This is why GTX590s blow up. Infrared thermography: GTX590 vs HD6990

I remember why I stopped reading the [H] video card section.

jeremyshaw said:
bash bash. HAHA, I own nVidia cards, so that means all AMD GPU can suck my minime.

Now for me to make a fanboi comment:

reviewer much idiot. munchie munchie. 7 other reviewers fried cards? pffsshshhh. Those stupid idiots. They are all incompentent idiots. Only reviewers who didn't blow up GPU matter in my eyes. Luckily, I'm not a TPU forumer, otherwise, I'd slit my fucking wrists. :rolleyes:

HA! You have AMD in your sig! OMG! That means you are an AMD fanboi!! :eek: out, OUT!! ARg, crap, this isn't the nVidiot section! :mad: WTH was the OP thinking!? :rolleyes:
Click to expand...

QFT
 
jeremyshaw said:
bash bash. HAHA, I own nVidia cards, so that means all AMD GPU can suck my minime.

Now for me to make a fanboi comment:

reviewer much idiot. munchie munchie. 7 other reviewers fried cards? pffsshshhh. Those stupid idiots. They are all incompentent idiots. Only reviewers who didn't blow up GPU matter in my eyes. Luckily, I'm not a TPU forumer, otherwise, I'd slit my fucking wrists. :rolleyes:

HA! You have AMD in your sig! OMG! That means you are an AMD fanboi!! :eek: out, OUT!! ARg, crap, this isn't the nVidiot section! :mad: WTH was the OP thinking!? :rolleyes:
Click to expand...

QFT, I went AMD and you went nvidia funny how the universe balances itself out. Anyway you sound as sick of the fanboyism as I am.
 
What I'm getting from all of this is that apparently, the Power Limit function doesn't cope very well with bioses that allow voltage adjustments over 1.0v. This is what you would call a 'Perfect Storm'.

Since Afterburner will only allow you to adjust voltages up to the maximum allowed in the bios, and maybe (10-20%), the best advice is to NOT use Afterburner with ANY of the 267 drivers on the GTX 590. Did AIBs even test if the power limiter was compatible with the modified bioses? If Nvidia advised that the voltage should not be increased over 0.938v then why in the hell are the AIBs releasing cards that allow you to go way beyond that?

Edit: I'm waiting to see what ASUS puts out in the form of the ROG MARS II/ARES II... 3x8pin for 525W or 4x8pin for 675W??
 
Cards blowing up have damaged Nvidia's rep, this negative PR for wrong or right reasons will hurt them, the fact that's not justified probably won't matter. When you add the bumpgate issues and previous driver killing cards you tend to see the large amount of negative sentiment towards Nvidia spill out, you might no like it but the perception is not going to change anytime soon.
 
Nvidia should just not allow any clock/voltage adjustment, we need to cater to idiots.
 
I was under the impression that people have over-volted AMD cards, when was the last time one failed in the way the 590 has? You have to be blind not to accept the fact that Nvidia made bad design decision(s) which allowed this to happen.
 
darkstarcow said:
I was under the impression that people have over-volted AMD cards, when was the last time one failed in the way the 590 has? You have to be blind not to accept the fact that Nvidia made bad design decision(s) which allowed this to happen.
Click to expand...

I think the dilemma comes from the fact they released it to the early testers vs the public (or at least to my understanding). Releasing a driver that allows you to do something you shouldn't be doing to some testers getting free product is different than blowing up someone's $700 vid card. How many people that actually BOUGHT the video card have blown them up?

I still think this was a mistake by Nvidia to release the card in this matter to testers. Whether it was on accident, oversight, etc.. Personally i don't care, if i was in the market for one of these, i'd like to see a little bit of long term tests for reliability and performance before plunking down some $$$$

p.s. Non nvidia fanboy here, feel free to flame still :mad:
 
Nvidia fanboys in denial.

Finding excuses, blasting reviewers, calling those guys working hard and working hours to write those reviews and help us ''newbies'' stupid'' blah blah blah only to defend their ''green-honor-e-peen-gospel'', without never going public, and never attacking them directly on their sites. If you can do better then them, why don't you do a review of the card? Go do your own IR metering instead of attacking their method, and hiding behind your keyboards in your basement! Start your own site! Stop hiding under the skirts of your HardOCP geeen-team friends!

If you're better, and know everything, why don't you go and tell those ''bad'' reviewers directly? Start a thread on their site! Start a new review site! Will see if you can do better!

Typical on HardOCP.

590 is a FAIL product. That's all. Nvidia will probably do better next time.
 
Last edited:
What a cluster fuck. I would never buy a 590/6990 so I don't care if they have issues, but that's just a poor showing to have cards blowing like that. I hope we don't see too many people buying them with the same problem.

An 800 dollar card should not be doing that. End of story.
 
chithanh said:
The TPU author directly addresses this and your other claims in the ensuing forum thread. Regarding voltage bumps e.g. here:
http://www.techpowerup.com/forums/showthread.php?p=2235832#post2235832
Click to expand...


O RLY?

Only dual GPU TPU ever overvolted for a review was 5970.

http://www.techpowerup.com/reviews/ASUS/Radeon_HD_6990/25.html
http://www.techpowerup.com/reviews/Inno3D/GeForce_GTX_295_Platinum/32.html
http://www.techpowerup.com/reviews/Sapphire/HD_4870_X2/27.html

5970 was a much less power hungry design, but lets see:
"Wow! 1.337V on the GPUs resulted in 980 MHz Core, 1300 MHz Memory."

So assuming 1.1 default voltage on 5970, that's a 35% overclock with a 21% overvoltage.

On a GTX 590 he already had 815MHz on 1V - a nice 34% overclock with only a 6% overvoltage.

Yet after that he went straight for 1.2V, a 28% overvoltage - and started "to heat up the card"
If he had only bothered to RTFM that was kindly supplied to him by ASUS:

It is not advised to exceed the 1.050 to 1.065 vcore range as this begins to meet the limits for the OCP/OVP mechanism on the card. Exceeding these values without disabling OCP/OVP. or having superior cooling could affect the lifespan and functionality of the card/gpu.

Tell that to your newb and unprofessional behaving boss who gives a card 7.1 and writes "EPIC FAIL" after he destroys other people's property
 
Last edited:
I don't really see a big market for this card anyways. The 6990 is a better product. It might be louder, but it runs cooler, is faster, has more memory, can be much more safely overclocked, and will be more useful at the resolutions a dual GPU board is actually designed to play at.

The 580 is a fantastic product but the 590 is a bit of a waste.
 
HohnyF said:
O RLY?

Only dual GPU TPU ever overvolted for a review was 5970.

http://www.techpowerup.com/reviews/ASUS/Radeon_HD_6990/25.html
http://www.techpowerup.com/reviews/Inno3D/GeForce_GTX_295_Platinum/32.html
http://www.techpowerup.com/reviews/Sapphire/HD_4870_X2/27.html

5970 was a much less power hungry design, but lets see:
"Wow! 1.337V on the GPUs resulted in 980 MHz Core, 1300 MHz Memory."

So assuming 1.1 default voltage on 5970, that's a 35% overclock with a 21% overvoltage.

On a GTX 590 he already had 815MHz on 1V - a nice 34% overclock with only a 6% overvoltage.

Yet after that he went straight for 1.2V, a 28% overvoltage - and started "to heat up the card"
If he had only bothered to RTFM that was kindly supplied to him by ASUS:

It is not advised to exceed the 1.050 to 1.065 vcore range as this begins to meet the limits for the OCP/OVP mechanism on the card. Exceeding these values without disabling OCP/OVP. or having superior cooling could affect the lifespan and functionality of the card/gpu.

Tell that to your newb and unprofessional behaving boss who gives a card 7.1 and writes "EPIC FAIL" after he destroys other people's property
Click to expand...

where is this manual you refer to?
 
Rolled into a joint?
Mailed to ASUS along with their busted card?
Your guess is good as mine.
 
the 6990's i have here a the house run load at 76-83c card1/2 the 590gtx i have asus model loads at 69c in same case i have the fan set for 72-77% all the time no up down time but seems like they run at same temps 6990-590gtx :)
 
btw the 590gtx i love becasue it has 3 d-dvis so i get perfect 3 way lcd setup not that dp or hdmi are ot good but dvi with my dispaly seems to be best and all get to share the same output
 
HohnyF said:
O RLY?
Click to expand...
YA RLY.

You claimed "(TPU rarely if ever uses voltage bumps when showcasing an overclock.)" which is not true. Note how they overvolted the GTX 580 to 1.21V and it survived. This probably motivated them to try 1.2V on the GTX 590. They don't overvolt every card, and especially not budget/mainstream cards though.

You claimed "AND using old beta drivers, not even bothering to check if WHQL is out." which has been shown to be not the only cause of card death, as cards with 267.71 drivers died too (Google translation).

You claimed "AND being trigger happy with 30% overvoltage on 590 because he "couldn't wait to try bumping the voltage". " Actually 27.9% (0.938->1.2V) which is a large but not an extreme overvolt. The same TPU did on the 580 and not much more than the 23.4% (1.175->1.45V) TPU did on the 6970. Besides even non-overvolted cards went pop. So while it is likely that the overvolt accelerated the card death, maybe it was not the cause.

You claimed "AND not increasing fan profile" which is irrelevant too, as the card died when it was idle, so overheating cannot be blamed.

Regarding your claim "AND not disabling OCP/OVP as per guide the card was supplied with" that is maybe true, I have neither seen the guide nor the OCP/OVP setting.
HohnyF said:
Only dual GPU TPU ever overvolted for a review was 5970.
Click to expand...
That is correct if you don't count the switch on the 6990 card, which raises GPU voltage from 1.12V to 1.175V.

HohnyF said:
Tell that to your newb and unprofessional behaving boss who gives a card 7.1 and writes "EPIC FAIL" after he destroys other people's property
Click to expand...
Users should be thankful for reviewers pointing out possible hazards for the card. The choice of words may be suboptimal, but it is going to save many people a lot of money.
 
I really think that Nvidia should of implemented a better failsafe on a card of this caliber. The biggest problem I see is Nvidia put 2 580's on one PCB. People are going to assume you can treat the card as a 580 when it comes to modding and over clocking. If they had put 2 fully clocked 570's on with a bump in memory and a lower price tag to compensate, they would be sitting in a much better position than what has resulted from this fiasco.
 
chithanh said:
Users should be thankful for reviewers pointing out possible hazards for the card. The choice of words may be suboptimal, but it is going to save many people a lot of money.
Click to expand...

Wait, isn't pretty fucking basic that overvolting something by 30% can be dangerous? I've seen MOSFETs blow up before. I've had white hot pieces fly off when someone wired one in revers and ran current through....if you exceed the designed thresholds, it's kinda expected that it might fail. I again point to my gun/ammunition analogy. I should totally write a gun review and point out "When I put rounds in twice the pressure designed for the gun, the gun blew up." or "I tried putting a 3006 round into a gun chambered for .308 win...and the gun blew up" - stuff like that is...well, yeah. No effing shit.
 
chithanh said:
YA RLY.

You claimed "(TPU rarely if ever uses voltage bumps when showcasing an overclock.)" which is not true. Note how they overvolted the GTX 580 to 1.21V and it survived. This probably motivated them to try 1.2V on the GTX 590.
Click to expand...

So he overclocked single GTX 580 from 1.05 to 1.21, and that motivated him to try 0.938 -> 1.2V on a dual GPU card? Although he already had respectable 815MHz on 1.0V?
Shouldn't OCP kicking in at 1.2V really be a no brainer?

Because TPU themselves measured 305W on a single GTX [email protected]

So WTF did W1zzard think that same GF110 core, this time x2, would pull when overvolted to 1.2V?

Nvidia does some magic with their dual gpu cards power, but really...

1xGF110 @1.05V = 305V
2xGF110 @1.2V = ?

chithanh said:
You claimed "AND being trigger happy with 30% overvoltage on 590 because he "couldn't wait to try bumping the voltage". " Actually 27.9% (0.938->1.2V) which is a large but not an extreme overvolt. The same TPU did on the 580 and not much more than the 23.4% (1.175->1.45V) TPU did on the 6970. Besides even non-overvolted cards went pop. So while it is likely that the overvolt accelerated the card death, maybe it was not the cause.
Click to expand...
Actually, "I couldn't wait to try bumping the voltage" is what W1zzard said.

Besides I thought you wanted to discuss how W1zzard burned ASUS GTX 590 that was sent to him. Sure, we can talk about the other revievers burning their 590's,
but it's TPU destroying other people's property we're focused on right now.

chithanh said:
You claimed "AND not increasing fan profile" which is irrelevant too, as the card died when it was idle, so overheating cannot be blamed.
Click to expand...

"I went to heat up the card and then *boom*" are W1zzard's words.


chithanh said:
Users should be thankful for reviewers pointing out possible hazards for the card. The choice of words may be suboptimal, but it is going to save many people a lot of money.
Click to expand...

Yeah. Unlike his /gtx500ocp GPU-Z switch.
 
HohnyF said:
So he overclocked single GTX 580 from 1.05 to 1.21, and that motivated him to try 0.938 -> 1.2V on a dual GPU card? Although he already had respectable 815MHz on 1.0V?
Shouldn't OCP kicking in at 1.2V really be a no brainer?

Because TPU themselves measured 305W on a single GTX [email protected]

So WTF did W1zzard think that same GF110 core, this time x2, would pull when overvolted to 1.2V?

Nvidia does some magic with their dual gpu cards power, but really...

1xGF110 @1.05V = 305V
2xGF110 @1.2V = ?


Actually, "I couldn't wait to try bumping the voltage" is what W1zzard said.

Besides I thought you wanted to discuss how W1zzard burned ASUS GTX 590 that was sent to him. Sure, we can talk about the other revievers burning their 590's,
but it's TPU destroying other people's property we're focused on right now.



"I went to heat up the card and then *boom*" are W1zzard's words.




Yeah. Unlike his /gtx500ocp GPU-Z switch.
Click to expand...

to stop the cards from throttling when too much voltage is applied you have to mod the bios. You can bump the voltage all you want but the card sets itself to a safe mode when the pc boots so it can always boot. Theres a table for the performance and a table for the voltage. Doesn't matter what you do you need to hide that tables like Asus do to stop the cards from throttling and your voltage to be used at boot.
Now who fucked with the cards bios?
Was it a reference one or was it one from a compiler like Asus, gigabytes etc etc.
 
Spooony , I don't believe ppl who say they fried their card at boot/idle/default voltage or because of a driver crash. But that's just me.
 
HohnyF said:
Spooony , I don't believe ppl who say they fried their card at boot/idle/default voltage or because of a driver crash. But that's just me.
Click to expand...

that was just to add to your post. I agree with you. wizard did his own junk on the card
 
rampantandroid said:
Wait, isn't pretty fucking basic that overvolting something by 30% can be dangerous?
Click to expand...
Well, people see how much you can overvolt a 580 and a 6970, and they think it is safe to do with a 590 too. This is similar to how experience with Intel's indestructible Conroe CPUs made some people apply dangerous voltages to their 980X.
HohnyF said:
Besides I thought you wanted to discuss how W1zzard burned ASUS GTX 590 that was sent to him. Sure, we can talk about the other revievers burning their 590's, but it's TPU destroying other people's property we're focused on right now.
Click to expand...
You are right that we are talking about TPU. But I have to back up my claims (drivers weren't at fault, overvolt was at most contributing to a faster death but not causal) with facts.
HohnyF said:
"I went to heat up the card and then *boom*" are W1zzard's words.
Click to expand...
Not sure how much he succeeded in heating up the card there. In the forum thread he later said that the card died within two minutes of raising the voltage.
 
Can't we all just agree the thermals on the initial page don't look nice and call it a day? There's a lot of red and some high load temperatures. The 6990 performs a bit better, has a bit more ram and as a result, might be a tad bit better for triple-monitor setups which is what anyone thinking of buying either card should be running.

With regards to the card exploding when over-volted. Yes, over-volting cards is dangerous and stands a high chance of damaging a card if over-volted too far. Realistically though, every review by virtually any website over the past 5 years has included a section labeled 'overclocking results'. Its a fairly standard industry wide process to try overclocking cards in reviews. Even if you are an nVidia or ATI fan, it is hard to ignore this key point. A company --should-- expect reviewers to do some overclocking and quite possibly some over-volting on your card when you send it to reviewers.

Not checking/testing the driver/cards to ensure they can survive some overclocking/overvolting or sending a driver out with flaws that might cause the power limit features to trip, is a bit of a quality-control/driver-team mistake. I'm sure one could argue that nobody expected reviewers to use the first driver they were given, but, its like a new toy. As soon as you get it, you use the drivers you have available. You generally don't wait six months for stable drivers to be released, refined, refined again and refined once more before installing and overclocking your new gf and beginning to write your review.

This should be evident by over 7 different reviewers blowing up their cards. Not 1 or 2 reviewers but 7 reviewers with what I'm sure adds to years upon years of combined experience. Green team fans calling these reviewers idiots or newbs won't help damage the message sent. Attacking the message would be more effective than attacking the person sending it when there's many different reviewers saying the same thing.

Regardless, a failure came from the Green development's team. If only because of the bad PR that's enused and that 'nVidia cards explode' will be ammunition used by the Red team for probably the next six month to two years until the trolls old enough to remember the event have grown up and matured. Sure, new drivers are out and the likelihood this will occur to an end-user customer is nearly negligible at this point, but, the PR damage is already done.

Getting back to the original topic, the thermals on the initial page don't look nice but don't look much worse than the 4570? 4750? cross fire thermals that also showed of 113 to 119ish degrees. I'd personally fear a bit ever considering Quad-SLI or 2x590 looking at those thermals though. I think I'd be asking for the top card to repetitively fail. Unless I was happy with 1x590 for 6 months of the year and 2x590 when one of the card wasn't in some stage RMA.
 
The thermals on the previous pages are all BS, and are not worth the time discussing, and this has been proven.
 
happymedium said:
The thermals on the previous pages are all BS, and are not worth the time discussing, and this has been proven.
Click to expand...

I agree, no clue how he go those crazy high temps.. to be honest I am seeing alot of people just trying to pile on. We all know the 6990 is the top with the GTX 590 about 2-3% behind it overall. This fanboy crap has honestly gotten out of control and people just need to grow up.
 
I just recently got home insurance, and a new field sort of surprised me.

"Do you own a GTX 590 video card?"

was one of the questions on the application :D
 
God this thread is such garbage. No wonder I stopped reading the video card subforum.
 
This just goes to show why the GPUs onboard the GTX 590 are clocked so low. Clocking the fan down to reduce noise was probably to save face in light of all the flak they've given AMD over the last few years for their extremely loud cooling solutions.
 
It's nice to see that when AMD "f**ks up" all the nvidia fanboys make fun of it and stuff like that ...

When Nvidia does the same ... everyone gets really sensitive ...

Meh ... and i have a Nvidia ... before you start calling me a fanboy too... eheheh :p

Just lets admit that Nvidia's try to get the two fully featured/unblocked GF110's outcome was not that exciting as people expected ...

Although it's still a pretty fast card imo! :)
 
Postmodum said:
It's nice to see that when AMD "f**ks up" all the nvidia fanboys make fun of it and stuff like that ...

When Nvidia does the same ... everyone gets really sensitive ...

Meh ... and i have a Nvidia ... before you start calling me a fanboy too... eheheh :p

Just lets admit that Nvidia's try to get the two fully featured/unblocked GF110's outcome was not that exciting as people expected ...

Although it's still a pretty fast card imo! :)
Click to expand...

As a single solution its compelling for anyone not wanting to go with a 6990 because they have issues with AMD cards or drivers for whatever reason. Short of that I don't think it's a good buy. Cards like that are nice when you need the power of dual GPUs in a small form factor but generally speaking the high end single GPU cards are where it's at. SLI / Crossfire a pair of those or more, for the best results possible.
 
Dan_D said:
As a single solution its compelling for anyone not wanting to go with a 6990 because they have issues with AMD cards or drivers for whatever reason. Short of that I don't think it's a good buy. Cards like that are nice when you need the power of dual GPUs in a small form factor but generally speaking the high end single GPU cards are where it's at. SLI / Crossfire a pair of those or more, for the best results possible.
Click to expand...

Definitely agree with that. Neither company's "dual gpu on a stick" solution is particularly compelling, IMHO.
 
In any case I think it's quite apparent this card is an example of poor engineering. Failures have been reported from "bad" drivers or high voltage to simply starting up the card at boot at stock.

Tolerances are simply not very high - any little difficulty and it has a chance of going *pop*.
 
Dan_D said:
As a single solution its compelling for anyone not wanting to go with a 6990 because they have issues with AMD cards or drivers for whatever reason. Short of that I don't think it's a good buy. Cards like that are nice when you need the power of dual GPUs in a small form factor but generally speaking the high end single GPU cards are where it's at. SLI / Crossfire a pair of those or more, for the best results possible.
Click to expand...

I think both 590 and 6990 are great solutions. I can't say i'm not thinking about one myself. But that's because i just like to have 1 card on my water loop.

I had the Asus GTX590 for beta testing and i can say it was quick ... but for those, like me, who like tweaking everything, it was a continued frustration :)

well, as for myself, i'm just trying to decide if i'll go with the 6990 or, just get the 580 Lightning and forget about the Dual gpu's on single cards ...

I don't really trust any of the solutions from either companies ... don't know why but i just don't feel safe with these ... and that's the first time i fell like this ...
 
You must log in or register to reply here.
Back
Top