Thread: Some quick Opteron 32-bit/64-bit results
I just finished upgrading the OS on our Opteron 148 from Redhat9 to Fedora FC2 X86_64 with full recompiles of Postgres/Apache/Perl/Samba/etc. The verdict: a definite performance improvement. I tested just a few CPU intensive queries and many of them are a good 30%-50% faster. Transactional/batch jobs involving client machines (i.e. include fixed client/networking/odbc overhead) seem to be about 10%-20% faster although I will need run more data through the system to get a better feel of the numbers.
Hi Willian,
Which are the GCC flags that you it used to compile PostgreSQL?
Best regards,
Gustavo Franklin Nóbrega
Infraestrutura e Banco de Dados
Planae Tecnologia da Informação
(+55) 14 3224-3066 Ramal 209
www.planae.com.br
> I just finished upgrading the OS on our Opteron 148 from Redhat9 to
> Fedora FC2 X86_64 with full recompiles of Postgres/Apache/Perl/Samba/etc.
>
> The verdict: a definite performance improvement. I tested just a few CPU
> intensive queries and many of them are a good 30%-50% faster.
> Transactional/batch jobs involving client machines (i.e. include fixed
> client/networking/odbc overhead) seem to be about 10%-20% faster
> although I will need run more data through the system to get a better
> feel of the numbers.
>
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
> (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)
>
Biggest speedup I've found yet is the backup process (PG_DUMP --> GZIP). 100% faster in 64-bit mode. This drastic speed might be more the result of 64-bit GZIP though as I've seen benchmarks in the past showing encryption/compression running 2 or 3 times faster in 64-bit mode versus 32-bit. William Yu wrote: > I just finished upgrading the OS on our Opteron 148 from Redhat9 to > Fedora FC2 X86_64 with full recompiles of Postgres/Apache/Perl/Samba/etc. > > The verdict: a definite performance improvement. I tested just a few CPU > intensive queries and many of them are a good 30%-50% faster. > Transactional/batch jobs involving client machines (i.e. include fixed > client/networking/odbc overhead) seem to be about 10%-20% faster > although I will need run more data through the system to get a better > feel of the numbers.
I gave -O3 a try with -funroll-loops, -fomit-frame-pointer and a few others. Seemed to perform about the same as the default -O2 so I just left it as -O2. Gustavo Franklin Nóbrega wrote: > Hi Willian, > > Which are the GCC flags that you it used to compile PostgreSQL? > > Best regards, > > Gustavo Franklin Nóbrega > Infraestrutura e Banco de Dados > Planae Tecnologia da Informação > (+55) 14 3224-3066 Ramal 209 > www.planae.com.br > > >>I just finished upgrading the OS on our Opteron 148 from Redhat9 to >>Fedora FC2 X86_64 with full recompiles of Postgres/Apache/Perl/Samba/etc. >> >>The verdict: a definite performance improvement. I tested just a few CPU >>intensive queries and many of them are a good 30%-50% faster. >>Transactional/batch jobs involving client machines (i.e. include fixed >>client/networking/odbc overhead) seem to be about 10%-20% faster >>although I will need run more data through the system to get a better >>feel of the numbers. >> >>---------------------------(end of broadcast)--------------------------- >>TIP 2: you can get off all lists at once with the unregister command >> (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) >> > > > > > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) >
William Yu <wyu@talisys.com> writes: > Biggest speedup I've found yet is the backup process (PG_DUMP --> GZIP). 100% > faster in 64-bit mode. This drastic speed might be more the result of 64-bit > GZIP though as I've seen benchmarks in the past showing encryption/compression > running 2 or 3 times faster in 64-bit mode versus 32-bit. Isn't this a major kernel bump too? So a different scheduler, different IO scheduler, etc? -- greg
Greg Stark wrote: > William Yu <wyu@talisys.com> writes: > > >>Biggest speedup I've found yet is the backup process (PG_DUMP --> GZIP). 100% >>faster in 64-bit mode. This drastic speed might be more the result of 64-bit >>GZIP though as I've seen benchmarks in the past showing encryption/compression >>running 2 or 3 times faster in 64-bit mode versus 32-bit. > > > Isn't this a major kernel bump too? So a different scheduler, different IO > scheduler, etc? > I'm sure there's some speedup due to the kernel bump. I really didn't have the patience to even burn the FC2 32-bit CDs much less install both 32-bit & 64-bit FC2 in order to have a more accurate baseline comparison. However, that being said -- when you see huge speed increases like 50% 100% for dump+gzip, it's doubtful the kernel/process scheduler/IO scheduler could have made that drastic of a difference. Maybe somebody else who has done a 2.4 -> 2.6 upgrade can give us a baseline to subtract from my numbers.
I ran quite a few file system benchmarks in RHAS x86-64 and FC2 x86-64 on a Sun V40z - I did see very consistent 50% improvements in bonnie++ moving from RHAS to FC2 with ext2/ext3 on SAN. On Sun, 2004-11-14 at 23:51 -0800, William Yu wrote: > Greg Stark wrote: > > William Yu <wyu@talisys.com> writes: > > > > > >>Biggest speedup I've found yet is the backup process (PG_DUMP --> GZIP). 100% > >>faster in 64-bit mode. This drastic speed might be more the result of 64-bit > >>GZIP though as I've seen benchmarks in the past showing encryption/compression > >>running 2 or 3 times faster in 64-bit mode versus 32-bit. > > > > > > Isn't this a major kernel bump too? So a different scheduler, different IO > > scheduler, etc? > > > > I'm sure there's some speedup due to the kernel bump. I really didn't > have the patience to even burn the FC2 32-bit CDs much less install both > 32-bit & 64-bit FC2 in order to have a more accurate baseline comparison. > > However, that being said -- when you see huge speed increases like 50% > 100% for dump+gzip, it's doubtful the kernel/process scheduler/IO > scheduler could have made that drastic of a difference. Maybe somebody > else who has done a 2.4 -> 2.6 upgrade can give us a baseline to > subtract from my numbers. > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org