Originally Posted by Artikbot
ES, second silicon revision...
Hasn't got the final microcode, hasn't got the final silicon revision, and even in that case, the models do not concur.
I call this not accurate. Might not be a fake, but that ES is veeeery old.
This is the most likely scenario, they're not fake, just not accurate, I'm hoping so for OUR sake.
There's gotta be bugs that need fixing, perhaps these runs aren't running with the Updated BIOS needed?
Originally Posted by SCollins
Umm, really what instructions doe it use ? Because Intel is hush about it and its one of those benchmarks that Intel consistently outperforms AMD in, which suggest hanky up front.
It also makes heavy use of FP and maybe thanks to Intel Compiler be using the worst possible code paths available.
IF these bench's are real. In fact if the CPU dispatch instructions are not flagging x87 which would be the default, it could have to run allot of this on integer which would completely hose performance.
There is only one group of people who benefit from this type of press, and it sure ain't AMD.
Here's a problem with your theory. Check out this thread:
Look up Phenom II scores.
980X @ 4.4 gets 91.5 GFLOPS
1090T @ 4.3 gets 82.3 GFLOPS
930 @ 4.2 gets 60 GFLOPS
1090 @ 3.9 gets 74.6 GFLOPS
955 @ 3.8 gets 49 GFLOPS
( these are the closest clock for clock comparisons )
Those are all pretty linear results. They're ALL performing then Bulldozer? Blame Linpack, or face the reality? Otherwise, a 1090T wouldn't be pushing 80 GFLOPS at 4.4. Twice the GFLOPS as BD at 4.5, with 2 less cores.
It's not a conspiracy, BD is just under performing right now, and needs a fix, what ever it is, AMD has not said. What ever that BIOS is they are talking about better fix the issue.
Originally Posted by JCPUser
If you look at some of the early leaks and statements from AMD, they were saying things like 50% more performance for 33% more cores. And now... it seems like they will have a hard time equaling the performance of a theoretical Phenom II x8 (which in itself is unacceptable because it implies little to no IPC increase). Clearly, AMD ran into some major problems.
Like you said, a new uarch and a die shrink is very risky and that AMD was bound to have problems. Thus I had no issues defending the early BD performance leaks since AMD still had time to fix issues and optimize performance. But now it is confirmed that we are 1 week from launch...... people who are still defending these results thinking AMD has some magic BIOS or microcode they plan on releasing at launch are just silly. Come on, the writing is on the wall.
My guess is AMD is just going to release whatever they have (even if it is slower than PH II), make as much money as they can, and move to Piledriver (fixed BD) as quickly as possible. That would explain why Trinity was originally using BD cores but now will use PD cores AND that fact that 8 Piledriver cores will be released on AM3+ (Vishera) and NOT
Seems like AMD is trying to fix new uarch and get the manufacturing up to snuff before moving on -- hence the quick move to Piledriver.
Hopefully, PD will rock so I can switch back to my preferred company, AMD.
Seeing your new avatar is kinda weird, but suits you well.
Looks like they'll get it right Next Gen, or hopefully fix what ever issues they have now, and improve the performance.
ALL year, all signs point to BD having issues. It's a new arch, I'm sure AMD will sort it out. I remember reading an article saying that a new stepping will be out in January, and THAT's the one to wait for.
Originally Posted by reflex99
it would make sense that Intel is planting these "leaks" to capitalize on AMD delaying their product.
It wouldn't cost them much to create some fake screenshots, and they have enough money to pay off any judge/corporation that found out.
Think of all the people that have already jumped ship to SB because of all these "leaks". Major gain for intel there.
kinda a stretch, but possible
It's funny how far people will go to justify the under performance being seen here. Intel has already had to pay $1 Billion+ to AMD for shady business practices.
They have SB, the 990X, SB-E out next month. They don't need to do anything shady since they're already holding the performance crown, why would they jeopardize it?
How many leaks have we seen all year? Can they really "bribe" that many people to leak "fakes"?
Face the reality. All year we've seen under performing results, delays, LN2 and Liquid Helium "demos" it all adds up to one thing.
BD is not up to par right now, and under performing.
HOPEFULLY AMD will resolve it quickly, especially if it's something as simple as a BIOS fix that unleashes BD's true performance. Otherwise, it's not going to be pretty for all of us.