This week, IBM announced that it has received the highest TPC-C performance result ever achieved by a 4-processor server. And it’s running Linux. Red Hat Enterprise Linux 5 Advanced Platform to be exact. The IBM System x 3850 M2 server and 64-bit IBM DB2 9.5 on Red Hat Enterprise Linux 5 Advanced Platform has set a new record for 4-processor performance, achieving 516,752 tpmC on the TPC-C online transaction processing benchmark. (IBM System x3850 M2 with the Quad-Core Intel Xeon Processor X7350 2.93 Ghz (4 processors/16 core/16 threads) 516,752 tpmC, $2.59 USD/tpmC, availability of March 14, 2008.)

The solution includes a combination of IBM, Intel and Red Hat solutions:

  • IBM’s exclusive fourth-generation Enterprise X-Architecture x3850 M2 server
  • Quad-core Intel Xeon Processor X7350 at 2.93 GHz with 2x 4MB L2 cache per processor (4 processors /16 core / 16 threads)
  • DB2 9.5 (64-bit)
  • Red Hat Enterprise Linux 5 Advanced Platform

Red Hat Enterprise Linux 5, providing a significant technology advance for commercial open source solutions, together with the new DB2 9.5, delivering significant performance improvements with its shift toward threaded engines, provides customers with the value of a reliable, flexible and scalable solution with record performance results.

To view TPC benchmark results, visit here.

Addendum: On November 1, IBM submitted SPECcpu2006 and SPECjbb2005 results, which highlight another aspect of the Linux scalability story, in this case several CPU-intensive results for Linux. IBM submitted results on our latest Red Hat Enterprise Linux 5.1 release for IBM’s newest p 570 POWER6 line of systems, with leadership results and near-linear scalability demonstrated from 4-core, 8-core, and 16-core systems for SPECint_rate2006, SPECfp_rate2006, SPECjbb2005 and Linpack HPL. See the below chart and footnote for the submission details.

Overall Position Speed Final RHEL 5.1 p 570 POWER6 4.7GHz
Best POWER Result SPECint 2006 21.7
BoB, BoL SPECfp 2006 22.5
     
4-Core Position 4-Core Final RHEL 5.1 p5 570
BoB, BoL SPECint_rate 2006 122
BoB, BoL SPECfp_rate 2006 116
BoL, #2, 4-Core SPECjbb2005 BOPS 169304
Great Result Linpack HPC compared with non-super computing systems 60.37
BoB RISC SMP Stream Standard RISC/SMP 29404
     
8-Core Position 8-Core Final RHEL 5.1 p5 570
BoB, BoL SPECint_rate 2006 243
BoB, BoL SPECfp_rate 2006 216
BoL, #2 8-Core SPECjbb2005 BOPS 335424
Great Result Linpack HPC comparied with non-super computing systems 116.4
     
16-Core Pisition 16-Core Final RHEL 5.1 p5 570
BoB, BoL SPECint_rate 2006 484
BoB, BoL SPECfp_rate 2006 430
BoB, BoL SPECompM2001 94,350
BoL SPECjbb2005 BOPS 664,167
Great Result Linpack HPC compared with non-super computing systems 229.4

*BoB = Best of Breed
*BoL = Best of Linux