Hi Fengwei,

I did not see such big performance regression on Juno, followings are my result:

geekbench 3.3.2:
                    single    multi
w/o opt. memcpy        857     2221
with opt. memcpy       855     2263
performance gain       -0.2%      1.9%

Quadrant 2.1
w/o opt. memcpy    9662
with opt. memcpy   9563
performance gain    -1.0%




--
潘穎軒 Yingshiuan (Peter) Pan


2015-07-06 11:16 GMT+08:00 yfw <fengwei.yin@linaro.org>:
Hi Yingshiuan,
That's fine. Do we have updated patch for memcpy? Did you get the
Quadrant/Geekbench3 data in your side? Thanks.

Regards
Yin, Fengwei

On 2015年07月06日 09:24, Yingshiuan Pan wrote:
Hi Fengwei
Sorry, my carelessness, you're right.


--
潘穎軒 Yingshiuan (Peter) Pan


2015-07-03 19:51 GMT+08:00 yfw <fengwei.yin@linaro.org
<mailto:fengwei.yin@linaro.org>>:

    Hi Yingshiuan,
    https://android-review.googlesource.com/#/c/147915/ is for memchr.
    while
    http://review.android.git.linaro.org/#/c/15529/ is for memcpy. Could you
    please double check whether you have correct link? Thanks.

    Regards
    Yin, Fengwei

    在 2015/7/3 14:04, Yingshiuan Pan 写道:

        Hi Fengwei,

        Thanks for the report, there is a new revision here:
        https://android-review.googlesource.com/#/c/147915/
        Although only single instruction is changed, someone said he did see
        measurable improvement.
        Would you give a try? I'll also run those benchmarks to see if I can
        reproduce the result.

        btw, what is the platform you are using?

        --
        潘穎軒 Yingshiuan (Peter) Pan


        2015-07-03 13:59 GMT+08:00 yfw <fengwei.yin@linaro.org
        <mailto:fengwei.yin@linaro.org>
        <mailto:fengwei.yin@linaro.org <mailto:fengwei.yin@linaro.org>>>:

              From their report, they used:
             Quadrant Professional(v2.1.1)
             GeekBench3 (v3.3.2)

             I suppose the methodology to run these benchmarks is very
        known.


             Regards
             Yin, Fengwei

             在 2015/7/3 11:53, Tom Gall 写道:

                 Hi Fengwei,

                 Thanks for the report.  I think we would be interested in
                 knowing the
                 nature of the test being used.

                 I've added Bero to the CC as I'm sure he'll be
        interested as well.

                 On Thu, Jul 2, 2015 at 9:35 PM, yfw
        <fengwei.yin@linaro.org <mailto:fengwei.yin@linaro.org>
                 <mailto:fengwei.yin@linaro.org

        <mailto:fengwei.yin@linaro.org>>> wrote:

                     Hi,
                     About the patch
        http://review.android.git.linaro.org/#/c/15529/, we got
                     an issue report from member. It could improve the micro
                     benchmark but
                     hurt general benchmark actually. Here is their test
        result:

                           bionic-benchmarks32 BM_string_memcpy:
                           Size  AOSP    Linaro  Perf Improvement
                           8     450.7   329.5   -37%
                           64    1715.7  2225.3  23%
                           512   3100.7  4253.0  27%
                           1K    3271.8  4652.3  30%
                           8K    2460.9  3174.7  22%
                           16K   2267.4  2953.2  23%
                           32K   2222.9  2800.7  21%
                           64K   2204.3  2640.1  17%

                                           without patch   with patch
        perf delta
                           Quadrant mem    3400            2100        -38%
                           GeekBench3
                             single-core   436             260         -40%
                             multi-core    434             358         -18%

                     I am wondering whether this is an known issue or
        not for
                     this list.
                     Thanks in advance.

                     Regards
                     Yin, Fengwei
                     _______________________________________________
                     linaro-android mailing list
        linaro-android@lists.linaro.org
        <mailto:linaro-android@lists.linaro.org>
                     <mailto:linaro-android@lists.linaro.org
        <mailto:linaro-android@lists.linaro.org>>
        https://lists.linaro.org/mailman/listinfo/linaro-android




             _______________________________________________
             linaro-android mailing list
        linaro-android@lists.linaro.org
        <mailto:linaro-android@lists.linaro.org>
        <mailto:linaro-android@lists.linaro.org
        <mailto:linaro-android@lists.linaro.org>>
        https://lists.linaro.org/mailman/listinfo/linaro-android