELCIUS INCREASE BY SIMPLY UPGRADING FROM AN NVIDIA REFERENCE 45.32 driver issue.to A NVIDIA FORCEWARE 52.16 driver set?

thats all.

thanks guys

[arcaine25]

To reply to Beep99 without sounding like a major dick, myabe I should repost this message that came DRIECTLY from Nvidia, as you will notice what they say, it says that the new drivers FIXED a long standing issue on how the GPU was used, what I gather from this is the fact that quite possibly the GPU was not being used to it's full advantage.

I have had 2 FX 5900 cards, one was a Jaton FX 5900 standard (I recommend against this card period, as the VGA core clock was only 300 mhz!)

And my recent purchase which is an ASUS V9950 Ultra. both cards had higher temps when installing the 52.16 drivers.

It is not a temp reading issue VIA the drivers, it is merely the fact that the new drivers have fully unlocked the potential of the FX chip.

here is the Nvidia post again, this was emailed DIRECT from Nvidia to me when I emailed them about what I thought at the time was a problem.

"Thanks for the feedback, but it's actually not a bug or problem. The newer drivers did not change the clock speeds at all -- they merely fixed a long-standing bug in how they used various GPU features. In some scenarios it might lead to increased GPU temperature due to the change in workload, but the temperatures still are well within the the operating tolerances NVIDIA designed for the GPU product"

I hope this clears things up a bit...

[arcaine25]

One other thing that Nvidia says about the temps.

It is well within spec even if your GPU reached 100 Degrees celsius (although that is hot, it is still ok by ther standards)

The GPU core slowdown threshold is at 120 degrees for the FX5900, so it will operate normally until it reaches that temperature, then it slows the core clock down to protect the chip.

My ASUS has not gotten over the high 80's when playing games, and according to Nvidia, it is within spec.

[Beep99]

good lord...that high and that's ok?

interesting....ok...thanks arcaine...youre kind reply is muchly appreciated here in beepland....