Server 08 runs as good as Linux based HPC

[iMav]

Ferrari Takes Microsoft’s New High-Performance Computing for a Spin

Relying on consistent, breakthrough innovation in all aspects of automotive technology, Ferrari brings to market awe inspiring cars that simply take your breath High performance computing (HPC) has long been a strategic asset that has helped their engineers develop the right technology components – engines, aerodynamics, gears, braking etc. – that help make great cars. Therefore, when Ferrari decided to deploy Windows HPC Server 2008 to power the same demanding simulation workloads that were being run on Linux based HPC, one thing was clear, the notion of performance being a question mark on the Windows platform for HPC was banished forever. And I’m talking about real-world-performance-you-can-use, type performance, not a synthetic benchmark workload that is the fixation of everyone in the business.

Source | [via EverythingMS]

 

[RCuber]

Looks like its time for microsoft

 

[IronManForever]

I doubt those raising a question mark on the Performance of Windows in Mission-Critical tasks. But that doesnt mean Linux has less to offer. It rocks in tweakability due to its open nature. Windows will be good for single deployment fixed task scenario. Linux has a more dynamic nature...

 

[NucleusKore]

^ +1 @ ironmanforever
Not to mention that it belongs to nobody yet everybody

 

[Faun]

MS ftw

 

[IronManForever]

Not to mention that it belongs to nobody yet everybody

Yeah, everybody is a winner, whether Windows or Linux or Macintosh or Solaris or BSD or OS/2 p) ; neither can get better without the other.

 

[chandru.in]

As Iron Man said the real power Linux in HPC is not in speed alone it is in its flexibility and consequently portability. The performance of HPC clusters depend as much on the hardware as the OS. Will Windows HPC 2008 run on the CPUs of Roadrunner? But yeah running Windows HPC gives a huge benefit, you can soon find MCSEs at low salaries to maintain them and they'll keep using the term "network failure" when it is actually the network stack of some Windows node which has frozen.

YAWN!

Can someone wake me up when a Windows HPC based cluster enters the first 5 ranks at *top500.org which is currently has a Linux-powered 1.026 petaflop/s beast at #1?