First, we have AMD run a test between Opterons and Barcelonas which seems to indicate that (after taking into account the extra cores) Barcelonas are at best only minimally better than Opterons. No clock speeds, no nothing about the specs of the two machines.
Then Intel gussies up an 8-core Cloverton system, and it (apparently) whips the 16-core Barcelona system by about 25%. Yes, they were using 3GHz processors, and the AMD Barcelonas (we hope) were running slower than the speeds that will go on sale, but still, eight shouldn’t beat sixteen.
This looks pretty bad, especially when people are looking for any kind of indication as to how well Barcelona will do.
What does AMD have to say about it? They whine that the other machine had more RAM.
I’m sorry, but when Intel does something like that to you, and you have something to fire back with, you do fire back, immediately. You don’t let Intel get that favorable first impression. If memory really makes a big difference, you load your 16-core with 16GB, and you mention that score.
You might also mention the clockspeed of the Barcelonas being benchmarked. If it ran slowely, people can extrapolate what a higher clockspeed would do to the numbers. If the benchmark can be played with to yield a higher score, AMD ought to mention that, too.
What you do is respond to this with hard cold numbers, not whine and make vague allusions.
Because only fanboys are going to see something good out this, everyone is going to assume the opposite, because even after making a generous allowance for increased clockspeed and scaling, a 16-core Barcelona system still looks like it would just nudge that Xeon 8-core. That would be much worse than even the most rabid AMD hater would have believed before all this. Even in the comparison that AMD put together, there doesn’t seem to be any real improvement to Barcelona over Opteron other than two extra cores per socket, which will still leave it trailing the Clovertons.
Until they get a much better explanation than they’ve gotten so far, people are going to assume based on the numbers they have that Barcelona isn’t going to be a lot better, or maybe not even as good, as the Intel stuff. Granted, it’s probably not as bad as the Pov-ray numbers would suggest, but bad enough.
What’s The Problem?
I think this quote from the AMD article gives us a clue as to the problem:
“At some point before our launch you can plan on us showing a demo of our parts vs. Intel’s high-performance processors.”
Maybe this is reading tea leaves, but that sure doesn’t sound like Computex to me. Rather, it sounds like they don’t know when they’re going to do it, which sounds like they’re not ready for prime time yet.