On Mon, Feb 8, 2010 at 22:06, Gqcy <gmufasa01@xxxxxxxxx> wrote:
I'm just curious...
How do you benchmark an "x" (intel) , against POWER?
Java works well. Synthetic and real world benchmarks. WebFocus
consists of native and Java code, and there's an IBM i version
available - you can also use the Windows version and connect it to
your IBM i (licenses required). All of our WebFocus deployments run on
Windows, because there we actually have CPU power to run Java
efficiently.
Of course, the results are fairly obvious - a 20k$ System x versus a
20k$ IBM Power system is a curb stomp - tons of RAM, lots of disk
arms, BBWC, 8 cores, 16 threads versus 1 core, 2 threads (i believe),
two mirrored disks and 2GB of RAM.
For fun and games, you can also test SQL Server against DB2 with a
20k$ IBM Power system against a 3k$ ThinkPad. Use 6GB of test data -
with 2GB of RAM and two 15kRPM disks on the IBM i side and 8GB of RAM
and an SSD on the ThinkPad side, the winner is also obvious.
Cue people yelling these tests are unfair - they're not. We're
comparing price/performance here.
midrange.com
