ATI Removes NDA on Performance Previews for 4850 (#'s Inside)

MrMike said:
Anand's review states it's under NDA

http://www.anandtech.com/video/showdoc.aspx?i=3338

I can't find another review that specifically states information with reference to the architecture.

I'm just going to unsubscribe from this thread now because you're not interested in actually defending your posts with any real evidence. Take it easy.
Click to expand...
did you even read the title of the thread you are in?? its NOT under NDA and if you cant figure that out then why should I go through all the trouble of finding info that you can easily find for yourself?
 
cannondale06 said:
did you even read the title of the thread you are in?? its NOT under NDA and if you cant figure that out then why should I go through all the trouble of finding info that you can easily find for yourself?
Click to expand...

Don't spoon feed him. Plenty of info around.
 
This reviewer got it up to the 700 max.

And much of the architecture is indeed still under NDA. Yesterday (Thursday) ExtremeTech.com had a article with lots of information about the architecture with official (I think) block diagrams, but today it was pulled. Apparantly previews with benches are allowed but we won't be getting the finer details until next week.
 
The extremetech review mentioned 800 SPs, 40 TMUs, removal of the ringbus in favor of a hub and new distributed bus concept. The overclocking may well be the result of low voltages. Rumors on the 4870 show significant bumps in voltage and it is running at 750MHz if rumors are correct.

Also when overclocking, achieving 4870 speeds with a 4850 would probably be difficult. Odds are memory bandwidth is a significant bottleneck for the 4850 and there is likely only so much of a gain to be had. The 4870 could very well be bordering on double the bandwidth.

Memory Clocks:
4850(GDDR3) -> 1800MHz (effective)
4870(GDDR5) -> 3600MHz (effective)
 
Anarchist4000 said:
The extremetech review mentioned 800 SPs, 40 TMUs, removal of the ringbus in favor of a hub and new distributed bus concept. The overclocking may well be the result of low voltages. Rumors on the 4870 show significant bumps in voltage and it is running at 750MHz if rumors are correct.

Also when overclocking, achieving 4870 speeds with a 4850 would probably be difficult. Odds are memory bandwidth is a significant bottleneck for the 4850 and there is likely only so much of a gain to be had. The 4870 could very well be bordering on double the bandwidth.

Memory Clocks:
4850(GDDR3) -> 1800MHz (effective)
4870(GDDR5) -> 3600MHz (effective)
Click to expand...
well I hope the heat issues are just a result of a slow fan that can easily be adjusted. to me 95 degrees under load is ridiculous for a 55nm core at 625mhz.
 
cannondale06 said:
well I hope the heat issues are just a result of a slow fan that can easily be adjusted. to me 95 degrees under load is ridiculous for a 55nm core at 625mhz.
Click to expand...

It is the slow fan. At least one review mentioned the fan only running at 14% of max speed. ATI set it to the lowest it could go without hurting the card, so it would be almost silent. I bet setting it at 30% or 40% would not increase the noise by much at all, but would provide much better cooling.
 
snowysnowcones said:
The past week, I've heard more about this "microstutter" then I ever have before.. ever. Someone care to explain what exactly it is? And can you actually notice it? From the time I've played with two 8800Ultras, I've never noticed it - nor have I ever read reviews mentioning it.
Click to expand...

So, several people have replied with discussions on the stuttering that can occur when assets are loaded swapped in and out of the page file when there is insufficient physical memory for all of the data. This is stuttering, but it's not "microstuttering".

The issue is that when your CF or SLi system is in AFR mode there's the possibility that the frames won't be drawn in a particularly smooth manner. If Card B displays its frame just after Card A, then there will be a relatively long gap before Card A draws the third frame. Obviously, the inter-frame delay is only fractions of a second longer than it usually would be (hence microstutter) but our eyes can still perceive the difference in 'smoothness' of the display, even though the number of frames drawn in a second might be the same as a configuration without microstutter.

I've never seen it firsthand and I have no idea if it actually matters at all unless you're looking for it quite intently. Certainly nothing as bad as the Ram-stuttering described earlier. Check out this thread for more information.

Edit: the chart at this link shows microstutter well. The line has a lower slope (higher overall framerate) but is less smooth. There's a video there too but I certainly can't tell the difference and I highly doubt its at all possible to see the difference on video that isn't at very high framerate.
 
Kaldskryke said:
So, several people have replied with discussions on the stuttering that can occur when assets are loaded swapped in and out of the page file when there is insufficient physical memory for all of the data. This is stuttering, but it's not "microstuttering".

The issue is that when your CF or SLi system is in AFR mode there's the possibility that the frames won't be drawn in a particularly smooth manner. If Card B displays its frame just after Card A, then there will be a relatively long gap before Card A draws the third frame. Obviously, the inter-frame delay is only fractions of a second longer than it usually would be (hence microstutter) but our eyes can still perceive the difference in 'smoothness' of the display, even though the number of frames drawn in a second might be the same as a configuration without microstutter.

I've never seen it firsthand and I have no idea if it actually matters at all unless you're looking for it quite intently. Certainly nothing as bad as the Ram-stuttering described earlier. Check out this thread for more information.

Edit: the chart at this link shows microstutter well. The line has a lower slope (higher overall framerate) but is less smooth. There's a video there too but I certainly can't tell the difference and I highly doubt its at all possible to see the difference on video that isn't at very high framerate.
Click to expand...

its funny... the people who are on here complaining about SLI/X-fire microstuttering either dont use or have ever used Sli
 
I use Crossfire now in my third rig, and have tried multi-gpu and SLi.

I've seen "micro stutter" and wild FPS fluctuations/crashes. Some times just panning around quickly in a game will produce this behavior.
 
Blacklash said:
I use Crossfire now in my third rig, and have tried multi-gpu and SLi.

I've seen "micro stutter" and wild FPS fluctuations/crashes. Some times just panning around quickly in a game will produce this behavior.
Click to expand...
Ive seen that plenty of times with single cards too especially in older games. I think a lot of these issues have nothing to do with multi gpu setups.
 
Blacklash said:
I use Crossfire now in my third rig, and have tried multi-gpu and SLi.

I've seen "micro stutter" and wild FPS fluctuations/crashes. Some times just panning around quickly in a game will produce this behavior.
Click to expand...

and u never seen that with single cards?... i know i have
 
To me, people who deny the effects of microstuttering are in the same boat with "I can't tell the difference higher than 30 fps" and "my LCD doesn't have input lag or ghosting" people. Maybe they have poor vision, maybe they are not seeing what they don't want to see on their hard earned possessions, maybe they are just very non-discriminant. Or maybe I'm just very sight sensitive after wasting inordinate amounts of time playing online FPS ever since qtest. But the fact is, if you sat me down in front of two setups with and without SLI/crossfire, I'd tell you which one is which a hundred times out of a hundred based on microstuttering, and to me that's not a good thing.

But hey, if you can stand playing games at 30fps on your TN panel, then this won't be an issue in the least for you.
 
How does this card in CF go against 8800GT's in SLI?

Edit: This will be @ 24" (can't remember the res off the top of my head).
 
jonjon0nline said:
How does this card in CF go against 8800GT's in SLI?

Edit: This will be @ 24" (can't remember the res off the top of my head).
Click to expand...

2 4850's should easily beat 2 8800GT's, except maybe in Crysis, which doesn't seem to like crossfire much, but there's always new drivers, so who knows?!
 
Aix said:
To me, people who deny the effects of microstuttering are in the same boat with "I can't tell the difference higher than 30 fps" and "my LCD doesn't have input lag or ghosting" people.
Click to expand...

To me, people who demand the problems are there, to people who state they don't have any of the problems, are in the same boat as the "McCarthyism" and "Elitist" people.
 
Aix said:
To me, people who deny the effects of microstuttering are in the same boat with "I can't tell the difference higher than 30 fps" and "my LCD doesn't have input lag or ghosting" people. Maybe they have poor vision, maybe they are not seeing what they don't want to see on their hard earned possessions, maybe they are just very non-discriminant. Or maybe I'm just very sight sensitive after wasting inordinate amounts of time playing online FPS ever since qtest. But the fact is, if you sat me down in front of two setups with and without SLI/crossfire, I'd tell you which one is which a hundred times out of a hundred based on microstuttering, and to me that's not a good thing.

But hey, if you can stand playing games at 30fps on your TN panel, then this won't be an issue in the least for you.
Click to expand...

hey if it makes u feel better i can pretend i see microstutter and it ruins my game play experience if that helps?? i really dont know what to tell u
 
You must log in or register to reply here.
Back
Top