LOL@Evga response to 8800GT fan issue

S

sluzbenik

Limp Gawd
Joined
Aug 9, 2006
Messages
166
So if you have a default 8800GT with the fan stuck at 29% until it hits 100c, you have nothing to worry about according to JacobF:

"The 8800GT runs at 29% fan speed by default, it will increase when the the card deems necessary, usually this is over 100c.

The card is rated for over this temp, and the fan speed only increases when it is absolutely necessary for the stability of the card.

Increasing the fan speed will definetely (sic) lower the operating temp, however, it should not be necessary for the stability of the card.

Thanks, "
source: http://www.evga.com/forums/tm.asp?m=166784#

So to all of you who haven't changed the fan settings yet and are experiencing lockups, artifacts, and crashes when temps get over 80c, it's clearly not because of temperature. And if you turn up the fan and it stops doing such things, it's just a coincidence.

I'm pretty sure they said the same thing about the ATI 1900 series...Oh don't worry, it's rated for that. Riiiiiiiiiiiiiiiiiight.
 
sluzbenik said:
So if you have a default 8800GT with the fan stuck at 29% until it hits 100c, you have nothing to worry about according to JacobF:

"The 8800GT runs at 29% fan speed by default, it will increase when the the card deems necessary, usually this is over 100c.

The card is rated for over this temp, and the fan speed only increases when it is absolutely necessary for the stability of the card.

Increasing the fan speed will definetely (sic) lower the operating temp, however, it should not be necessary for the stability of the card.

Thanks, "
source: http://www.evga.com/forums/tm.asp?m=166784#

So to all of you who haven't changed the fan settings yet and are experiencing lockups, artifacts, and crashes when temps get over 80c, it's clearly not because of temperature. And if you turn up the fan and it stops doing such things, it's just a coincidence.

I'm pretty sure they said the same thing about the ATI 1900 series...Oh don't worry, it's rated for that. Riiiiiiiiiiiiiiiiiight.
Click to expand...

So it's a feature not a bug. :p Just kidding....

Seriously though, has anyone read or heard anyhing official from Nvidia about this fan speed thing? I have read a ton of posts on this and people have said everything from, it's a driver bug, or it's a bug in the VGA-BIOS, or it's not a bug at all and that's how it's suppose to work as the email you received from EVGA indicates. Personally, I don't mind using Rivatuner to keep the card running cooler since I would be using it anyway to overclock the card.

Edit: I have an EVGA SC by the way, and it's fan ramps-up to 35% way before it hits 100c (without Rivatuner)
 
THats actually pretty true. The 8800GT runs hotter than any of the other video cards mainly because of the single slot cooling solution and low fan speed. I really dont know whether the 29% fan speed is really true but if jaocb says thats what it is i trust him more than some random review site. Videocards are quite literally his life/job. I would just bumpup the fan speed in rivatuner if its getting too hot for you. thats what ive been suggesting to people.
 
sluzbenik said:
So if you have a default 8800GT with the fan stuck at 29% until it hits 100c, you have nothing to worry about according to JacobF:

"The 8800GT runs at 29% fan speed by default, it will increase when the the card deems necessary, usually this is over 100c.

The card is rated for over this temp, and the fan speed only increases when it is absolutely necessary for the stability of the card.

Increasing the fan speed will definetely (sic) lower the operating temp, however, it should not be necessary for the stability of the card.

Thanks, "
source: http://www.evga.com/forums/tm.asp?m=166784#

So to all of you who haven't changed the fan settings yet and are experiencing lockups, artifacts, and crashes when temps get over 80c, it's clearly not because of temperature. And if you turn up the fan and it stops doing such things, it's just a coincidence.

I'm pretty sure they said the same thing about the ATI 1900 series...Oh don't worry, it's rated for that. Riiiiiiiiiiiiiiiiiight.
Click to expand...

No, you're clearly trying to stir shit up.

Everything he said is spot on. The thermal profile on the 8800GT doesn't have the fan ramping up until the GPU temp goes past 100C, and that's completely normal for this card.

If you've got a card that is artifacting with the stock fan configuration then it's DEFECTIVE and you should seek out an RMA. Regardless of wether the artifacting stops when you increase the fan speed, or if it's artifacting all the time - it's DEFECTIVE. It's not a design flaw as a few shit-stirrers would like to make you believe... it's just a bad card and you had the same chance of getting a bad 8800GTS or 8800GTX or any card before it.
 
Blue Falcon said:
If you've got a card that is artifacting with the stock fan configuration then it's DEFECTIVE and you should seek out an RMA. Regardless of wether the artifacting stops when you increase the fan speed, or if it's artifacting all the time - it's DEFECTIVE. It's not a design flaw as a few shit-stirrers would like to make you believe... it's just a bad card and you had the same chance of getting a bad 8800GTS or 8800GTX or any card before it.
Click to expand...
QFT

you shouldn't have to tweak *anything* to run your card at stock settings.
 
i think this is all stupid bullshit and I hate all the bitches whining about the single slot cooler.

runnning sli my cards have never got over 74c with 60% fan with an OC and thats when I play crysis with a HOT ROOM

usually I am in the mid 60's LOAD .... thats cooler than any of the gtx or gts I had ran...
 
Why would the 8800GT's silicon hold up to water-boiling temps but CPU's can't? 100C can't be good for any processor can it?
 
I would like to know how long lifespan Nvidia predicts for these cards :confused:

Logic dictates that the hotter a computer component runs the shorter its life span is. Running a card 100c frequently for long perioids of time does not sound too healthy in my mind.
 
Yeah seriously. Something just tells me running the GPU at 100C is not the smartest thing to do.
 
when i game I just up the fan speed to 55 percent, my card runs at about 55-60c when I do this.
 
100c is absurd, no matter how you look at it.

The processor may well be able to run at that temp, what concerns me is the memory. Overheating memory can frequently lead to small errors that become bigger with time, until you get crashes or artifacts in game...

It also does not happen overnight. You might be able to run with those temps for quite some time before an error occurs, a little glitch here or there, an artifact that disappears.
It's a proven fact heat degrades chips over time.

Anyway, if you read the forums on Guru3d on how to use the auto fan functions, you can set it so it spins up and down according to temperature. I do this and below 55% or so it is very unobtrusive. By setting it so 80c results in 100% fan rpms the fan has never had to go over 62%. In this case I think it's set that way because they know how much noisy fans annoy people, but they could have optimized their formula more to improve the lifespan of their cards.

The thing is these cards have very short development cycles, and the #1 priority is performance, not card longetivity, or noise/cooling optimization.
 
That's what keeps the aftermarket heatsink sales going strong.
Thermalright has made a ton of money, thanks to Nvidia:D
 
No sale

Considering how burning hot the entire card gets when they hit 80C+, I find it a bit hard to
believe all the components are fine running under these conditions.
 
Battleneter2 said:
No sale

Considering how burning hot the entire card gets when they hit 80C+, I find it a bit hard to
believe all the components are fine running under these conditions.
Click to expand...



Maybe not on this card (time will tell,especially for the overclockers),but on most yes the tolerance for heat is very,very high.
 
Note the response mentions only stability. Life is not mentioned. Frankly as a EE and former designer I am horrified. The "value added" overclocked cards with stock cooling are a particular concern. I am very pleased the majority of the thread comments indicate good common sense.

My issue would be that the CPU is not the only component on the board. So if the CPU is at 95C and the fan is at 29% speed wth temp are the MosFETs running at ? Yes, MosFETs are typically rated for 120C but we have no way of learning their temp unless you do direct measurements and they still live longer if cooler ) At what temp are the caps cooking at ? What is the temp rating of the caps on the board ? 80C is standard, 105C for very high quality. What is the lifetime/operating temp derating curve for the components on the board. With the low fan speed at what temp does the heatsink start conducting/radiating heat INTO the other components due to thermal saturation?

Unless I had a lifetime warranty on the card, (and even then who wants to go through the hassle of an RMA), I would be running the fan to keep it less than 60C (not gpu core temp but that would be lowered as well) enviromental/ambient under that metal shroud to ensure long life of ALL the components. Heat kills electronics, end of story. I did not realize it was so difficult to ramp fan speed up based on a temp curve, oh wait, its not.

sluzbenik +1

So run it "stock" and return it if it artifacts if you want. I would rather keep my stuff as cool as is reasonable so I do not have to deal with failures caused by design or cost compromises and avoid the hassle.

Just my opinion and it is worth what you paid for it.
 
Maybe not on this card (time will tell,especially for the overclockers),but on most yes the tolerance for heat is very,very high.
Click to expand...

I am not sure what you mean by this. If you are saying the components on the board are in some way more tolerant to heat than normal, I would like to see the documentation. I am considering buying one of these cards and have inspected the high resolution photographs on XS of several areas of the card especially around the PLL clock chip and voltage supply circuitry. I see no indications the board components are anything other than standard commercial quality parts. For instance the IC packages would be ceramic not the epoxy/plastic if indeed they were high temp rated parts. For example the PLL chip specs are not public but specs on the previous version is, http://www.primarion.com/PrimarionMain/Documents/PDB-3544.pdf, and the new version uses the same package.

Packaging & Part Number
Part Number Ambient Package
PX3544BDSG 0 to 85°C 48-lead 7mm x 7mm MLF

85C is NOT a high temp part. Now that is not an operating temp, which could be higher, but the point is the same, thats not a high temp spec part.

The data sheet for the most commonly used dram on these boards is here:
http://www.qimonda.com/promopages/index.html
Notice the DC functional operational parameters specify a Tc of less than 95C.

Note I am not talking about any specific "brand" of card, but the reference design in general. I would love to have information on a card specifically engineered for high temp operation.
 
Met-AL said:
Wouldn't you think that a company that offers a legit lifetime warranty would automatically be concerned about life of the card?
Click to expand...

Good point, one would both think and hope so. But I have yet to see the engineering proof and my experience with electronics leads me to be very wary (actually "horrified" is the exact word I used) at the advice given out by the person on the forum. As mentioned above I have been investigating these cards (not just EVGA) intensely. I am basing my conclusions on the datasheets and what other slim technical information available and my experience and education.

Do no misunderstand me, I like these cards, very much. I like EVGA above the rest for their customer support. Back when they had issues with (i forget the number) some KO series of cards and REPLACED everyone's card, I knew they were a stand up, better than the rest company. I will most likely buy an EVGA card, regardless of my findings, they are no worse than any other reference card design and very likely better than most. However, it is what it is. And based on my investigation blah blah blah the comment about the "stability" of the card may be true, but from an engineering standpoint concerning life and reliability it does not make any sense at all.

More:
I know it seems like I am throwing rocks at these cards, that is really not the case. EVGA as mentioned above has the best customer service policy I have ever seen. However these are reference card designs and here is the point, A fantastic level of performance at a great price. To get that great price engineering compromises have to be made. It is the nature of all products. For example I paid top dollar for the brand new P965-DQ6 two days after it was released because of the VRM design. I then spent the entire weekend "fixing' the shoddy heatpipe thermals and mounting including fabrication of a stiffener to keep board flex from moving the VRM mosfets away from their heatsink/pipe when attaching my waterblock. If Gigabyte had engineered such a solution and implemented it in the factory it would have added $100 to the cost and not everyone (read most people) are going to OC the board to its maximum and need such modifications and more importantly would not be willing to pay for it. Everything is a compromise, cost vs quality, profits vs cost of returns, etc. etc. Now this is not wussycomputers4you.com and so here I am with my very high standards and critical nature making a nusiance of myself. /shrug.

But I am having real difficulity with the forum "response". It just seems so easy to run the fan faster to cool the card properly. Even if not set that way at the factory, we have the the ability to do it ourselves and I just cannot swallow the advice to not worry about it until the gpu gets over 100C. Its almost negligent advice imo.
 
i've been getting display shutdowns at around 80C. but i ziptied a fan over the open PCI slots and it brought the temps down by 10C. the A05B is a nice case but it sucks for airflow for anything above the CPU and RAM. least now i don't have to hack open the case with a dremel though i prolly should....

idle at 50C and full load at 75C with ATITool. i'm happy now.
 
So, to sum it up. People think nVidia does not know what it's doing. Even though they designed the card, the chip, the cooler and probably tested the heck out of it, they don't know what they are doing. nVidia is so naive that they are not seeing an obvious problem that will cost them millions of dollars in failed products?

Maybe the card is just fine and that people need to realize that if you want to overclock it, you need to "overclock" the fan as well. It isn't like they sold you a card who's fan is at it's maximum performance just to cool the card at stock speeds.
 
Met-AL said:
So, to sum it up. People think nVidia does not know what it's doing. Even though they designed the card, the chip, the cooler and probably tested the heck out of it, they don't know what they are doing. nVidia is so naive that they are not seeing an obvious problem that will cost them millions of dollars in failed products?

Maybe the card is just fine and that people need to realize that if you want to overclock it, you need to "overclock" the fan as well. It isn't like they sold you a card who's fan is at it's maximum performance just to cool the card at stock speeds.
Click to expand...



Nvidia just sells the chips and provides (for a fee probably) the reference design. I am not sure who bears the cost of a burnt chip and if Nvidia makes good a replacement or credit to the card manufacturer but the major cost of the repairs would be borne by the manuf, not Nvidia. It is also highly likely IMO the failure would not be the actual GPU (it oribaky can survive at 100C) but one of the other components on the board. I thought this point was made clear and fairly well documented as much as can be expected. If you have one of the cards it is easy enough to pull the cover off and look at the caps to find a temp rating. The ICs can be looked up as well. OEMs have the option to specify better or worse quality parts and modify the cooling. For example on the reference card there are provisions for a third GPU voltage power phase. Typically not implemented to keep cost down. However I have seen at least two OEMs implement it, Galaxy and Palit. in addition to making other changes. And of course an OEM can completely redesign the card if they want to. This bring us to the concern I made passing mention of earlier, namely OEM overclocked from the factory cards with reference card cooling solutions. But that is off topic, remember it is the forum response to a end user query on a default card that is being questioned.

Er, no mention at all of overclocking the card in the forum response and I think nothing said in this thread supports your summation. We are discussing stock cards with stock cooling under "normal" operating conditions that are not manufactured by Nvidia. You seem to keep missing the point. Based on the recommendations given in the forum response and the apparently poor fan performance curve or lack of same, a lot of us (apparently) [ http://hardforum.com/showthread.php?t=1244257 ] feel the onboard components other than the GPU are at significant risk of life reduction due to unnecessary heat because of the fan speed profile. I have no idea if this specific fan speed profile is stock Nvidia or of EVGA origins. I replied to your post because it was posed as a question and this is my response. Anyway if that is how you feel about it, fine, your opinion is just as valuable as mine.
 
Personally I don't buy a video card without buying some after market heatsink as well. Even if I run the card at stock speeds, I still put one on. I feel a whole lot better when the temp drops 10-15 degrees. Hell, my OC 8600GTS idles at about the same temp my processor does (35*C CPU, 39*C GPU currently). Should I have to do this? No. Will the video card manufacturers change the reference design enough to make it work? No.

The fact of the matter is that GPU's put out more heat than a CPU these days. However if you compare heatsinks, the sizes are like night and day. I think there are a few things that cause this:

1. PCI slot/space requirements. Most cards have to fit within one slot. I think it's time that the industry realizes this isn't going to work anymore and makes a standard for more space around the PCI-E slot. Do this and heatsinks can grow and cool GPU's better.

2. Max profit/low cost. They want to do it cheap...and that's exactly what gets strapped on the card. A cheap cooling solution.

3. Lack of cooling innovation by the reference design maker (Nvidia, ATi) These guys keep using the same old crappy designs over and over. All of the innovation happens at the vendor/brand level. BFG and Gigabyte are good examples of this.

Bottom line is GPU cooling needs a major overhaul.
 
If you look into the bios of the 8800gt, the only time it ramps up to 100% is at 100celcius. The bios is broken - there no setting you can change in the bios to go to 75% at say 80 celcius.

Since evga has lifetime warranty on this why worry.
 
Met-AL said:
So, to sum it up. People think nVidia does not know what it's doing. Even though they designed the card, the chip, the cooler and probably tested the heck out of it, they don't know what they are doing. nVidia is so naive that they are not seeing an obvious problem that will cost them millions of dollars in failed products?

Maybe the card is just fine and that people need to realize that if you want to overclock it, you need to "overclock" the fan as well. It isn't like they sold you a card who's fan is at it's maximum performance just to cool the card at stock speeds.
Click to expand...

7900gt.

(that is my only response to your naivety.)
 
I hate to reply posts like "what the man said" but in my mind BillParrish is spot on.

I have the mental image that Nvidia provides only the chips to the boards which the card manufacturers then assembly according to Nvidias reference design. The sad part in my mind is that the card manufacturers seem to follow the reference design without much thought. I dunno why this is so, probably due to extreme competition? And looking into the history, it becomes clear that the ATI/Nvidia reference designs are far from perfect at times, 7900GT is the reacent reminder of that.

So no, I would not be totally shocked to find out that the Nvidia reference design in 8800GT case had faults and since the card manufacturers basicly just copy the reference design, their cards too have the same faults. And about the Evga warranty, would it not be weird if their 8800GT did not have lifetime warranty when other cards do have it? What would that tell you about the card.

I have considered 8800GT card for my next card, but even if I could get one (avaibility is horrid), I would slap on a better 3rd party cooler on the card immidiately just to be safe.

ps. I would really love to know how far goes the chip manufacturers responsibility in these matters. Common sense would say that they are only responsible for the chip and not the other parts.

So if this is so, its the card manufacturers who are going to feel the heat when cards start to fail. Nvidia says, hey its not the chip that brakes up, its the other parts, we take no responsibility. Its not our fault if you guys followed our reference design, you could have designed a better card on your own. Ok this is highly speculative and in no way I claim this to be even remotely the truth :)
 
RussianHAXOR said:
THats actually pretty true. The 8800GT runs hotter than any of the other video cards mainly because of the single slot cooling solution and low fan speed. I really dont know whether the 29% fan speed is really true but if jaocb says thats what it is i trust him more than some random review site.
Click to expand...

That's not the issue. The issue isn't one of trust where the choices are between him and review sites. The issue is one where people are experiencing lock-ups and other over-heating related issues with their stock cards because the fan doesn't ramp up soon enough to cool the card properly. If there's a trust choice, it's between the customers who actually own the product and are experiencing real problems and this eVGA rep who is contradicting real-world data.

The reality is it's more often the stock-overclocked cards - MEANING CARDS THE RESELLER (eVGA in this case) OVERCLOCKED, Met-AL, NOT THE END USER. They oc'd them and are selling them that way, which means they believe the card is capable of the given speeds while using the reference design cooler. The reality is they are being increasingly proven wrong as more people report problems.
 
I'm sorry but this "it's OK to have your component run scorching hot" is not acceptable.

Gee.....let's use ntune.......ooops, it is the worst softwarz ever invented.:eek:

My advice, and I did this right after I installed my son's 8800 GT......RivaTuner......
its so easy to apply a profile to ramp up the fan speed.
Unfortunately, this card sounds like a leaf blower at 75%, but what else are you going to do in the short run.

I'm searching for a good aftermarket cooler......otherwise the card is superb.
(I got an XFX, BTW, not an EVGA; and its a stock model)
 
I modded my heatsink with an 80mm fan without removing the heatsink from the card.
The top pops off the heatsink with a little effort.
I then took a coolermaster 80mm fan and its clear plastic box.

The clear box was cut to fit over/around the fin array so it still has an air guide.
The 80mm fan is strapped on top of the old fan (as its hard to remove without taking off the whole heatsink). The old fan has a really thick sticker which raises the 80mm fan too high so I removed the little sticker before putting the 80mm on top.
The 80mm fan is an almost perfect fit, its as though they realised people would want to strap an 80mm fan there!

The net result is quite a leaky design as I havent sealed the gaps yet, I'll finish today, but...
Even with the leaks, it cools silently at about the same rate as the original fan at 60% which was noisy.
An added bonus is the air guide I made can direct the air towards a removed blanking plate on the next slot so the hot air can be mostly funnelled out of the PC.
I'll hopefully finish making that today as well.
 
magoo said:
I'm sorry but this "it's OK to have your component run scorching hot" is not acceptable.
Click to expand...
It is if it works just fine at those temperatures. If you want to keep the card quiet, then you allow it to run at 29% and hit those extremes. If you want to expand the lifetime, then you can use nTune or whatever to have the fan run faster and keep the temperature lower.

I wish NVIDIA hadn't gone with a single-slot reference HSF design, but that's life, I guess. The single-slot coolers just aren't viable for performance cards anymore, but I think the problems many are having may have to do with case airflow (or lack thereof) than what the diode on the GPU is reading.
 
BillParrish said:
Nvidia just sells the chips and provides (for a fee probably) the reference design. I am not sure who bears the cost of a burnt chip and if Nvidia makes good a replacement or credit to the card manufacturer but the major cost of the repairs would be borne by the manuf, not Nvidia.
Click to expand...

98% of all the 8800GT cards are manufactured by Flextronics under direct contract with nVidia. And in the case of the 6800, 7800, 7900, 8800GTS, and 8800GTX boards 100% of the boards are manufactured by Flextronics. The AIB partners have only been free to use nVidia GPUs on custom boards in the case of the mid to value oriented market for years now... presumably to give nVidia a lot more control over the quality of the product that bears it's name. Anyway nVidia most definately ends up bearing the cost of a bad board when all is said and done...

Only now with the 8800GT do we see nVidia giving some slack and allowing third party designs, and sofar only one or two AIBs have been brave enough to manufacture their own card. (The EU based Zotac and their new 8800GT card comes to mind)

I personally find it hilarious to see the number of people in this thread that feel so strongly that they know BETTER THAN NVIDIA how the card should be cooled and what temperature it should operate at. The intarweb gall of some people simply astounds me... :p

Now don't you extreme overclockers get your panties in a wad... I'm not referring to you guys that want higher clocks. This IS [H]ardOCP after all.... ;)
 
I'd rather it run cooler so I'm using rivatuner. At idle I have the fan speed set to 40% and I can't hear the fan. Idle temps are around 55C at that speed.

It ramps up to 80% at 60C and 100% over 70C. Works for me.
 
People might be throwing wild guesses in this thread, but atleast one should not be NAIVE enought to actually think that Nvidia, or be it which ever component manufacturer out there, really cares about customer after the purchase has been made.

To be realistic, in my mind, 3D card needs to work for about 2 years or so. Even if it brakes after that, nobody cares as the cards really are outdated by that time. Now nobody here is saying that 8800GT design is flawed. There are signs that people are getting artifacts etc, which are no doubt linked to overheating. 8800GT would have been much better off with dual slot cooler, why Nvidia did not implement it is a mystery to me. And this is all the thread is about. Sure its ok that the card runs 100c by Nvidia, but what if the cards start to drop like flies after 6 months or so?
 
Dethred said:
Why would the 8800GT's silicon hold up to water-boiling temps but CPU's can't? 100C can't be good for any processor can it?
Click to expand...

Get it through your thick skull, they are designed to run like this.

sam0t said:
People might be throwing wild guesses in this thread, but atleast one should not be NAIVE enought to actually think that Nvidia, or be it which ever component manufacturer out there, really cares about customer after the purchase has been made.

To be realistic, in my mind, 3D card needs to work for about 2 years or so. Even if it brakes after that, nobody cares as the cards really are outdated by that time. Now nobody here is saying that 8800GT design is flawed. There are signs that people are getting artifacts etc, which are no doubt linked to overheating. 8800GT would have been much better off with dual slot cooler, why Nvidia did not implement it is a mystery to me. And this is all the thread is about. Sure its ok that the card runs 100c by Nvidia, but what if the cards start to drop like flies after 6 months or so?
Click to expand...

That's what warranties are for.
 
well I remember that every 10 degrees a chip goes up in deg c, it halfs its life, so at 0 deg its 1000years, 10 its 500years, 20 its 250, 30 125, 40 67.5, 50 38.25, 60, 19.125, 70 9.5125yearss, 80 4.xx, 90, 2,xx, 100, 1.xx!

not sure what the initial life of the silicon is >_> but you get the idea, 100deg is bad not to mention what thats going to do to your ram and cpu sitting directly above the card where heat loves to rise,

I on the other hand would love one :) It would warm up my room nicely
 
You must log in or register to reply here.
Back
Top