HPE 9000 and HPE e3000 Servers
1753717 Members
4499 Online
108799 Solutions
New Discussion юеВ

Re: request benchmark info for PA8900 vs PA8800 on rp4440

 
SOLVED
Go to solution
Steve Post
Trusted Contributor

request benchmark info for PA8900 vs PA8800 on rp4440

Does anybody have performance information comparing an rp4440 running PA8900 cpus versus PA8800 cpus?

We are considering getting 2 rp4440 as sybase database servers. I know the PA8900 is faster. But it would be nice to see some actual performance data on it.


3 REPLIES 3
Sameer_Nirmal
Honored Contributor

Re: request benchmark info for PA8900 vs PA8800 on rp4440

The following white paper mention significant advantages of PA8900 processor
in terms of power , scalability and compares with PA8800 processor.

http://h71028.www7.hp.com/ERC/downloads/5982-4172EN.pdf

I am not sure if the benchmarking for Sybase on PA8800 and PA8900 is done. I would contact HP and confirm.

Re: request benchmark info for PA8900 vs PA8800 on rp4440

Steve,

The important thing to point out here is that the PA8900 is FASTER than the PA8800 and CHEAPER as well!

Not really a lot of point in buying a PA8800 now (unless you're in the market for 2nd user equipment)

Incidentally unless you absolutely have to squeeze every last drop of performance out of the box, go for 1.0GHz rather than 1.1GHz as they're nearly as performant, but quite a bit less expensive.

HTH

Duncan

I am an HPE Employee
Accept or Kudo
Ted Buis
Honored Contributor
Solution

Re: request benchmark info for PA8900 vs PA8800 on rp4440

On the rp4440, the clock rates options are the same for the pa8900 as the pa8800, 1GHz and 800MHz, so the only difference for it is the 64MB L2 cache for the pa8900 versus the 32MB L2 cache of the pa8800. Since this is the only significant refinement, I would not expect to see much of an performance increase. Memory is the same, I/O is the same and the clock rate is the same, so unless you are really lucky and have an application that gets a big boost from the larger cache the improvement will be minimal. Benchmark performance improvement would be very dependent on the specific benchmark. You could always write one to make the improvement look large or small, but I think real applications will see a very minor improvement. If you are looking for real performance improvements per core, consider Itanium if you have recompiled code for it. Unfortunately, you will have to wait for Montecito to get 8 Itanium cores in that entry level system, or go the the rx7620.
Mom 6