From: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
To: ffmpeg-devel@ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH 3/4] checkasm/h263dsp: test dct_unquantize_{intra, inter}
Date: Sat, 6 Jul 2024 21:10:02 +0200
Message-ID: <AS8P250MB07449A44AFE65C39B4A187F68FD82@AS8P250MB0744.EURP250.PROD.OUTLOOK.COM> (raw)
In-Reply-To: <20240701191328.466433-3-remi@remlab.net>
Rémi Denis-Courmont:
> ---
> tests/checkasm/h263dsp.c | 57 +++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 56 insertions(+), 1 deletion(-)
>
> diff --git a/tests/checkasm/h263dsp.c b/tests/checkasm/h263dsp.c
> index 2d0957a90b..26020211dc 100644
> --- a/tests/checkasm/h263dsp.c
> +++ b/tests/checkasm/h263dsp.c
> @@ -18,13 +18,65 @@
> * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> */
>
> +#include <stdbool.h>
> #include <string.h>
>
> #include "checkasm.h"
>
> -#include "libavcodec/h263dsp.h"
> +#include "libavutil/avassert.h"
> #include "libavutil/mem.h"
> #include "libavutil/mem_internal.h"
> +#include "libavcodec/h263dsp.h"
> +#include "libavcodec/mpegvideodata.h"
> +
> +static uint_fast8_t mpeg_qscale_rnd(void)
> +{
> + int n = rnd(), q = (n >> 1) & 31;
> +
> + if (n & 1)
> + return ff_mpeg2_non_linear_qscale[q];
> + else
> + return q << 1;
This branch reminds me of the q_scale_type check from the
mpeg2-unquantize functions. Why is it here?
(For H.263, qscale is simply a value in 1..31. This code will test
values outside of this range.)
> +}
> +
> +typedef void (*unquantizer)(int16_t *, ptrdiff_t, int, int);
> +
> +static void check_dct_unquantize(unquantizer func, bool inter)
> +{
> + const char *name = inter ? "inter" : "intra";
> +#define LEN 64
> + LOCAL_ALIGNED_16(int16_t, block0, [LEN]);
> + LOCAL_ALIGNED_16(int16_t, block1, [LEN]);
> + size_t len = rnd() % LEN;
> + const int qscale = mpeg_qscale_rnd();
> + const int qmul = qscale << 1;
> + const int qadd = (rnd() & 1) ? (qscale - 1) | 1 : 0;
> +
> + declare_func(void, int16_t *, ptrdiff_t, int, int);
> +
> + for (size_t i = 0; i < LEN; i++) {
> + int r = rnd();
> +
> + block1[i] = block0[i] = (r & 1) ? (r >> 1) : 0;
This will potentially set all elements to some nonzero value. In
reality, all elements not processed in the (current) C version are zero;
in other words, one could add
for (; i <= 63; i++)
av_assert0(!block[i]);
at the end of dct_unquantize_h263_(inter|intra)_c. This is what makes it
possible to switch to a do-while loop. It would also make it possible to
unroll the loop/omit special handling for the tail when vectorizing, so
this test is a hindrance to potential optimizations.
> + }
> +
> + if (check_func(func, "h263dsp.dct_unquantize_%s", name)) {
> + call_ref(block0, 0, qmul, qadd);
> + call_new(block1, 0, qmul, qadd);
> +
> + if (memcmp(block0, block1, LEN * sizeof (int16_t)))
> + fail();
> +
> + av_assert0(len < LEN);
> + call_ref(block0, len, qmul, qadd);
> + call_new(block1, len, qmul, qadd);
> +
> + if (memcmp(block0, block1, LEN * sizeof (int16_t)))
> + fail();
> +
> + bench_new(block1, LEN, qmul, qadd);
> + }
> +}
>
> typedef void (*filter)(uint8_t *src, int stride, int qscale);
>
> @@ -56,6 +108,9 @@ void checkasm_check_h263dsp(void)
> H263DSPContext ctx;
>
> ff_h263dsp_init(&ctx);
> + check_dct_unquantize(ctx.h263_dct_unquantize_intra, false);
> + check_dct_unquantize(ctx.h263_dct_unquantize_inter, true);
> + report("dct_unquantize");
> check_loop_filter('h', ctx.h263_h_loop_filter);
> check_loop_filter('v', ctx.h263_v_loop_filter);
> report("loop_filter");
_______________________________________________
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".
next prev parent reply other threads:[~2024-07-06 19:10 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-07-01 19:13 [FFmpeg-devel] [PATCH 1/4] lavc/h263dsp: add DCT dequantisation functions Rémi Denis-Courmont
2024-07-01 19:13 ` [FFmpeg-devel] [PATCH 2/4] lavc/mpegvideo: use H263DSP dequant function Rémi Denis-Courmont
2024-07-06 15:23 ` Andreas Rheinhardt
2024-07-06 16:10 ` Rémi Denis-Courmont
2024-07-06 16:20 ` Andreas Rheinhardt
2024-07-06 16:47 ` Rémi Denis-Courmont
2024-07-06 18:27 ` Andreas Rheinhardt
2024-07-01 19:13 ` [FFmpeg-devel] [PATCH 3/4] checkasm/h263dsp: test dct_unquantize_{intra, inter} Rémi Denis-Courmont
2024-07-06 19:10 ` Andreas Rheinhardt [this message]
2024-07-01 19:13 ` [FFmpeg-devel] [PATCH 4/4] lavc/h263dsp: R-V V " Rémi Denis-Courmont
2024-07-06 10:54 ` [FFmpeg-devel] [PATCH 1/4] lavc/h263dsp: add DCT dequantisation functions Rémi Denis-Courmont
2024-07-06 15:17 ` Andreas Rheinhardt
2024-07-06 16:10 ` Rémi Denis-Courmont
2024-07-06 16:19 ` Andreas Rheinhardt
2024-07-06 17:28 ` Rémi Denis-Courmont
2024-07-06 18:23 ` Andreas Rheinhardt
-- strict thread matches above, loose matches on Subject: below --
2024-06-12 4:47 [FFmpeg-devel] [PATCHv5 " Rémi Denis-Courmont
2024-06-12 4:47 ` [FFmpeg-devel] [PATCH 3/4] checkasm/h263dsp: test dct_unquantize_{intra, inter} Rémi Denis-Courmont
2024-06-12 18:39 ` James Almer
2024-06-12 19:15 ` Rémi Denis-Courmont
2024-06-12 19:39 ` James Almer
2024-06-12 19:52 ` Rémi Denis-Courmont
2024-06-09 16:23 [FFmpeg-devel] [PATCHv3 1/4] lavc/h263dsp: add DCT dequantisation functions Rémi Denis-Courmont
2024-06-09 16:23 ` [FFmpeg-devel] [PATCH 3/4] checkasm/h263dsp: test dct_unquantize_{intra, inter} 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=AS8P250MB07449A44AFE65C39B4A187F68FD82@AS8P250MB0744.EURP250.PROD.OUTLOOK.COM \
--to=andreas.rheinhardt@outlook.com \
--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