AMD: R9xx Speculation

[CarstenS]

After seeing the inconsistent text style and alignment of the Unigine benchmark screenshot, I would say that this myth is busted.

Uploaded with ImageShack.us

Interestingly, this screenshot mostly is from a GTX 480 with tessellation at normal instead of extreme (this and the model-field are the only thing changed (and even without the aggressive JPEG-artifacting) - and it delivers almost the same perf as the supposed HD 6800 series card. Coincidence?

 

[rpg.314]

Uploaded with ImageShack.us

 

[mczak]

I was speculating along the same lines. The lowest priced HD 5850 has fallen to $269 at newegg, with HD 5870 at $349. So it seems reasonable enough for the new parts to be priced at $299 and $399. Im betting both the parts will be equipped with 2GB memory standard

I'd agree the fastest model will probably come with 2GB standard. I'm not so sure on the slower one, I could imagine there both 1GB and 2GB being standard versions.

2Gbit GDDR5 chips were also slated for introduction in Q3 so i'd expect Barts to stick with 128 bit so that it could use 4 2Gbit chips to achieve 1 GB. This would be the most cost effective. Juniper was stuck using 8 1Gbit memory chips which increased costs.

I think that'll depend on the availability/pricing of these chips. If a 2gbit chip doesn't cost more than 2 1gbit chips sure that's more cost effective. OTOH I don't think using twice the chips increases costs a lot, clamshell mode makes this easy and shouldn't complicate pcb too much.

 

[thop]

Im sorry for the offtopic, im new, but what exactly is Dave Baumann's job at Amd? is he into marketing.. chip designing.. or something else?

Legend, and Product Manager.

 

[caveman-jim]

So, is this next release going to be more like to HD 2900 -> HD 3870, than HD 3870 -> 4870? Except without the reduced power part, cos we don't have a magical new process node.

 

[Kaotik]

So, is this next release going to be more like to HD 2900 -> HD 3870, than HD 3870 -> 4870? Except without the reduced power part, cos we don't have a magical new process node.

I have my fingers crossed for northern islands @ 40nm due the codenames in drivers and some rumors before those appeared.
So bigger change than HD2k > 3k, 3k>4k or 4k>5k

 

[caveman-jim]

Bigger change than 4k to 5k, eh? which delta, 4870 to 5870; or 4890 to 5870? or 4870 x2 to 5970?

 

[no-X]

HD5870 was 50% faster than HD4890. I'd expect similar performance jump for HD5870->6870

 

[Kaotik]

Was thinking architecture wise, not performance

 

[Robert Varga]

HD5870 was 50% faster than HD4890. I'd expect similar performance jump for HD5870->6870

Wow, but we are on the same old 40nm this time. 50% looks too much for me.

 

[homerdog]

Wow, but we are on the same old 40nm this time. 50% looks too much for me.

RV770 was a massive jump over RV670, on the same process node. Granted, RV770 was 33% bigger and I don't expect RV970 to be quite that much bigger than RV870, but it is not impossible.

 

[Mianca]

http://news.ati-forum.de/index.php/...termin-und-weitere-details-der-hd6000er-serie

If those details are true (i.e. that Barts will launch first and consume slightly below 200 watts and already need heat-pipe-cooling), wouldn't it be possible that:

Barts = 6770 = high-mid-range NI part "ported" from 32nm to 40nm = October 2010 launch

Cayman = 6870 = high-end NI part "ported" from 32nm to 28nm = H1 2011 launch?

 

[neliz]

If those details are true (i.e. that Barts will launch first and consume slightly below 200 watts and already need heat-pipe-cooling), wouldn't it be possible that:

Barts = 6770 = high-mid-range NI part "ported" from 32nm to 40nm = October 2010 launch

Cayman = 6870 = high-end NI part "ported" from 32nm to 28nm = H1 2011 launch?

Or you could speculate like we did months ago that Barts is where it's at for 2010 and Cayman doesn't show up on 28nm 'till next year.

So let's say Cayman isn't ready for production yet, who would believe Barts is delivering the numbers discussed here?

 

[iMacmatician]

RV770 was a massive jump over RV670, on the same process node. Granted, RV770 was 33% bigger and I don't expect RV970 to be quite that much bigger than RV870, but it is not impossible.

TDP also went up 42% from 3870 to 4870. I don't think it's impossible either though.

 

[Broken Hope]

If they don't fix their AF implementation I'll be disappointed. If it in fact is a hardware bug/issue.

 

[cal_guy]

TDP also went up 42% from 3870 to 4870. I don't think it's impossible either though.

I think you can chalk that up to the immature GDDR5 technology in RV770, the HD 4850 had significant improvements over the HD 3870 with only a slight increase in TDP.

 

[ShaidarHaran]

TDP also went up 42% from 3870 to 4870. I don't think it's impossible either though.

A mere 42% bump in TDP for a 100% or more bump in performance is well worth the trade off for anyone concerned with performance.

 

[Man from Atlantis]

 

[rpg.314]

384 bit me bus, along with the rumored ~33% higher clocked memory, Cayman must be quite big then.

 

[GZ007]

384 bit me bus, along with the rumored ~33% higher clocked memory, Cayman must be quite big then.

Or they didint clocked the memory higher and just increased the bus width. The 6.4 GHz gddr5 gpu-z shots are fake it seems.
Maybe its cheaper to buy mass production 5GHz chips in the end(also pcb and gpu is less complex if it doesnt need to run on such high frequency, and ati already mastered 5GHz). Thats still 240 GB/s on just 5 GHz if its true.