NVIDIA GF100 & Friends speculation

CPU i7 965
MB RAMPAGE II EXTREME
RAM 3G DDR3

cjgd1.jpg


http://www.mobile01.com/topicdetail.php?f=298&t=1475786&p=1
 
195% - gtx480
183% - 5870 1GB
161% - gtx470
153% - 5850

But between the gtx285 and the 5850, difference is high.
Therefore, the tested games a little bit like better Ati.;)

If that were true I'd expect to see the 4870x2 a lot closer to the gtx295, and the 4890 beating the gtx275.
 
If there's any truth in that graph then it's 50-60% faster than a 285. Not great, but not a disaster either. Kinda boring after the long wait but I'll reserve final judgment until I see how it fares in the titles where it matters.
 
If there's any truth in that graph then it's 50-60% faster than a 285. Not great, but not a disaster either.
Not that this is surprising now, but I'd say it's disappointing. You are right it's not a disaster, but considering the complexity of the chip compared to gt200, I can't see much of the supposed efficiency gains. Kinda like G92->gt200, a chip with twice the complexity but only 50% more performance. Not that Cypress had perfect scaling in that sense from rv770, but it certainly doesn't look like nvidia could make that perf/transistor deficit they have since rv7xx any smaller this generation (not even the "missing" shader cluster or slightly higher clock would change things significantly).
 
Not that this is surprising now, but I'd say it's disappointing. You are right it's not a disaster, but considering the complexity of the chip compared to gt200, I can't see much of the supposed efficiency gains. Kinda like G92->gt200, a chip with twice the complexity but only 50% more performance. Not that Cypress had perfect scaling in that sense from rv770, but it certainly doesn't look like nvidia could make that perf/transistor deficit they have since rv7xx any smaller this generation (not even the "missing" shader cluster or slightly higher clock would change things significantly).

Maybe the problem is the missing TMUs, according to Neliz?
 
mao5 said "synthetic" for the record. What that graph is supposed to tell me is that a 5970 is on average by nearly 90% faster than a 1GB/5870 yeahrightsureok.

mao5 said "synthetic score based on single tests", and when I assumed it means similar to Tom's "sum of all fps", he said "seem correct"
Which IMO indicates that the benches aren't synthetic, just the score as it's not representative of any actual game, but all games put together
 
If there's any truth in that graph then it's 50-60% faster than a 285. Not great, but not a disaster either. Kinda boring after the long wait but I'll reserve final judgment until I see how it fares in the titles where it matters.

Yeah, definitely not great if true across the board. But it's not that bad either. I do want to see the tesselation results in actual games though, since that is apparently where it should show its muscle.
 
Not that this is surprising now, but I'd say it's disappointing. You are right it's not a disaster, but considering the complexity of the chip compared to gt200, I can't see much of the supposed efficiency gains. Kinda like G92->gt200, a chip with twice the complexity but only 50% more performance. Not that Cypress had perfect scaling in that sense from rv770, but it certainly doesn't look like nvidia could make that perf/transistor deficit they have since rv7xx any smaller this generation (not even the "missing" shader cluster or slightly higher clock would change things significantly).

Another way of looking at it is that nv paid off a *lot* of it's technical debt in one shot with fermi, while AMD will be paying it off incrementally over the next few generations. However, it does appear that there is a certain level of efficiency gap between the two families which remains till now.

Regarding the relative efficiency improvements from the cache hierarchy, I think they'll be be lost by the time NI rolls around.
 
One should remember, too, that the transistor budget they put solely to geometry performance improvements isn't really showing much gains with todays games
 
One should remember, too, that the transistor budget they put solely to geometry performance improvements isn't really showing much gains with todays games
How big is that transistor budget anyway? All the EG chips has two tesselation engines in them and they're perfectly fine with it.
 
One should remember, too, that the transistor budget they put solely to geometry performance improvements isn't really showing much gains with todays games

I'm sure that that will confirmed, with the newer games that should use heavy tesselation.
 
I'm sure that that will confirmed, with the newer games that should use heavy tesselation.

I doubt any game in next year or so will use tesselation heavily enough to give significant benefits for GF100. More likely it'll be like how X1900's beat the 7900's nowadays, but it's well past their age.
 
Back
Top