ASUS GeForce GTX 590 Video Card Review @ [H]

Great review as always guys. I can't wait to possibly see a Quad SLI vs CrossfireX.
 
Hakuren said:
Wait for fully clocked 590 with some non-reference cooling and most importantly 3rd 8 pin PEG connector for additional power.

C'mon MSI give it some TwinFrozrIII treatment! :D
Click to expand...

The Review said:
Worth noting is that all of the GTX 590 cards brought to market will be built by NVIDIA, so you will not be seeing any "special sauce" in these cards. ASUS and EVGA will be exclusively handling sales of the GTX 590 in North America.
Click to expand...
(Emphasis mine)

It doesn't seem like MSI will have the opportunity.
 
XacTactX said:
The true potential of this card won't be known until review sites examine it's overclocking potential. It's quite obvious that the only thing differentiating the 590 from 580 SLI is clock speeds, so if the 590 can reach 580 clocks the 6990 will not be as good. The 6900 tends to gain ~10% of its stock performance after overclocking, the 590, I predict will gain more. In other words, my point is, the 6990 is better than the 590 when both are stock, the 590 will probably be better than the 6990 when both are overclocked (I mean that on-average).

One thing I want to mention, is that you guys neglected to mention the memory bus on the 590. It doesn't just have the shading potential of 580 SLI with 512 shaders, it has a 384-bit bus, it is 580 SLI, except with a muzzle :p.
Click to expand...

Although, I believe overclocking to extreme levels will help the card, in a review like this using surround resolutions I do not believe the Vram on the GTX 590 will allow it to push ahead. Perhaps at 2560x1600 with moderate IQ or below but not in surround. Also it's 384bit bux x 2.
 
This is what I was talking about in my post ^^^. TPU got that on stock volts.

gpuz_oc.gif

perf_oc.gif
 
Zarathustra[H];1037020984 said:
Yeah, I based that comment off of Anands review.

The [H] uses triple head (which is appropriate for these cards) so the results Are going to be different, that being said this doesn't explain the full difference...
Click to expand...

I think Anandtech is still using canned benchmarks and the like as well, and we ALL know that does not equate to real world gameplay.

Derek Wilson, Anandtech – 01/28/08: “ I'll still stand by the fact that it is not necessary to look at gameplay situations in order to build an accurate picture of the relative performance of a graphics card.”
Click to expand...
 
Lord_Exodia said:
Although, I believe overclocking to extreme levels will help the card, in a review like this using surround resolutions I do not believe the Vram on the GTX 590 will allow it to push ahead. Perhaps at 2560x1600 with moderate IQ or below but not in surround. Also it's 384bit bux x 2.
Click to expand...

I don't think VRAM will be a limitation at 57 x 12. 580 SLI has enough VRAM to be the fastest setup available, and it has 1.5GB. I think raw performance is the limitation, and the 590 will gain a lot.
 
I don't see why people would complain too much about this card. it delvers 570 sli performance for about the same price as 570 sli at MSRP would cost. plus you have more vram than 570 sli would offer. it also allows for surround with just one monitor and is not offensive when it comes to noise. not to mention you can still oc for more performance of course. if you want even MORE performance then pay 300 bucks MORE and get gtx580 sli.
 
I had a suspicion it was finally going to be a dual gpu on single card set up, and my other same suspicion held true as well, its nerfed, The nerfing is obvious as their was no way to stick a true 580 in dual onto a single card as it stood, as the power/ heat on a single card no way to maintain it withing specs of usability for a normal system with dual gpu on a single PCB.

take a current 580 card then x2 it and you get something unusable due to power / heat issues beyond what most normal computer systems are built for.

So they took the only options they could to bring it within specs to be normalized for a standard PC that can be sold retail to most endusers. which case we see here the memory and voltage are way lowered to reduce that key point power / heat.

and at the obvious cost of performance as this article implies.

I knew this to be the exact case once i saw the teaser yesterday, even before i knew they had a dual gpu in the works, as any other card would not be so major since nvidia never put out a dual gpu on single PCB in the past.

They basically failed to deliver on performace, but they did deliver on a card for dual in 1 so to speak so they have that now for the 1st time, amd has had them for a while now.
just time will tell if nvidia can make a bin with enough performace in the specs while not going out of range on power/heat that their current gen of cards deal with.

AMD has done this with their cards as their power / heat is within specs and range, amd has dual gpu single cards for a while now, so thier designs are much imporved as time goes on.

nvidia needs a 600 series bin with better power managment smaller die cast to correct this, but not lose the performace in the process, its been the difficulty both vendors are working hard at.
 
XacTactX said:
I don't think VRAM will be a limitation at 57 x 12. 580 SLI has enough VRAM to be the fastest setup available, and it has 1.5GB. I think raw performance is the limitation, and the 590 will gain a lot.
Click to expand...

I dunno I'm fairly certain the AA levels will not be able to be cranked as high in surround but that depends on the game. If the game utilizes a small amount of vram then the GTX 590 will be fine. Either than or they'll scale lower than 5760x1200 like they did in this article. The extra mhz on the core and shader aren't going to do that much to help Vram limited scenarios that can easily present themselves @ 5760x1200 with 4x AA or above.
 
I only read the conclusion but thanks for the insights kyle and brent. Theres a typo in one of the paragraphs "gtx 570 and gt x580"

So I wonder if the price fell decently if u could get a quad 570 setup with two 590s. If nothing else they should have put 3gb or at least 2 gb per gpu on the card.
 
Hey Kyle, great review!

Small typo I found

"We have seen the Silverstone Raven cases quiet the noisy beasts that are the GTX 480 and GT X580." Should be GTX 580 :) Keep up the great work!
 
really though how many [H] readers game at 5760x1200. i think you need to include what he majority of your readers probably game at realistically 1920x1080 , 1900x1200


but good review. it hit right where i expected it to..
 
Grimham said:
I find more value in "Apples to Apples" than "highest playable"

I understand the "real world gaming results" just fine. The problem is what you think is highest playable settings, and what I think are highest playable settings are very subjective. Given the choice of a higher res vs. enabling AA, I will always go with the higher res. AA is the very last thing I would turn on in any game if i found I had the overhead.

Either way, you cover both bases, so great job.
Click to expand...

I understand that well but disagree. for instance my tolerance to lower frame rates in crysis I is Lower (that is I need higher frame rates) then theirs is. but simple bench marks still give me far less of an idea where and what I can do for myself. it is subjective and that is the inherent limitation to real world game reviews. but you simply cannot extrapolate real world from canned benchmarks, there is simply too much that doesn't show.
 
Filter said:
really though how many [H] readers game at 5760x1200. i think you need to include what he majority of your readers probably game at realistically 1920x1080 , 1900x1200


but good review. it hit right where i expected it to..
Click to expand...

Not necessary.

If it provides acceptable framerates at 5760x1200 then you know it will work well at lower resolutions.

Traditional bar charts are kind of dumb really. Who cares if a new video card gets 160fps at 1080p vs the 120fps of the older card. They are both playable and there won't be a notable difference.
 
Last edited:
Zarathustra[H];1037021226 said:
Not necessary.

If it provides acceptable franerates at 5760x1200 then you know it will work well at lower resolutions.

Traditional bar charts are kind of dumb really. Who cares if a new video card gets 160fps at 1080p vs the 120fps of the older card. They are both playable and there won't be a notable difference.
Click to expand...

not true becasue this is showing the 6990 is the winner per kyle at those resolutions. but if you go to the other review sites. the 590 is winning in some of the lower resolutions. so again depends on the game you play. LIke WOW which alot of people play the 590 owns the 6990.
 
Filter said:
not true becasue this is showing the 6990 is the winner per kyle at those resolutions. but if you go to the other review sites. the 590 is winning in some of the lower resolutions. so again depends on the game you play. LIke WOW which alot of people play the 590 owns the 6990.
Click to expand...

Then your gripe really is with the games used for testing, not with the test methodology :p

And - quite frankly - as log as you are getting min framerates in the 30-60fps range it doesn't matter which card "wins". That's why tests at lower resolutions than the max playable are a waste of time.

This is not an FPS competition, it's a "can you play a game at these settings or not" competition.

Otherwise you've got the lame reviews where they claim one card is better than another because it got 270fps in Quake 3 instead of 250fps...
 
As long as your minframerate is above 60, it doest matter if you get a bagillion FPS or if you get 70 FPS. It's still a tie. It can play the game.
 
Zarathustra[H];1037021329 said:
Then your gripe really is with the games used for testing, not with the test methodology :p
Click to expand...


i would say resolution falls under test methodology then games tested.

Zarathustra[H];1037021344 said:
As long as your minframerate is above 60, it doest matter if you get a bagillion FPS or if you get 70 FPS. It's still a tie. It can play the game.
Click to expand...

no cause 60fps can be 30fps real quick.
 
Hakuren said:
Yes I do need SLI on one card. With couple RAID cards it is the only option really. And because I find whole 2/3/4 SLI[CF] with 2-3-4 cards idiotic.

Honestly I'm not surprised at all with reference GTX590, I was expecting that from the start. It was obvious that both chips and mem must be seriously under clocked to keep the power drain within some acceptable levels (<-375W). Wait for fully clocked 590 with some non-reference cooling and most importantly 3rd 8 pin PEG connector for additional power.

C'mon MSI give it some TwinFrozrIII treatment! :D
Click to expand...
I find it a bit ironic that you find 2 video card setups "idiotic" but you have multiple RAID cards in one box. :p

You would fit the ideal for this card then, but you are far from the norm.

Also, unrelated but what do you do with your system that you need both SLI GPUs and multiple RAID cards?
 
Filter said:
not true becasue this is showing the 6990 is the winner per kyle at those resolutions. but if you go to the other review sites. the 590 is winning in some of the lower resolutions. so again depends on the game you play. LIke WOW which alot of people play the 590 owns the 6990.
Click to expand...

A-W-E-s-o-m-e dude!

Uber-niche twitchers like WOW are THE reason why Nvidia RULZ and AMD sucks mario's balls!!
 
Filter said:
not true becasue this is showing the 6990 is the winner per kyle at those resolutions. but if you go to the other review sites. the 590 is winning in some of the lower resolutions. so again depends on the game you play. LIke WOW which alot of people play the 590 owns the 6990.
Click to expand...

Uh... 6990 gets roughly 80-90fps at 1680, 1920, and 2560 (all at 8xAA). GTX590 gets between 105-110 at those same resolutions. Yea, 590 'owns' 6990 there, but what's the difference? You're maxing out settings either way, and both are far above 60fps.

Also, those numbers aren't changing much between the different resolutions. Seems like something else is holding things back.

So, yea, ok, 590 is faster at the lower resolutions, but both are already so fast it doesn't really matter.
 
I have to admit, I expected a bit worse, at least it is competitive on the performance side and it is quieter. But seems that the GTX 590 is already pushed to the limits in terms of power envelope and circuitry, which means that the HD 6990 still having a bigger margin for overclocking and better performance overall. This is the first time ever that the two GPU vendors are evenly matched with their top dual GPU solutions.

I think that the 40nm process can't give anymore. It affected both GPU vendors, thanks to AMD smaller die approach, they had lost the performance per watt efficiency compared to the HD 5x00 series, but they were able to fit two big GPU's on a reasonable TDP, nVidia couldn't get any better with the GTX 590 which at its limits just to keep up with the stock HD 6990 which has a bigger overclocking margins and hence, better performance.
 
Pyroja said:
Uh... 6990 gets roughly 80-90fps at 1680, 1920, and 2560 (all at 8xAA). GTX590 gets between 105-110 at those same resolutions. Yea, 590 'owns' 6990 there, but what's the difference? You're maxing out settings either way, and both are far above 60fps.

Also, those numbers aren't changing much between the different resolutions. Seems like something else is holding things back.

So, yea, ok, 590 is faster at the lower resolutions, but both are already so fast it doesn't really matter.
Click to expand...

dont know what review you looked at but it was 121 to 151fps. and thos FPS can dip real low when raiding.

http://www.techpowerup.com/reviews/ASUS/GeForce_GTX_590/18.html

wow_1920_1200.gif


wow_2560_1600.gif
 
Just something to point out- The stock clocked 590 is using ~ the same power as the overclocked BIOS switch setting uses on the 6990. I find that pretty telling about how the 2 companies measure the TDP. The 590 claims 365W, and if AMD had used the same way to report their TDP, they could have actually run 2 full clocked 6970s as the stock setting rather than the OCed mode! Props to AMD for amazing efficiency.
 
Zarathustra[H];1037021329 said:
Then your gripe really is with the games used for testing, not with the test methodology :p

And - quite frankly - as log as you are getting min framerates in the 30-60fps range it doesn't matter which card "wins". That's why tests at lower resolutions than the max playable are a waste of time.

This is not an FPS competition, it's a "can you play a game at these settings or not" competition.

Otherwise you've got the lame reviews where they claim one card is better than another because it got 270fps in Quake 3 instead of 250fps...
Click to expand...

The problem is in the definition of "max playable" - different people have different opinions of what is playable (or more correctly, enjoyable). Also, just because a card runs High settings at 2560, doesn't necessarily mean it'll run Very High or Ultra at 1920 (for example). So there is a place for both kinds of testing.
 
Just went off the first thing that came up in google. Interesting that these guys are getting significantly lower FPS.

WoW%201680.png

WoW%201920.png

WoW%202560.png
 
Hakuren said:
Yes I do need SLI on one card. With couple RAID cards it is the only option really. And because I find whole 2/3/4 SLI[CF] with 2-3-4 cards idiotic.

Honestly I'm not surprised at all with reference GTX590, I was expecting that from the start. It was obvious that both chips and mem must be seriously under clocked to keep the power drain within some acceptable levels (<-375W). Wait for fully clocked 590 with some non-reference cooling and most importantly 3rd 8 pin PEG connector for additional power.

C'mon MSI give it some TwinFrozrIII treatment! :D
Click to expand...

sadly, but all Gtx590 will be made by nVidia, so expect no Glh,Gs, twin snizzler editions. Aib's will only deliver slightly oc'ed reference cards.

(And maybe possibly an uber psycho ARES or whatever, but that is an other league)

NB

Another site, this time romanian, had their 590 die on them:
Article, read conclusion
 
Last edited:
Pyroja said:
Just went off the first thing that came up in google. Interesting that these guys are getting significantly lower FPS.
Click to expand...

to me looks like hes cpu limited. could be wrong. since the 480 sli is up there.
 
OMG here we go with how does it do at 1680x1050 or 1920x1080. If you SPENT $700 on a video card you DAMN well have spent 2x that on your monitor. IF your still playing on somthing you got from walmart Get your ass off the [H] forums.
 
Can this fit into a netbook? I play competitive Angry Birds, and any dip below 500 FPS could cause me to lose a crucial point.
 
Makeroflostsouls said:
OMG here we go with how does it do at 1680x1050 or 1920x1080. If you SPENT $700 on a video card you DAMN well have spent 2x that on your monitor. IF your still playing on somthing you got from walmart Get your ass off the [H] forums.
Click to expand...

A lowly 550 Ti + 2560 is worse than 590 @ 1280 ... :D
 
You must log in or register to reply here.
Back
Top