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

[Deleted member 13524]

I don't think that's always the case though, only if it knows how to read it. In particular it might not be true for new gpus which might return their basic configuration data differently than previous chips. There's also always the possibility of misinterpretation if the meaning of such data changes (for instance the chip might not return CU count but CU groups and hence if the group size suddenly changes and you don't take that into account you'll get wrong results obviously). And I'm quite sure you can't read TMU count for instance, you just have to know how many TMUs there are per CU.

While I don't know exactly how GPU-Z detects the number of shaders, I do know that shader count is almost always (except for photoshopped results, of course) correct in new cards.
Moreover, w1zzard usually gets colaboration directly from AMD in order to update GPU-Z.
FWIW, GPU-Z is the tool being used to prove world record overclocks.

 

[trinibwoy]

Any word on launch/review date for Cape Verde? Feb 15 still good? Hints have been dropped that reviewers are under NDA but no date confirmed.

 

[Alexko]

Any word on launch/review date for Cape Verde? Feb 15 still good? Hints have been dropped that reviewers are under NDA but no date confirmed.

No word yet, as far as I'm aware; apart from the rumored specs, of course. Either it's not on Feb 15, or AMD is keeping a very tight lid on everything, including pictures of the cards themselves, which is quite unusual as those tend to leak out pretty easily.

 

[CarstenS]

Didn't the rumor mill churn out alleged HD 7700 images quite early already?

 

[Alexko]

Didn't the rumor mill churn out alleged HD 7700 images quite early already?

Oh, it looks like you're right, my bad, I guess I just missed it: http://www.tanscomputer.com/latest/amd-radeon-hd-7700-series-details-revealed

Maybe we'll see Cape Verde in 2 days after all.

 

[CarstenS]

Cannot vouch for accuracy, obviously, but there have been some alleged pics and benchies last year already. Right here on B3D.
http://forum.beyond3d.com/showthread.php?p=1609212#post1609212

 

[mczak]

While I don't know exactly how GPU-Z detects the number of shaders, I do know that shader count is almost always (except for photoshopped results, of course) correct in new cards.
Moreover, w1zzard usually gets colaboration directly from AMD in order to update GPU-Z.

Ok I just thought it's mostly guesswork for new gpus. But if he gets help from AMD directly I guess it will indeed read the right results even for new gpus.

FWIW, GPU-Z is the tool being used to prove world record overclocks.

Oh I don't doubt the results are correct once a gpu is fully supported (and maybe they always are...).

Cannot vouch for accuracy, obviously, but there have been some alleged pics and benchies last year already. Right here on B3D.
http://forum.beyond3d.com/showthread.php?p=1609212#post1609212

Yes but they are soooo yesteryear...
So old driver for instance.
That said I think the results still look reasonable. Above 6770, below 6850.

 

[iMacmatician]

The "stimulating promotional video" for Verdetrol is now up on the Verdetrol website, this embedded YouTube video.

 

[DarthShader]

Had a feeling they would go in that direction when I saw the blue pills.... the disclaimer at the end is great though.

 

[Broken Hope]

Gibbo confirming his previous 7770 specs were incorrect, looks like it's going to be more expensive than the 6850 and yet also slower.

http://forums.overclockers.co.uk/showpost.php?p=21263436&postcount=36

 

[mczak]

Hmm fudzilla thinks the HD7770 is faster than HD6850:
http://www.fudzilla.com/home/item/25936-radeon-hd-7770-scores-p3535-in-3dmark-11
Don't know though how they came to that conclusion when even the only benchmark they show is actually below HD6850.
I'd be quite impressed if it could equal a HD6850 (well except in some carefully selected tests where this should easily be possible).

 

[AlphaWolf]

anything with tesselation it should gain

 

[mczak]

anything with tesselation it should gain

That'll fit my "carefully selected test".
I'm not sure though this will be always be true. Pure tesselation benchmark, sure.
Anyway some new results have surfaced:
http://chinese.vr-zone.com/10056/amd-radeon-hd-7770-02142012/
So HD7770 3dmark11 extreme gpu score is indeed just about identical to HD6850 (with reference clocks).
(Can't quite figure out what the Lost Planet 2 / Street Fighter IV scores are worth.)

 

[CNCAddict]

WTF Eric Demers gone

http://www.brightsideofnews.com/news/2012/2/14/breaking-eric-demers2c-cto-graphics-leaves-amd.aspx

Hope he doesn't go into the mobile gpu field.....

 

[nAo]

Hope he doesn't go into the mobile gpu field.....

What's wrong with that?

 

[mczak]

WTF Eric Demers gone

http://www.brightsideofnews.com/news/2012/2/14/breaking-eric-demers2c-cto-graphics-leaves-amd.aspx

Yeah more people leaving AMD, too bad. If you didn't know he actually used to post here (sireric) though not recently.

 

[Arty]

 

[LordEC911]

Pulled from radier at XS

 

[Man from Atlantis]

^^ 56% more shading power than 7750 at same bandwidth..

 

[fellix]

Looking from the optimistic side, it's more like 36% less shading power on the same bandwidth.