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

[Hao Chen]

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.
>     In addition, clobbering memory is really not a necessary 
> operation. I will remove it.