Linked by David Adams on Sat 6th Sep 2008 14:36 UTC
Thread beginning with comment 329437
To view parent comment, click here.
To read all comments associated with this story, please click here.
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE: Calculating correctly would be useful too
by evangs on Sat 6th Sep 2008 18:33
in reply to "Calculating correctly would be useful too"
RE: Calculating correctly would be useful too
by bnolsen on Sat 6th Sep 2008 19:12
in reply to "Calculating correctly would be useful too"
RE[2]: Calculating correctly would be useful too
by daschmidty on Sat 6th Sep 2008 22:25
in reply to "RE: Calculating correctly would be useful too"
While I agree that I wouldn't be suggesting windows as a high-performance computing environment. The top 500 supercomputers reference should also be taken with a grain of salt considering most supercomputers are pretty exotic hardware that you likely couldn't even make windows boot on. Alas, the comment regarding the divergence is a scary thought, since the current trend for many large corporations in industry using such software for dynamic simulation is away from aging UNIX boxes and toward pc workstations.
Edited 2008-09-06 22:26 UTC
RE[2]: Calculating correctly would be useful too
by jayson.knight on Sun 7th Sep 2008 18:19
in reply to "RE: Calculating correctly would be useful too"
Umm...considering Microsofts virtual non showing on the top 500 super computers you actually think their products can be used for serious computing?
MS practically owns the TPC list, which is a much more relevant benchmark for computers as they related to real world processing within the business realm, which is where the vast majority of computers are actually used. The giant multi-thousand core machines on the Top500 list are niche products.






Member since:
2006-01-19
As I had to discover the hard way, Windows cannot calculate correctly when the machine is under heavy load.
I am using 3 of four processor cores to solve a multi-body dynamic simulation (of a car engins), and using the 4th core to prepare the next model, using ABAQUS.
Whenever I do that, the simulation diverges after some time (but every time slightly differently). If I let the computer alone, the simulation runs through.
None of this happens with HPUX, IRIX, AIX or Linux. There I can run the simulation with all 4 cores, and still work with ABAQUS, effectively overloading the machine, but the simulation runs as it should.
It would be even more helpful than speed.