Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
From: "Rémi Denis-Courmont" <remi@remlab.net>
To: ffmpeg-devel@ffmpeg.org
Subject: [FFmpeg-devel] [PATCH 5/6] checkasm: allow run-time fallback to AV_READ_TIME
Date: Wed, 19 Jul 2023 22:55:39 +0300
Message-ID: <20230719195540.46961-5-remi@remlab.net> (raw)
In-Reply-To: <2113654.OBFZWjSADL@basile.remlab.net>

On Linux RISC-V, depending on kernel version and configuration, the
cycle counters may be available:
- via Linux perf (as on Arm),
- directly with RDCYCLE (which is considered legacy),
- neither (if admistratively disabled), or
- or both (in legacy compatibility mode).

This allows to try Linux perf, with a fallback to RDCYCLE. Without
this, benchmarking would breakdown in upcoming Linux RISC-V kernel
releases. This also paves the way for falling back from the cycle
counter on Arm or RISC-V, whence they too are disabled.
---
 tests/checkasm/checkasm.c | 32 +++++++++++++++++++-------------
 1 file changed, 19 insertions(+), 13 deletions(-)

diff --git a/tests/checkasm/checkasm.c b/tests/checkasm/checkasm.c
index fad660a924..3c8aa8631e 100644
--- a/tests/checkasm/checkasm.c
+++ b/tests/checkasm/checkasm.c
@@ -714,37 +714,43 @@ static int bench_init_linux(void)
     state.close = checkasm_bench_linux_perf_close;
     return 0;
 }
-#elif CONFIG_MACOS_KPERF
+#endif
+#if CONFIG_MACOS_KPERF
 static int bench_init_kperf(void)
 {
     ff_kperf_init();
     state.start = state.stop = ff_kperf_cycles;
     return 0;
 }
-#else
+#endif
+#ifdef AV_READ_TIME
 static int bench_init_ffmpeg(void)
 {
-#ifdef AV_READ_TIME
     printf("benchmarking with native FFmpeg timers\n");
+    state.start = state.stop = AV_READ_TIME;
     return 0;
-#else
-    fprintf(stderr, "checkasm: --bench is not supported on your system\n");
-    return -1;
-#endif
 }
 #endif
 
 static int bench_init(void)
 {
+    int ret = -1;
 #if CONFIG_LINUX_PERF
-    int ret = bench_init_linux();
-#elif CONFIG_MACOS_KPERF
-    int ret = bench_init_kperf();
-#else
-    int ret = bench_init_ffmpeg();
+    if (ret < 0)
+        ret = bench_init_linux();
+#endif
+#if CONFIG_MACOS_KPERF
+    if (ret < 0)
+        ret = bench_init_kperf();
 #endif
+#ifdef AV_READ_TIME
     if (ret < 0)
-        return ret;
+        ret = bench_init_ffmpeg();
+#endif
+    if (ret < 0) {
+        fputs("checkasm: --bench is not supported on your system\n", stderr);
+        return -1;
+    }
 
     state.nop_time = measure_nop_time();
     printf("nop: %d.%d\n", state.nop_time/10, state.nop_time%10);
-- 
2.40.1

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

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

  parent reply	other threads:[~2023-07-19 19:56 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-07-19 19:55 [FFmpeg-devel] [PATCHv3 0/6] RISC-V Linux perf run-time enablement Rémi Denis-Courmont
2023-07-19 19:55 ` [FFmpeg-devel] [PATCH 1/6] checkasm: make perf macros functional Rémi Denis-Courmont
2023-07-19 19:55 ` [FFmpeg-devel] [PATCH 2/6] checkasm: use pointers for start/stop functions Rémi Denis-Courmont
2023-07-19 20:34   ` Lynne
2023-07-19 19:55 ` [FFmpeg-devel] [PATCH 3/6] checkasm: remove unused variables Rémi Denis-Courmont
2023-07-19 19:55 ` [FFmpeg-devel] [PATCH 4/6] checkasm: make bench clean-up also a function pointer Rémi Denis-Courmont
2023-07-19 19:55 ` Rémi Denis-Courmont [this message]
2023-07-19 19:55 ` [FFmpeg-devel] [PATCH 6/6] configure: enable Linux perf on RISC-V by default Rémi Denis-Courmont

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230719195540.46961-5-remi@remlab.net \
    --to=remi@remlab.net \
    --cc=ffmpeg-devel@ffmpeg.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
		ffmpegdev@gitmailbox.com
	public-inbox-index ffmpegdev

Example config snippet for mirrors.


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git