![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
GPU temperature has nothing to do with how much heat gets dumped into your room/house.
Heat/cooling cycling is what effects solder joints/BGA balls not total temp so long as that temp is well below the melt point of the solder which @ 95C the R9 290X will be.
Supposedly this is due to reference voltage regulation + lack of available over volting on reference cards.
See post #118
If there wasnt any heat cycling you would have a point.
They go hand in hand.
Then why call it out?
You are being pedantic for the sake of it.
The 480GTX ran at similar temps under load (93-95C) since that card/GPU was designed to run at those temps. "High temp" is a relative term that means nothing in of itself and must be applied on a case by case basis for it to have any meaning.nenu said:
I'll spell it out for you.
A quick search on google reveals it is not insignificant.
It should be considered.
Apparently it was needed since you've now shifted goal posts from claiming high heat was going to cause solder reliability issues to now claiming its heat cycling that is going to be the problem....due to its high temp. Which doesn't make any sense. Also I said you were technically correct WRT to heat cycling being the cause of solder fractures, not that it was going to be an issue with the R9 290/x.
Uh no it isn't. Your google also shows 5850's, 8800's, and various other video cards being baked to try and fix them. You'd need a study or paper showing that 95C will cause [x] amount of failure rate with the same or similar solder using the same or similar packaging that is used for the GPU in the R9 290/x to have a point. I don't think anything like that publicly exists unfortunately. Same thing goes for 480GTX failure rates though if you'll accept googling forum posts there is plenty of anecdotal evidence that they were low for that card.
Based on what though? Because it runs at 95C? That is only about 10C higher than my 4890's used to run at, which for solder that usually has a flow point in the hundreds of degrees matters not at all, its nearly a rounding error!
My goal posts havent changed at all.
I havent trolled you once, only defended myself from you.
I said it needs to be considered, not that it will happen for definite.
Those that use the oven trick no longer have a warranty so the chance the failure is down to a badly soldered joint or bad QC is vastly reduced.
Yes, based on heat cycling causing solder joint fracturing and hotter temps dont improve the situation, they make it worse.
Hahah denial and "NO YUO BAD" won't convince anyone of anything. You've gone from "heat causes GPU failure" to "heat and heat cycling go hand in hand" to "no its heat cycling" to "well it only needs to be considered, heat/heat cycling might not be a problem...maybe" in the space of a few posts.
You linked a google for 480GTX oven baking which isn't the same thing as proof of "fractured solder due to heat cycling" and many cards from different generations popped up in that search too, some of which were not known for being particularly "high temp" cards. We have no clue exactly what is going on here. My personal WAG is these issues are due to the lead free solder that has to be used now since years ago before that switch occurred these failures were seeing today were uncommon. However that is just a WAG. No one really has a good answer. Certainly no AIB's or IHV's have bothered to give a reason publicly why this keeps happening. Oven baking works...sometimes. Sometimes not. This issue is likely lots more complex than "just" thermal stress cracking solder on BGA's of the GPU.
Too much of a blanket statements to be useful + assumes this is an issue that can't be engineered around. Which is silly since there are devices out there which work fine at temps higher than 95C that are soldered, things like satellites and such. Those things have temp swings of hundreds of degrees daily and yet work for years on end. For all we know AMD had higher temp brazing solder used or had a custom solder paste that they know will hold up to the thermal stress. None of that is impossible or unlikely.
Its clear reading isnt your strong point.
I trust the statements of many of them over you.
Now you admit that it is an issue
You profess to know what you are talking about and then equate the manufacture of graphics cards to satellites.
Note the use of specialised components.
Yes.
Also, in my brief time with the 290s, they don't overclock for shit.
The best I can squeeze out of my MSI card is 1100, my XFX won't even do 1075.
There's no voltage control on these things. All you can do is push the fan and power sliders in the CCC and up the clocks using Afterburner.
It's not heat, because the fans scale and the temp never went over 93C.
Maybe with water and an overall lower temp I can squeeze a bit more.
HardOCP said:
Lack of SLI scaling is of course a driver problem, what else should it be? (Well it could be that the game saturates the SLI link @4K, remember why AMD ditched the CF connector? But that's purely speculation on my part.) Which would be forgivable if Metro: LL was an AMD GE title that came out yesterday. But it has been out for a while, and NVidia had an extra couple of months to optimize the game and their driver for each other through the TWIMTBP program.First of all....thats 4K. I think Brent mentioned that he thought nvidia had some driver issues at that resolution.
Look, this is [H]. A site for high end gaming and enthusiasts. This site benchmarked 5760x1200 triple screens when other, lesser sites still thought that 1024x768 fps numbers were somehow relevant for the performance evaluation of a graphics card. I think that the 4K numbers are very relevant to the readers of this site.
