[FFmpeg-devel] [PATCH v3 7/7] avutil/la: Add function performance testing

[Hao Chen]

在 2023/5/24 下午7:03, Rémi Denis-Courmont 写道:
>
> Le 24 mai 2023 10:39:59 GMT+03:00, Hao Chen <chenhao at loongson.cn> a écrit :
>> 在 2023/5/20 下午5:38, Rémi Denis-Courmont 写道:
>>> Le lauantaina 20. toukokuuta 2023, 10.27.19 EEST Hao Chen a écrit :
>>>> From: yuanhecai <yuanhecai at loongson.cn>
>>>>
>>>> This patch supports the use of the "checkasm --bench" testing feature
>>>> on loongarch platform.
>>>>
>>>> Change-Id: I42790388d057c9ade0dfa38a19d9c1fd44ca0bc3
>>>> ---
>>>>    libavutil/loongarch/timer.h | 48 +++++++++++++++++++++++++++++++++++++
>>>>    libavutil/timer.h           |  2 ++
>>>>    2 files changed, 50 insertions(+)
>>>>    create mode 100644 libavutil/loongarch/timer.h
>>>>
>>>> diff --git a/libavutil/loongarch/timer.h b/libavutil/loongarch/timer.h
>>>> new file mode 100644
>>>> index 0000000000..44ed786409
>>>> --- /dev/null
>>>> +++ b/libavutil/loongarch/timer.h
>>>> @@ -0,0 +1,48 @@
>>>> +/*
>>>> + * Copyright (c) 2023 Loongson Technology Corporation Limited
>>>> + * Contributed by Hecai Yuan <yuanhecai at loongson.cn>
>>>> + *
>>>> + * This file is part of FFmpeg.
>>>> + *
>>>> + * FFmpeg is free software; you can redistribute it and/or
>>>> + * modify it under the terms of the GNU Lesser General Public
>>>> + * License as published by the Free Software Foundation; either
>>>> + * version 2.1 of the License, or (at your option) any later version.
>>>> + *
>>>> + * FFmpeg is distributed in the hope that it will be useful,
>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>>>> + * Lesser General Public License for more details.
>>>> + *
>>>> + * You should have received a copy of the GNU Lesser General Public
>>>> + * License along with FFmpeg; if not, write to the Free Software
>>>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301
>>>> USA + */
>>>> +
>>>> +#ifndef AVUTIL_LOONGARCH_TIMER_H
>>>> +#define AVUTIL_LOONGARCH_TIMER_H
>>>> +
>>>> +#include <stdint.h>
>>>> +#include "config.h"
>>>> +
>>>> +#if HAVE_INLINE_ASM
>>>> +
>>>> +#define AV_READ_TIME read_time
>>>> +
>>>> +static inline uint64_t read_time(void)
>>>> +{
>>>> +
>>>> +#if ARCH_LOONGARCH64
>>>> +    uint64_t a, id = 0;
>>> Initial value is never used.
>>>
>>>> +    __asm__ volatile ( "rdtime.d  %0, %1" : "=r"(a), "=r"(id) :: "memory"
>>>> ); +    return a;
>>>> +#else
>>>> +    uint32_t a, id = 0;
>>>> +    __asm__ volatile ( "rdtimel.w  %0, %1" : "=r"(a), "=r"(id) :: "memory"
>>>> ); +    return (uint64_t)a;
>>>> +#endif
>>> Why do you clobber memory here?
>>>
>>>> +}
>>>> +
>>>> +#endif /* HAVE_INLINE_ASM */
>>>> +
>>>> +#endif /* AVUTIL_LOONGARCH_TIMER_H */
>>>> diff --git a/libavutil/timer.h b/libavutil/timer.h
>>>> index d3db5a27ef..861ba7e9d7 100644
>>>> --- a/libavutil/timer.h
>>>> +++ b/libavutil/timer.h
>>>> @@ -61,6 +61,8 @@
>>>>    #   include "riscv/timer.h"
>>>>    #elif ARCH_X86
>>>>    #   include "x86/timer.h"
>>>> +#elif ARCH_LOONGARCH
>>>> +#   include "loongarch/timer.h"
>>>>    #endif
>>>>
>>>>    #if !defined(AV_READ_TIME)
>>>>
>>>      Thanks for your advice.  As described in loongarch's instruction manual, the rdtime.d instruction is used as follows:
>>> rdtime.d rd, rj. The rj register stores the counter ID. In this application, the value of counter ID is equal to 0.
> You're setting a value, zero, to a variable `id`, that is then used as output operand. As far as the compiler is concerned, the value zero is never used and the initialisation can be elided. The value of register %1 is unspecified.
>
> If you meant for `id` to be an input operand, the constraints are incorrect.
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".


You are right! Thank you very much for your reminder. I will correct it.

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel at ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".