Starcraft AA benchmarks: ATi vs nVidia

[happy medium]

So I'm almost at the end and the game's unplayable even with MSAA at 1920x1080. I am literally getting 1-2s pauses every few seconds with regular FPS <10. The amount of units is crazy and all those graphs showing 60FPS with 40MIN are bullshit. The last several missions have been like that and there's no way I can play them with any AA at all.

So happy, stop reading silly reviews and stop basing your opinion on them. Most of them are very wrong. People have been telling you that yet you continue to ignore gameplay facts. Or better yet, buy the game and see for yourself.

Are you serious? SO you are saying Xbit Labs and you are right and don't go by review sites? WHat people besides you? Mabe I missed it.

 

[Voo]

Are you serious? SO you are saying Xbit Labs and you are right and don't go by review sites? WHat people besides you? Mabe I missed it.

Once again, have you actually played the game seriously? I.e. the whole campaign or some large multiplayers with lots of units? And if you did, you really got similar results to pcpers and co in the large battles?

I somehow doubt it.

 

[GodisanAtheist]

Are you serious? SO you are saying Xbit Labs and you are right and don't go by review sites? WHat people besides you? Mabe I missed it.

-Seriously tho, I'm someone who actually owns a GTX460 coming off of an HD5770 and while the 460 is faster its definitely not WHOAMG faster, and it sure as hell is not as fast as whatever the hell metric those reviews were using.

Maybe you can say all these cards make your shit smell nice if you have a bleeding edge Core I7 980X $4000 dollar test bench, but for those people that actually buy $200 dollar cards to upgrade their Core 2 Quad system or something you're not going to be seeing holy hell numbers.

There, you now have people besides him.

[Edit: Don't get me wrong though, I'm really loving my 460, it looks like its going to sit alongside my 4850, 9800Pro and 4600TI in my hall of fame. To be fair, I had borrowed a friend's extra GTX470 (he went SLI 480 the nut) while I was between cards (see sig) and let me tell you that card is a piece of crap. Yeah its fast and all, but holy shit it was so loud and so hot that it actually DISCOURAGED me from using my computer for a solid 2 weeks until I settled on the GTX460. GTX470, card so awesome it'll make you go outside and enjoy the world... After that experience, ballpark performance without the baggage might be biasing my opinion in favor of the 460.]

[Edit2: This might also be a good time to restate the question I asked earlier in the thread that was either laughably stupid or so inciteful that it could only be bulldozed by fanboy dick-waving: Why does Xbit insist on turning Catalyst AI OFF in their reviews when Nvidia ostensibly does not even provide the end user the option to turn off their driver optimizations? Is this some strange form of fairplay experienced only in backwards land or is there a meaning to it that I'm simply missing?]

 

[Qbah]

Are you serious? SO you are saying Xbit Labs and you are right and don't go by review sites? WHat people besides you? Mabe I missed it.

Grooveriding to name another. And yes, I am telling you how the game runs in campaign on an OCed HD5850 and a 3.2GHz C2Q and the numbers shown on hardwarecanucks or pcp aren't even close to actual performance in-game. Xbit ones are a lot closer, though it can still get lower than that. The terrain animations and fire everywhere is what lowers your FPS even more. There's lava everywhere, flamethrowers, burning corpses and a swarm of zerglings and other crap too.

 

[Aristotelian]

Grooveriding to name another. And yes, I am telling you how the game runs in campaign on an OCed HD5850 and a 3.2GHz C2Q and the numbers shown on hardwarecanucks or pcp aren't even close to actual performance in-game. Xbit ones are a lot closer, though it can still get lower than that. The terrain animations and fire everywhere is what lowers your FPS even more. There's lava everywhere, flamethrowers, burning corpses and a swarm of zerglings and other crap too.

Have you done any 4v4s on Battlenet? As long as you aren't rushed in the first few minutes, it can be a very taxing test. I think that cards should be tested in the most rigorous aspect of a game, such as 4v4 200 food armies colliding: battlecruisers, carriers, tons of stalkers warping in and out, zerglings everywhere and so on.

Perhaps a separate test for single player would be good as well, since some people aren't interested in Battlenet. The final mission I found to be very taxing, even more than the lava mission. A lot of micro is required, and there's just tons of x everywhere (x, so as not to spoil). Framerates can start to plummet in scenarios such as those I've mentioned.

 

[happy medium]

Grooveriding to name another. And yes, I am telling you how the game runs in campaign on an OCed HD5850 and a 3.2GHz C2Q and the numbers shown on hardwarecanucks or pcp aren't even close to actual performance in-game. Xbit ones are a lot closer, though it can still get lower than that. The terrain animations and fire everywhere is what lowers your FPS even more. There's lava everywhere, flamethrowers, burning corpses and a swarm of zerglings and other crap too.

That could be the difference. Read the pc perspective review on taxing your cpu.

http://www.pcper.com/article.php?aid=958

 

[happy medium]

Grooveriding to name another. And yes, I am telling you how the game runs in campaign on an OCed HD5850 and a 3.2GHz C2Q and the numbers shown on hardwarecanucks or pcp aren't even close to actual performance in-game. Xbit ones are a lot closer, though it can still get lower than that. The terrain animations and fire everywhere is what lowers your FPS even more. There's lava everywhere, flamethrowers, burning corpses and a swarm of zerglings and other crap too.

I dont see his posts, and he has 2 gtx 480 in sli , I believe.

 

[Cerb]

That could be the difference. Read the pc perspective review on taxing your cpu.

http://www.pcper.com/article.php?aid=958

...which is kind of the point trying to be made. It's no mystery that the game's detail settings depend on the CPU. The game even tells you that, when you go to adjust them. Stressing the CPU needs to be a core part of the review.

A proper performance review will, at the least, make a test that seriously stresses an i7 at Ultra, and then test at different detail levels, and give you the replay for your own verification/critiquing. The last bit is an unforgivable flaw for many of the reviews, IMO, since SC2 automatically creates replays, with usable file names, in an easily locatable directory tree under My Documents. PCPer, FI, did a test that did not stress the i7 at all, even on Ultra; despite actual users finding scenarios which can really slow them down with such high-end hardware.

While none have been great so far, the game just came out, and it certainly looks like Xbit and TechSpot tried to make meaningful tests, with no canned benchmark in the game. Hopefully, AT or somebody will take their time on one, focusing on metrics that really affect performance, or somebody like Xbit or TechPowerUp will have a more detailed and accurate 2nd run of a review; taking discussions like these as a guide for how to test, and what to focus on.