Re: filesystem and x86 vs. x86_64 benchmarking...
> If they got 100% cpu then user+system should add up to real. If the
> system was otherwise idle then the missing time must have been spend
> on waiting for the disk.
> The large difference in time would mean different FS have drastically
> different seek behaviour. What else could create more IO waits?
My assumption was that the I/O wait time was not being reported by the bash
"time" builtin in either "user" or "sys". I initially had a column showing
them added up. They basically were identical for all cases, so I assumed
the missing time was I/O wait time since the machine was quiescent other
than the tests.
> PS: try using -pipe unless you notice that it causes swaping.
Good idea. ./configure --disable-static has a noticeable effect as well
;-) (Half the compilations!)
> Register allocation is also exponential with the number of registers
> iirc. And amd64 has a lot more than i386.
Right. I did file a bug; we'll see if the gcc people have something to say
about it.
> > not a bug. It would be interesting to know if you need double the memory
> > for comparable (slightly improved?) performance for C++ compilation in
> > x86_64. I hadn't really considered that. Hopefully it's a bug ;-)
>
> Every pointer is double the size and you have a lot of those in
> gcc. On top of that you get more registers, more opcodes, more
> inlining, ...
Well, if it really takes twice the memory to get a few percent speed
increase, I'd like to know. Compilation isn't the only interesting test,
but it _is_ interesting.
> 64bit userspace yes. But comparing 32/64 bit kernel with 32bit
> userspace doesn't show a winner on your tests yet. You can#t realy
> decide what kernel to use going by those tests.
I don't have enough memory to run 64 bit userspace, I think we agree
there. There is a clear speedup running a 64/32 setup over 32/32 using
xfs, right? So why might 32/32 be better? Sorry to be dense.
I'm assuming that CPU bound things like "lame" and "mencoder" will see a
decent speedup in x86_64 mode, too, so I like the idea of being able to run
certain things in 64 bit mode.
> I expect some differences in the char speed. The block speed should be
> mostly disk bound and show little to no change.
I'm running it in all three modes on xfs right now.
> But a lot of real live read/writes will be to/from cache. The char
> access shows how fast system calls and the in memory VFS are and
> should reflect on the cache speed (consider the char speed as cache
> accesses per second).
Interesting, I hadn't thought too much about this.
> You can make it up with doing those extra bonnie tests. :) Never
> installed a 32bit kernel here so I could never test this.
See the attachment. (Now I _really_ hope that there is g++ bug ;-))
Take care,
Dale
--
Dale E. Martin - dale@the-martins.org
http://the-martins.org/~dmartin
Version 1.03 ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
xfs 64/64 2G 49933 87 52891 8 22838 4 39790 68 52517 4 283.0 0
xfs 64/32 2G 39172 92 49659 12 20257 4 33729 75 49165 4 284.9 0
xfs 32/32 2G 40419 94 47372 10 21362 5 34045 75 50770 4 252.3 0
------Sequential Create------ --------Random Create--------
-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
xfs 64/64 16 3507 15 +++++ +++ 3190 14 3444 16 +++++ +++ 1196 5
xfs 64/32 16 3421 98 +++++ +++ +++++ +++ 3540 99 +++++ +++ 12430 99
xfs 32/32 16 3280 98 +++++ +++ +++++ +++ 3430 99 +++++ +++ 14181 99
xfs 64/32,2G,39172,92,49659,12,20257,4,33729,75,49165,4,284.9,0,16,3421,98,+++++,+++,+++++,+++,3540,99,+++++,+++,12430,99
Version 1.03 ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
xfs 32/32 2G 40419 94 47372 10 21362 5 34045 75 50770 4 252.3 0
------Sequential Create------ --------Random Create--------
-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 3280 98 +++++ +++ +++++ +++ 3430 99 +++++ +++ 14181 99
xfs 32/32,2G,40419,94,47372,10,21362,5,34045,75,50770,4,252.3,0,16,3280,98,+++++,+++,+++++,+++,3430,99,+++++,+++,14181,99
Reply to: