On Tuesday 11 October 2005 02:05 pm, S Destika wrote:
[b][u][i]Solaris has always been good under load, something Linux falls
with typically. It really wasn't until recentely that Linux could handle 2
CPUs, and the only reason it can *marginally* utilize 2 CPUs is due to
Moore's Law, IMO.
2. File a bug with a specific, repeatable benchmark
with clear results
comparing Solaris to insert OS here.
I am waiting for the existing ones to be fixed - some are dated 1999. I don't
think there is point in running duplicate benchmarks and filing new ones. Most
of them are quite
[ Please send mail from a valid email address ]
SGI claims it scales *linearly* and there aren't any major problems
remaining in scalability area. You can google for actual benchmark
numbers and claims. And of course work is in progress to addre ss the
minor scalability issues (page fault
Lots of them are available here - http://www.spec.org but see the ones here -
http://www.spec.org/jbb2000/results/res2005q3/jbb2000-20050913-00388.html.
Linux/IA32
8-way - 296463 ops/s
1-way system 51395 ops/s.
In an ideal world on an 8-way the number should be 8 times 1-way number. In
reality
On Monday 10 October 2005 13:23, S Destika wrote:
Linux/IA32
8-way - 296463 ops/s
1-way system 51395 ops/s.
...
Now Solaris/AMD64 - Mind you 64bit CPU, Kernel and JVM
2-way - 85967 ops/s
4-way - 142789 ops/s
The best test would be on exactly the same hardware. Linux has a 64-bit
kernel,
Alan DuBoff wrote:
On Monday 10 October 2005 13:23, S Destika wrote:
Linux/IA32
8-way - 296463 ops/s
1-way system 51395 ops/s.
...
Now Solaris/AMD64 - Mind you 64bit CPU, Kernel and JVM
2-way - 85967 ops/s
4-way - 142789 ops/s
The best test would be on exactly the same hardware.