TheGreatGrapeApe :
No it's not!
Once again you don't bother to actually look at the benchmarks in depth.
QuadSLi GX2 is faster at the lower resolution, at the higher resolution the X2 in CF beats it. so the GX2 gets you 1 or two frames in the easier mode, but then move to higher resolution when these aren't CPU bound, and the X2 CF is ahead.
And once again you pull out DX9 Crysis results. I don't know why you or reviewers even bother with those. It's like running games in 16bit mode to show higher fps, pointless with these setups that can handle full DX10 mode, or well the GTX and HD4K can handle full DX10 Very High, the GX2, not so much.
And many of these posts you're putting up counter you previous GX2 vs GTX280 argument, where HL2, UT3, and Quake wars all have the GTX well out in front and in the DX9 Crysis benchmarks both the GTX280SLi and even GTX260SLi well outperform the GX2 Quad SLi.
Next time actually look at what you're posting.
Then go beyond the 1920x1200 range and the GX2 falls of the end of the earth running out of VRAM, making quad SLi very limited for it.
Seriously you try to pimp the hell out of the GX2 like you have stock in them.
No offense, but you're the one that haven't looked at the benchmarks, not me.
Look closer, the benchmarks include 2 bars for each game. The higher one is for 1920x1200, the lower one for 2560x1600. In each and every case other than Half Life 2, 9800gx2 quad perform higher than 4870x2/4870 cf on both resolutions respectively. I'm not sure where "Then go beyond the 1920x1200 range and the GX2 falls of the end of the earth running out of VRAM, making quad SLi very limited for it" comes from. While you're certainly right that the lower vram will tank performance for 9800gx2 at some point, 2560x1600 just isn't quite there yet, not to mention the smaller 1920x1200. While there are resolutions larger than 2560x1600, it's safe to say most monitors can't handle those. 2560x1600 resolution is hardly "cpu bound," especially since their test rig use an "extreme" quad cpu. And those games do not run in 16bit mode.
As for dx9 vs dx10, honestly, I haven't noticed that the benchmarks are dx9 and not dx10. They didn't say on the test pages. Side to side comparisons of 4870x2 vs 9800gx2 quad sli is extremely rare. They're hard enough to find to begin with, so I was happy to be able to come across one at all, dx9 or dx10. As for your claim that 4870x2 will outperform 9800gx2 quad sli, I believe you. You're a mod and been here for a long time, your word is good enough for me.
But to say that dx9 benchmarks are somehow invalid is just an elitist viewpoint. Most benchmarkers still use either dx9 or a mixture of dx9 and dx10 instead of dx10 only. Most games still use either dx9 or both dx9 and dx10 dual mode, no game in existence so far use dx10 only. Majority of gamers also still run on dx9. To say that the only side to side benchmark that include this rare setup I could find is invalid because it happen to use dx9 is like saying unless you drive a Mercedes, you're not a driver. It's unfair. Nothing wrong with the still mainstream dx9.
Just as a side note, where in the benchmark did they actually say the tests are done in dx9 and not dx10? I reread it and weren't able to catch that part. They did say they use Vista, and 3dmark Vantage test is certainly done in dx10, since it doesn't run in dx9. They just didn't say if the games are done in dx10 Vista or if they forced dx9 mode.
As for this:
"And many of these posts you're putting up counter you previous GX2 vs GTX280 argument, where HL2, UT3, and Quake wars all have the GTX well out in front and in the DX9 Crysis benchmarks both the GTX280SLi and even GTX260SLi well outperform the GX2 Quad SLi."
I never posted that 9800gx2 quad sli outperforms gtx280 sli, in this thread or anywhere else. I know you're working and you're busy, but please try to keep track of things and don't say things that doesn't actually exist. All I said in other threads is that a single 9800gx2 outperform a single gtx280 or single 4870 (when 9800gx2 wasn't affected by the horrible quad scaling), which is true, using those benchmarks:
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=13
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=14
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=15
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=16
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=17
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=18
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=19
This particular benchmark does use dx10, including Very High shader setting for Crysis. I try to use dx10 benchmarks whenever possible. It's the new thing, after all. Benchmarks comparing 4870x2 and 9800gx2 quad sli side to side are just far too rare, and the one is all I could find, so there's no choice there.
For HL2 and UT3, the earlier benchmarks didn't include them on its lineup. Many benchmarkers didn't include them either, probably because they are slightly older and so well optimized that you max out with hundreds of fps. But you're right, those are popular games, and should be included in benchmarks. As for Quake Wars, which was included in both benchmarks, the quad benchmark use OpenGL instead of dx, which might have contributed to the difference in performance. Although I guess it's really better and more logical that way since it runs on OpenGL natively and dx is only a matter of sound and compatibility.
I knew from the beginning that going against conventional wisdom is going to result in taking some heat, even if backed up by benchmarks. I just didn't expect so much of it from a level headed and well respected poster like you.
Since you're from Canada, in a time zone close to my own, and your post is timestamped 4:27 am, I'll just assume you're fatigued.
I still respect you very much. But please try to look at things more carefully and check for possible errors before hitting the submit button.
Who's ever tired on a long weekend? It's afterwards returning to work that would be the issue, and then again I have the rest of the week off too.
Facebook
Twitter
Instagram