AMD: Southern Islands (7*** series) Speculation/ Rumour Thread

slide30fnp54.jpg


from obr

EDIT: lol all slides



i see AF fix slide!!!
 
From my understanding there shouldn't be a performance difference for a single card from PCIe 2 to PCIe 3. Any proof showing otherwise?
I always thought that PCIe 3 would be something for those PCIe SSD cards (forget the name of them).
 
ECH said:
From my understanding there shouldn't be a performance difference for a single card from PCIe 2 to PCIe 3. Any proof showing otherwise?
I always thought that PCIe 3 would be something for those PCIe SSD cards (forget the name of them).
Click to expand...

There shouldn't, or only marginal (in some cases, PCIe 1.x x16 / 2.x x8 is limiting the performance already on single chip cards by couple %'s)


DegustatoR said:
Yay, paper launch.
Click to expand...
Is it? HD5 for example had quite limited quantities in beginning, but it wasn't paper launch by any definition
 
DieH@rd said:
Even though there will not be a lot of them on the market, at least we will get a tons of benchmarks and user impressions.
Click to expand...
I'm all for an early launch for the same reason, what better way to spend the holidays going through benchmark data, but I fear for Charlie's savage reaction to a soft launch!

Also, according to the slide, 2048 steam processors, up only from 1560? Isn't that's a bit disappointing for a full node jump?

I understand that it's supposed to be more efficient (for compute loads at least) and there must be additional overhead due to ECC and various other compute features, but still. A trade-off between area for the 384-bit MC and shader core area? Going by the GTX 560 Ti 448 and GTX570 benchmarks, a lot of performance seems to be determined only by bandwidth, so this may be a sensible choice after all.
 
silent_guy said:
Also, according to the slide, 2048 steam processors, up only from 1560? Isn't that's a bit disappointing for a full node jump?

I understand that it's supposed to be more efficient (for compute loads at least) and there must be additional overhead due to ECC and various other compute features, but still. A trade-off between area for the 384-bit MC and shader core area? Going by the GTX 560 Ti 448 and GTX570 benchmarks, a lot of performance seems to be determined only by bandwidth, so this may be a sensible choice after all.
Click to expand...
Yes, compute features do matter here much more. And don't forget that DP performance increase is much more that the plain difference in ALU count.
 
Is it sane to soften my expectations about a 20%-40% jump in performance rather than a 50%-70% performance jump?
 
silent_guy said:
Also, according to the slide, 2048 steam processors, up only from 1560? Isn't that's a bit disappointing for a full node jump?
Click to expand...
Completely different architecture, you can't just compare counts ;)

Arty said:
Is it sane to soften my expectations about a 20%-40% jump in performance rather than a 50%-70% performance jump?
Click to expand...
I'd expect closer to GTX590 than 580, but definately not above 590
 
You must log in or register to reply here.

Similar threads

I
AMD: Pirate Islands (R* 3** series) Speculation/Rumor Thread
140 141 142
Replies
3K
Views
621K
kalelovil
K
N
AMD: Volcanic Islands R1100/1200 (8***/9*** series) Speculation/ Rumour Thread
141 142 143
Replies
3K
Views
685K
pMax
pMax
S
AMD: Sea Islands R1100 (8*** series) Speculation/ Rumour Thread
69 70 71
Replies
1K
Views
390K
AnarchX
A
Back
Top