Palit GeForce GTX 470 Dual Fan

[W1zzard]

Palit's GeForce GTX 470 is one of the GTX 470 versions that comes with a custom PCB design and non-reference thermal solution. Since the card is $20 cheaper than the reference design we take a close look to find out whether Palit cheaped out on their card or if it's worthy for consideration.

Show full review

 

[wahdangun]

wew, thanks palit you make it more hot, more power hungry and more noisy.

well at least, you make it $20 cheaper

 

[GSquadron]

Dont forget that it makes the card a great overclocker.

 

[W1zzard]

wew, thanks palit you make it more hot, more power hungry and more noisy.

well at least, you make it $20 cheaper

did you even read the review ?

 

[LAN_deRf_HA]

Seems slightly better in all regards, and cheaper... so this will probably never be in stock or get price jacked due to high demand.

 

[Bjorn_Of_Iceland]

Wow, I just saw this on the Palit site, and Wiz got a review of it.. very fast ^^. Cool looking card. Good review

 

[araditus]

I recently have been laughing at all of the green camps offerings as of late, as I have been chillin in the red camp for the 5000 series enjoying the price, low heat and low power consumption, but I had to applaud this card for the managable power consumption, slightly better performance than its competitor (5850 execpt for at resolutions above 1080p) in that price range, however I feel that the true shine of this card is the 20% overclock, very nice for a GPU and by the benchmark, at those speeds its sniffing around at the 5870 territory of performance.

I do have 1 request wizz, what was the power numbers on the overclocked cards, some hardware is exponential it seems when the stock is increased, some are linear and some are less than expected, I feel that if you can get away with the 20% overclock, its stays cool, and the power numbers dont sail away, this might be the best thing nvidia has out right now

 

[W1zzard]

what was the power numbers on the overclocked cards, some hardware is exponential it seems when the stock is increased

sorry i don't have that number, might be something worth looking into for future reviews. but how do you define power draw when overclocked ? idle ? load avg? load peak? furmark?

 

[newtekie1]

Sweet, does this mean we might see an SLi GTX470 review!?

Please please please please please please please please please please please...

sorry i don't have that number, might be something worth looking into for future reviews. but how do you define power draw when overclocked ? idle ? load avg? load peak? furmark?

I would say just grabbing load avg and load peak would be good since those are the most relevent. Idle should be the same thanks to the clock profiles switching, and furmark power usage isn't really anything you would see under "normal" use.

 

[GSquadron]

2 GTX470 = 5850+5870 (simply put them in a hydra chip mobo, like crosshair extreme IV )

 

[araditus]

sorry i don't have that number, might be something worth looking into for future reviews. but how do you define power draw when overclocked ? idle ? load avg? load peak? furmark?

Same as what newtekie descirbed, the avg and peak when running 3dmark, just so that you can get an idea of what power requirements you need if you wish to get 1 or 2 of these cards, and the power supply needed.

or since you seem to always run the overclocked card in COD4 if you already have the power measuring equipment hooked up the difference when playing a game, whatever is convient for you.

I also think that since in this case the overclocked card was only 1 degree celsius higher than stock clock, if you were to find that the power draw increases to about iunno for numbers sake, 150 to 225 yet only saw 1 degree temp increase, in your conclusion you could now emphasize how good the non reference cooler is.

 

[LAN_deRf_HA]

How exactly do you figure out the highest memory clock? You'd need a very consistent benchmark to notice when the performance starts dropping due to errors.

 

[newtekie1]

How exactly do you figure out the highest memory clock? You'd need a very consistent benchmark to notice when the performance starts dropping due to errors.

I'm pretty sure the artifact scanner gives a consistant enough framerate to tell when the card becomes unstable.

 

[xtremesv]

Reading all those reviews about non-reference GTX470's, I notice how they're getting better and better. In essence (watts apart), GF100 is a very good GPU mounted in not so good boards.

With a good PCB design and cooling solution, Nvidia could bring a powerful dual GF100 card to compete with 5970.

 

[BUCK NASTY]

Wow, great review W1zz. I'm holding out for the right GTX470 and this may be it. Now if I could only find it in stock. Anyone seen it available online yet?

 

[wpupkin]

Metro 2033

Why don't you use Metro 2033 for test? NVidia has the significant advantage bei Tessellation. It would be fair to use this game for Review. The game optic is the best one for now.

 

[W1zzard]

Why don't you use Metro 2033 for test? NVidia has the significant advantage bei Tessellation. It would be fair to use this game for Review. The game optic is the best one for now.

looks worse than crysis and is slower, and worse programmed. but i'll use it after the next rebench

 

[REALIN]

looks worse than crysis and is slower, and worse programmed. but i'll use it after the next rebench

But it has DirectX 11 and awesome lightning effects!

 

[erixx]

dark games suck in any respect and if the engine is worse, get back to crysis beach summertime!

 

[wahdangun]

did you even read the review ?

yes, wizz, i read it, why you asked it?

Why don't you use Metro 2033 for test? NVidia has the significant advantage bei Tessellation. It would be fair to use this game for Review. The game optic is the best one for now.

no, don't metro, it's use physix crap by default and can't be changed. it's unfair for ati card

use AVP 3, Dirt 2, or better yet use BF :BC2 if you want to see DX 11 performance

 

[$immond$]

2 GTX470 = 5850+5870 (simply put them in a hydra chip mobo, like crosshair extreme IV )

I call B.S.

Have you read any of the reviews of a 5870 + 5850 crossfire?

http://www.legitreviews.com/article/1107/1/

http://www.overclockersclub.com/reviews/xfx58705850/

Very Sub-par performance for the cost.
How can you base an assumption about performance involving 2 Nvidia cards with 2 different ATI cards?
For one the architecture is completely different and a single 470 GTX performs at a level inbetween a 5850 and 5870.

 

[Bl4ck]

Seems to me like i have a good card to replace my "old" GTX275, marginal differences in power usage but faster card , only thing is that at that the end user probably never will see a good price for these cards.

 

[Wile E]

Hey w1z, do you think you could use a different test to show OCed performance differences. COD4 isn't very strenuous on the GPUs anymore. Maybe Metro or even Crysis?

Still, excellent review as always, w1z.

 

[DanishDevil]

Here's the first one.

 

[newtekie1]

yes, wizz, i read it, why you asked it?

I'm pretty sure he is asking becuase the only accurate thing out of your statement was that it is more power hungry. It runs cooler, and there is no noise difference.

no, don't metro, it's use physix crap by default and can't be changed. it's unfair for ati card

The PhysX in Metro 2033 that can't be disabled runs on the CPU, regardless of the presents of an nVidia or ATi card. So it is not unfair to ATi cards.

use AVP 3, Dirt 2, or better yet use BF :BC2 if you want to see DX 11 performance

AVP maybe, but Dirt2 and BC2 aren't DX11 titles...well they say they are...but they aren't, they are DX9 titles with a few DX10 features added, and tessellation added to one or two insignificant objects.