The Fermi Thread - Part 3

[jaredpace]

it168.com - GTX 480/470 review

Edit : Translated version

thanks!

[Chickenfeed]

it168.com - GTX 480/470 review

Edit : Translated version

Very thorough review actually, thanks

[GoldenTiger]

wow first review

http://vga.it168.com/a2010/0326/865/...65774_20.shtml

Meets/beats a bit, but trounces the 5870 in some like Dirt 2, Far Cry 2, Battlefield Bad Company 2, Street Fighter 4 (although that runs so fast it doesn't matter at that point), and Batman... it's hard to tell though from this review as no minimums are provided, and they use PhysX on batman/etc. They also only use DX9 for NFS: Shift unfortunately which makes it hard to tell on that one.

Still, great review/link, thank you! Needless to say, I think we all can be quite happy with the performance!

[Chickenfeed]

Mixed bag, but trounces the 5870 in some like Dirt 2, Far Cry 2, Battlefield Bad Company 2, Street Fighter 4 (although that runs so fast it doesn't matter at that point), and Batman... it's hard to tell though from this review as no minimums are provided, and they use PhysX on batman/etc. They also only use DX9 for NFS: Shift .

Still, great review/link, thank you!

Yeah they forced PhysX on with ATI hardware, I just noticed that... misleading much, same with Dark Void They lost my respect now.

[GoldenTiger]

Yeah they forced PhysX on with ATI hardware, I just noticed that... misleading much, same with Dark Void They lost my respect now.

Still, the non-PhysX game results are pretty darn impressive regardless: the 480 just stomps on the 5870 in the important ones. I think we'll see this play out across reviews...

[Chickenfeed]

Still, the non-PhysX game results are pretty darn impressive regardless: the 480 just stomps on the 5870 in the important ones. I think we'll see this play out across reviews...

Important? Perhaps in BFBC2, I'll give you that but in the rest All TWIMTBP stuff other than Dirt (and I'm greeting this as suspect, thats 2 reviews that show roughly 60fps, I get 70 something with the ingame bench so thats fishy). I always expected them to do much better in those titles in the first place. It goes to show though looking at CoP how much a companys support / intervention can do to the game. Everything would indicate the 480 would do better in Stalker but it doesnt which is quite odd...

[Manicdan]

so it looks like its about 20% stronger on average, but the sway is very massive. loosing in some, and dominating in others. the chiphell numbers seemed very close to this review, so i guess they are accurate. and yes lol at the physx games, thanks for really showing us what the difference is, it was so helpful /sarcasm

[Motiv]

Still, the non-PhysX game results are pretty darn impressive regardless: the 480 just stomps on the 5870 in the important ones.

well it doesn't actually.

between 0-10% in dx9/10 and a wider range with DX11, although it loses in stalker.

hopefully we'll see some cat 10.3 reviews soon, and the entire physx stuff is plainly retarded.

Not sure about uniheaven, would be nice to see dx9, 10 and 11 mark; maybe the previous version also.

What it does show is that in DX11, with heavy tesselation, it is pretty damn fine.

I would certainly bag it over the 5870 for this reason alone.

Saying that, it's 1 review and it seems a little biased with all of the nvidia demos and physx stuff. Hopefully a more legitimate site soon...

[Dalten]

lol @ google's translation of that batman title, "Batman: Forrest Gump madhouse"

[Yvese]

thanks!

I seriously hope other review sites don't force PhysX on ati cards. That's just silly.

Also, were they using 10.3 drivers? Where are the temps/power consumption in that review?

And another thing, those BC2 results look odd. I get higher FPS than 47 avg, regardless of map on max settings, 32 players.

Regardless it looks like it's totally not worth it over a 5870 unless you fold. Hell, even the 5870 isn't worth it. The 5850 is king. It OC's past 5870 speeds and matches the GTX480 for nearly half the price.

[Chickenfeed]

lol @ google's translation of that batman title, "Batman: Forrest Gump madhouse"

Yeah I saw that too lol didn't know wether to laugh or not Those incencitive Asians and their crazy languages!

@ Yvese

They used 10.2 drivers.

As far as BFBC2, without an ingame benchmark, you can't hope to reproduce results so don't put too much weight on it. It is near impossible to measure high player count online games given the performance can vary by a huge amount.

[jaredpace]

cat 10.2 in it168's review

[-Sweeper_]

PhysX finally acceptable with 1 GPU?

[overclock]

Really, in the wake of this review, the GTX480 does not impress me : in 2560x1600, my resolution, and on the most strenuous games (Crysis Warhead, Stalker), the GTX480 and the 5870 are about the same. In Metro 2033, the 5870 results are very low which seems strange : a driver problem ? Indeed, in xbitlabs review, the 5870 was at around 36 fps on Metro 2033.

So, it will depend on the price : 100 $ more expensive than the 5870 and the GTX480 does not worth it

[Motiv]

Really, in the wake of this review, the GTX480 does not impress me : in 2560x1600, my resolution, and on the most strenuous games (Crysis Warhead, Stalker), the GTX480 and the 5870 are about the same. In Metro 2033, the 5870 results are very low which seems strange : a driver problem ? Indeed, in xbitlabs review, the 5870 was at around 36 fps on Metro 2033.

So, it will depend on the price : 100 $ more expensive than the 5870 and the GTX480 does not worth it

Physx...

[tajoh111]

Performance is a mixed bag and the use of physX in the benchmarks is really unprofessional. Should have used equal settings on both and none phsyX. Regardless PhysX games are superior, atleast experience wise on NV hardware, so those games will be a given regardless.

Up and down like crazy those. It looks like a 20% performance gap on average, but directX 11 seems to favour NV this time around. The gtx 470 looks pretty disapppointing those.

[Olivon]

cat 10.2 in it168's review

Yeah, plus PhysX

[HelixPC]

i dont get it, nVidia took 6 months to bring Fermi out, and when its released it does nothing but "tie" the 5870 which has been out for 6 months, seems like ATi is ahead of the game, only if the drivers were released faster with more bug fixes, ATi would be whipping the floor with nVidia. If those numbers are true, nVidia is in alot of trouble, more power than 5870, more expensive, but same performance, gee i wonder which one i would buy

[Motiv]

Performance is a mixed bag and the use of physX in the benchmarks is really unprofessional. Should have used equal settings on both and none phsyX. Regardless PhysX games are superior, atleast experience wise on NV hardware, so those games will be a given regardless.

Up and down like crazy those. It looks like a 20% performance gap on average, but directX 11 seems to favour NV this time around. The gtx 470 looks pretty disapppointing those.

I agree, it looks a good card for dx11 stuff, although as you say rather unprofessional with the cat 10.2's and physx.

[cegras]

if you remove all unplayable results (<30fps) from the charts and keep only the game benchmarks; you end up with ~28.7% increase in performance over HD 5870.

granted that this card is overkill for 1680x1050, and also 1920x1200 for most games, even with 8xAA.

On a slightly off topic note, 'overkill' is a very subjective word. When I bought a 9800 GTX / 4850 for gaming at 1440 x 900, every review out there said I could run every game out at 8xAA if I so happened to want to. It turns out that TF2 has unacceptable minimum dips with 4x AA on both cards - forcing me to choose 2xAA (and even then it gets laggy in a full server on dustbowl). The only card that really let me play smoothly, no matter what, was the 4870 - surely 'overkill.'

I think there's a lot that goes unsaid in the reviews, but I haven't picked up on it yet. I highly suspect it's related to my processor, but meh.

I agree, it looks a good card for dx11 stuff, although as you say rather unprofessional with the cat 10.2's and physx.

The first 10 pages was an opus magnum PR stunt.

[Motiv]

i dont get it, nVidia took 6 months to bring Fermi out, and when its released it does nothing but "tie" the 5870 which has been out for 6 months, seems like ATi is ahead of the game, only if the drivers were released faster with more bug fixes, ATi would be whipping the floor with nVidia. If those numbers are true, nVidia is in alot of trouble, more power than 5870, more expensive, but same performance, gee i wonder which one i would buy

not sure where you say it ties with the 5870. It's quicker that's for sure (in max fps), but with the wrong drivers for ati and an unprofessional inclusion of physx, the results will get closer with other reviews.

[grimREEFER]

basically, it looks like in dx10 the 5870 is better than a 470, but in dx11 its the other way around

[HelixPC]

not sure where you say it ties with the 5870. It's quicker that's for sure (in max fps), but with the wrong drivers for ati and an unprofessional inclusion of physx, the results will get closer with other reviews.

I hope, atleast for nVidia's sake, we dont want another FX 5900

[FischOderAal]

basically, it looks like in dx10 the 5870 is better than a 470, but in dx11 its the other way around

Begs the question what DX11 adds visually? If (right now) developers use DX11 mostly for Tesselation it's clear why GF100 is faster.

[Baron]

Physx...

Which is the main point of Metro is it not.