Re: More thougths on DOD

2003-01-08 Thread Leopold Toetsch
Mitchell N Charity wrote: Summary: its slower :-( :( Yep Calculating the flags position in the pool in pobject_lives() and free_unused_pobjects() takes more time then the smaller cache foot_print does gain. Two reasons: positions have to be calced twice and cache is more

Re: More thougths on DOD

2003-01-08 Thread Dan Sugalski
At 6:15 PM -0500 1/6/03, Mitchell N Charity wrote: +pool_pmc[i] = memalign(ALIGN, SIZE*sizeof(PMC)); This is the only problem--memalign's not universal unless we build with the malloc we provide. Have we looked into whether we can mix this malloc with the current memory allocation sys

Re: More thougths on DOD

2003-01-08 Thread Mitchell N Charity
?) Message-ID: <[EMAIL PROTECTED]> Date: Wed, 08 Jan 2003 15:00:38 +0100 From: Leopold Toetsch <[EMAIL PROTECTED]> To: [EMAIL PROTECTED] Cc: P6I <[EMAIL PROTECTED]>, Dan Sugalski <[EMAIL PROTECTED]> Subject: Re: More thougths on DOD References: <[E

Re: More thougths on DOD

2003-01-08 Thread Leopold Toetsch
Mitchell N Charity wrote: The attached patch adds a scheme where: - gc flags are in the pool, and - pmc->pool mapping is done with aligned pools and pmc pointer masking. Observations: - It's fast. (The _test_ is anyway.) I did try it and some more in realiter. Summary: its slower :-( Ca

Re: More thougths on DOD

2003-01-07 Thread Mitchell N Charity
Attached test program shows some additional effects of PMC size and timing. [...] Nifty. The attached patch adds a scheme where: - gc flags are in the pool, and - pmc->pool mapping is done with aligned pools and pmc pointer masking. Observations: - It's fast. (The _test_ is anyway.) P

Re: More thougths on DOD

2003-01-07 Thread Leopold Toetsch
Mitchell N Charity wrote: The attached patch adds a scheme where: - gc flags are in the pool, and - pmc->pool mapping is done with aligned pools and pmc pointer masking. Thanks for that. Observations: - It's fast. (The _test_ is anyway.) Perhaps 4x random, 10x+ linear. I see ~8x/~1

Re: More thougths on DOD

2003-01-06 Thread Leopold Toetsch
Dan Sugalski wrote: At 1:30 PM +0100 1/6/03, Leopold Toetsch wrote: 1) for linear access half sized PMCs give double mark speed. This is in good relation to stress.pasm ... A region of the PMC pool that's entirely mark area would up the cache density by a factor of three or four. s/three

Re: More thougths on DOD

2003-01-06 Thread Dan Sugalski
At 1:30 PM +0100 1/6/03, Leopold Toetsch wrote: Attached test program shows some additional effects of PMC size and timing. A PMC is 32 byte, a SPMC is 16 byte, matching current and minimal PMC sizes for i386 (or a typical 32 bit system). 1) for linear access half sized PMCs give double mark spe

More thougths on DOD

2003-01-06 Thread Leopold Toetsch
Attached test program shows some additional effects of PMC size and timing. A PMC is 32 byte, a SPMC is 16 byte, matching current and minimal PMC sizes for i386 (or a typical 32 bit system). 1) for linear access half sized PMCs give double mark speed. This is in good relation to stress.pasm 2) A