On 10/10/2013 11:29 PM, Xin Li wrote:
> On 10/10/13 20:18, Richard Yao wrote:
>> Thanks for letting us know about this. I have a few comments:
> 
>> 1. We could eliminate a branch entirely by doing this:
> 
>> mlen = MIN(d_end - dst, mlen); while (--mlen >= 0) *dst++ = *cpy++
> 
> I don't think this eliminates the branching as MIN is usually a macro
> that expands to a > b ? b : a.

My mistake. I was thinking of generic swap routines. I do think that
using the MIN() macro is more readable though.

>> 2. This optimization could become a generic optimization pass in 
>> the system compiler. If anyone wants to be helpful, I would
>> suggest sending this patch to the developers of both LLVM and GCC.
> 
> I would ask a LLVM developers for that.  If you know any GCC
> developer, it would be great to let them know as well.

I will ping them on freenode.

>> 3. Have any benchmarks been done to quantify the performance 
>> improvement? That should not be required to merge this, but if any
>>  benchmarks have been done, I would like to see them for my own 
>> curiosity.
> 
> Well to tell the truth I didn't benchmarked the code because looking
> at the assembler, the change is quite clearly an improvement (smaller
> code inside the loop and the actual working code is quite simple).
> The author did some benchmarks and claims to have 6% to 10%
> improvement over the previous version of code with various different
> types of data, but it is still interesting if someone would do a
> scientific benchmark.

I agree that it is an obvious improvement.

>> On 10/10/2013 10:39 PM, Xin Li wrote:
>>> Hi,
>>>
Received: from [140.211.166.183] (helo=smtp.gentoo.org)
        by node002.open-zfs.net
        with esmtp (HybridCluster distributed mail proxy)
        (envelope-from <[email protected]>); Fri, 11 Oct 2013 03:39:41 -0000
>>> FYI, I think you may be interested in one of our (FreeBSD)
>>> recent change to lzjb decompression.
>>>
>>> FreeBSD r256132:
>>>
>>> Improve lzjb decompress performance by reorganizing the code to 
>>> tighten the copy loop.
>>>
>>> Received: from [64.62.153.212] (helo=anubis.delphij.net) by 
>>> node002.open-zfs.net with esmtp (HybridCluster distributed mail 
>>> proxy) (envelope-from <[email protected]>); Fri, 11 Oct 2013 
>>> 02:39:10 -0000 Submitted by:        Denis Ahrens <denis h3q com>
>>>
>>> Should I file a ticket for this for tracking?
>>>
>>> Cheers,
>>>
>>>
>>>
>>> _______________________________________________ developer
>>> mailing list [email protected] 
>>> http://lists.open-zfs.org/mailman/listinfo/developer
>>>
> 
> 
> 
> 
> 
> 
> -------------------------------------------
> illumos-zfs
> Archives: https://www.listbox.com/member/archive/182191/=now
> RSS Feed: https://www.listbox.com/member/archive/rss/182191/24010604-91e32bd2
> Modify Your Subscription: 
> https://www.listbox.com/member/?member_id=24010604&id_secret=24010604-1b6b7e03
> Powered by Listbox: http://www.listbox.com
> 



Received: from [140.211.166.183] (helo=smtp.gentoo.org)
        by node002.open-zfs.net
        with esmtp (HybridCluster distributed mail proxy)
        (envelope-from <[email protected]>); Fri, 11 Oct 2013 03:39:41 -0000

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
developer mailing list
[email protected]
http://lists.open-zfs.org/mailman/listinfo/developer

Reply via email to