Printable View
http://en.expreview.com/img/2008/07/31/score.png
every single numbers up there is average FPS after ten runs, it is enough to eliminate margin of errors.Quote:
177.79 is the first 177 driver that not only supports GTX 200 series but also supports 8800 and 9800 series. Finally Nvidia kept their Unified Driver promise.
Link;)
Nice. Should make comparing the series of cards very easy since they're all the same revision.
Aren't the most of the differences withing the margin of error(~3%) in these measurements??:confused:
I think I'll stick to this 175.95 laptop driver for a while still as I've had bad results with the other 177 drivers and 175.95 is definitely faster than 175.19 so I guess they're about equal but the sharp detailed textures is what I really like in them, best driver for UT3 so far IMO.
Welcome to the real world software tweaking. :p:Quote:
Originally Posted by informal
Yes. At lower resolutions, even 5% is within the margin of error.Quote:
Originally Posted by informal
The TREND that all of them are an improvement indicates it is not a statistical aberration.
If these were truly due to random chance, then you would expect 50% of the benches to show an improvement, and 50% a decrease.
The odds of EVERY SINGLE benchmark showing an improvement, by random chance, is astronomically small (like 2^24).
all the numbers are average FPS of ten runsQuote:
Originally Posted by informal
so most of the margin of error are eliminated
Average of 2,5% performance gain; few frames advantage over 175.19 drives in games. But a gain is a gain, even tho it's a little one.
Solid perfomance gain? Where? Well, it's a solid 2% that's true :lol:
177.7x fixed the SLI Citadel/Tower bug in Mass Effect for me... Finally :)
Minor framerate gains all around, but improved smooth feeling with SLI in a few games. Good driver in my book.
Most of those results are within the margin of error. The others that are beyond the margin are ones specifically mentioned in the release notes.
Perkam
These results between both drivers prove no performance gain, actually...
True :up:Quote:
Originally Posted by perkam
But the bug I am talking is not within margin of error: 3-11 fps vs 34-60 fps v-sync'ed (v-sync really doesn't matter, only way to get around the bug, is to disable SLI completely).
Still a good driver in my book, as I didn't break anything but solved something for me personally :)
Has anyone tested these with older cards? Probably don't even support my 7800's anymore.
These also break 2D/3D clock switching on GT200 cards. They perform similar overall to 177.66. No Low Power clock switching makes my overclocked 280 run at 45c. It's nice and quiet when clocks are switched correctly and it's running at 34c.
Blah! They make the 175.19 driver perform almost 10% worst than the 175.16 driver and now this one offers what? 1% increased performance? I'm definitely not impressed with that.
Edit: Here http://benchmarkreviews.com/index.ph...d=194&Itemid=1
Nvidia does not list any 6, 7, or 8 series cards as supported in their 177.79 release notes.Quote:
Originally Posted by Nasgul
Perkam
That's nice, but I already replaced my 8800GTS g92 with a Radeon 4870 :D Maybe next time. :)
And BTW, has anybody with a GTX 280 tried the new drivers yet?
I'm using it with an 8800 gtx no need for a modded inf.Quote:
Originally Posted by perkam
I haven't done much testing yet.
Ewww. i tried them out on server 2008 64x and they are totally slower just within the OS. It might be some other side effect, but on 175.19 everything runs smoothly, 177.79 and now firefox is stuttering when I scroll and music streams skip when I switch windows. Back to 175.19 I go.
I have been playing around with this series for a couple days I have especially noticed this as well, with both the 177.70 and the new 79s.Quote:
Originally Posted by Seraphiel
I noticed that too but it does have them listed here under the appropriately named "Products Supported" tab. Plus I didnt need a modified .inf for my 8800GTX.Quote:
Originally Posted by Perkam
The margin of error is mostly eliminated after 10 runs though.Quote:
Originally Posted by perkam
Quote:
Originally Posted by mikeyakame
See above post by me. 2D/3D clock domain switching is pretty much broken on my GTX280 and nearly everybody else using them too. The switching initially works until you play a game or similar then it gets stuck in 3D Performance level and doesn't switch back. It's a decent driver just I honestly like my clock domain switching! In terms of performance though I find 177.66 (which I'm running now) to be around the same ball park.Quote:
Originally Posted by gOJDO
I love you too ;>Quote:
Originally Posted by LOE
Edit: You may now spread the love back. Doesn't look big and bold to me though. Perhaps your font size scaling is busted.
"Add" in CP program settings is broken...
If you want 10% more gain, go upgrade your graphic cardQuote:
Originally Posted by LOE
the driver is free after all..
Call me a cynic, but I bet if ATI/AMD came out with a driver revision that improved performance 1 - 5% across the board, you wouldn't get all these posts about 'margin of error' but instead praised the driver team for coming up with the performance enhancements. ;)
From experience, if you are testing in a controlled state, such as a timedemo, the margin of error is very low, usually less than 1%.
I've tried this driver with my 8800GT G92, fps increased in my benchmarks, about 2-4 fps. Not bad. :)