A few people had solaris 10 working and and noted the rack kits do not work with old sun racks, which is what we have at work. Thanks to everyone and sorry for the delay. Here are two more emails: Hi. Here we've bought a v20z in december. The box is running under sol10-b72 since few months without pbs. The box makes some Automatic Translation and Natural Language Processing with good performances. See www.spec.org if you have a doubt... pc Hi Eric, We threw various benchmarks at a couple of v20z systems. One system was running RedHat Enterprise Linux 3, gcc 3.4. The other was running Sol10 plus gcc 3.3 and Studio10. Clock speeds were not the same on both systems so we had to fudge the results by assuming that the performance scales linearly with the clock speed. The RHEL system was a 244 system while the Sol10 system was a 248. We compiled our benchmarks with various types of optimisation, taking the best results we could get. Note that we understand gcc and Portland (see later) a lot better than Studio10, so please bear in mind that an expert in the latter might be able to get better results than we have. When using gcc the performance of the Solaris10 system was terrible. When using Studio10 it was much better and seriously outperformed the RHEL-gcc box. There is however a catch: We obtained a third, dual opteron (242) system from another vendor (so not a v20z) with Suse Pro 9.1. Because this machine was entirely under our control, We were able to install gcc, Portland, Intel and Pathscale compilers. When using the Portland and Pathscale compilers in particular, this 242 system was generally able to match the Solaris10 248 system *without* any scaling for clock speed. If we scaled the results as before, the Sol10-248 system was usually soundly beaten. So we have concluded that if we're happy to use proprietary compilers, our benchmarks give better results on a Linux system than they do under a _similar_ Solaris 10 system. gcc is also much better on Linux than Solaris, but far behind Sun's own compiler on Solaris. Hope that's useful, _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagersReceived on Fri Mar 4 12:36:30 2005
This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:43:44 EST