Additionally, the single vs Dual show very well how little Dual Channel impacts AMD's core
I don't know... Some of the tests showed there being little difference. Other benchmarks however really seemed to lag more than just one processor speed grade when in single-channel.
Sad to say, but the way things are according to such benchmarks (remember that SSE2 IS enabled on K8), AMD is done.
I had rather feared that would be the case. The benchmarks did a rather good job of showing just how well SSE2 was implemented by AMD, and it's sad to see just how poor of a job AMD did. SSE2 will definately <i>not</i> be Opteron's saving grace.
The ondie memory controller does seem to help out quite at bit at least. The latency is significantly better than the AXPs. (Though Intel seems to have done a very good job with the i875.) I have to wonder how much better the latency will be for A64 with DDR400.
What's scary though is how much this doesn't matter for AMD. Low latency is not that big of a concern. With their clock speeds so low, AMD's CPUs aren't affected by latency nearly as much as Intel's CPUs. Imagine if Intel put an ondie memory controller into Prescott... Now <i>that</i> would be scary.
All-in-all, I was rather unimpressed with the Opteron bencies. (Not the quality of the site, but AMD's scores.) Prescott is definately going to be putting a crimp in AMD's exuberance if these are any indications of AMD's future performance.
And one last thought ... whatever happened to AMD's decision that Opteron was for multi-CPU only and A64 was for single-CPU only? Now we have single-CPU Opterons??? Not that I mind, but why can't AMD even make up its mind on something so simple?
"<i>Yeah, if you treat them like equals, it'll only encourage them to think they <b>ARE</b> your equals.</i>" - Thief from <A HREF="http://www.nuklearpower.com/daily.php?date=030603" target="_new">8-Bit Theater</A>