The Fermi Thread - Part 3

[annihilat0r]

well 90% of them are Nvidias TWIMBP titles. I'm hoping that there would be more titles that are a little bit more neutral.

I'm not saying people aren't playing far cry 2 at the moment, or crysis warhead, which they probably aren't, but it just looks a little odd having so many titles from nvidias PR dept in.

yeah but most big PC titles are TWIMTBP anyway. to do a review, you don't pick "neutral" games, you pick popular titles that people are playing, because people don't care about how their cards perform on a "neutral" title, they care how they perform on a game they actually play.

[SocketMan]

In case it's been pulled or no one can get onto Hexus, I've been fast and uploaded everything.

Many thanks!

[Manicdan]

150W extra so if I game at 100% GPU speed 24/7 for a month my electric bill will increase $4-$5..... So realistically I'm looking at 0.25-0.50 increase per month. I think I have this covered.

i did the math a few pages back, its alot more than you think if your gaming 24/7. try 15$ a month

[tajoh111]

One more thing we can look at is that Fermi actually beats its old generation dual card upon launch. Something that the 5870 failed to do in regards to the 4870x2 and is why I considered it a lackluster card upon launch.

Although the 5870 is likely faster now(I am guessing here) than the 4870x2, if we are looking at a strictly performance perspective(not in power, which it is a huge failure), the gtx480 has the potential to look more attractive from a performance perspective in the future.

The 5970 might be a better performer, but like all cards it has dual card issues, e.g profiles, inconsistent performance, microstutter etc. Its one of the reason I want to move away from my current setup. If Fermi pickss up say 15% percent more performance with drivers, it will be in spitting distance of the 5970. Enough atleast that being a 100 dollars cheaper and being a single chip(although with 10-15 percent less performance) would make it the better buy.

[eric66]

WiC, RES 5 and BAA are included in most of the reviews done before so I don't know why not in this one?

Which other games that are much more popular than the titles in that benchmark would you recommend, instead of just general bashing?

cod battlefield bioshock 2 mass effect 2 dragon age assassins creed ... well there are tons of games around

[Motiv]

yeah but most big PC titles are TWIMTBP anyway. to do a review, you don't pick "neutral" games, you pick popular titles that people are playing, because people don't care about how their cards perform on a "neutral" title, they care how they perform on a game they actually play.

perhaps cod4:mw2, dow2, bc2, avp, bioshock 2, lfd2

[Chickenfeed]

Why does it seem everyone is using the same benchmarks. We could at least see cod:mw2, dow2, team fortress 2 (and oldie). it just seems everyone is benching the same.

Good to see just cause 2, it seems a little different to other reviews so far.

Totally agree. I'm looking forward to see JC2 results. I don't believe it has been benched yet. It has 3 built in benchmarks (desert, jungle/rain, city ) which I feel fairly represent different areas of the game ( for the record for those who have only played the demo, the content in the demo performs 2-3x better than some other zones in the game as the desert area is not demanding at all ) I'm a good 20 hours in to the game ( 25% completion ) and really have enjoyed it. I'll hazard a guess that the 480 will do quite well in it...

[Yvese]

WiC, RES 5 and BAA are included in most of the reviews done before so I don't know why not in this one?

Which other games that are much more popular than the titles in that benchmark would you recommend, instead of just general bashing?

I'd like to see Just Cause 2, Assassin's Creed II, AvP, Metro2033 ( PhysX off or with the ATi PhysX hack ) and Shattered Horizon.

[Motiv]

Totally agree. I'm looking forward to see JC2 results. I don't believe it has been benched yet. It has 3 built in benchmarks (desert, jungle/rain, city ) which I feel fairly represent different areas of the game ( for the record for those who have only played the demo, the content in the demo performs 2-3x better than some other zones in the game as the desert area is not demanding at all ) I'm a good 20 hours in to the game ( 25% completion ) and really have enjoyed it. I'll hazard a guess that the 480 will do quite well in it...

I think it will do good too, and I will say presumably because it's TWIMTBP. Still, it's nice to see something different.

however, having games in the current top 10s, big games (MW2) and different genres (dow2) would give a more varied and more neutral view.

[annihilat0r]

cod battlefield bioshock 2 mass effect 2 dragon age assassins creed ... well there are tons of games around

MW2 and BC2 is in the list AFAIK, he just didn't write it there.

on the other hand none of the other games prove a challenge for either card, yes BAA or RE5 doesn't too, that's why the most important results should (still) be Crysis and Metro 2033, i.e. titles that current cards have problems on.

Also Just Cause 2 is challenging too as far as I know so it'll be important too. Yeah it's a nvidia title get over it

I'd like to see AvP too.

[skycrane]

ummm we wait over six monts for this ??? for fermi.....

hexus had this up, but brought it down

http://www.xtremesystems.org/forums/...d.php?t=248255

[96redformula]

MW2 and BC2 is in the list AFAIK, he just didn't write it there.

on the other hand none of the other games prove a challenge for either card, yes BAA or RE5 doesn't too, that's why the most important results should (still) be Crysis and Metro 2033, i.e. titles that current cards have problems on.

Also Just Cause 2 is challenging too as far as I know so it'll be important too. Yeah it's a nvidia title get over it

I'd like to see AvP too.

Having played through and beat Metro 2033, the game is just terrible optimized....It is supposed to be multi-core optimized according to the developers, but it gets the same FPS on 2 or 4 cores....

[orangekiwii]

I'd like to see Shattered Horizon. A good looking game, and performs horribly.

[Chickenfeed]

Having played through and beat Metro 2033, the game is just terrible optimized....It is supposed to be multi-core optimized according to the developers, but it gets the same FPS on 2 or 4 cores....

In fairness the majority of modern titles / engines are threaded to take advantage of multiple cores. The reality is however they don't tend to scale much if at all past 2 ( yes theres exceptions ) Try playing a healthy selection of newer games on a single core system ( or a multi core system with cores disabled ) and tell me they aren't threaded for multiple cores Technically by defintion multiple is merely more than 1 so they aren't lieing per say. It would be a strech to claim there are noticeable gains past 2 cores though... ( which I'm guessing was your point )

[Solus Corvus]

I'd like to see Just Cause 2, Assassin's Creed II, AvP, Metro2033 ( PhysX off or with the ATi PhysX hack ) and Shattered Horizon.

I'd like to see Shattered Horizon. A good looking game, and performs horribly.

I'd love to see some Shattered Horizon benches. Unfortunately nobody benchmarks it (even though it's one of the few games that could actually take advantage of all this new hardware ) and not enough people play it (which I don't understand because it's one of the best games I have played, and I have been playing PC games since before the original Wolf3D).

[RaZz!]

Jen-Hsun Huang.

lol, that makes sense. now i know why there was such a supply shortfall at 5800 series' launch: huang bought them all!

i'll wait for other reviews before i jump to a conclusion, but the hexus benchmarks indicate nothing overwhelming... ok'ish performance (the least you could expect from a 6 month difference) with ridiculous high power consumption.

it's a shame that nvidia's "fastest single gpu graphics card IN DA WORLD!!!" still sux big time in crysis and doesn't offer ANY performance increase at all.
i'm more interested in bad company 2 performance though.

I'd love to see some Shattered Horizon benches. Unfortunately nobody benchmarks it (even though it's one of the few games that could actually take advantage of all this new hardware ) and not enough people play it (which I don't understand because it's one of the best games I have played, and I have been playing PC games since before the original Wolf3D).

why? just bench with 3dmark vantage. in the end, it's the same engine, isn't it?

[ChinStrap]

i think cat 10.3 is WHQL too. with luck, reviewers will use them tonight when NDA is up.

[phill]

I'm just rather glad of the choices I've made Roll on the September release of the new ATI cards

[MrMojoZ]

Having played through and beat Metro 2033, the game is just terrible optimized....It is supposed to be multi-core optimized according to the developers, but it gets the same FPS on 2 or 4 cores....

Well techincally 2 is multi-core.

[jogshy]

I don't trust in benchmarks done with beta or very immature drivers though....

[96redformula]

Well techincally 2 is multi-core.

Found it for you

PCGH: By now multi-core CPUs have become very popular and the number of players with such machines is rapidly increasing. Did you integrate multi-core support into the engine from the beginning?

Oles Shishkovstov: Yes, the engine was architected to be multi-threaded from the start. That's the only sensible route to go. You just cannot add multi-threading later in the development cycle or it will be horribly sub-optimal.

PCGH: How many core are supported and what is the expected performance gain from 2, 4 or even 8 cores?

Oles Shishkovstov: We support at least two cores, and up to whatever count you have. You should expect linear scaling with the number of cores, when you aren't GPU limited.

Some of the game looks great, while some of it is very MEH. When you play it the framerates are all over the place, sometimes happens when there doesn't seem to be alot going on GPU wise.

[Solus Corvus]

why? just bench with 3dmark vantage. in the end, it's the same engine, isn't it?

It is the same engine. But it isn't the same scenes, effects, etc. Are they directly comparable? Maybe, but we won't know until someone benches it and finds out.

[cegras]

I might as well repeat myself: per watt, ATI's architecture is more efficient.

[-Sweeper_]

liveblog: http://blogs.nvidia.com/ntersect/201...-liveblog.html

30 minutes left!

[alcachofa]

http://translate.google.fr/translate...sl=zh-CN&tl=en

Introducing the Air Force Kit