Christopher Petrilli wrote:
When the cache is not in use StoryServer appears quite slow. Even with the cache they only claimed 1M hits/day using a dual CPU Ultra 2 sparc, which sounds much less than Zope.
While we're not a big SPARC shop, a minor projection of current benchmarks we've done with ZEO indicate that on a dual CPU box (with 2 ZEO clients, and a storage server) that we could achieve somewhere around 3.2M hits assuming a minimal level of dynamicism. Because of our strong dynamic nature, the more dynamic, the less throughput you get---it's just more cycles per request. I would say a HIGHLY dynamic site on such a box would perform around 1-1.5M hits live, as opposed to cached.
I would be much interested on the performance of python on some high-end SPARC box.
From my limited testing (using pystone.py), the SPARCs appear to be _much_ slower for python code.
my home Celeron450 runs at [hannu@kodu hannu]$ /usr/lib/python1.5/test/pystone.py Pystone(1.1) time for 10000 passes = 1.71 This machine benchmarks at 5847.95 pystones/second and dual PentiomIIIXeon 500/512 IBM Netfinity at: [hannu@nf hannu]$ /usr/lib/python1.5/test/pystone.py Pystone(1.1) time for 10000 passes = 1.65 This machine benchmarks at 6060.61 pystones/second It would be nice to get some comparison numbers from other chip architectures. --------------------- Hannu