Search
Saturday, 10th of November 2018, 14:36:40 UTC
14:58:11
stassats
ok, os or not os bug, can it be worked around?
15:05:36
stassats
where does the OS deallocate the stack when it's an os-provided one?
15:14:27
stassats
http://openbsd-archive.7691.n7.nabble.com/library-5771-rthreads-concurrent-pthread-exit-and-pthread-join-can-segv-tt189977.html
15:28:39
stassats
i think even when left to the OS to deallocate the stack it still can fail
15:29:11
pfdietz
I don't think it's an accident that two of the most popular CLs are also the fastest to build (sbcl, ccl).
15:29:28
pfdietz
Not sure how long cmucl takes now.
15:29:48
stassats
if you can call what cmucl does as "build"
15:33:43
pfdietz
Ok, kicked off tests in four processes.
15:34:00
stassats
what hardware are you using?
15:35:48
pfdietz
Intel Core i7-6700, 3.4 GHz, 32 Gb memory
15:36:38
stassats
i imagine the new 16 core cpus could speed up testing
15:37:19
pfdietz
I think this has four physical cores and 8 cores if you count hyperthreading.
15:37:57
pfdietz
Yes, the testing is highly parallelizable.
15:37:57
stassats
i'm still using i7-4770K, other than having 4 cores it's still as capable as five years ago (moore's law, where art thou)
15:38:41
pfdietz
Moore's law is over. Computer architecture is becoming a hot topic again. This has implications for programming languages and compilers.
15:39:33
pfdietz
Gnashing of teeth and crying from the singularitarians, though.
15:39:37
stassats
and i'm really into that much parallel processing, so if all they do is go up to 128 cores, i think i'm good for another five years...
15:40:32
pfdietz
SBCL has ARM64 as its secondary target, right? After X86-32/64.
15:41:40
stassats
well, it's the port i did, so i care about it
15:42:20
stassats
but not really using it as i had planned
15:43:25
pfdietz
Not a waste of your time, I think.
15:43:39
stassats
well, people are using it, so there's that
15:44:25
pfdietz
x86 has energy efficiency issues, and that will become more important, especially as the number of cores goes up.
15:44:26
stassats
and doing a full port is a good way to acquaint yourself with the backend and the runtime
15:44:42
pfdietz
For some applications energy cost over the lifetime of the hardware > hardware cost.
15:45:49
stassats
but really time consuming, that's why i don't want to repeat it for PPC64
15:46:42
pfdietz
Does PPC have much of a future?
15:52:55
pfdietz
To get some idea of performance: this machine is building sbcl from scratch (using the last sbcl release) in almost exactly 2 minutes.
15:54:48
stassats
i use a safety 0 sbcl for building, in 1:49
15:54:59
stassats
and around a minute when using multiple threads
15:55:48
stassats
and always using SBCL_MAKE_JOBS=-j8
15:55:50
pfdietz
I think I tweaked dynamic space size up, but that was about it for customization.
15:56:03
stassats
(that's for the c runtime only)
15:56:34
stassats
SBCL_MAKE_PARALLEL=8 is for make-host-2, but it's not as good when something goes wrong
15:57:49
pfdietz
Let's try those (testing jobs are niced).
16:00:52
pfdietz
Still about 2 minutes. I'll try it again later with no contention from testing.
16:14:19
pkhuong
pfdietz: if you send me a public key, I'll give you ssh on my xeon phi
16:14:57
pkhuong
256 threads that look like slowish x86-64
17:53:53
flip214
real 1m38.045s including contribs for both SBCL_MAKE_*=6, Debian amd64 on i7-4810MQ
18:20:14
pfdietz
I'm ok with this, thanks.
Sunday, 11th of November 2018, 2:36:40 UTC