AMD: Southern Islands (7*** series) Speculation/ Rumour Thread
- Thread starter UniversalTruth
- Start date
I'm sure it will be an ogl extention eventually.
We already have an OGL extension for it (though I don't know if it is expeosed in the initial drivers).
Broken Hope
Regular
Stock levels seem to be fine, still expecting these to be like £500 though in the UK..
http://img20.imageshack.us/img20/5976/ocuk7970pyramid.jpg
http://img20.imageshack.us/img20/5976/ocuk7970pyramid.jpg
OH dear, it seems even people in Russia might be able to buy this the week after it launches!
DegustatoR
Veteran
Highly doubtful considering that we're on national holidays from 31st of December till 10th of January (meaning that customs and other such things are closed). And while I appreciate your concern, I still don't see how this makes 22th Dec launch any less "paper".
Hey, if Finland has cards in stock before the "release-for-sales-date", Russia has to have too
Humm, that's a very good point. Unfortunately I didn't have a chance to look at code in-depth, otherwise I hopefully would have caught that. It would certainly explain why the "Time" results are more consistent than the "Kernel Time" results.So "Time" is a poor measure of performance since it takes the total set-up time and adds in the average time per iteration. That means if you have a higher set-up time but much faster performance per iteration, you might still "lose". setupTime should be divided by the number of iterations as well if you care about the average performance per iteration as the set-up time is a one time cost so that should spread out over all the iterations equally, or you can just look at the "[Transfer+Kernel]Time" if you don't care about set-up costs.
Yet another bug to file against the samples
ATI/AMD dethrone Nvidia for Top overclocking record.
http://kingpincooling.com/forum/showpost.php?p=19111&postcount=35
http://kingpincooling.com/forum/showpost.php?p=19111&postcount=35
Hm? Shamino went 1700MHz core 2000MHz mem on Tahiti already last year, so 1600/1900 is hardly doing anything
Hm? Shamino went 1700MHz core 2000MHz mem on Tahiti already last year, so 1600/1900 is hardly doing anything
Difference is this is Quad-Fire! It's much trickier to get 4 cards OC to the same maximum level so you're bound to get a bit less MHz from 2,3,4 card setup. Anyway both Shamino's and Andre's results are amazing. My question now is: if this is TSMC 28nm HP process, how hight we will be able to go with future refresh of Thaiti (possibly HD8xxx) on TSMC 28nm SHP designed specificity for GPU's? Will 2GHz on LN2 be finally broken?
PS. This was funny from Shtal's link:
Oo, didn't notice the quad-fire
Anyway, SHP? First I've heard of such, where did you read from such process?
The way I understood it, Tahiti was first rumored to use HPL which was ready before HP, which is supposed to be the "highest performance process", but as it now seems, Tahiti uses HP and not HPL
Man from Atlantis
Veteran
i thought SHP was GloFo 
Oo, didn't notice the quad-fire
Anyway, SHP? First I've heard of such, where did you read from such process?
The way I understood it, Tahiti was first rumored to use HPL which was ready before HP, which is supposed to be the "highest performance process", but as it now seems, Tahiti uses HP and not HPL
Sorry, mixed with GL process, anyway do we have any source about Thaiti being made on HP? I thought it's HPL ...
HPL was mentioned in the very same rumors as XDR2 was
There was some talk about it in some of the HD7-threads here, the voltages, even if they can't be used as direct reference, suggested IIRC HP rather than HPL (based on the processes respective reference voltages)
HPL was mentioned in the very same rumors as XDR2 was
There was some talk about it in some of the HD7-threads here, the voltages, even if they can't be used as direct reference, suggested IIRC HP rather than HPL (based on the processes respective reference voltages)
Thanks!
I know GPU manufacturers never stick to process reference voltages but with HD7970 1.17vGPU you're probably right in saying it's HP. We will have to wait for smaller members of S.I. to see if they can be pushed higher. I doubt they will brake 2GHz though. It doesn't even matter really except for record breaking clock :smile:
nonono, it's the other way around. hpl/lp uses higher voltage. the standard voltage for 28nm should be around 0.85v, while hpl is 1.0v, lp is 1.05v.
so it's most likely hpl. considering amd uses lpg@40nm, and hpl is romored to have been more mature.
The idle voltage of Tahiti, on the other hand, is just that 0.85v if my memory serves me right
Broken Hope
Regular
It's great living in the UK, it's looking like we're going to have to pay around £460, or $709 for the Americans on the boards.
Similar threads
