Official GTX 590 Review Thread (23 reviews at this time)

[Martimus]

'As a first step, I increased the voltage from 0.938 V default to 1.000 V, maximum stable clock was 815 MHz - faster than GTX 580! Moving on, I tried 1.2 V to see how much could be gained here, at default clocks and with NVIDIA's power limiter enabled. I went to heat up the card and then *boom*, a sound like popcorn cracking, the system turned off and a burnt electronics smell started to fill up the room. Card dead! Even with NVIDIA power limiter enabled. Now the pretty looking, backlit GeForce logo was blinking helplessly and the fan did not spin, both indicate an error with the card's 12V supply.
After talking to several other reviewers, this does not seem to be an isolated case, and many of them have killed their cards with similar testing, which is far from being an extreme test.



With the card dead, I looked for any damage and found one of the 12V input resistors acting as fuses to be damaged (1st pic, no visible damage), so I soldered a bridge on it to bypass it, the card then "exploded" again, this time one of the MOSFETs blew up as shown in the second picture above. This was just an experimental fix to potentially revive the card. If only the resistor was damaged then it should be working fine again, looks like something else was broken and the "fuse" resistor merely blew to protect the card.

Thanks to the kind people at ASUS, I was lucky enough to have two cards for an SLI review - for which NVIDIA could still not provide a Quad-SLI supporting driver. So at least I could finish the rest of this review. I most strongly advise anyone to stay away from overclocking this product and use extremely conservative settings, maybe up to 650 MHz and no voltage adjustments.

According to NVIDIA this should not happen. In their official reviewer driver (which I used), the NVIDIA Power limit is designed to be active for all applications, not only Furmark.'


Card go BOOM. Maybe they don't like extra voltage like the reference 570s.

The person who wrote that does not understand how electronics work very well. First a resistor does not act as a fuse, and would not blow up anyway. Second that was definitely NOT a MOSFET.

 

[OCGuy]

Another idiot who used the original driver than put 1.2v through the card. I have a feeling the number of websites getting page-clicks from nV GPU reviews is going to be smaller for Kepler.

 

[pcm81]

The person who wrote that does not understand how electronics work very well. First a resistor does not act as a fuse, and would not blow up anyway. Second that was definitely NOT a MOSFET.

Resistor can act as a fuse. V=(R1+R2)*I now give it a fixed V and fixed R1 and R2, the resistor will only let no more than "I" current through. R1 is your "fuse" resistor and R2 is the resistance of the rest of the circuit which is in series with R1. In this case R1 is much bigger than R2.

Now, Bump up V higher, so you also increase I since R1+R2 is constant. With that you also increase the voltage drop across R1. Now, if power Rating of R1 is less that V/(R1+R2)*V then R1 melts... which is one of the things that happens. Ofcourse this is a MUCH MUCH simplified model of what actually happens in GTX590 power circuit.

 

[Grooveriding]

http://translate.google.com/transla...ulletin/showpost.php?p=3960849&postcount=352'

http://img845.imageshack.us/img845/2889/quemado.jpg

Another one bites the dust.

 

[TerabyteX]

it was all good till you got to the properly designed part!!! again back to the beginning!!! Is not the 6990 crippled too? oh its not as crippled? Thats just a backwards view. How come you think its possible for nvidia to surpass the very power limitations that AMD was bound to? Its impossible, this perfect design thing isnt real. You are imagining something then saying nvidia didnt reach that. Its not possible, the 590 is the max nvidia could get under their power restrictions, period. Nothing faulted anything. It was a limit they had to fit their card in, you cant make it any other way. why dont you focus on why AMD couldnt clock their 6990 cores the same as the 6970 to be able to beat nvidias best instead? Its the same line of thinking.

IDK how we get back to this point but its not likely you get what i am saying, again.

Lol

You are right on this one. Radeon HD 6990 was crippled due to the TDP limitations. As everyone can remember, the Radeon HD 3870X2 had its core clocked higher than the stock HD 3870. So in overall, AMD's smaller die approach worked well.

http://translate.google.com/transla...ulletin/showpost.php?p=3960849&postcount=352'

http://img845.imageshack.us/img845/2889/quemado.jpg

Another one bites the dust.

Damn, that's some serious issues. Noticias3D (News3D) is a famous spanish technology site. I think nVidia pushed the GTX 590 too far.

 

[formulav8]

They are still the same GPUs.

Does the GPU on my 6950 change into a different one when its at 2d clocks?

Yeah but does your 6950 GPU become a 6970 when you flash it. :sneaky:

 

[formulav8]

Another idiot who used the original driver than put 1.2v through the card. I have a feeling the number of websites getting page-clicks from nV GPU reviews is going to be smaller for Kepler.

OK, why would someone be an idiot for using an nvidia approved driver? :\

 

[formulav8]

I think nVidia pushed the GTX 590 too far.

This

 

[hdfxst]

http://translate.google.com/transla...ulletin/showpost.php?p=3960849&postcount=352'

http://img845.imageshack.us/img845/2889/quemado.jpg

Another one bites the dust.

I read he was using the driver that came with the card because the new driver throttled the card to much.

 

[Tempered81]

that's like over a dozen dead cards now

 

[Baasha]

Damn, you guys don't let up do you?

The GTX-590 just came out. It seems to have some serious OC'ing issues but almost all of the cases where the GPU died have been due to over-volting, not OC'ing with stock voltages.

Plus, once the drivers mature for the GTX-590, it should perform well. Perhaps NVidia did skimp out on the PCB and consequently the VRMs, but the card is still an absolute monster and beats tri-SLI GTX-580 when in Quad-SLI mode. That is quite an achievement IMO. At high resolutions (2560x1600), the Quad-SLI GTx-590 equals or bests the Tri-SLI GTX-580.

Proof: http://www.evga.com/forums/tm.aspx?m=925437

The ultimate point is, you guys are way too quick to judge the card. Yes, it could have been as long as the AMD 6990 and had more VRMs to solve the power circuitry issue but let us wait for a few weeks while people are still taking delivery of their cards and see what they can achieve in terms of stable OC's, at stock voltages as well as over-volted.

 

[tincart]

The ultimate point is, you guys are way to quick to judge the card.

I didn't know there were rules for judging video cards. How long should we wait?

Is it okay to judge the 6990 since it's been around a bit longer, or do we have to wait on that one too?

 

[formulav8]

The ultimate point is, you guys are way to quick to judge the card. Yes, it could have been as long as the AMD 6990 and had more VRMs to solve the power circuitry issue but let us wait for a few weeks while people are still taking delivery of their cards and see what they can achieve in terms of stable OC's, at stock voltages as well as over-volted.

Huh? So this stuff happening to the 590's should be viewed as minor and just part of the maturing process? I would definitely tell you no thanks for your advice when I think about buying one. :\

 

[Grooveriding]

The GTX-590 just came out. It seems to have some serious OC'ing issues but almost all of the cases where the GPU died have been due to over-volting, not OC'ing with stock voltages.

Bolded for the issue. The cards are faulty.

Plus, once the drivers mature for the GTX-590, it should perform well.

The 470,480,570,580,590 are all using the same architecture. Any improvements seen on any of the cards coming before will be seen on the cards coming after and vice versa.

Perhaps NVidia did skimp out on the PCB and consequently the VRMs, but the card is still an absolute monster and beats tri-SLI GTX-580 when in Quad-SLI mode. That is quite an achievement IMO. At high resolutions (2560x1600), the Quad-SLI GTx-590 equals or bests the Tri-SLI GTX-580.

Proof: http://www.evga.com/forums/tm.aspx?m=925437

It beats the 580 Tri-Sli by a few frames. After you overclock those setups, and hopefully with neither or both of the 590s blowing up, the 580 Tri-SLI is going to be faster.

That's 3 GPUs faster than 4. I would take the 3 580s in a heartbeat over the two 590s, I think most sensible enthusiasts would as well. The 580s are good high quality cards, the 590s are looking to be turds.

You have 580s right ? Good on you, no need to upgrade :thumbsup:

I think the lack of enthusiasm you see is with new launches we want to see something new. The 590s are slower than the 6990s whether you are using one or two of each.

Not much to get excited about, but the blowing up is pretty interesting I'm looking forward to see how it's handled going forward. If they start a recall or release a re-designed quality card.

 

[Baasha]

I didn't know there were rules for judging video cards. How long should we wait?

Is it okay to judge the 6990 since it's been around a bit longer, or do we have to wait on that one too?

My tendency is to wait about six-eight weeks after the release of the card to judge its true potential. By that time, many people would have received their cards, tried to OC them (with and without over-volting), and we will have some real-world results.

Now, if a MAJORITY of the samples of any GPU is found to flounder, then, it can be unanimously declared as a POS. If the GTX-590s die at STOCK voltages, NVidia will have a serious issue of a massive recall on their hands. The simple fact is that this is NOT the case. There have been a couple that have supposedly died with STOCK settings, but remember, the GPU does not operate in a vacuum. Other components could be at play although the GPU also could be faulty.

All I'm saying is, let us wait for more people to receive their cards, install them, benchmark them, test them thoroughly with a wide range of GPU-intensive applications, and then post the results for us all to judge. At that point, we can make the call on whether the GTX-590 is intrinsically defective in its hardware design.

I think the few reviewers who over-volted their GPUs and posted the videos/reviews online exacerbated the already mind-numbing fanboyism between AMD and NVidia and now, the problem (the VRM/power circuitry) seems to be blown way out of proportion.

As far as the AMD 6990 goes, once people get their waterblocks on them and test them, we will clearly have a winner.

However, at high resolutions and at STOCK CLOCKS, we already know that the AMD 6990 is better than the GTX-590. There is no point harping on that issue ad nauseum.

 

[Baasha]

It beats the 580 Tri-Sli by a few frames. After you overclock those setups, and hopefully with neither or both of the 590s blowing up, the 580 Tri-SLI is going to be faster.

Well, the single GPU solutions always seem to fare better than their dual-GPU counterparts. However, in the case of not having enough PCI-E slots, dual-GPU solutions make sense.

Nobody is saying that the GTX-590 OCs better than the GTX-580 or even equally as well as the 580. If I can get Quad-SLI in 4 PCI-E slots as opposed to EIGHT (which most people don't have), it's a good deal. Plus, the Quad-SLI 590 is about $500 cheaper than the single GPU configuration. Of course, at that point, the price of the cards is hardly the issue. We do have to take into account all the factors though.

You have 580s right ? Good on you, no need to upgrade :thumbsup:

Yes, I was waiting for EVGA to release their 3GB 580s but I have only five PCI-E slots available (one for sound card and one for RAID controller for a total of 7).

I think the lack of enthusiasm you see is with new launches we want to see something new.

Agreed. There was simply too much hype surrounding the GTX-590 and its mythical performance and ownage of the 6990. That didn't happen. On top of that, several of the 590s have died. NVidia is facing a blowback and then some.

If they start a recall or release a re-designed quality card.

That is quite interesting. If they do re-release the card with a longer PCB, for instance, NVidia can redeem themselves but no company in their right frame of mind will admit design defects; that could potentially be a multi-million dollar class-action lawsuit on their hands. As someone else suggested, a non-reference PCB might be the silver lining for the GTX-590. However, I don't see that happening for a card that is supposed to be the "fastest card in the world".

 

[Grooveriding]

Nobody is saying that the GTX-590 OCs better than the GTX-580 or even equally as well as the 580. If I can get Quad-SLI in 4 PCI-E slots as opposed to EIGHT (which most people don't have), it's a good deal. Plus, the Quad-SLI 590 is about $500 cheaper than the single GPU configuration. Of course, at that point, the price of the cards is hardly the issue. We do have to take into account all the factors though.

I figured they would be close to the same price ? Two 590s is about $1400 and three 580s should be around $1500 ?

Or you're referring to buying four 580s ? Well that is madness in my book, lol. But I would buy the 3 580s over 2 590s in a quick fashion.

 

[Baasha]

I figured they would be close to the same price ? Two 590s is about $1400 and three 580s should be around $1500 ?

Or you're referring to buying four 580s ? Well that is madness in my book, lol. But I would buy the 3 580s over 2 590s in a quick fashion.

I was referring to buying 4 GTX-580s (3GB versions). Three of them still will take up 6 slots so might as well fill the damn MoBo and get four!

 

[tincart]

My tendency is to wait about six-eight weeks after the release of the card to judge its true potential. By that time, many people would have received their cards, tried to OC them (with and without over-volting), and we will have some real-world results.

Now, if a MAJORITY of the samples of any GPU is found to flounder, then, it can be unanimously declared as a POS. If the GTX-590s die at STOCK voltages, NVidia will have a serious issue of a massive recall on their hands. The simple fact is that this is NOT the case. There have been a couple that have supposedly died with STOCK settings, but remember, the GPU does not operate in a vacuum. Other components could be at play although the GPU also could be faulty.

All I'm saying is, let us wait for more people to receive their cards, install them, benchmark them, test them thoroughly with a wide range of GPU-intensive applications, and then post the results for us all to judge. At that point, we can make the call on whether the GTX-590 is intrinsically defective in its hardware design.

I think the few reviewers who over-volted their GPUs and posted the videos/reviews online exacerbated the already mind-numbing fanboyism between AMD and NVidia and now, the problem (the VRM/power circuitry) seems to be blown way out of proportion.

As far as the AMD 6990 goes, once people get their waterblocks on them and test them, we will clearly have a winner.

However, at high resolutions and at STOCK CLOCKS, we already know that the AMD 6990 is better than the GTX-590. There is no point harping on that issue ad nauseum.

You're confusing me here. You say you wait to judge a cards "true potential" which suggests to me that when you say people are judging too quickly you mean judging the whole package too quickly. However, you then go on to suggest that this only applies to the exploding card issue and whether or not the card is "intrinsically defective".

To my mind, that is an issue separate from "true potential". A card which explodes does not have any true potential. It would be recalled. I don't judge the performance of faulty products. I write them off.

Then, you start talking about water blocks and overclocking and all that. That doesn't tell us anything about true potential except over a large sample size with consistent cooling setups and similar systems. All a high-end enthusiast overclock tells you is what that particular card can do with that setup. It doesn't tell us anything about the "true potential" of a certain series of cards in general.

In brief: I have no idea what you are trying to say.

 

[krumme]

Hitler does a review of the GTX590.
http://www.youtube.com/watch?v=QFqxn804Vmk

 

[Grooveriding]

Hitler does a review of the GTX590.
http://www.youtube.com/watch?v=QFqxn804Vmk

lol. The 590 debacle has earned its self a hitler meme.

 

[distinctively]

Now, if a MAJORITY of the samples of any GPU is found to flounder, then, it can be unanimously declared as a POS. If the GTX-590s die at STOCK voltages, NVidia will have a serious issue of a massive recall on their hands. The simple fact is that this is NOT the case. There have been a couple that have supposedly died with STOCK settings, but remember, the GPU does not operate in a vacuum. Other components could be at play although the GPU also could be faulty.

From what I have seen, its the VRMs that have been frying. It seems as though a longer board to accommodate more/better VRMs may have helped.

 

[distinctively]

I am actually very curious to see a non-reference design, simply because it has a potential of being MUCH better than 6990, but it will be very costly performance in terms of electric power and cooling.

That would be a very interesting scenario. I wonder though, how it will perform if its being throttled at the higher power draws. If the card is being throttled by the drivers, OEMs might not be able to gain many advantages with custom designs.

 

[3DVagabond]

Remember the people who told you to not unlock the 6950 because it was too risky? Now, same folks are telling you there's no problems with the 590. One of them is even running around betting on O/C performance for the 590.

Of course, any dead/malfunctioning 6950 cards could not possibly been caused by the owner, either. All of these dead 590's? It's user error. Couldn't possibly be a fault with the card.

 

[3DVagabond]

I am actually very curious to see a non-reference design, simply because it has a potential of being MUCH better than 6990, but it will be very costly performance in terms of electric power and cooling.

I don't see the big upside to the 590 over the 6990. At resolutions where these cards are most useful, the 580 doesn't enjoy any big advantage in performance over the 6970. Given the added efficiency of the 6990, I think what we're seeing now could very well be the status quo.