in defense of the 1900 series, mine runs fine @ default fan settings
Hey Hey no defending here
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
LOL@Evga response to 8800GT fan issue
- Thread starter sluzbenik
- Start date
Hey Hey no defending here
Hm... lemme think through this... I was looking at the physical change in the state of the transistors, but got hung up on only looking at the case of an increase in potential energy as a result of a state change, ignoring that it could be a decrease as well XD
You're right that isn't going to matter... as you're not constantly increasing the potential energy of the system (wouldn't that be scary!).
So you've just got the waste heat... and... I guess some waste RF as well (although that's going to be rather minor I'd imagine)
Brain fart ftw!
Roberty
Extremely [H]
- Joined
- Nov 30, 2001
- Messages
- 44,953
I just had to RMA my new EVGA 8800GT SSC. Video corruption and lockups within a few minutes of starting a game. Never had this problem with my old 8800 GTS. Just for shits and giggles I took off my side cover and placed a small fan blowing straight on the card and it helped a lot so it's got to be a heat issue. It didn't cure it completely but it would last about 30-45 minutes before it gave any problems after that. I have 6 fans in my case so airflow isn't an issue. I ended up ordering one of the XFX 8800GT cards at Dell for $250 shipped and sold the new one that EVGA sent me on Ebay for $450 shipped.
I ended up ordering one of the XFX 8800GT cards at Dell for $250 shipped and sold the new one that EVGA sent me on Ebay for $450 shipped.I agree. They should have made the goal of the cooler to keep the card below 65C by spinning up and down when needed. That how I plan to configure mine when I get it. Personally I would rather have a smart videocard that I can vocally hear it doing it's job than a quiet dumb videocard that burns itself out because it wasn't programmed not to.
TheRapture
Supreme [H]ardness
- Joined
- May 31, 2000
- Messages
- 6,885
No, the heat leaving your case is EXACTLY the same regardless of what the speed of the fans inside it are running. If the internals are producing 300w of heat, it's still 300 watts of heat, air moving around might cool better or worse due to speeds, but power output is the same.
TheRapture
Supreme [H]ardness
- Joined
- May 31, 2000
- Messages
- 6,885
No, a tiny fraction, however small, is translated into work of some kind. Even digital information takes some power to create it/move it around.
I don´t want to sound like an ass, but I will anyways, sorry about that. But If the 3D card manufacturers really took everything account, we would not have defective cards on our hands right? In my mind, after HDD and PSU, it is the 3D card that is most likely the component in your system to fail. And like I said before, I really think this is probably due to the fact that, the 3D card manufacturers seem to follow the GPU manufacturers reference card designs blindly. Somebody correct me if Iam wrong in this.
Also I guess you don´t have very long memory in these matters, one does not have to look too far into the history to see all sorts of error where 3D card manufacturers have not "taken things in account".
Defective 7900GT cards that dropped dead like cherries on prom nigh is probably the best example of this. I remember that many people in these forums went through 2-3 RMA with their 7900GT cards.Thats one of the most drastic examples I remember from reacent history, but Iam sure there are plenty more when you start digging. So no it would not amaze me if the all knowing card manufacturers would screw up again. After all they are bunch of humans after all.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,118
even if the card will run @ 110deg, if you NEED a single slot cooler, chances are you dont want to run this card @ 110deg since it will be close to something else, thats still alot of heat to dump into your case, even if its only 70Deg coming off of the hs..
why is it so hard for companies to make a GOOD cooling solution? even if its single slot there has to be somethinggggggggggg
why is it so hard for companies to make a GOOD cooling solution? even if its single slot there has to be somethinggggggggggg
Emission
Supreme [H]ardness
- Joined
- Dec 6, 2005
- Messages
- 4,420
The good news is that carbon nanotube solutions will make single slot solutions 5-10x as effective as the best dual slot solutions today. OCZ is pretty far along on this concept with its Hydrojet, now the concept has to be applied to single slot graphics card coolers.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,118
well how hard would it be to make the single slot cooler exhaust outside of your case? (same goes for both ati/nv in this case)
but at least ATI went an extra step with the 3870(dual slot though) and extended the plastic shroud to the end of the case
Scouras626
Weaksauce
- Joined
- Jul 2, 2003
- Messages
- 65
Actually, not really.
The resistivity of metals increases as temperature increases. So as the card gets hotter, all the circuitry on the board is getting more resistive, which makes the card require more power, which increases the temperature, which makes it more resistive, etc.
Now I'm not sure how significant this is on these cards, but it definitely happens.
The 10°C = halved life would seem to result from the chemistry "rule of thumb" that a chemical reaction rate coefficient doubles with temperature. The reaction rate coefficient usually rises exponentially with temperature following the Arrhenius law expression but in lieu of actual data, the "rule of thumb" is actually a reasonable estimate for rough calculations.
Treating the thermally induced degradation of silicon is a chemical reaction, I could see how this generalization gets tossed around in this context.
However, keep in mind that if the rate coefficient at reference temperature is very very small to begin, it will probably still be small enough to not matter at the elevated temperature within the practical operating service life.
Disclaimer: I am a MS Chem Eng and not an EE. I am also not defending that the original reference to this expression was necessarily appropriate.
Doubles with.... what? A doubling in absolute temperature?
If that's the case, to halve its life, a 50C piece of silicon would have to be raised to over 370C.
Where does the "10C" come from?
honestjohn
Gawd
- Joined
- Jul 28, 2004
- Messages
- 864
I'm sorry but I disagree.
Trepidati0n
[H]F Junkie
- Joined
- Oct 26, 2004
- Messages
- 9,269
Care to elaborate on this? (hint, it's a trap).
I should have better stated a temperature increase of 10°C doubles the reaction rate coefficient. For example, if the temperature is increased from 10 to 20 °C, the rate coefficient doubles which in turn doubles the rate of reaction.
Again, I am not making a broad claims to the specific applicability of this rule of thumb to silicon. I am simply suggesting where this common claim probably originated.
Search google for "reaction rate doubles 10C" and numerous hits on university chemistry sites will come up or find a good reaction kinetics textbook.
The poster you quoted was theorizing as to where the "10°C = halved life" figure that's been brought up in many previous posts comes from... that in Chemistry it's used as a way to estimate change in reaction rates.
So going from 50C to 60C doubles the rate at which a reaction occurs and going from 60C to 70C doubles it again.
The problem with using this rule of thumb is that while it's a good rule of thumb for chemical reactions... electron migration isn't a chemical reaction
Although the person you quoted never claimed it was... he was just theorizing as to where people were getting the 10C = halved figure from.As for my qualifications... I spent 3 years working on a Chem Eng degree before getting my CSE degree.
To diverge just a bit... I didn't say anything stupid in my reply back to you did I? I'm just wanting to make sure I've got things straight in my head
Excuse me for not reading this whole thread but their response is probably only relevant if you leave the card at stock speeds and don't overclock. EVGA may be overclocker friendly in their return policy but their cards (and fan speed recommendations) only apply to stock speeds. I don't see what's so absurd about the response.
honestjohn
Gawd
- Joined
- Jul 28, 2004
- Messages
- 864
The poster you quoted was theorizing as to where the "10°C = halved life" figure that's been brought up in many previous posts comes from... that in Chemistry it's used as a way to estimate change in reaction rates.
So going from 50C to 60C doubles the rate at which a reaction occurs and going from 60C to 70C doubles it again.
The problem with using this rule of thumb is that while it's a good rule of thumb for chemical reactions... electron migration isn't a chemical reaction
As for my qualifications... I spent 3 years working on a Chem Eng degree before getting my CSE degree.
Those were my thoughts exactly, so thank you for saving me the posting time.
As for my qualifications... I spent 5 years working as an Amtrak Eng before getting my BS degree.
It was a factory OCed card...
Although thier response seemed to be simply trying to get more information about the problem, not suggesting a permanent solution (which is how some people seemed to take it)
Soul.Survivor
Limp Gawd
- Joined
- Feb 13, 2004
- Messages
- 139
This is why I bought a GTS 640 with 112sp. I don't need a space heater in my PC and I'm DONE buying aftermarket cooling for every damned part in my PC. It adds another $100-$150+ to the price of a PC to buy a gpu cooler, cpu cooler, nb cooler, heatsinks for every little mosfet and it's a hassle to install all that junk. I've got a box of this crap that I thought was a good idea at the time but turned out to be more trouble than it's worth. I also killed a nforce motherboard trying to replace the NB heatsink cause the chip was so brittle.
If there really are a certain percentage of these cards failing because of inadequate cooling then it's a design flaw as far as I'm concerned and I'm not willing to take the chance of having to remove my graphics card for a couple of weeks to RMA it when the product should've been cooled better to start with.
If there really are a certain percentage of these cards failing because of inadequate cooling then it's a design flaw as far as I'm concerned and I'm not willing to take the chance of having to remove my graphics card for a couple of weeks to RMA it when the product should've been cooled better to start with.
For all those who are using Rivatuner, etc to adjust the fan speed to keep the card cooler . . .
You can argue all day about whether the card should run at 100C before ramping fan speed up. The bottom line is that EVGA person told you it was designed that way. That means that the fan was not designed to run anywhere near 100% for very long.
I'll bet you your fan dies a lot quicker that it would have.
I keep hearing people argue that it may be fine for the chip to run at 100C but not the supporting components. That is bull. No Manufacturer would offer a lifetime warranty on a design that wasn't rigorously tested. If their was a design problem with this card, you would have hundreds of reports of failures instead of the handful that will always occur in any mass produced product.
You can argue all day about whether the card should run at 100C before ramping fan speed up. The bottom line is that EVGA person told you it was designed that way. That means that the fan was not designed to run anywhere near 100% for very long.
I'll bet you your fan dies a lot quicker that it would have.
I keep hearing people argue that it may be fine for the chip to run at 100C but not the supporting components. That is bull. No Manufacturer would offer a lifetime warranty on a design that wasn't rigorously tested. If their was a design problem with this card, you would have hundreds of reports of failures instead of the handful that will always occur in any mass produced product.
That sounds really nice. Key word sounds, if you've ever worked in the real world on anything from airplanes to key chains you know that products get released early because some one is yelling at them and that they don't get tested enough before they get out the door. Some person answering the phones or e-mails is not a thermal engineer, he is usally getting paid 6$ an hour to read a prompt. Bridgestone release tires that killed a bunch of people before they admitted there was a problem. That was a huge company, which was dealing with a product that could (and did) kill people. We all like to believe that the designer is some god who knows everything, but they're not. That's like saying well Windows Vista was designed and throughly tested before it was released. Also read the fine print on most "lifetime warantee"s. Most of them say simply that it will be free from manufactoring defects, not that it will last forever.
This thread has so many retarded posts on so many different levels by so many people.
Thanks for the flame. I would completely agree with you on the retarded posts statement, but have to take exception to you including mine. Their is no design flaw on these cards and the proof is the lack of massive returns thus far. We can argue all day, but the fact is that if their was a thermal design flaw, many times more failures would have been reported by now.
How is that retarded?
Granted the first part of my statement may appear naive, EVGA has a history of good support and to my knowledge hasn't burned anyone with a "bad" design to date.
You are not reading my post in the context of this thread. This is about EVGA telling a customer that the card is designed to run up to 100C before the fan runs at 100%
My post is in regards to EVGA specifically and their offering of a warranty. Also, I did not state that they guarantee the card will work for life. You don't "accidentally" set the fan to run at 29% up to 100C and train your support people to tell everyone that is correct. That was an intentional design decision made by them. I was merely trying to point out that that decision WAS intentional and if it were as flawed as some people around here think it was, massive failures in the field would be the result. Since we aren't seeing those failures, the design is sound.
I have worked for over 10 years in an electronics manufacturing company and I am well aware of how the "real world" works. Your statement describes exactly why we do occasionally see flaming piles of shit get sold to consumers. They key here is that those products are proven by the returns to be defective. Their is no proof that this card design is defective.
We are talking about a temperature issue here. You can run this card at 100C but you sacrafice the life expectancy. We WILL have to wait to see how much it impacts that. In a year when a sigma of these cards have failed only then can we raise consern?
Your right, and every keeps saying the fan is louder than hell right? So how much of a stretch is it to say this was a decision that was made by (or highly influenced by)marketing. The way to make a cheaper card is to keep reusing designs from the past and keep trying to push them farther and farther. There comes a point where it's pushed too far. Is this one of those places? Time wil tell.
The problem with your statements is this is EVGA, and they have a track record of tweaking reference design for factory overclocking experience. That means they have experience is adjusting fan speed, etc. and will back that experience with the warranty. They are not stupid enough to intentionally lower the fan speed just to make a marketing guy happy.
You are correct that time will tell. I will bet good money that a year from now if you look back, this card will be proven as the great design that it is and EVGA will not have massive returns.
All I'm saying is that EVGA as a company does not have the track record of making stupid design decisions.
Time will tell as you said.
Emission
Supreme [H]ardness
- Joined
- Dec 6, 2005
- Messages
- 4,420
Another fun thing to consider is that the G92 is brand new silicon, designs based on it are extremely new and will have problems in the beginning. The 8800GT hasn't been out for more than a month yet.
Correct if Iam wrong, but when was the last time EVGA tweaked the reference design of a card? Do they use different components in their cards than the other manufacturers? All I have ever seen they do is sell cards which are factory overclocked and very rarely there has been a different heatsink on a card.
In my mind they do what all the 3D card manufacturers do today, they buy their cards from a firm like flextron or foxxcon, slap their own sticker on the card, apply their warranties, bundles and thats it. Granted EVGA is known to have excelent customer service and their step up program is unique, but other than that, I have very hard time believing that their cards really different from the Nvidia reference designs that much. Perhaps they cherry pick the cards or something? As I have not heard that EVGA owns any card manufacturing fabs.
ps. Your praise of EVGA made me think of the movie 300. "This is not MADNESS, this is EVGAAAHHHH"
The problem with using this rule of thumb is that while it's a good rule of thumb for chemical reactions... electron migration isn't a chemical reaction
First, thanks for eloquently clarifying what I was trying to say.
Interesting topic. A quick search turned up references that suggest electromigration is a diffusion-limited process with an activation energy just like chemical reactions. The diffusion coefficient relates to temperature via an Arrhenius expression.
As an example, the following source on pg 4 paragraph 6 (http://www.cadence.com/whitepapers/4095_Electromigration_WP.pdf) suggests that for aluminum interconnects, the rate of electromigration doubles with every 20°C increase in temperature. This is within the same order of magnitude as what was suggested earlier and is not that different from a chemical reaction.
Now, I am guessing that nvidia IC's are made using more advanced materials for the interconnects (copper etc..) that are more resistant to electromigration. But, I would be surprised if the temperature dependence of the diffusion rate was significantly different.
If I get some time, I will do a technical literature/journal search to see if I can turn up any data on the temperature dependence of the rate of electromigration induced diffusion in other IC materials.
. . .ps. Your praise of EVGA made me think of the movie 300. "This is not MADNESS, this is EVGAAAHHHH"
That was too funny
Actually what I think they tweak is things like fan/thermal settings in bios. I have seen several posts on the web about peoples fans ramping up at different temperatures depending on make and model. I would be willing to bet that they have the tools to make custom configurations for the various versions of the cards they sell to ensure proper performance at the rated speeds.
If I were a company like EVGA and I offered products (I did not physically manufacture) at different price points depending on their clock speeds and I also offered a lifetime warranty, I would have a binning process set up that tested boards at the rated speed I'm selling them at and I would tweak bios configurations to optimize cooling when under load.
P.S. I realized reading my posts here that I really sound like an EVGA fanboy. The funny thing is that I've never owned a card from them. I have read a LOT of posts when researching purchases over the years and a lot of people are very happy with them.
Sorry for the "fanboyishness" of my posts though.
When I first got my 8800GT SSC I noticed it ran hot, and I also saw other posts complaining about this as well when doing some research before buying the card. I even saw one post of a guy saying his fan speed wouldn't change on load, thinking it was broke so he RMAed it. Probably like everyone else he didn't wait for it to reach 100c+ before letting the fan kick in.
I've never ran it at 29% speed, right off the bat I set it to 50%, then from there 60%, artifacted once in Crysis (though after days of playing this never happened, must of been change in room temps, parents had the heat at full blast that night). I then set it to 70% just to be safe, also this is the level of noise I can handle as I don't much hear a huge difference in noise, mostly what I'm use to from my older computers. Anything above this level though is too loud for me.
At first, it would idle at 50+ and reach like 70-80+, no exact numbers as I've forgot. However now after putting a side panel fan blowing on the card and took out all the empty PCI covers and put a fan there exhausting out the back and my fan speed at 70% my temps have dropped to 45c idle not seen over 68c load (from what I've seen while playing Crysis for 2+ hours).
For the people that are having overheating problems are you playing Crysis? I have a friend who bought a BFG version, he can play any other games no problem but for some reason even at stock settings (the 600 core stock ones) he is still artifacting and locking up in Crysis at random. Sometimes hours of playing then quit and play again for 5-10 mins and then a lock up. Must be a Driver + Crysis programming issue/conflict.
For now I'm comfortable with 45c idle 68-70c load I guess, I would be even more comfortable if it was below 60c. Personally 100c+ is kinda scary for me, however the cards may be able to handle that heat so I may be worrying for no reason.
I've never ran it at 29% speed, right off the bat I set it to 50%, then from there 60%, artifacted once in Crysis (though after days of playing this never happened, must of been change in room temps, parents had the heat at full blast that night). I then set it to 70% just to be safe, also this is the level of noise I can handle as I don't much hear a huge difference in noise, mostly what I'm use to from my older computers. Anything above this level though is too loud for me.
At first, it would idle at 50+ and reach like 70-80+, no exact numbers as I've forgot. However now after putting a side panel fan blowing on the card and took out all the empty PCI covers and put a fan there exhausting out the back and my fan speed at 70% my temps have dropped to 45c idle not seen over 68c load (from what I've seen while playing Crysis for 2+ hours).
For the people that are having overheating problems are you playing Crysis? I have a friend who bought a BFG version, he can play any other games no problem but for some reason even at stock settings (the 600 core stock ones) he is still artifacting and locking up in Crysis at random. Sometimes hours of playing then quit and play again for 5-10 mins and then a lock up. Must be a Driver + Crysis programming issue/conflict.
For now I'm comfortable with 45c idle 68-70c load I guess, I would be even more comfortable if it was below 60c. Personally 100c+ is kinda scary for me, however the cards may be able to handle that heat so I may be worrying for no reason.
BillParrish
Supreme [H]ardness
- Joined
- Aug 25, 2006
- Messages
- 7,519
Bingo Marketing drives engineering, remember the "Dust Buster", what was it ? 6800 something series ? Perfect Nvidia and Perfect EVGA got there asses reamed by customers over just the noise issue. Well bad example, whatever.
The 10deg C reduction , double the life (which is the way I heard it) is an old old electronics rule of thumb. This rule of thumb was first discoverd by Elmer Clodstomper of Hotashell Alabama in 1843 who determined that if he let his mule rest in the shade for 10 minutes every 2 hours they only fell over dead 1/2 as often during the spring plowing. Since then it has been applied to many fields of human endevor with good basic results, the burning of witches being a glaring exception.
Look at the temp derating curve for an aluminum electrolytic capacitor. They are much better than they where 20+ years ago and light years better than the paper ones used before electolytics. There should be service life vs temp as well. Mallory used to have it available. Reams of stuff on electronic componet life vs operating temp. We had shelves and shelvesof data manuals full of the stuff.
I just bought an EVGA 8800GT this morning. Tools exist to fix the fan speed or not as you wish, I am going to water cool it anyway. Carry on.
As was stated much earlier in the thread, if your getting errors in ANY application (artifacting, crashing, etc) then RMA the card as that is not normal. Forget the fact that you "know" how to adjust the fan speed. You shouldn't have to adjust anything to get trouble free gaming. If everyone who has these kinds of issues would RMA instead of finding "fixes", we'd have a much better idea if their is a design flaw.
I played the Witcher on my BFG 8800GT for 6 hours straight with rivatuner monitoring temps and it ran at about 90C during the gaming but I had no errors, artifacts, etc. I had two 1900XTX in crossfire mode for over a year running in the 70-90's with no errors or artifacts. i was told with those cards two that I didn't need to worry until temps reached 120C. They both still work fine.
People need to RMA cards that have issues promptly so design problems with new silicon can be caught early on.
Maybe its better for you guys in the states but every time that I RMA something it cost me a fortune.
For example I shoot a product from Canada over to the USA, it cost $20 bucks for shipping, then they send it back, using UPS/Fedex and I get charged duty/tax (normally around 10% of the item) and the brokerage from UPS/Fedex. For a $300 dollar card it might cost me $40-50 bucks to get it back as well I don't have a part for about 3 weeks. I am lucky (sorta) to have many spare parts kicking around. However if I did not, I would just buy another part rather then wait the 3 weeks and pay the costs.
RMAing a part is not a solution to a overheating problem. Could you imagine I a company like Western digital would say that too you. "Oh ya, don't worry about losing all your data and not having a working computer for 3 weeks...you get a new drive at the end of it"
For example I shoot a product from Canada over to the USA, it cost $20 bucks for shipping, then they send it back, using UPS/Fedex and I get charged duty/tax (normally around 10% of the item) and the brokerage from UPS/Fedex. For a $300 dollar card it might cost me $40-50 bucks to get it back as well I don't have a part for about 3 weeks. I am lucky (sorta) to have many spare parts kicking around. However if I did not, I would just buy another part rather then wait the 3 weeks and pay the costs.
RMAing a part is not a solution to a overheating problem. Could you imagine I a company like Western digital would say that too you. "Oh ya, don't worry about losing all your data and not having a working computer for 3 weeks...you get a new drive at the end of it"