View Full Version : R600 3DMark06 bench and pics.
Xion X2
03-20-07, 09:36 PM
Woah, because we all know that 3d mark is a perfect representation of every game out there and supports dx10!!!:rolleyes:
It's not the end-all-be-all, but it's been a very reliable performance indicator in the past.
Take 3dMark06 and benchmark runs between the 1900XT and 7900GTX, for example. The GTX always came out just a little on top and was a slightly faster card in most applications. To denounce it completely would be dumb.
It's not the end-all-be-all, but it's been a very reliable performance indicator in the past.
Take 3dMark06 and benchmark runs between the 1900XT and 7900GTX, for example. The GTX always came out just a little on top and was a slightly faster card in most applications. To denounce it completely would be dumb.
QFT. People love to hate on 3DMark06, but truth is it is a quick, easy way to gauge a graphics cards performance, and for the most part, fairly accurate.
QFT. People love to hate on 3DMark06, but truth is it is a quick, easy way to gauge a graphics cards performance, and for the most part, fairly accurate.
3dmark tends to put more of an emphasis on the CPU side of performance compared to what most games need. Obviously more so on the older versions.
If you pop in a QX6700 you'll get an instant 10% boost on your overall 06 score .vs a similarly clocked C2D, but that 10% doesn't translate into any real world performance except for maybe in SupremeCommander.
all the waiting, just for 200pts... :)
where is my 8900GS!!!!!!? :D
The 3DMark06 difference isnt quite worth it :o
methimpikehoses
03-21-07, 04:21 PM
all the waiting, just for 200pts... :)
where is my 8900GS!!!!!!? :D
I got tired of waiting and finally bought a EVGA 8800GTX yesterday. If the new stuff really pwns it, which I doubt, I'll stepup to a 8900 or Ultra or whatever you call it.
It doesn't sound like the new ATi's will be in anyone's home rig for some time...
stevemedes
03-21-07, 04:53 PM
I'm pretty sure I'm going to eventually go with the 8800 GTS 640MB. With an O/Ced C2D and that GPU you should hit ~10k in 3dmark06, which is plenty and is a nice increase over my 6500ish on my 7900gtx. Hopefully that card will drop in price to $350, then I will buy it. I'm planning on selling my 7900gtx for like 200 bucks then using that money to buy the GTS. Shuld be pretty sweet, unfortunately its still a ways off. So while I wait for a) NV's drivers to mature and b) prices to drop and c) me to get money, I'm gonna buy a new case and power supply.
Xion X2
03-21-07, 09:25 PM
3dmark tends to put more of an emphasis on the CPU side of performance compared to what most games need. Obviously more so on the older versions.
If you pop in a QX6700 you'll get an instant 10% boost on your overall 06 score .vs a similarly clocked C2D, but that 10% doesn't translate into any real world performance except for maybe in SupremeCommander.
All that is true, but if you take the same config and run it on two different graphics cards, then the difference in scores should be very indicative of the difference in graphics power between the two cards.
SlieTheSecond
03-21-07, 10:41 PM
I'm pretty sure I'm going to eventually go with the 8800 GTS 640MB. With an O/Ced C2D and that GPU you should hit ~10k in 3dmark06, which is plenty and is a nice increase over my 6500ish on my 7900gtx. Hopefully that card will drop in price to $350, then I will buy it. I'm planning on selling my 7900gtx for like 200 bucks then using that money to buy the GTS. Shuld be pretty sweet, unfortunately its still a ways off. So while I wait for a) NV's drivers to mature and b) prices to drop and c) me to get money, I'm gonna buy a new case and power supply.
I have that. The highest score I've hit is 10660.
E6400 @ 3.4 Ghz. 660/1800 on the 8800 GTS 640 mb.
Hitting 1100 would be very lucky with a single gts. And you probably need a 4Ghz overclock. I can't see anyone getting passed 1100 (excluding dropping visual quality with drivers :) )
All that is true, but if you take the same config and run it on two different graphics cards, then the difference in scores should be very indicative of the difference in graphics power between the two cards.
I agree with you on that one. However, comparing your 3dmark score with someone else is pretty pointless when considering real world gaming performance.
Look at these two '06 results:
13156 marks - Geforce 8800 GTX @ 620/1000mhz - QX6700 @ 3.33 ghz
10848 marks - Geforce 8800 GTX @ 648/1026mhz - E6300 ES @ 2916 MHz
Which one do you think is faster in every game when not running at silly resolutions (besides Supreme Commander)?
Tough call IMO, but it's definitely not a 18% difference like 3dmark suggests.
Tygerwoody
03-22-07, 07:59 AM
why doesn't image quality matter to anyone besides me?
I guess because everyone knows ati isn't going to fix the problems they already have, which just screw up the whole thing for me, I don't know about any of you.
I'll be damn impressed if they get trilinear filtering and angle invariant af, and fp16 HW filtering. I also want to be able to force trilinear filtering in old games.
But since people only care about the benchmarks and b/c ati makes really dumb decisions, that won't ever happen. I think they actually totally want there to be aliasing on games played in their hardware. Nvidia 1/2 wants aliasing when games are played on their hardware.
What isn't fast enough about doom3 engine games when the frame rate is capped at 60 anyway? You could not only have the highest settings on 2 gtx in sli w/ doom3 engine games, but you could have 2.5-3x the image quality load on the "highest" current setting and it wouldn't go under 60fps (at least not much, and even if it did it would still be perfectly playable.)
I can't, for the life of me, comprehend why benchmark scores need to be higher.
Don't flame me, I'll never be able to comprehend why in any way.
I was just born to care about image quality and to be able to be satisfied with framerates of 60+ fps.
you do realize that Nvidia is not only a tremendous amount faster, but also has a HUGE lead in image quality when compared to ATI/AMD? Just checking
Xion X2
03-22-07, 10:54 AM
I agree with you on that one. However, comparing your 3dmark score with someone else is pretty pointless when considering real world gaming performance.
There are some cases where it's not a reliable indicator, like with quad-cores and 8800GTX SLI for example, but if you use 3dMark in the right way, then it can be a very reliable tool to gauge performance. But not in the way that you're trying to do it.
It doesn't take the entire app and multi-thread it like games do today (after all, it came out, what, 1.5-2 years ago?)--it simply runs a CPU test to test the strength of your processor in conjunction with everyone else's. It does this by rendering graphics solely on the CPU. Therefore, someone with an X2 4400+ will get a better overall score, with the same config as the test bench, as someone else who has a 4000+ single-core.
It may not be accurate down to the exact percentage, but the dual-core machine will come out on top every single time.
Gaming is at a crossroads right now between dual-threaded and quad-threaded applications. As such, you can't compare 3dM scores to real-world gaming performance because applications are just starting to become multithreaded. But that's not the fault of 3dM. There is no benchmark on this earth that can accurately measure real-world performance like real-world gaming benchmarks. The two (multithreaded apps/3dM) are just out of sync right now. By the next release of 3dM, that may change.
But again, the most important thing to recognize here is that the original tests between the R600 and G80 were ran on the same test bench. Same processors. Therefore they are pretty accurate measures of what we can expect. What does seem to lack in accuracy pretty severely right now are test runs with single 8800GTXs and test runs with 8800GTX SLI. I scored 12,203 on a single GTX and 15,941 on SLI. That's not indicative at all of real-world performance, obviously. The difference may have to do with the fact that 3dMark doesn't multi-thread all of the runs--just the CPU benchmark. So the graphics rendering in SLI is all done on a single-core, possibly?
There is no benchmark on this earth that can accurately measure real-world performance like real-world gaming benchmarks.
Exactly my point.
The two (multithreaded apps/3dM) are just out of sync right now. By the next release of 3dM, that may change.
Yeah it is always a cat and mouse game with Futuremark, but I also think the games are becoming more diversified due to multi threading. Futuremark should release genre benchmarks. RTS and FPS being the most prominent ideas that come to mind because they tend to stress subsystems (and core loads) differently.
But again, the most important thing to recognize here is that the original tests between the R600 and G80 were ran on the same test bench. Same processors. Therefore they are pretty accurate measures of what we can expect.
Agreed.
What does seem to lack in accuracy pretty severely right now are test runs with single 8800GTXs and test runs with 8800GTX SLI. I scored 12,203 on a single GTX and 15,941 on SLI. That's not indicative at all of real-world performance, obviously.
Another reason why I always say 3dmark leans heavy in the CPU department, especially since most run at the lower res.
The difference may have to do with the fact that 3dMark doesn't multi-thread all of the runs--just the CPU benchmark. So the graphics rendering in SLI is all done on a single-core, possibly?
GPU drivers are multi threaded already, but I am not sure how efficient they are at balancing the load across cores with SLI, so the futuremark app could make a significant difference.
Man i'm starting to look forward the R600 just because of STALKER. if it plays the game @1600x1200,16xHQAF smoothly with no slowdowns .. i'm all for it. & if it can get AA to work with deffered lighting (even if DX10 only) at smooth framerate then i'll get it the day its out!
LORD-eX-Bu
03-23-07, 01:16 AM
I loved my 1900XTX but sadly it died..
then I got a 8800 GTS and that was freaking nuts!!
then vista came and soured it all up... and it died.
hope AMD can deliver decent drivers at launch cuz I will probably be getting the new XTX:D
Xion X2
03-23-07, 02:03 AM
Man i'm starting to look forward the R600 just because of STALKER. if it plays the game @1600x1200,16xHQAF smoothly with no slowdowns .. i'm all for it. & if it can get AA to work with deffered lighting (even if DX10 only) at smooth framerate then i'll get it the day its out!
Why?
Does your 8800GTX not play Stalker at a smooth framerate with high settings? If not, then something's wrong somewhere.
Why are you so sold on the R600 when it's a whopping 200 points better than the 8800GTX in 3dM and we've yet to see any DX10 benchmarks between the two cards?
Why?
Does your 8800GTX not play Stalker at a smooth framerate with high settings? If not, then something's wrong somewhere.
Why are you so sold on the R600 when it's a whopping 200 points better than the 8800GTX in 3dM and we've yet to see any DX10 benchmarks between the two cards?
At 1600x1200,16xAF its not entirely smooth. outdoors framerate is around 30-50fps & in some heavy firefights it goes below 30fps.
in indoor levels its worse due to the high number of dynamic lights & other special effects i believe , its goes in 20s.
other people played the game with same & better configs than mine confirmed this so its not me & i'm pretty careful on my settings & readings.
remember all that without FSAA.
why i'm sold ? well i've already stated my conditions to approve the R600 .. i dont care about the 3dmark score if it plays STALKER better than the 8800GTX. & again these are still not the official numbers & hence its the rumor mill!
EDIT : for more info check the STALKER Feedback thread.
Xion X2
03-23-07, 05:20 AM
Something is wrong if Stalker's running in the 20's and 30's. That game doesn't even look all that good.
It's probably a coding issue. The 8800GTX should be spinning that game around like it's nothing.
Something is wrong if Stalker's running in the 20's and 30's. That game doesn't even look all that good.
It's probably a coding issue. The 8800GTX should be spinning that game around like it's nothing.
The games does look good especially the indoors .. heres some screens posted in the feedback thread :
http://www.nvnews.net/vbulletin/showpost.php?p=1202822&postcount=135(indoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1202970&postcount=153(indoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1202805&postcount=133(outdoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1202947&postcount=152(outdoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1203119&postcount=165
3dmark tends to put more of an emphasis on the CPU side of performance compared to what most games need. Obviously more so on the older versions.
If you pop in a QX6700 you'll get an instant 10% boost on your overall 06 score .vs a similarly clocked C2D, but that 10% doesn't translate into any real world performance except for maybe in SupremeCommander.
yes, but if you run one card, remove it, slap in the other card, and run it on the same system at the same settings, it is a good way to gauge which card is better.
to the guy who said that drivers will increase the performance 2-fold? get real, thats the most ridiculous thing ive ever heard.
and if the 8800GTX only loses by a couple hundred points, the 8900GTX will stomp a mudhole in the R600. i was really lookin forward to the R600, but if it's only on par with the 8800, then ill probably go with an 8800GTS.
Zelda_fan
03-23-07, 11:53 AM
BWAHAHAHAHAHA
Nvidia -> :nutkick: <-AMD
I am getting sick and tired of this fanboy crap(and this is comming from someone who owns an 8800).
It's a 3DMARK SCORE for crying out loud. Not only that, we don't even know if that score is legit. Wait for real benchmarks.
Dreamingawake
03-23-07, 01:26 PM
Any card 8800 + all the way to the 8900 range and Ati's new Dx10 cards will play Crysis and Alan Wake, UT3 JUST FINE.
Xion X2
03-23-07, 03:35 PM
The games does look good especially the indoors .. heres some screens posted in the feedback thread :
http://www.nvnews.net/vbulletin/showpost.php?p=1202822&postcount=135(indoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1202970&postcount=153(indoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1202805&postcount=133(outdoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1202947&postcount=152(outdoors)
http://www.nvnews.net/vbulletin/showpost.php?p=1203119&postcount=165
The indoors definitely looks better than the outdoors. The outdoors is kinda "Meh" for me, but to each his own. :)
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.