Anandtech's 7900GTX Review is not so great (fixed)

[tuteja1986]

http://www.rage3d.com/board/showpost.php?p=1334215962&postcount=105

1st : WHat drivers did they use
2nd : Why did they test only 4 games :? comeon more is expected from one the worlds biggest Computer Hardware review.
3rd : FEAR benchmark is a little fishy

Older review http://www.anandtech.com/video/showdoc.aspx?i=2679&p=12
FEAR 4x AA @ 1920x1400

x1900xtx crossfire Scores @ 55
7800GTX 512MB sli Scores @ 36

Newer Review http://www.anandtech.com/video/showdoc.aspx?i=2717&p=7
FEAR 4x AA @ 1920x1440

x1900xtx crossfire Scores @ 44 << SCORE DROP
7800GTX 512MB sli Scores @ 55 << WTF ? HUGE JUMP
7900GTX SLI Scores @ 58


http://accessories.us.dell.com/sna/Prod...053YR&category_id=4009&c=us&l=en&cs=19

anandtech review needs some changing :! they need widescreen bechmarking
More games and Ultra high res :!

Anyways Anandtech review ain't so awesome :!

I think one of the best 7900GTX currently is from firing squad
http://firingsquad.com/hardware/nvidia_geforce_7900_gt_gtx_performance/

 

[MBrown]

 

[dug777]

explain yourself mate

looks more like something is wrong with you

 

[tuteja1986]

lol :! I pressed enter by mistake :! i was like "OHH F$#$ "

 

[5150Joker]

You're right, the newest AT reviews crossfire losing over 10 fps with a minor bump in resolution (1400 vs 1440) and AA setting..wtf? 55 fps vs 44 fps in their latest review...hmmm.

 

[fbrdphreak]

AT's video reviews have been downhill for quite some time...

 

[imported_Finny]

Yeah, minus 20, and you'll be about right.

Anywho, yeah, it would've been nice to know what kind of drivers (Although I doubt that'd make THAT much of a difference), but it is odd that the 512 GTX would suddenly score THAT much higher this time... Drivers wouldn't help that much.

My best guess is that either A, they're using different areas, resulting in different scores, or B, they're being a little "generous" with the GTX512. I'm kinda with the latter; I can understand a 1 FPS difference on the ATi part (C'mon, a 1 FPS difference?), but the GTX512...

Edit: Aha, re-read the reviews. It is a 10 FPS difference... Curiouser and curiouser...

 

[fbrdphreak]

Why is this a bad thread? Its true - AT's video reviews are slipping. Have they ever heard of formatting? A couple pages there are like books of just plain text, no bold or anything. And the grammar is slipping too. Plus everything already mentioned.

At least that new kid didn't write this and spend two pages describing the box art.

 

[BenSkywalker]

x1900xtx crossfire Scores @ 44 << SCORE DROP
7800GTX 512MB sli Scores @ 55 << WTF ? HUGE JUMP
7900GTX SLI Scores @ 58

Not sure on the ATi score drop but nV was holding back some driver optimizations for FEAR for quite a while; the FiringSquad review you linked to shows the same thing- ATi and nV quite close running FEAR. Actually, FiringSquad shows that the 7900GTX has a higher minimum framerate then the 1900xtx running FEAR.

why even bother benchmarking at 1920x1440 :?

People who own gaming monitors- they pretty much all go up to 1920x1440 and a lot of them go to 2048x1536.

 

[Genx87]

The 84.14 beta's have some fixes for FEAR.

 

[tuteja1986]

Originally posted by: BenSkywalker

x1900xtx crossfire Scores @ 44 << SCORE DROP
7800GTX 512MB sli Scores @ 55 << WTF ? HUGE JUMP
7900GTX SLI Scores @ 58

Not sure on the ATi score drop but nV was holding back some driver optimizations for FEAR for quite a while; the FiringSquad review you linked to shows the same thing- ATi and nV quite close running FEAR. Actually, FiringSquad shows that the 7900GTX has a higher minimum framerate then the 1900xtx running FEAR.

why even bother benchmarking at 1920x1440 :?

People who own gaming monitors- they pretty much all go up to 1920x1440 and a lot of them go to 2048x1536.

yeah i get the point . but you miss understood me :! i was saying talking about that particular res :!

anyways about the fear score :! your talking about the 1280x1204 res :!

but when you when bechmarked higher res ATI clearly wins

2048x1536x32 4xAA 16xAF
X1900XTX : 35
7900GTX : 28
7800gtx 512MB : 27

Anyways thoes weren't sli or crossfire.

 

[imported_Finny]

Actually, now that I've looked more into it, apparently the 84.17 drivers add almost 20% performance increase, so it might actually be plausible for the GTX512 to have gotten such a boost.

This little problem could've been avoided had the reviewer simply stated what drivers he was using.

 

[Matthias99]

Originally posted by: Finny
Actually, now that I've looked more into it, apparently the 84.17 drivers add almost 20% performance increase, so it might actually be plausible for the GTX512 to have gotten such a boost.

This little problem could've been avoided had the reviewer simply stated what drivers he was using.

And settings. Soft shadows, for instance, make a huge difference in how the game performs (and in THG's reviews, give the X1900XTX a HUGE lead over the 7900s rather than a small one).

THG

FEAR -- 1600x1200 noAA noAF no SoftShadows
X1900XTX -- 87
7900GTX -- 77
7900GT -- 56
(they didn't bench the X1900XT or X1800XT...)

FEAR -- 1600x1200 noAA noAF w/SoftShadows

X1900XTX -- 62
7900GTX -- 38
7900GT -- 29

I'm guessing this makes the test much more shader-limited, which is why the 7900GTX then performs at about half the speed of the X1900XTX. This is with the 'Forceware 84' drivers (doesn't indicate if it's 84.17 or not...)

 

[ElFenix]

Originally posted by: 5150Joker
You're right, the newest AT reviews crossfire losing over 10 fps with a minor bump in resolution (1400 vs 1440) and AA setting..wtf? 55 fps vs 44 fps in their latest review...hmmm.

if you look at the chart below the graphs, the chart lists 1920x1440. i think 1920x1400 is a typo. in which case there is no change in resolution

 

[Ackmed]

Id like to know why the ATi scores went down so much. Pretty bad not listing what they used.

 

[tuteja1986]

Originally posted by: ElFenix

Originally posted by: 5150Joker
You're right, the newest AT reviews crossfire losing over 10 fps with a minor bump in resolution (1400 vs 1440) and AA setting..wtf? 55 fps vs 44 fps in their latest review...hmmm.

if you look at the chart below the graphs, the chart lists 1920x1440. i think 1920x1400 is a typo. in which case there is no change in resolution

amm i think your right :! I went back to older review of firing squad and compared it with their latest driver and found that cat 6.3 does a performance decrease in fear

Latest Benchmarking with the latest Driver : ForceWare 84.17 and Cat 6.3
http://www.firingsquad.com/hardware/nvi...rce_7900_gt_gtx_performance/page14.asp

Older Benchmarking with Driver version ForceWare 81.98 and driver Driver version sample_8-203-3-060104a-029367E
http://www.firingsquad.com/hardware/ati_radeon_x1900_xt_preview/page13.asp

Anyways i don't recommend the cat 6.3 driver :! cat 6.2 you will get a much better performance in fear.

 

[BFG10K]

why even bother benchmarking at 1920x1440 :?

Because middling/low resolution tests like 1600x1200 and below are useless for testing GPUs.

 

[otispunkmeyer]

there was a driver issue wit fear an AA with nv cards wit the new drivers