Hi Xiang:
Got it and thanks.
Will follow the rules in next PS.
Chenxi
On 5/17/19 2:35 PM, Gao Xiang wrote:
> Hi Chenxi,
>
> Some words about the patch format.. not important tho.
>
> One suggestion is that the subject line should be better written
> as "[PATCH v2/v3/...] title" since it's
Hi Chenxi,
Some words about the patch format.. not important tho.
One suggestion is that the subject line should be better written
as "[PATCH v2/v3/...] title" since it's more clear to know
which patch is the latest patch among these emails.
On 2019/5/17 13:56, Chenxi Mao wrote:
> FAST_DEC_LOOP
FAST_DEC_LOOP was introduced from LZ4 1.9.0[1]
This change would be introduce 10% on decompress operation
according to LZ4 benchmark result on X86 devices.
Meanwhile, LZ4 with FAST_DEC_LOOP could get improvements on ARM64,
however clang compiler has downgrade if FAST_DEC_LOOP enabled.
So
FAST_DEC_LOOP was introduced from LZ4 1.9[1].
This change would be introduce 10% on decompress operation
according to LZ4 benchmark result on X86 devices.
Meanwhile, LZ4 with FAST_DEC_LOOP could get improvements,
however clang compiler has downgrade if FAST_DEC_LOOP enabled.
So FAST_DEC_LOOP only
Hi Xiang:
I checked my patch, this issue fixes already included in my first E-mail.
I am working on the patch PS2, I will send it later.
Chenxi
On 5/16/19 10:10 AM, Gao Xiang wrote:
>
>
> On 2019/5/16 10:07, Mao, Chenxi wrote:
>> Hi Xiang:
>>
>> I checked the deliver history.
>> There is only
It seems you just sent the same email twice, please check my previous reply.
Thanks,
Gao Xiang
On 2019/5/16 10:11, Mao, Chenxi wrote:
> Hi Xiang:
>
> I checked the deliver history.
> There is only below delivery related with lz4.c
> Pull request:616
> 4e3accc Fix Dict Size Test in
On 2019/5/16 10:07, Mao, Chenxi wrote:
> Hi Xiang:
>
> I checked the deliver history.
> There is only below delivery related with lz4.c
> Pull request:616
> 4e3accc Fix Dict Size Test in `LZ4_compress_fast_continue()`
> 535636f Don't Attach Very Small Dictionaries
> This 2 changes seems like
Hi Xiang:
I checked the deliver history.
There is only below delivery related with lz4.c
Pull request:616
4e3accc Fix Dict Size Test in `LZ4_compress_fast_continue()`
535636f Don't Attach Very Small Dictionaries
This 2 changes seems like ONLY bug fixes for dictionary fix baesd on 1.8.3
Based on
Hi Xiang:
I checked the deliver history.
There is only below delivery related with lz4.c
Pull request:616
4e3accc Fix Dict Size Test in `LZ4_compress_fast_continue()`
535636f Don't Attach Very Small Dictionaries
This 2 changes seems like ONLY bug fixes for dictionary fix baesd on 1.8.3
Based on
Hi Yann and Xiang:
For this FAST_DEC_LOOP change, I only pick up decompress related patches to
current kernel LZ4 implementation(based on 1.8.3).
Here is my cherry-pick list:
2589c44 created LZ4_FAST_DEC_LOOP build macro
605d811 enable LZ4_FAST_DEC_LOOP build macro on aarch64/GCC by default
On 2019/5/16 7:48, Mao, Chenxi wrote:
> Hi Yann and Xiang:
> For this FAST_DEC_LOOP change, I only pick up decompress related patches to
> current kernel LZ4 implementation(based on 1.8.3).
> Here is my cherry-pick list:
> 2589c44 created LZ4_FAST_DEC_LOOP build macro
> 605d811 enable
Hi Yann,
On 2019/5/16 1:03, Cyan wrote:
> Re-posted,
> it seems the previous message was rejected by the linux-kernel server
> due to some kind of format limitation (no html).
>
>
> Le mer. 15 mai 2019 à 09:56, Cyan a écrit :
>>
>> The v1.9.0 version has a bug which makes it read a few bytes
Re-posted,
it seems the previous message was rejected by the linux-kernel server
due to some kind of format limitation (no html).
Le mer. 15 mai 2019 à 09:56, Cyan a écrit :
>
> The v1.9.0 version has a bug which makes it read a few bytes out of bound in
> certain cases.
> This was fixed in
Hi Xiang:
Thanks for your reply, I will have a stress test on my device later.
I didn't have chance to test LZ4 with clang build because of device limitation.
I think I could do it later.
I guess the clang performance downgrade is caused by some compiler optimization
options.
I will double
Hi Chenxi,
On 2019/5/15 8:43, Chenxi Mao wrote:
> FAST_DEC_LOOP was introduced from LZ4 1.9.
> This change would be introduce 10% on decompress operation
> according to LZ4 benchmark result on X86 devices.
> Meanwhile, LZ4 with FAST_DEC_LOOP could get improvements,
> however clang compiler has
FAST_DEC_LOOP was introduced from LZ4 1.9.
This change would be introduce 10% on decompress operation
according to LZ4 benchmark result on X86 devices.
Meanwhile, LZ4 with FAST_DEC_LOOP could get improvements,
however clang compiler has downgrade if FAST_DEC_LOOP enabled.
So FAST_DEC_LOOP only
16 matches
Mail list logo