New malloc ready, take 42
David Xu
davidxu at freebsd.org
Wed Dec 28 21:39:24 PST 2005
Jason Evans wrote:
>
> On Dec 28, 2005, at 9:06 PM, David Xu wrote:
>
>>
>> I am using patch file jemalloc_20051222c.diff. I don't use
>> MALLOC_OPTIONS and don't create link malloc.conf in /etc.
>> I have tested super-smack with following command:
>>
>> %super-smack ./select-key.smack 10 5000
>> Query Barrel Report for client smacker1
>> connect: max=4ms min=1ms avg= 2ms from 10 clients
>> Query_type num_queries max_time min_time q_per_s
>> select_index 100000 0 0 15895.76
>
>
> Can you please try again with MALLOC_OPTIONS=aj, and (similar to the
> tests I ran):
>
> MALLOC_OPTIONS=aj super-smack ./select-key.smack 4 10000
>
> It will be helpful to know whether your hardware behaves differently
> for the same test. Also, I don't have any working hardware right now,
> so I can't try to reproduce your results.
>
> Thanks,
> Jason
>
>
OK, I have linked /etc/malloc.conf to aj, this changes the performance:
%super-smack ./select-key.smack 4 10000
Query Barrel Report for client smacker1
connect: max=23ms min=0ms avg= 5ms from 4 clients
Query_type num_queries max_time min_time q_per_s
select_index 80000 0 0 20480.54
%super-smack ./select-key.smack 4 10000
Query Barrel Report for client smacker1
connect: max=1ms min=0ms avg= 0ms from 4 clients
Query_type num_queries max_time min_time q_per_s
select_index 80000 0 0 19734.14
%super-smack ./select-key.smack 4 10000
Query Barrel Report for client smacker1
connect: max=1ms min=0ms avg= 0ms from 4 clients
Query_type num_queries max_time min_time q_per_s
select_index 80000 0 0 19846.49
%super-smack ./select-key.smack 4 10000
Query Barrel Report for client smacker1
connect: max=1ms min=0ms avg= 0ms from 4 clients
Query_type num_queries max_time min_time q_per_s
select_index 80000 0 0 20045.10
%super-smack ./select-key.smack 4 10000
Query Barrel Report for client smacker1
connect: max=0ms min=0ms avg= 0ms from 4 clients
Query_type num_queries max_time min_time q_per_s
select_index 80000 0 0 19544.33
%
but it still can not beat phkmalloc. :-)
More information about the freebsd-current
mailing list