I don't understand why people produce shitty and irresponsible benchmarks OR why SLASHDOT would give this CRAP any attention. Several things are clear from this data. 1) Several of the system had there ide drives in MODE 4 at best. Not only is this PAINFULLY obvous from the test scores nobody in there right mind is going to put a piece of via shit in a production server let alone some door stop of an IDE drive. If SCSI drives had been used in the first place this would have even gotten around the shitty VIA chipset. 2) When the solaris boxes turn in the worst NFS scores alarm bells should be ringing (setting aside the fact x86 solaris is a bastard), HELLO remember sun... they invented this little thing called NFS. Disk I/O obvously was fucking things up and I'm sure that the solaris system tuning was non existant. May I remind you that many a consultant makes there living tuning solaris boxes and there are several books dedicated to the subject. 3) If the intent is to show scalability why was a uniprocessor system used? I'm typing this on a linux/sparc box but crap man do really think that every buisness in the world that uses solaris over linux/bsd is smoking crack? (well crack would explain NT's use but I'm side tracking). Solaris does have higher system requirements then bsd/linux due to overhead in things like the ip/stack but everything is fully threaded and will leave everyother platform in the dust (even on x86) when scaling to multiple processors.
It was totaly FUCKING STUPID to have even released these OBVOUSLY WRONG benchmarks. I hope that everyone see's it for the PISS that it is.
Slashdot Mirror
Drink the coolaid.
http://www.whosyourdaddyinc.com/
I don't understand why people produce shitty and irresponsible benchmarks OR why SLASHDOT would give this CRAP any attention. Several things are clear from this data. 1) Several of the system had there ide drives in MODE 4 at best. Not only is this PAINFULLY obvous from the test scores nobody in there right mind is going to put a piece of via shit in a production server let alone some door stop of an IDE drive. If SCSI drives had been used in the first place this would have even gotten around the shitty VIA chipset. 2) When the solaris boxes turn in the worst NFS scores alarm bells should be ringing (setting aside the fact x86 solaris is a bastard), HELLO remember sun... they invented this little thing called NFS. Disk I/O obvously was fucking things up and I'm sure that the solaris system tuning was non existant. May I remind you that many a consultant makes there living tuning solaris boxes and there are several books dedicated to the subject. 3) If the intent is to show scalability why was a uniprocessor system used? I'm typing this on a linux/sparc box but crap man do really think that every buisness in the world that uses solaris over linux/bsd is smoking crack? (well crack would explain NT's use but I'm side tracking). Solaris does have higher system requirements then bsd/linux due to overhead in things like the ip/stack but everything is fully threaded and will leave everyother platform in the dust (even on x86) when scaling to multiple processors.
It was totaly FUCKING STUPID to have even released these OBVOUSLY WRONG benchmarks. I hope that everyone see's it for the PISS that it is.