BM Seer Unofficial thoughts from an anonymous Sun employee

« ...more on Sun &... | Main | Tired Fud & Selectiv... »

Selective Vision

Friday Jan 20, 2006

Not much writing lately... busy times. That must only mean one thing, more results soon. Actually if I worked at a different company it would probably mean that systems are slower than Sun

But I had to comment on this, Lisa Vass's blog was clearly well feed by the IBM folks. In it Lisa says (by the way it is UltraSPARC IV+)...

“There are no Sun UltraSparc IV results in the top 10 TPC-C list (transaction processing benchmark). Pseries servers hold 6 of the top 10 results in TPC-C,” said an IBM spokesman. “There are no Sun UltraSparc IV results in the top 10 SAP SD 3-tier benchmark results,” he continued. “DB2 on pSeries holds the No. 1 spot in that benchmark.”

As I've pointed out in the past TPC-C has issues that don't make it a predictor of performance these days. For more on TPC-C please see: http://blogs.sun.com/roller/page/bmseer?entry=tpc_c_how_old_is

But if you want to see comparisons between Sun UltraSPARC IV+/Ultra SPARC T1 vs. IBM Power5/Power5+ you need to look at:

TPC-H
SPECjAppServer2004 Dual Node
SPECjAppServer2004 Multiple Node
SPECjbb2005
SPECweb2005 (wow: 2RU T2000 vs. costly 4RU IBM p5 550 is instructive)
SPECint_rate
Lotus Domino R6iNotes
Manugistics NetWORKS Fulfillment v7
Informatica
SAP-SD (the 2RU T2000 vs. costly 4RU IBM p5 550 is instructive)
SPECjbb2000(retired but another comparison)
RSA/DSA Performance
static web performance
...The list goes on but I hungry and need to eat food...

[2] Comments

Like this post?

Posted at 06:16AM Jan 20, 2006 in Benchmarks
Tags:

Comments:

You do realize that several of the benchmarks that you have listed, which ostensibly show superior performance by Sun machines are in fact entirely non-comparable, right? For instance, you cannot compare manugistics 7.1 to 7.2. Unless of course your goal is to analyze the different performance characteristics of the software. SAP SD scores vary by a huge amount (anywhere from 20-100% variance) between version 4.7 and 5.0. You simply cannot make comparisons across different software revisions. A benchmarking guru should know this...

Posted by Anonymous Coward on January 22, 2006 at 03:50 PM PST #

Actually you need to look at the Manugistics press release for the versions that were in fact comparable. Sun was running the harder version and winning. Also on the SAP 5.0 versions, The statement I made above is head-to-head with IBM's version 5.0 result. Correct 4.7 is lighter weight, if you take that into account T2000 also beats some HP results by a lot.

Posted by BM Seer on January 26, 2006 at 02:32 PM PST #