Martin Albrecht wrote:
>> It would appear this cache program is not particularly good at guess cache
>> sizes, but what is probably more important is that it is able to optimise
>> the code so it works best on a particular processor. It is pretty
>> irrelevant what it thinks that processor may o
> It would appear this cache program is not particularly good at guess cache
> sizes, but what is probably more important is that it is able to optimise
> the code so it works best on a particular processor. It is pretty
> irrelevant what it thinks that processor may or may not be. It's more
>
Martin Albrecht wrote:
> Hi there,
>
> in the thread
>
> http://is.gd/4EMT8
>
> David and I agreed that M4RI should have a better, more robust and cross-
> platform mechanism to check for cache sizes.
>
> Thus I wrote a little C program to check for the cache size using timings of
> rand
OK, one more. This might be interesting or amusing - I am running
your cache program with the virtualbox image of sage-4.1.2. Timings
were somewhat inconsistent, probably because I was doing other tasks
the first time I ran it. Second time it gave better results:
8 0.032 3.071
1
Maybe a more interesting machine: Dual-Core Intel Xeon with a total of
4 cores, 32K of L1 cache, 4MB L2 cache per core.
./cache
8 0.010 2.045
16 0.019 2.000
32 0.041 2.084
64 0.096 2.368
128 0.336 3.491
256 0.790 2.352
512 1.7
I have a core 2 duo, 32k L1 on each, 3 MB L2, from the sage install
this agrees:
sage-4.2: grep "cache size" install.log
checking the L1 cache size... 32768 Bytes
checking the L2 cache size... 3145728 Bytes
and:
./cache
8 0.017 1.877
16 0.024 1.353
32 0.045 1.91
Laptop:
Intel(R) Core(TM)2 CPU T5300 @ 1.73GHz
8 0.023 7.611
16 0.046 2.018
32 0.093 2.034
64 0.289 3.097
128 0.641 2.216
256 0.744 1.160
512 1.542 2.073
1024 3.394 2.202
2048 7.826 2.305
4096
On Tue, Oct 27, 2009 at 9:51 PM, Martin Albrecht
wrote:
> I'd appreciate if those people on this list who have different hardware or
> software (OSX, Solaris, PPC, Sparc) could compile the program and report back
> whether the program reports correct sizes for them. To compile and run:
CPU info
Martin Albrecht wrote:
> Hi there,
>
> in the thread
>
> http://is.gd/4EMT8
>
> David and I agreed that M4RI should have a better, more robust and cross-
> platform mechanism to check for cache sizes.
>
> Thus I wrote a little C program to check for the cache size using timings of
> rand
Ubuntu 8.10 64bit Intel Q6600 (Core2Quad)
./cache
8 0.024 7.431
16 0.050 2.085
32 0.102 2.026
64 0.251 2.465
128 0.468 1.866
256 0.557 1.192
512 1.151 2.065
1024 3.055 2.654
2048 6.898 2.258
409615.7
It seems the walltime is always zero for you, I'll see if I can reproduce it
somewhere.
Martin
--
name: Martin Albrecht
_pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99
_otr: 47F43D1A 5D68C36F 468BAEBA 640E8856 D7951CCF
_www: http://www.informatik.uni-bremen.de/~malb
_jab: mar
On Tuesday 27 October 2009, Michael Welsh wrote:
> I dunno. How do I find out?
Mhh, good question you could send the exact version number of your CPU and I
could start googling. Or, if you installed Sage from source you could:
m...@road:~$ grep "cache size" /usr/local/sage-4.1.2/install.log
I've got 32K/32K (which may mean 64K) according to
http://www.everymac.com/systems/apple/imac/stats/imac-core-2-duo-2.66-20-inch-aluminum-early-2009-specs.html
On 28/10/2009, at 10:18 AM, Martin Albrecht wrote:
> does your C2D have 64K of L1? (IIRC there are no C2D with 64K data
> L1, but I
>
I dunno. How do I find out?
On 28/10/2009, at 10:18 AM, Martin Albrecht wrote:
> does your C2D have 64K of L1? (IIRC there are no C2D with 64K data
> L1, but I
> do get confused by Intel's marketing from time to time)
--
http://yomcat.geek.nz
--~--~-~--~~~---~--~-
On Tuesday 27 October 2009, Michael Welsh wrote:
> OSX Dore 2 Duo.
>
> Gordon:Downloads yomcat$ ./cache
> 8 0.004 1.944
> 16 0.008 2.010
> 32 0.019 2.413
> 64 0.066 3.501
>128 0.226 3.405
>256 0.479 2.121
>512 0.983
Martin Albrecht wrote:
> Hi there,
>
> in the thread
>
> http://is.gd/4EMT8
>
> David and I agreed that M4RI should have a better, more robust and cross-
> platform mechanism to check for cache sizes.
>
> Thus I wrote a little C program to check for the cache size using timings of
> rand
OSX Dore 2 Duo.
Gordon:Downloads yomcat$ ./cache
8 0.004 1.944
16 0.008 2.010
32 0.019 2.413
64 0.066 3.501
128 0.226 3.405
256 0.479 2.121
512 0.983 2.055
1024 1.997 2.031
2048 4.043 2.025
409
17 matches
Mail list logo
