AMD: R7xx Speculation
- Thread starter Unknown Soldier
- Start date
- Status
Slappi,
How do you know that? Do you have some actual numbers to support that that you can share with us or is this your personal opinion?
It seems there are getting more and more people stating as fact performance superiority of one product over another. I understand if that is your opinion which may or may not be backed by history on older architectures.
Is R700 Crossfire... good chance but there is a possibility that we may have some type of surprise. I am not sure either way. If its Crossfire, then what yyou state is probably correct but it may be something new. We can always hope as that represents progress.
How do you know that? Do you have some actual numbers to support that that you can share with us or is this your personal opinion?
It seems there are getting more and more people stating as fact performance superiority of one product over another. I understand if that is your opinion which may or may not be backed by history on older architectures.
Is R700 Crossfire... good chance but there is a possibility that we may have some type of surprise. I am not sure either way. If its Crossfire, then what yyou state is probably correct but it may be something new. We can always hope as that represents progress.
Reading your post history leads me to believe you arent a press member to have signed NDA, so most likely its an educated guess ..
Have they solved the weird FPS drops and stuttering too? 3870X2 is a benchmark card, but not a true gamer card.
I saw a discussion on this matter on a Czech forum. Perhaps the stuttering on the HD 3870 X2 is caused by insufficient PCIe bandwidth - each chip effectively has half of the 16 lanes available and the PLX bridge supports only PCIe 1.1. According to a user who has two 3870's in CrossFire on an X38 board, this setup fares better in terms of minimal FPS. R700 is supposed to have a PCIe 2.0 compliant bridge so maybe that'll solve the issue.
The 3870X2 is too weak of a base, too unbalanced in its config, too rushed of its launch, too lame in front of the pillars of AA/AF. And yet most of the time it was good enough for the price, and the performance ratio it brought was nice too.
The 8.3s were weak sauce with a lot of reviews, which directly just used DX10 which wasn't particularly optimized with it. Since 8.4, the Vantage hotfix, and 8.5, a lot has changed in this area and I wouldn't be surprised if the chip comes off a lot more positively this time in the hands of understanding reviewers.
The 8.3s were weak sauce with a lot of reviews, which directly just used DX10 which wasn't particularly optimized with it. Since 8.4, the Vantage hotfix, and 8.5, a lot has changed in this area and I wouldn't be surprised if the chip comes off a lot more positively this time in the hands of understanding reviewers.
Ok see yah in a few weeks.
I remember you from R3D. No one believed you there either
IOW: if you have something to back up your statements - feel free to prove us wrong.
Last edited by a moderator:
annihilator
Newcomer
I have absolutely no reason to believe 4870x2 is going to produce nice latency frames with a synchronization like a single GPU system. And unless it actually does that, no one can show me a FRAPS benchmark showing 60 FPS for 4870x2 and 50 FPS for GTX 280 and claim that it proves 4870x2 is actually "faster" than GTX 280.
Well, if R700 is actually a UMA architecture this should help AFR rendering performance tremendously and basically eliminate synchronization issues. Of course, it's only a rumor that R700 is more than just Crossfire on a card.
I remember you from R3D. No one believed you there either
IOW: if you have something to back up your statements - feel free to prove us wrong.
You don't understand me or where I am coming from.
See yah in a few.
Note that from the "common" multi-gpu rendering schemes AFR, SFR, and SuperTiling (well conceptually SuperTiling is a form of SFR) only AFR really has latency issues. There may be some hope for SuperTiling (even if it means doubling the vertex work), though noone seems to really know much... Though if you believe the "built for multi-gpu setups" rumours, maybe it's not AFR - AFR is really a solution for dual-gpu at best (ok I know you can get triple-SLI and quad-crossfire using AFR but it's next to useless imho, at least for interactive stuff).
LOL, and a statement like this is going to change what, precisely?
If you have something to share, please do. Otherwise, you're just blowing hot air, so to speak.
annihilator
Newcomer
Well, this is the second time I'm using multi GPU; my former multi-GPU rig was an X1900 XT Crossfire, and now I have 8800GT SLI. I've tried most of the different modes.
SFR's performance is very low and if someone's willing to pay extra for another GPU, they definitely want the AFR numbers, not SFR. SFR's performance vastly needs to be improved to be an actual incentive for a multi-GPU setup.
I don't really see Nvidia or ATI move their butts to dedicate some engineering resources into improving SFR as long as all major hardware review sites include AFR scores for SLI without mentioning (or maybe even without knowing) that in games such as Crysis, Call of Juarez or Lost Planet, nearly all of the extra frames per second you see (than the single-gpu setups) have no effect in improving the smoothness of game play.
JasonCross
Newcomer
I literally just got out of a meeting with ATI, and honestly, I'd just advise everyone to calm down and wait for direct info from ATI/AMD.
A lot of these rumors and speculation are just plain wrong, some of the numbers floating around out there are made up, and the ones that aren't are often based on older drivers that aren't performing as well as the newer stuff.
I'm under NDA so I obviously can't share any performance numbers with you, so don't ask. I wouldn't believe them anyway, because I never believe the performance numbers ATI or Nvidia gives me at face value. I believe the numbers I get when I test things myself.
A lot of these rumors and speculation are just plain wrong, some of the numbers floating around out there are made up, and the ones that aren't are often based on older drivers that aren't performing as well as the newer stuff.
I'm under NDA so I obviously can't share any performance numbers with you, so don't ask. I wouldn't believe them anyway, because I never believe the performance numbers ATI or Nvidia gives me at face value. I believe the numbers I get when I test things myself.
- Status
