He screwed up his makefile on the TSCP benchmark. (It fails to apply the -O2 option during compile) The 3500+ should score about ~300K, not 150K, beating the Nocona by a significant margin.
He copied the wrong data over on the MySql test-- the 3500+ wins both.
It's unclear he built the other synthetic tests correctly.
Some of the other tests are 32bit, not 64bit.
ubench is known to be buggy, as explained at another site.
This "review" is pure garbage.
The fact that he used a 3500+ instead of an Opteron 150 is a minor issue. The major issue is that he doesn't know what he's doing when he attempts to build benchmarks from source.
He copied the wrong data over on the MySql test-- the 3500+ wins both.
It's unclear he built the other synthetic tests correctly.
Some of the other tests are 32bit, not 64bit.
ubench is known to be buggy, as explained at another site.
This "review" is pure garbage.
The fact that he used a 3500+ instead of an Opteron 150 is a minor issue. The major issue is that he doesn't know what he's doing when he attempts to build benchmarks from source.