Re: native threads

Your "thread pools" approach sounds wise. While I don't (but should) 
know what CCL does right now w.r.t. threads, the "modern thing to do" is 
for Lisps to use native OS threads, which I can easily imagine take some 
time to launch, but then take good advantage of symmetric 
multiprocessing (SMP).  That's what Franz does anyway.  I assume, based 
on the compiler output I've studied, that CCL supports SMP threads. I 
would not expect that CCL runs its own scheduler anymore. [Wizards: 
please correct me if I'm wrong].

--Tim

On 1/5/24 10:54 AM, David McClain wrote:
> I have spent more than a decade implementing various versions of 
> Actors systems. Some were along the lines you describe, launching 
> native threads to perform the work of Actors.
>
> I finally concluded that, having just run a good race, you ought not 
> kill the race horses, but rather retire them to the stable for another 
> race. IOW, I now use thread pools, permanently launched and waiting 
> for dispatch to another task. Thread startup entails a huge amount of 
> overhead - allocating stack space and CPU registers, registering with 
> OS, etc.
>
> - DM
>
>> On Jan 5, 2024, at 07:32, Taoufik Dachraoui 
>> <[email protected]> wrote:
>>
>> why ccl is much slower?
>>
>> taoufik@Ankbot:~/workspace/ccl/actor$ sbcl
>> This is SBCL 2.3.11, an implementation of ANSI Common Lisp.
>> More information about SBCL is available at <http://www.sbcl.org/>.
>>
>> SBCL is free software, provided as is, with absolutely no warranty.
>> It is mostly in the public domain; some portions are provided under
>> BSD-style licenses.  See the CREDITS and COPYING files in the
>> distribution for more information.
>> * (time (prog1 t (loop for i from 0 upto 100000000 collect i)))
>> Evaluation took:
>>   1.379 seconds of real time
>>   1.379015 seconds of total run time (0.856185 user, 0.522830 system)
>>   [ Real times consist of 1.143 seconds GC time, and 0.236 seconds 
>> non-GC time. ]
>>   [ Run times consist of 1.146 seconds GC time, and 0.234 seconds 
>> non-GC time. ]
>>   100.00% CPU
>>   2,913,031,106 processor cycles
>>   1,600,427,200 bytes consed
>>
>> T
>> * (quit)
>> taoufik@Ankbot:~/workspace/ccl/actor$ ccl
>> Clozure Common Lisp Version 1.12.1 (v1.12.1-22-g6b1f1d3a) LinuxX8664
>>
>> For more information about CCL, please see http://ccl.clozure.com.
>>
>> CCL is free software.  It is distributed under the terms of the Apache
>> Licence, Version 2.0.
>> ? (time (prog1 t (loop for i from 0 upto 100000000 collect i)))
>> (PROG1 T (LOOP FOR I FROM 0 UPTO 100000000 COLLECT I))
>> took 11,946,514 microseconds (11.946514 seconds) to run.
>>      11,147,722 microseconds (11.147722 seconds, 93.31%) of which was 
>> spent in GC.
>> During that period, and with 24 available CPU cores,
>>      11,632,370 microseconds (11.632370 seconds) were spent in user mode
>>         292,428 microseconds ( 0.292428 seconds) were spent in system 
>> mode
>>  1,600,000,032 bytes of memory allocated.
>>  397,400 minor page faults, 0 major page faults, 0 swaps.
>> T
>> ?
>>
>> On Fri, Jan 5, 2024 at 3:28 PM Taoufik Dachraoui 
>> <[email protected]> wrote:
>>
>>     It looks like it is not because of the threads:
>>
>>     taoufik@Ankbot:~/workspace/ccl/actor$ sbcl
>>     This is SBCL 2.3.11, an implementation of ANSI Common Lisp.
>>     More information about SBCL is available at <http://www.sbcl.org/>.
>>
>>     SBCL is free software, provided as is, with absolutely no warranty.
>>     It is mostly in the public domain; some portions are provided under
>>     BSD-style licenses.  See the CREDITS and COPYING files in the
>>     distribution for more information.
>>     * (time (reduce #'+ (loop for i from 0 upto 100000000 collect i)))
>>     Evaluation took:
>>       1.836 seconds of real time
>>       1.835223 seconds of total run time (1.351394 user, 0.483829 system)
>>       [ Real times consist of 1.184 seconds GC time, and 0.652
>>     seconds non-GC time. ]
>>       [ Run times consist of 1.184 seconds GC time, and 0.652 seconds
>>     non-GC time. ]
>>       99.95% CPU
>>       3,876,850,246 processor cycles
>>       1,600,427,200 bytes consed
>>
>>     5000000050000000
>>     * taoufik@Ankbot:~/workspace/ccl/actor$ ccl
>>     Clozure Common Lisp Version 1.12.1 (v1.12.1-22-g6b1f1d3a) LinuxX8664
>>
>>     For more information about CCL, please see http://ccl.clozure.com.
>>
>>     CCL is free software.  It is distributed under the terms of the
>>     Apache
>>     Licence, Version 2.0.
>>     ? (time (reduce #'+ (loop for i from 0 upto 100000000 collect i)))
>>     (REDUCE #'+ (LOOP FOR I FROM 0 UPTO 100000000 COLLECT I))
>>     took 13,091,079 microseconds (13.091079 seconds) to run.
>>          11,036,666 microseconds (11.036666 seconds, 84.31%) of which
>>     was spent in GC.
>>     During that period, and with 24 available CPU cores,
>>          12,734,283 microseconds (12.734283 seconds) were spent in
>>     user mode
>>             336,005 microseconds ( 0.336005 seconds) were spent in
>>     system mode
>>      1,600,000,032 bytes of memory allocated.
>>      397,400 minor page faults, 0 major page faults, 0 swaps.
>>     5000000050000000
>>     ?
>>
>>     On Fri, Jan 5, 2024 at 3:10 PM Taoufik Dachraoui
>>     <[email protected]> wrote:
>>
>>         Hi
>>
>>         In my current implementation of a classical actor model
>>         I found that using ccl processes is much slower than sbcl threads
>>
>>         to create a thread I use ccl:process-run-function, is there
>>         another way to
>>         create native threads that are much faster; I do not need the
>>         ccl scheduling,
>>         I want to create threads that are scheduled by the OS, I
>>         think that the ccl
>>         scheduler is the reason why my ccl tests are much slower than
>>         the tests run
>>         with sbcl
>>
>>         Regards
>>         -- 
>>         Taoufik Dachraoui
>>
>>
>>
>>     -- 
>>     Taoufik Dachraoui
>>
>>
>>
>> -- 
>> Taoufik Dachraoui
>>
>