For those hoping for 32nm Phenoms X8, sorry, it won't happen.
Just check ECS's documents for Stepping and Revisions, they perfectly match to old B0 Zambezi samples and these supposed Phenoms are nothing more than that.
The last K10 parts AMD is producing are Llano cores, after that it all will be Brazos and further cores and Bulldozer and further iterations for at least next 24+ months
Just check ECS's documents for Stepping and Revisions, they perfectly match to old B0 Zambezi samples and these supposed Phenoms are nothing more than that.
The last K10 parts AMD is producing are Llano cores, after that it all will be Brazos and further cores and Bulldozer and further iterations for at least next 24+ months
Well, um, yeah that was in the news and whatnot. Which is why I said X6s would be better.
Eric Demers makes an interesting comment here:
http://www.rage3d.com/interviews/amdchats/eric_demers_dec_2011/index.php?p=5
Who knows if he's preaching the partyline, or talking about just design and not performance.
http://www.rage3d.com/interviews/amdchats/eric_demers_dec_2011/index.php?p=5
Who knows if he's preaching the partyline, or talking about just design and not performance.
Demers had little negative to say about R600 too. He judges hardware differently than most of us I think, because of his perspective as an engineer, and he always considers the long term project aspect and reflects on why they made the decisions they did years ago and how they fit with current and future trends. Big picture vs short term semi-failure.
Plus he has a vested interest in talking AMD up, and in not burning bridges in general in public.
Plus he has a vested interest in talking AMD up, and in not burning bridges in general in public.
Considering how much of the R600 had to be fixed(Not altered, fixed), I'd say he's towing the party line. Given his position, it makes perfect sense. I can't blame him.
itsmydamnation
Veteran
apart from AA resolve what was broken in R600?
and on second though what you have said doesn't even make any sense in the context of what you are quoting?
In the minority of benchmarks its slower than the x6/x4 and in the minority of benchmarks its faster than the i5 2500k.
Its not a great chip at all , but if fixes are coming out with piledriver (trinity too) that give a 5-15% increase along with the windows patches that seem to also give a 5-15% increase from whats going on , the next verison of bulldozer can be a pretty decent chip
They'll need to pull a sort of RV770 to make Bulldozer compete with Intel who is definitely not sitting still and waiting. AMD just slips farther and farther back year after year so I'm not really holding up much hope.
Literally broken? Dunno. But it was such a nasty, inefficient mess that you had to wonder what the hell had happened. The performance was pretty unstable on launch too, even after the chip had been delayed for half a year and driver people should have had lots of time with it.
To be honest, the single biggest issue of R600 was the [strike]broken[/strike] missing hardware AA resolve, that contrasted in the benchmark results, together with the launch delays. Compute wasn't much of a setback in those days, too, at least for the consumer market. On the TDP front there was much to be desired. R600 SKUs were simply primitive from today's basic standards, with total lack of advanced power management features in hardware -- even the load/idle clock switch was driver-based with loose implementation, working in full-screen only.
Last edited by a moderator:
That is what I was referring to. So much of the R600 was a misstep...
R600 was radically different that what came before it. All the lessons learned from the R600 resulted in the R700 series which was a huge success.
So while it appears to be a huge mistake from a consumer point of view, I'm sure internally in AMD/ATI it was viewed as an important step forward.
Cheers
Looks like the performance increase was about what it was last time, which is to say somewhere between zero and a few percent.
