[Shootout-list] process creation & message passing

[Bengt Kleberg]

Isaac Gouy wrote:
...deleted
> imo That's misleading.
> The lowest N used for the procinst test was 32,000

i am not responsible for the n's used in procinst. please take that up 
with the right person.
i want a wide selction, so as to measure:
1 start up times (small n)
2 steady state times (medium n)
3 low memory times (large n)

i have sent in a new procinst (well, 2 since i thought somebody would be 
interested in the create&delete kind of test), but it has not yet made 
it into the shootout.


> ringmsg *specifies* 8,192 child processes

for chainmsg we use the latest specification that is posted to the mail 
list.
please use the latest ringmsg specification posted to the mail list, too.


>>i want to measure:
>>1 start up times (small n)
>>2 steady state times (medium n)
>>3 low memory times (large n)
>>
>>i do not want to exclude 1 and 3. do you?
> 
> 
> Yes. The Shootout uses broad brush-strokes.
> We're moving from one concurrency test (producer-consumer with it's 2
> threads) to three concurrency tests. 
> imo That's enough - there are other aspects of Shootout which need
> attention.

i am interested in 1, 2 and 3. i am willing to give the problem my 
attention. you are trying to stop me.

why do you do that, if you think other aspects need attention?


bengt