GPU Value Comparisons

Amazing work will definitely find this useful.

Just wondering though, how accurate the data truly is. I mean the GPU is not the sole factor in determining fps. Unless these stats are taken from similar system specs?

Though from the quick look that I've had so far, the fps scaling seems spot on.

Also would like to know if the cheaper brands are worth it(VTX3D, innovision) for example a asus HD 6850 or a VTX 3D HD 6870 for around the same price, better value?
 
The system specs & drivers used plays a major role in the determining framerate! I mean, my friend's Core 2 Duo E6750 wasn't even able to make use of 75%+ of my GTX 460 @ 725MHz!

Usually the sites that do these benchmarks use an oldish system that is heavily overclocked for all their benchmarks, so that they have the same reference throughout all their benchmarks.
eg. New AMD HD 6950 2GB vs old AMD HD 5850 test setup.
I'm not sure if their performance summaries just used the results of the previous setups, because then you'll see that the hard drives, power supplies, drivers and operating systems actually differs!

If VTX3D & Inno3D has the same reference design as the Asus/Gigabyte ones, then they do offer better value if they're quite a bit cheaper. The drawbacks are often shorter warranty periods, less accessories, old/faulty/no drivers on the CD and poor support with the cheaper brands.
I had issues with my Inno3D GTX 460 768MB card and it took about a month (or even more) to get it sorted out, where as with Gigabyte and Asus, which has local support, it would've probably just taken 1 week for replacements.

You can always contact like Wootware and see how good VTX3D's RMA/support is. Last when I asked them, which was sometime last year, they haven't had any returns on any of their VTX3D cards.
 
Thanks pada that explained a lot, but does this mean that oc will be the same for identical reference cards or will it differ depending on the brand and features.

Could I push higher clock speeds or would it simply run at cooler and more stable temps if anything at all

(Sorry for all the nooby questions)
 
Cant remember about all three sites but Guru3D uses a 6 core i7, so CPU does not play a factor at all, the FPS is purely GPU dependant. And since each site uses the same rig all the time the % differences should always be the same.

WRT to OC, stock cards will all OC roughly the same, only differences in the chip itself will make a (small) difference. Non reference models will OC differently.

Everyone else - will update charts when the 590 launches, simply been too busy
 
i wonder what brands was used, as that plays a big role as well, for instance, for those interested in some FYI... i have a gainward 6800GS 512md card, and every now and again i go and look for a better card for replacing it, (dont really need to as i have not found a game that i cant run full settings) and end up with having to pay more than a R1000 more than what i paid for mine, my price paid, R2000
 
6800GS? What games are you playing that you can run at max settings?

Any card for R1000 would beat your card. Badly.

Brand only matters for warranty, doesnt matter otherwise. Some companies use their own coolers which can add value or overclocking headroom though.
 
yodaspike said:
i wonder what brands was used, as that plays a big role as well, for instance, for those interested in some FYI... i have a gainward 6800GS 512md card, and every now and again i go and look for a better card for replacing it, (dont really need to as i have not found a game that i cant run full settings) and end up with having to pay more than a R1000 more than what i paid for mine, my price paid, R2000
Click to expand...

The brand makes zero difference if they are reference designs. The only time there is a difference is if it is a special OC edition and afaik everything in the charts are standard non-OC editions
 
perhaps, but i have before had 2 identical machines, only difference was the brand of the Graphics card, and difference in FPS was more than 20.
 
yodaspike said:
perhaps, but i have before had 2 identical machines, only difference was the brand of the Graphics card, and difference in FPS was more than 20.
Click to expand...

Sorry then there was another difference you were unaware of. Definitely.
 
+1 to what Ancalagon said.

Perhaps you had a GS version from the one brand and an Ultra version from the other one, or perhaps an LE version with unlocked shaders.

... now I know why yodaspike said that Win7 x64 was so slow :p
 
that was running a different machine and card.
any way, i even tried sparkle at one time a think it was a 6600 and it was worse than the 5900 i had before, and it should of been much better
 
very true, but still, its sad that a later much later model would be worse than the old one, and this would not be the first time i have come across a card that is quite old, but still faster than cards much newer. BUT the trick is, the core speed and ramdac speed and amount of ram, does not play such a big role as what bit the card is.
64-bit sow, 128-bit better/ok 256-bit good gaming card, anything above that, expensive, yet awesome.
 
Ancalagon said:
6800GS? What games are you playing that you can run at max settings?

Any card for R1000 would beat your card. Badly.

Brand only matters for warranty, doesnt matter otherwise. Some companies use their own coolers which can add value or overclocking headroom though.
Click to expand...

well, to answer this, F.E.A.R 1 and 2, crisis, new assassins creed as well, mmm, and with only 1 gig ram
 
P924 said:
At what resolution? 1024x768 or 800x600? :D
Click to expand...

more like 1x2 :p There is simply no way in hell that a 6800GS is powering Crysis 1 at max everything at high resolutions with FPS over I'd guess 5
 
Last edited:
Well, I suppose some people are happy running at 3fps...
 
resolution was 1024 perhaps higher cant remember, and everything ells was on max, i get all of yours disbelieve, many people do the same until they see it, now all of them want my card, and that is not happening.
 
yodaspike said:
very true, but still, its sad that a later much later model would be worse than the old one, and this would not be the first time i have come across a card that is quite old, but still faster than cards much newer. BUT the trick is, the core speed and ramdac speed and amount of ram, does not play such a big role as what bit the card is.
64-bit sow, 128-bit better/ok 256-bit good gaming card, anything above that, expensive, yet awesome.
Click to expand...

Its not as simple as that. Yes, memory interface matters, what really matters is the bandwidth, which is determined by both how wide it is in bits, but also the clockspeed and the memory technology. My 128 bit 5770 will beat your 6800GS quite badly in every single benchmark, for instance, even though it is only 128 bit and your card is 256 bit. It even beats my old 8800GTS, which had a 320 bit interface.

Core speed does matter, but again only in combination with the amount of ROPs, TMUs, shaders etc.

yodaspike said:
well, to answer this, F.E.A.R 1 and 2, crisis, new assassins creed as well, mmm, and with only 1 gig ram
Click to expand...

You play Crysis with a 6800GS? Wow.

Seriously though, a R1000 card would embarrass your card. Not being funny, it really would. Unless you have an 8800GTS or something?
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X