Interesting RTHDRIBL results @ Hexus
- Thread starter nelg
- Start date
Nothing direct that I can quote or use from NVIDIA yet, although I'm trying to talk to the right people, rather than product managers or regional PR.
I've got the last part of my 6800 testing nearly done (I've got the GT and non-Ultra I was missing) and the results are a bit different with the new driver.
I'll post some numbers here soon, if people are still interested in the RTHDRIBL render setup and how it's run on 6800.
Rys
I've got the last part of my 6800 testing nearly done (I've got the GT and non-Ultra I was missing) and the results are a bit different with the new driver.
I'll post some numbers here soon, if people are still interested in the RTHDRIBL render setup and how it's run on 6800.
Rys
What, my interest isn't enough to galvanize you to post results? 
I'm surprised that you couldn't get a straight/knowledgable answer out of nVidia after all this time, but even if they're not "publicly" addressing on the dribble issue directly, it's nice to know their drivers are improving performance (be it through general or specific optimizations). Have you gotten a response from dribble's author?
I'm surprised that you couldn't get a straight/knowledgable answer out of nVidia after all this time, but even if they're not "publicly" addressing on the dribble issue directly, it's nice to know their drivers are improving performance (be it through general or specific optimizations). Have you gotten a response from dribble's author?
I'm interested!!!! (Yeah, I've been lurking in this thread....I bet you didn't think I could ever keep me mouth shut in a thread for over a month!Rys said:
)
Pete said:What, my interest isn't enough to galvanize you to post results?
I've heard nothing from Masaki after I queried him on the shaders used, after 3DAnalyze showed them to be hand written.
As for performance, using 61.34, a 6800 GT (350/1000) is now faster (67.33fps) than an Ultra (400/1100) using 60.72 (66fps). I'll post my full numbers when I get a chance to throw the Ultra back in.
Rys
surfhurleydude
Newcomer
Rys, Ultras are clocked at 400/1100, not 1000. Typo?
Code:
6800 Ultra 60.72 - 66.000 - 58 - 77 - 400/1100
6800 Ult/E 60.72 - 67.800 - 60 - 79 - 450/1100
6800 GT 60.72 - 59.600 - 53 - 70 - 350/1000
6800 Ultra 61.34 - 71.917 - 64 - 85 - 400/1100
6800 non-U 61.34 - 52.367 - 46 - 63 - 335/700
6800 GT 61.34 - 66.467 - 58 - 78 - 350/1000
9800 XT 4.4 - 51.284 - 43 - 66 - 412/730
X800 XT 4.5B - 142.43 - 122 - 171 - 520/1120
X800 PRO 4.5B - 103.50 - 88 - 127 - 475/900The GT on 60.72 was emulated using the Ultra Extreme by downclocking, that may skew results a little due to memory timings.
Otherwise, there you go. Scores are Avg - Low - High
Notice the GT take on the older Ultra performance with the new driver and the non-Ultra with the new driver matching the 9800 XT, despite NV40 still being poor at this test.
Sorry for no non-Ultra results with 60.72, I can't make the driver work with it.
The R420 boards take a giant poop on the rest.
Rys
Thanks for that mental image! :?
Bouncing Zabaglione Bros.
Legend
Rys said:
But in a good way...
AlphaWolf said:
it's not really multisampling in the sense that is used as in "hardware multisampling" (the one that reduces edgies).
There is spatial and temporal multiple rendering that simulates both antialiasing and motion blurring at the same time (renders four frames at slight different position in time and space and blends them together).
The name can be misleading but multisampling is already quite overloaded in meanings.
That's why the huge loss..
Similar threads
