Re: ANN: linux-kernel-lzo-2.06.20120123 - update LZO to v2.06

2012-10-09 Thread David Sterba 
On Wed, Jan 25, 2012 at 02:36:18AM +0100, Andi Kleen wrote:
> I ran benchmarks on the new miniLZO and LZ4 on 64bit. LZ4 is generally slower 
> than snappy/lzo in the micro benchmarks.

And the reason why you measured worse speed for LZ4 although (AFAICT)
everybody else's measurements claim the opposite is quite simple:
likely due to a copy&paste error you did not benchmark LZ4 at all:

https://github.com/andikleen/snappy-c/blob/master/glue.c#L282

274 void test_lz4(char *map, size_t size, char *fn)
275 {
276 int i;
277 int err;
278 size_t outlen = size * 2;
279 char *out = xmalloc(outlen);
280 char *buf2 = xmalloc(size);
281
282 BENCH(fastlz, "lz4", fn, NULL);
  ^^

283
284 free(out);
285 free(buf2);
286 }

(LZ4 is on it's track towards kernel)

david
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ANN: linux-kernel-lzo-2.06.20120123 - update LZO to v2.06

2012-01-24 Thread Andi Kleen 
On Mon, Jan 23, 2012 at 05:19:40PM +0100, Markus F.X.J. Oberhumer wrote:
> Hi,
> 
> I've prepared a small package that updates the LZO version in the Linux
> kernel to LZO v2.06.

I ran benchmarks on the new miniLZO and LZ4 on 64bit. LZ4 is generally slower 
than snappy/lzo in the micro benchmarks. The new LZO is better than the old
one, but still loses to snappy most of the time (but often by very
small amounts only)

Will be worth checking the new LZO will the full distribution boot test.

I agree it's definitely a good idea to update the kernel version.
However I must say it would be a major project to bring it up
to kernel coding standards.

snappy is still interesting, but much less so than it was before.

-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ANN: linux-kernel-lzo-2.06.20120123 - update LZO to v2.06

2012-01-23 Thread Nitin Gupta 
Hi Markus,

Thanks for the patches!

On 01/23/2012 11:19 AM, Markus F.X.J. Oberhumer wrote:

> Hi,
> 
> I've prepared a small package that updates the LZO version in the Linux
> kernel to LZO v2.06.
> 
> Please get it from:
> 
>   
> http://www.oberhumer.com/opensource/lzo/download/Testing/linux-kernel-lzo-2.06.20120123.tar.gz
> 
> As stated in the README, its main purpose is to allow easy benchmarking of the
> latest LZO versions - these do feature some nice speed improvements, and while
> I have done a lot of synthetic benchmarking I'm really very curious and
> appreciate feedback on "real-world" performance numbers like usage in
> btrfs and zram.
> 

I will soon integrate them with zram and get some performance numbers.

Thanks,
Nitin



> 
> http://www.oberhumer.com/opensource/lzo/
> 
> On 2012-01-18 16:05, Markus F.X.J. Oberhumer wrote:
>> On 2012-01-13 01:28, Andi Kleen wrote:
>>> Here's a slightly updated version of the BTRFS snappy interface.
>>> snappy is a faster compression algorithm that provides similar
>>> compression as LZO, but generally better performance.
>>
>> I'd like to note that the LZO version in the current Linux kernel is
>> rather outdated - it seems to be based on the 2005 release.
>>
>> In fact the latest version LZO 2.06 does compress both slightly faster and
>> better than snappy 1.0.4 when benchmarking the Calgary and Silesia
>> compression corpus (tested with gcc 4.6 on Nehalem & Sandy Bridge).
>>
>> Furthermore please be aware that from a pure compression point of view
>> snappy et al. are very close cousins of LZO (strictly byte-aligned LZ77)
>> that mainly differ in implementation issues like using a table to
>> number of branches - and indeed similar optimizations could be applied
>> to any version.
>>
>> I'm not sure if there is an official kernel maintainer of LZO, but I'd
>> offer to assist you updating to the latest version and eliminating
>> any possible performance issues.
> 


--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html