Given the rumours of RV610 being a small, 64-bit chip and G86 being 128-bit, is that really the case?
The Official G84/G86 Rumours & Speculation Thread
- Thread starter Arun
- Start date
The 8500 GT will probably replace the old 7300 GT.
The others are just low-end stuff, without too much concern about game performance, although DX10 and image quality is there.
Wasn't the 8500GT on G84 like 8600GT/GTS?
Either way, what each chip competes against is more about the prices than anything else, if AMD will price some low end RV630 to the same point as 8500GT, then 8500GT will compete against RV630, not 610, etc.
No, 8500 GT is a G86 with souped up clocks and a 128bit bus.
As i said above, no word yet on the 8400/8300 range bus width.
Somebody got a hold of one and ran some benches - http://www.au-ja.de/review-club3d8600gts-4.phtml
3dmark06 4xAA, 8xAF
8600GTS - 5204
X1950 Pro - 4162
7900GT - 3991
Benchmarks on six games to follow.
Edit: Gets trounced in F.E.A.R, bandwidth ?
8600GTS - 36
X1950 Pro - 55
7900GT - 49
3dmark06 4xAA, 8xAF
8600GTS - 5204
X1950 Pro - 4162
7900GT - 3991
Benchmarks on six games to follow.
Edit: Gets trounced in F.E.A.R, bandwidth ?
8600GTS - 36
X1950 Pro - 55
7900GT - 49
Quite odd.
It appears to be much faster in 3DMark'06, but in F.E.A.R. it's slower, too slow to be accurate.
The 8600 GT preview had a similar tendency.
Is it a driver problem ? I haven't had much luck with the 100.95's, but on the other hand i still don't own any DX10 hardware...
A Geforce "8300 GT" just popped up, and it has a 128bit bus:
http://img267.imageshack.us/img267/6860/others012copyro5.jpg
Source:
http://my.ocworkbench.com/bbs/showthread.php?t=61654
http://img267.imageshack.us/img267/6860/others012copyro5.jpg
Source:
http://my.ocworkbench.com/bbs/showthread.php?t=61654
Did anyone spot the problem with this?
yes, 3dmark06 doesn't give a combined scores for G7x when AA is used because it can't do HDR+AA in GT3 and GT4
Last edited by a moderator:
While reading that thingy I was somewhat puzzled with the author's conclusion. The 8600GTS seems to show a weakness in FEAR and Doom3 and he attributes the difference to bandwidth? Of course can it also play it's role yet given that both applications go heavy on stencil shadows and the G84 seems to have (according to rumours) only two ROP partitions, the tests and relevant conclusions are rather sluggish.
All it would had taken from his side is to run all cards once more through FEAR for instance yet this time with reduced shadow settings. If in such a case the G84 still poses the same difference to the other two cards, it could be a driver bubble. If however the results would be as positive as in other tested games, then the first thing to point fingers at would be the ROPs and not by all means "bandwidth".
All it would had taken from his side is to run all cards once more through FEAR for instance yet this time with reduced shadow settings. If in such a case the G84 still poses the same difference to the other two cards, it could be a driver bubble. If however the results would be as positive as in other tested games, then the first thing to point fingers at would be the ROPs and not by all means "bandwidth".
No offence.. but that review was made in china
So what are the actual specs of the 8600GTS?
Here's complete 06 test, this one's from Australia:
http://forums.whirlpool.net.au/forum-replies.cfm?t=715719
The SM3.0 test score is really low(around the same you would get with X1950Pro paired with Athlon64 and they are using E6600), does the SM3.0 test use heavily stencils? Or is it because HDR rendering sucks bandwidth?
Last edited by a moderator:
No offence.. but that review was made in china
So what are the actual specs of the 8600GTS?
I don't see why I should be offended in any case. Would it had made any difference if it would had been written in Alaska or Germany?
No idea, but it sounds like a 4 cluster/2 ROP partition thingy to me.
Similar threads
