ASUS GeForce GTX 590 Video Card Review @ [H]

Wow this is rich, do you really want us to pull up some of your Caymen related posts?

Pull up whatever posts you want, Here is what I have said, nothing is fanboy stuff

1. Cayman Crossfire X hot and loud
Substantiated by 3 different reviews, this was a informative thread showing people the problems that can occur if you sandwich your 2 cards together in many cases without improving the airflow.

2 AMD is definitely cheating with catalyst degradations
Confirmed by guru3d

Since then I have bought a pair of Cayman GPU's and love them, I avoided sandwiching them because of the valuable information the 3 reviews showed, moreover AMD has removed the degredations in the default setting and I even stuck up for them.

I have also been very vocal about the limited Vram on the nvidia cards and them being overpriced, I also spoke negatively about nvidia and its greed when GTX 560Ti and 550Ti have launched as they were both overpriced.

The only party I'm guilty of being a fanboy of is my fellow forum peer. I always get shit from either side, I've been called an nvidia fanboy, an AMD fanboy.

Any thread you pull my post from I challenge you to post your own comments from the same thread too and lets see who the community thinks is really the fanboy after that.
 
How much faster would the 590 be than a 580 with a 25% overclock at 1920 x 1200?

At that resolution the 580 is all you need. If you plan to get a bigger monitor or go Surround, you might want the 590.

Save 200 bucks, unless you need a really big box and a new T-shirt. ( oh, I forgot the cool poster, but I think you get one of those with the 580):D
 
Both 590 and 6990 feel like an lazy effort probably due to no longer being limited by TDP.

5970 was impresive engineering effort those two are not.

Apart from boinc/folding/Cuda crowd i don't see much reason for those cards.
 
Pulled latest version of afterburner, will test in a bit. You guys think I'd be safe at 1000mV?
 
I'm at 975mV - 800core/1600shaders/1975RAM stable so far. auto fan puts me at 87C, max fan puts me at 75C.

Still testing
 
975mW
850core
1700shader
2000 memory

...was too much

returning to 975mW
835core
1670 shader
2000 RAM

....going to see if maybe RAM can go any higher at this voltage. not too interested in getting much higher in any category, especially if i have to go much higher on voltage. i have no intention of nuking my card
 
sabregen are you seeing throttling from OCP ? quite a few people on another forum are making a big stink about it cause it was shown in linus' youtube benchmarking video. there's even someone saying that anything over 700mhz on the gtx 590 is going to be riddled with OCP.
 
I'd be interested in seeing some benches of this thing at 772Mhz compared to a GTX580 SLI.

I am very curious to see what - if any - differences there are between this single board config and a dual board config at the same specs.
 
I'll be honest, up until getting the 590 I haven't OC'd a GPU really since my 260. Since then I've had an XFX 5850, then a second for Crossfire, then a 470, then 580 and now the 590. OCP and the idea of it is pretty new to me, and I don't profess to understand it fully, either.

Right now I'm just running Furmark and pushing to get to display driver crashes, so i know where my limit is. at 975mW, 835 core causes a display driver crash, and 2025 or higher on memory does the same.

Currently running Furmark with:

975mW
825core
1650 shader
2000 RAM
Fan on Auto

Going to see if it can go an hours or more at these speeds.
 
" With a card like GeForce GTX 590, that has so much latent performance, I couldn't wait to try bumping the voltage to see how much I could gain from it.

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.

So he thought he knows how OCP/OVP works on a brand new design /sigh

W1zzard, next time just RTFM.

An extract from ASUS’ GTX 590 reviewer’s guide:

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.

Indeed. :D
 
WTF is the point of running Furmark when OCP detects it and throttles down if power limit is exceeded?

Disable vsync and run Crysis, Metro 2033, 3DMark or any other game that utilizes your GPU 99%.

Go up slowly OC-ing and make sure you're getting more FPS by overclocking,

because OCP will kick in with latest drivers even in ordinary applications/games, therefore degrading your performance.

@ Rossi~ :p
 
I'll be honest, up until getting the 590 I haven't OC'd a GPU really since my 260. Since then I've had an XFX 5850, then a second for Crossfire, then a 470, then 580 and now the 590. OCP and the idea of it is pretty new to me, and I don't profess to understand it fully, either.

Right now I'm just running Furmark and pushing to get to display driver crashes, so i know where my limit is. at 975mW, 835 core causes a display driver crash, and 2025 or higher on memory does the same.

Currently running Furmark with:

975mW
825core
1650 shader
2000 RAM
Fan on Auto

Going to see if it can go an hours or more at these speeds.

i would strongly recommend you up the fan speed if you can. i'm working with a few people as we speak that are currently checking to see if the ocp is a heat issue or volt issue. someone i know has the OCP coming in on stock volts and core clock as soon as the card reaches 70*. running on auto fan. there's no proof yet but someone else with the watercooled 590 is saying that in his test so far at 840 he hasn't seen any throttling that he's noticed. he could be overlooking it/not knowing what to look for, but it could just be due to the heat.

would definitely explain why nvidia made sure to have waterblocks from not just evga(swiftech), but EK, dangerden, and koolance out on launch day. although koolance won't be out till the 31st. still alot sooner than normal
 
" worthless oc " without benches.

I bet that you'll be frustrated with the 800Mhz core results ...
From all the oc i saw to that card ... they fail to have a substantial increase, since OCP will kick in in no time and you'll find yourself having worst results that you got max oc with default voltage ;)
 
yeah i meant mV...sorry, a bit hungover. I'm going to consider this stable (3hours and 15 minutes):

Capture-4.jpg


more testing in a few.
 
WTF is the point of running Furmark when OCP detects it and throttles down if power limit is exceeded?

Disable vsync and run Crysis, Metro 2033, 3DMark or any other game that utilizes your GPU 99%.

Go up slowly OC-ing and make sure you're getting more FPS by overclocking,

because OCP will kick in with latest drivers even in ordinary applications/games, therefore degrading your performance.

@ Rossi~ :p

Working on it.
 
40fps?
My 460 stock does 65. So at least you know that OCP is working in Furmark :p

Why not dial back default voltage and find highest ingame usable non-overvolted clock that returns meaningful FPS gain.
 
OC settings for all test: 975mV, 815core / 1630 shader and 2000 memory. I had to dial back the GPU OC once I started testing, from 825.
afterburner.jpg


Futuremark apps... well I don't put much stock in them anyways, so what I'm about to say is not a real great loss. 3dmark 11 and Vantage were both very picky when the card was OC'd. As a matter of fact, I couldn't get the apps to even complete the first test with ANY kind of OC. Since I got everything else I tested to run, I'm going to just chalk it up to these being canned synthetic benches that i don't really give a shit about anyways. 3dmark 06 was run at complete stock settings, and only the default tests were run.

3dmark 2006 default clocks:
default.jpg


3dmark OC clocks:
oc.jpg



... and NO, those aren't accidentally reversed. The OC score IS lower.

Company of Heroes
settings:
settings.jpg


stock clocks:
resultsdefault.jpg


OC clocks:
resultsoc.jpg


What's interesting here isn't the additional 0.5FPS that the OC gets you, it's that the lowest FPS score goes up considerably, which equate to a smoother gameplay experience. This test was run 3x at each setting, because I couldn't believe that this wasn't a fluke...these results held up each time.

GTA IV

settings:
settings-1.jpg


stock clocks:
resultsdefault-1.jpg


OC clocks:
resultsOC-1.jpg

Resident Evil 5 DX10 Fixed

settings:
settings1.jpg

settings2.jpg


stock clocks:
resultsdefault-2.jpg


OC clocks:
resultsoc-2.jpg

Street Fighter IV

settings:
settings1-1.jpg

settings2-1.jpg


stock clocks:
resultsdefault-3.jpg


OC clocks:
resultsocd.jpg

Unigine Heaven DX11

I also could NOT get this benchmark to run with any amount of OC on the card, so here are the settings and the stock clocks results

settings:
settings-2.jpg


stock clocks:
resultsdefault-4.jpg
 
Last edited:
thanks sabregen. :) seems the oc works in games but not in benches? odd. even when it didn't pick up the average it picked up the min fps. looks like a driver issue.
 
Last edited:
sabregen can you do me a favor and test the card at a lower clock like 660-700 to see if it isn't throttling? someone else is saying that keeping it below 700mhz seems to not show throttling.

i think it's drivers but it would definitely rule out throttling this way
 
Man... Those are ALL "wrong" benchmarks.
Except Heaven. Provided that's extreme tess.those should be good runs,
but then again it's a Demo so I dunno.

CoH and GTAIV - just forget it.
3DMark 2006 - 25k? again too old and CPU sensitive. Seems LOW.
Dunno about RE5

Don't you have Metro, Crysis, Lost Planet 2, Just Cause 2, Call of Pripyat, Alien vs Predator, Vantage, 3DMark 2011...
 
Try using OCCT GPU Error test, it'll run it at 99% gpu usage, heat it up to it's max and show any sort of instability, it'll flag an error.
 
The sweet pot is the 700Mhz +- overclock. Anything higher will just result on OCP kicking over and over again ... and you get a weird performance ... or lack of it ... :x
 
How much faster would the 590 be than a 580 with a 25% overclock at 1920 x 1200?

25% faster than stock 590 clocks? That's achieveable. 25% over stock 580 clocks? You'll never get there. The RAM slider on Afterburner tops out at 2053MHz, you'd have to get close to 2500MHz to get to 25% overclock from stock 580 speeds. I dont think that the GPU core will hold either... you're looking for 910+ on GPU core, and the 590 just doesn't have it in it... I'm finding with a medium voltage increase youre going to be about 100 MHz lower.
 
CoH and GTAIV - just forget it.
3DMark 2006 - 25k? again too old and CPU sensitive. Seems LOW.
Dunno about RE5

Don't you have Metro
yes

yes

Lost Planet 2
no

Just Cause 2
yes

Call of Pripyat
yes

Alien vs Predator
no

Vantage, 3DMark 2011...

yes, but they wont run with ANY kind of OC for some reason


...and I'm out of town for the next week and a half for work, and unlikely to install games just to test my OC (all of the ones that I've listed I have... I have already beat, and therefore uninstalled them).

I dont think I'm going to se doing too much more testing other than possibly doing a few runs at 700MHz core... which seems retarded. if OCP kicks in at only ~80MHz Above stock, that's a pretty ridiculous decision on Nvidia's part.

Looks ike AthlonXP got one too. Maybe he'd be up for installing a few games. My limitation is operating in a mini-ITX chassis and only having the 120GB SSD. Space is at a premium for me. As soon as I'm done with a game, I uninstall it.

I do not agree with your GTA IV assessment though, that game is a system killer on PC. It could just be poorly coded...hellif I know. All I know is that on my x3440 and GTX470, my rig was on its knees.
 
3 drivers in 5 days means there's definitely driver issues to sort out with this card but at least nvidia is on it.
 
Look what Nvidia did with the voltage control! The previous auto setting on voltage control got the card up to 0.912mV. They've now capped it @ 0.900mV

Capture-5.jpg


gonna run some 700core/1400shader/2000ram benches
 
... hmmm remember to remove your old drivers before installing new ones ... these drivers are the worst launch of drivers i ever saw ... they send 1 to correct another, to correct another ... so after installing so many drivers you can end having some conflicting files ...

Those 900mv are hard to believe... even your vid will get higher than that in load
 
I uninstalled, ran driver cleaner, and did a clean install option in the nvidia drivers. they're definitely limiting us with this release. I even started MSI Afterburner, unchecked voltage control, closed down, reopened, rechecked, closed down and reopened again just to be sure. no change.

Also, OCCT GPU test is only loading each GPU 50%. So much for THAT test.

weird though... in the Afterburner program, I can only crank it to 0.900mV... but look at what's reported by GPU-Z Sensors:

Capture-6.jpg
 
Well no, I think they just limited the volts you can use.

But look at his afterburner screen and it's core/shader and memory clock, then look on teh right to GPU-Z. It looks like the new driver is forcing you to stock speeds, but it could just be reading it wrong... @Sabregen are your scores consistent with capped clocks?
 
Back
Top