One can only hope that Don is smart enough to head to HP's x86 business.

Ashlee...the new name for SPARC is SPANC....seems fitting since it gets spanked on all benchmarks

it's about time

My fellow friend, you must be stuck in a time void circa 1980 - jump in that tardis of yours and come back to 2008 and get some real SPARC processor power.

Yes the T processors have good benchmarks..they were created for benchmarks...eight low function cores will always beat two/four high function cores in the right benchmark.

Too bad the socket scalability is so bad...the Java performance increase on the two socket version was horrible. I hear Batoka right now has a negative peforrmance curve.

The T-series boxes were not created for benchmarks. They were originally created for network-edge workloads. then with the extra floating point and more threads to allow them to play in the middle tier.

The *right* Java workload would scale well on these boxes and that should be no surprise.

But - If the Java code has locks that are frequently required to be moved from one CPU to another, the change from "everything in one chip" to "moving from one to the other" will certainly deliver a performance boost of less than you might have been hoping for, but should not have expected anyway.

*ANY* SMP workload needs to be at least a little careful of what they do, and how they do it. That's why smart people get payed a lot of $$ to write snappy code to take advantage of leading edge tech as the Tx