virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2025-08-15 14:13:16 +02:00
Code Issues Releases Activity

lavu/tx: invert permutation lookups

Browse Source 
out[lut[i]] = in[i] lookups were 4.04 times(!) slower than
out[i] = in[lut[i]] lookups for an out-of-place FFT of length 4096.

The permutes remain unchanged for anything but out-of-place monolithic
FFT, as those benefit quite a lot from the current order (it means
there's only 1 lookup necessary to add to an offset, rather than
a full gather).

The code was based around non-power-of-two FFTs, so this wasn't
benchmarked early on.
This commit is contained in:
Lynne
2021-02-27 04:11:04 +01:00
parent 9ddaf0c9f0
commit 8e94b7cff0
3 changed files with 8 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
libavutil/tx.c
View File

@@ -91,7 +91,7 @@ int ff_tx_gen_compound_mapping(AVTXContext *s)
return 0; return 0;
} }
int ff_tx_gen_ptwo_revtab(AVTXContext *s) int ff_tx_gen_ptwo_revtab(AVTXContext *s, int invert_lookup)
{ {
const int m = s->m, inv = s->inv; const int m = s->m, inv = s->inv;
@@ -101,7 +101,10 @@ int ff_tx_gen_ptwo_revtab(AVTXContext *s)
/* Default */ /* Default */
for (int i = 0; i < m; i++) { for (int i = 0; i < m; i++) {
int k = -split_radix_permutation(i, m, inv) & (m - 1); int k = -split_radix_permutation(i, m, inv) & (m - 1);
s->revtab[k] = i; if (invert_lookup)
s->revtab[i] = k;
else
s->revtab[k] = i;
} }
return 0; return 0;

2
libavutil/tx_priv.h
View File

@@ -123,7 +123,7 @@ struct AVTXContext {
/* Shared functions */ /* Shared functions */
int ff_tx_type_is_mdct(enum AVTXType type); int ff_tx_type_is_mdct(enum AVTXType type);
int ff_tx_gen_compound_mapping(AVTXContext *s); int ff_tx_gen_compound_mapping(AVTXContext *s);
int ff_tx_gen_ptwo_revtab(AVTXContext *s); int ff_tx_gen_ptwo_revtab(AVTXContext *s, int invert_lookup);
int ff_tx_gen_ptwo_inplace_revtab_idx(AVTXContext *s); int ff_tx_gen_ptwo_inplace_revtab_idx(AVTXContext *s);
/* Also used by SIMD init */ /* Also used by SIMD init */

4
libavutil/tx_template.c
View File

@@ -410,7 +410,7 @@ static void monolithic_fft(AVTXContext *s, void *_out, void *_in,
} while ((src = *inplace_idx++)); } while ((src = *inplace_idx++));
} else { } else {
for (int i = 0; i < m; i++) for (int i = 0; i < m; i++)
out[s->revtab[i]] = in[i]; out[i] = in[s->revtab[i]];
} }
fft_dispatch[mb](out); fft_dispatch[mb](out);
@@ -738,7 +738,7 @@ int TX_NAME(ff_tx_init_mdct_fft)(AVTXContext *s, av_tx_fn *tx,
if (n != 1) if (n != 1)
init_cos_tabs(0); init_cos_tabs(0);
if (m != 1) { if (m != 1) {
if ((err = ff_tx_gen_ptwo_revtab(s))) if ((err = ff_tx_gen_ptwo_revtab(s, n == 1 && !(flags & AV_TX_INPLACE))))
return err; return err;
if (flags & AV_TX_INPLACE) { if (flags & AV_TX_INPLACE) {
if (is_mdct) /* In-place MDCTs are not supported yet */ if (is_mdct) /* In-place MDCTs are not supported yet */