Triple-buffering normally doesn't work with D3D games unless you use a third party tool like DirectX Tweaker. Are you sure you did this? If not, then SLI may not be the problem.
Elfear
Diamond Member
- May 30, 2004
- 7,126
- 738
- 126
I never tried DXTweaker to tell you the truth. I perused NvNews for awhile trying to find a solution to the tearing. The lack of success I found with trying to get DXTweaker to make Triple Buffering work with the games I played was disheartening. I did a lot of research in an effort to make my 7800GTs play nice, but I never got it resolved. If you do a search over at NvNews and here at AT, you can see it's definetely a limitation of SLI.
Guezz, thank you for that excellent post. Will you keep it updated for years to come, and where is the one main location where you think it will last the longest time, 3dmark forum?
Edited to add 2 observations and 2 questions.
Observation 1: I believe you should have the ATI X1900 XT/X as slightly better/faster than equivalently marketted nVidia 7800/7900 GTX's. Particularly in new shader heavy games like Oblivion where those market leading 48 pixel shader units seem to come in handy. I mean it doesn't quite make sense to me that you say the X1900XT is slightly slower while the X1900XTX is slightly faster than the 7900 GTX considering its very minor overclock of 25/50. For sure, the 7950 GX2 is slightly ahead of both current ATI flagships, (just keep in mind it's really 2 cards in 1 SLI card configuration and may suffer from any SLI bugs out there and extra power/heat). Also you may want to make a consideration for Image Quality where most consumers/reviewers seem to favor ATI in the IQ department.
Observation 2: Woot, according to that interesting PSU calculator, my system with a new X1900 XT should only need a PSU rated at 429w! Mine's an Antec TrueBlue 480w so I should be ok in theory (I was worried about this and did not want to waste more money). I hope it takes into account that most PSUs are about 80% effective...
Question 1: In all your research, do you believe that the general trend in overall performance can generally be attributed to the few variables listed underneath the cards? (pixel shader units, TMUs, ROPs, vertex shader units, core/mem speed). In other words, if there is a very strong correlation, I wonder if some simple formula or calculator can be devised to allow us to just plug in the values for the card we are considering and get a raw simple number that shows the relative speed/performance of the card.
Question 2: What does "* Games can be depended on profiles for optimal performance. This is the reason for still profile support despite all games being supported." mean?
Edited to add 2 observations and 2 questions.
Observation 1: I believe you should have the ATI X1900 XT/X as slightly better/faster than equivalently marketted nVidia 7800/7900 GTX's. Particularly in new shader heavy games like Oblivion where those market leading 48 pixel shader units seem to come in handy. I mean it doesn't quite make sense to me that you say the X1900XT is slightly slower while the X1900XTX is slightly faster than the 7900 GTX considering its very minor overclock of 25/50. For sure, the 7950 GX2 is slightly ahead of both current ATI flagships, (just keep in mind it's really 2 cards in 1 SLI card configuration and may suffer from any SLI bugs out there and extra power/heat). Also you may want to make a consideration for Image Quality where most consumers/reviewers seem to favor ATI in the IQ department.
Observation 2: Woot, according to that interesting PSU calculator, my system with a new X1900 XT should only need a PSU rated at 429w! Mine's an Antec TrueBlue 480w so I should be ok in theory (I was worried about this and did not want to waste more money). I hope it takes into account that most PSUs are about 80% effective...
Question 1: In all your research, do you believe that the general trend in overall performance can generally be attributed to the few variables listed underneath the cards? (pixel shader units, TMUs, ROPs, vertex shader units, core/mem speed). In other words, if there is a very strong correlation, I wonder if some simple formula or calculator can be devised to allow us to just plug in the values for the card we are considering and get a raw simple number that shows the relative speed/performance of the card.
Question 2: What does "* Games can be depended on profiles for optimal performance. This is the reason for still profile support despite all games being supported." mean?
Thanks!
I will try, and yes, use the guide at Futuremark.
Those 25 MHz and 100 MHz results in a improved fillrate of 0.4 GPixel/s and memory bandwith 3.2 GB/s - it's a bit faster. It's true that in reviews the difference between XT and XTX aren't large. When I rank cards it's to show the overall performance (D3D, OpenGL, AA, resolution, etc). A 7900 GTX can beat a X1900 XT in various D3D games and the overall OpenGL performance is better.
I have taken somewhat into concideration "Pixel shader intensive" games with this quote: "When more pixel shader intensive games are released an increase in performance is expected."
IQ
Have you read the ?Technologies? section?
A 7950 GX2 isn't just slightly faster than a X1900 XTX.
To be honest I don't know.
If I must give you a few formulas it must be these:
Pixel Fill-rate (pixel output): number of ROPs * core frequency
Texture Fill-rate: number of TMUs * core frequency
Memory bandwith: (width of bus * memory frequency) / 8
It should be noted that you can?t blindly trust these numbers (theoretical maximum) to give the correct real-life performance. Excellent example to this is X1900 XTX vs. X850 XT PE
X1900 XTX likes pixel shading intensive games while 7900 GTX likes texture (fill-rate) intensive games. It should be noted that I think pixel shading will play a larger role in the future.
nVidia is more efficient clock-for-clock - architectonical differences can result in different performance.
After the 8x. drivers you can use ?global settings: AFR, AFR2 or SFR? to force one SLI rendering to all games. If you choose SFR and a game?s performance scales better with AFR ? a driver profile is then used. Also some games may have redering issues if you use the ?incorrect? rendering.
Big update!
18 September 06: Added: X1650 Pro, X1900 XT 256MB, X1950 XTX, 7300 GT, 7900 GS and 7950 GT. TAAA for NV4X. Corrected: Terrible typo 7900 GTX vs. 7900 GT (30%, not 15%!) and updated X1900 GT performance.
18 September 06: Added: X1650 Pro, X1900 XT 256MB, X1950 XTX, 7300 GT, 7900 GS and 7950 GT. TAAA for NV4X. Corrected: Terrible typo 7900 GTX vs. 7900 GT (30%, not 15%!) and updated X1900 GT performance.
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Um...
32xAF? i dont think such a level of anistropic filtering exists.
Maybe you could add in SLi/Crossfire performance ratings.
Also its Transparency Anti Aliasing. There is no word "adaptive". The 6 series hence the nv4x series can use this feature as well.
For the G8x, you should put both CSAA and the 'perfect AF' under the heading of lumenex technology/engine.
Could also add in the quantum physics technology for G80. This havent been used yet since havokFX haasnt really been launched.
If youve added technologies such as 3dc, maybe you should add in the other stuff like ultra shadow 2 technology for nVIDIA for example. This is one of the reasons that nVIDIA performs much better than ATi cards when lots of stencil shadows are used.
Also, you could add stuff like the process these cards are on, the code names e.g NV40, and maybe thing that are important to a consumer, e.g heat/noise by taking in people's personal experience with the cards.
Other than this, good stuff~~~ :thumbsup:
32xAF? i dont think such a level of anistropic filtering exists.
Maybe you could add in SLi/Crossfire performance ratings.
Also its Transparency Anti Aliasing. There is no word "adaptive". The 6 series hence the nv4x series can use this feature as well.
For the G8x, you should put both CSAA and the 'perfect AF' under the heading of lumenex technology/engine.
Could also add in the quantum physics technology for G80. This havent been used yet since havokFX haasnt really been launched.
If youve added technologies such as 3dc, maybe you should add in the other stuff like ultra shadow 2 technology for nVIDIA for example. This is one of the reasons that nVIDIA performs much better than ATi cards when lots of stencil shadows are used.
Also, you could add stuff like the process these cards are on, the code names e.g NV40, and maybe thing that are important to a consumer, e.g heat/noise by taking in people's personal experience with the cards.
Other than this, good stuff~~~ :thumbsup:
Originally posted by: Cookie Monster
You're correct, and it was corrected months ago, in the former Norwegian edition ... Thank you for noticing it.
The suggestion is interesting and has crossed my mind earlier. I will reconsider the issue, although it should be noted that this would require a lot of work.
I will look at it, also re-read the section (the end of it) about NV4X ...
I don't think this is of importance since it's clearly G8X only technologies as presented by the guide, this is IMO enough.
Interesting, if so, also ATI's equivalent should be mentioned as well.
It's noted. Can you mention unbiased sites where this feature is discussed (not nVidia hype crap).
I will think about adding process and detailed code names.
Noise is a touchy issue since it?s hard to accurately quantify such a thing. I have earlier thought about it but was not adviced to do so by a physics teacher (a Beyond3D writer). Also, personal experience of sound is highly subjective so comparisons are impossible. Heat has too many unknown factors (case ventilation, room temperature, ect.).
Thanks, and your criticism aren't all that severe.
Thanks a lot for your reply. :beer:
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
In the Video Decoding section:
I suppose you meant market?
REad through this and maybe could write some more stuff to the section e.g pure video HD.
Im sure you dont need the pure video decoder anymore or am i wrong? someone correct me please.
So you can see that AVIVO is evenly matched with pure video as of today.
I suppose you meant market?
REad through this and maybe could write some more stuff to the section e.g pure video HD.
Im sure you dont need the pure video decoder anymore or am i wrong? someone correct me please.
So you can see that AVIVO is evenly matched with pure video as of today.
Great guide!
Side note - your Pure Video link is broken.
- G7X
* Very good decoding performance, also in the new formats: VC-1 and H.264.
* Purevideo costs: $20-50
* The complete list of supported formats
Side note - your Pure Video link is broken.
- G7X
* Very good decoding performance, also in the new formats: VC-1 and H.264.
* Purevideo costs: $20-50
* The complete list of supported formats
18 February 07: Added: 8800 GTS 320MB, X1950 GT and provided more information about G80 performance. Update: video decoding (e.g. added G8X).
happy medium
Lifer
- Jun 8, 2003
- 14,387
- 480
- 126
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 20K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 13K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Facebook
Twitter