Will in the case of the gtx they did reduce the settings but overall I can only guess it is because the game is so CPU limited. If you look right here at Tom's VGA charts you see the same thing happening. The gtx goes from 22.4 to 22.6 going from 1600*1200 to 1920*1200. Both sites could use some new tests with sp 1 which should be tasking the GPU more.
Yeah but if the pixel load is going UP then why does the XT's minimum FPS improve so much and the only thing hurt is the max FPS? Their 1600x1200 test show a completely different picture of the XT completely non-performant, yet move to Widescreen and the gap closes with less dips below 20fps even, and moves it closer to the others. To me it's just a poor test on a game that obviously has other areas of stress, and if it's just GPU bound then the fps shouldn't go up as a result of higher resolution it should stay bound. And if it was the XT alon that was causing performance issues then going from 16x12 to 19x12 should've dropped the performance even more. In all it's just an indication that it's a terrible game to be testing the 'my favourite setting' scenario. I would prefer a true apples-to-apples, but we can't even get that. Based on all the reviews that have come out since the [H] initial review, I value it less and less each subsequent review. Unfortunately it's one of the few with FSX tests.
I can't speak for them but I don't see too many sites testing 1920*1440.
True, but if you're going for 'max playable' then stopping at 16x12 for the 4:3 tests in every scenario, simply shows that it's a test limitation, not a card limitation. And of all the times when you want to compare them it's on a CRT that can scale to the best fit, unlike and LCD which is stuck at native resolution.
They tested at 1900*1200 which will take you up to a typical 24" lcd rather than some 19"crt at 1920*1440. The next stop would be the 30" panels which I personally would save for the sli/xfire tests.
The next stop is 20x15 which is another 4:3 resolution that most 21" and 22" gamers had on their Sony and Viewsonics. Remember that section is 4:3 gaming, so the 30" LCDs don't factor into it, but resolutions above 1600x1200 do for anyone who still appreciates CRT gaming and has the money to buy any of these cards. I think for a 'best fit' test, they aren't 'best fiting' the CRT gamers out there. Heck my old Phillips 17" was a 1600x1200 monitor, they couldn't find themselves at less a 19x14 let alone 20x15 monitor for testing?
If we were stuck at 1280x1024 testing with THG's tests we'd get a very different picture, and assuming the CPU bottleneck, not question what happened above 1280x1024 if we didn't have the tests?
http://www23.tomshardware.com/graphics_2007.html?modelx=33&model1=740&model2=722&chart=291
The GF7900GT playing in the field of the Gods should show that issue pretty clearly. Or moving down to 10x7 the GF7900GS beats both the GTS and GTX.
And if in their Apples to Apples 8XQ was actually the same performance as the 8XCSAA, then why not enable it instead of claiming it?
At 8X AA where you would think the benefit of ATI’s 512-bit ring bus memory subsystem would benefit more from seems to underperform the BFGTech GeForce 8800 GTS 640 MB. Note that this is using NVIDIA’s 8X CSAA mode, we did try 8xQ and found the performance delta shown here to be exactly the same still.
Yes, I read the section, and like I said, they should prove it not say it. WTF are they doing running CSAA at all if they say it doesn't match the same IQ setting? Why not just run ATi's wide tent AA and pretend it's the same as the CSAA? I know they tried to add that caveat, but the fact that they tested the poorer version and then said it performed the same as the better version simply shows lack of internal consistency/validity.
Till then I trust nothing with re. to dx10. DX10 has been a lot of missed deadlines and initially poor performance as an add on.
I understand that, but considering the similarity in performance right out of the bag, the areas that DX10 will benefit a game like FSX the most, and most importantly to me considering the statements made by the M$ developers about the G80 and R600 when making the DX10 patch, all I'm saying is that I would recommend the HD2900XT over a GTS as a starting point, but like I said, that's my recommendation. I wouldn't take the [H] DX9 benchmarks over the M$ DX10 comments when it comes to picking a card to play FSX. Of course others may differ.