diff options
| author | Mans Rullgard <mans@mansr.com> | 2011-07-28 20:21:33 +0100 |
|---|---|---|
| committer | Mans Rullgard <mans@mansr.com> | 2011-07-28 22:25:21 +0100 |
| commit | f5f004bc5a40e5a5de62bd5e2818ab5de7f6bf21 (patch) | |
| tree | bcd3206dee29abd1c26103fccbecee6bdf765bbc /libavcodec/x86/h264_i386.h | |
| parent | 6c3257654801d525aa61e6fb46022a2a3b12c074 (diff) | |
| download | ffmpeg-f5f004bc5a40e5a5de62bd5e2818ab5de7f6bf21.tar.gz | |
x86: cabac: don't load/store context values in asm
Inspection of compiled code shows gcc handles these fine on its own.
Benchmarking also shows no measurable speed difference.
Removing the remaining cases in get_cabac_bypass_sign_x86() does
cause more substantial changes to the compiled code with uncertain
impact.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Diffstat (limited to 'libavcodec/x86/h264_i386.h')
| -rw-r--r-- | libavcodec/x86/h264_i386.h | 32 |
1 files changed, 7 insertions, 25 deletions
diff --git a/libavcodec/x86/h264_i386.h b/libavcodec/x86/h264_i386.h index 0151cd59a8..eb77002ce0 100644 --- a/libavcodec/x86/h264_i386.h +++ b/libavcodec/x86/h264_i386.h @@ -45,23 +45,18 @@ static int decode_significance_x86(CABACContext *c, int max_coeff, int minusindex= 4-(intptr_t)index; int bit; x86_reg coeff_count; - int low; - int range; __asm__ volatile( - "movl %a11(%6), %5 \n\t" - "movl %a12(%6), %3 \n\t" - "2: \n\t" BRANCHLESS_GET_CABAC("%4", "%6", "(%1)", "%3", - "%w3", "%5", "%k0", "%b0", "%a13") + "%w3", "%5", "%k0", "%b0", "%a11") "test $1, %4 \n\t" " jz 3f \n\t" "add %10, %1 \n\t" BRANCHLESS_GET_CABAC("%4", "%6", "(%1)", "%3", - "%w3", "%5", "%k0", "%b0", "%a13") + "%w3", "%5", "%k0", "%b0", "%a11") "sub %10, %1 \n\t" "mov %2, %0 \n\t" @@ -85,13 +80,9 @@ static int decode_significance_x86(CABACContext *c, int max_coeff, "4: \n\t" "add %9, %k0 \n\t" "shr $2, %k0 \n\t" - - "movl %5, %a11(%6) \n\t" - "movl %3, %a12(%6) \n\t" :"=&q"(coeff_count), "+r"(significant_coeff_ctx_base), "+m"(index), - "=&r"(low), "=&r"(bit), "=&r"(range) + "+&r"(c->low), "=&r"(bit), "+&r"(c->range) :"r"(c), "m"(minusstart), "m"(end), "m"(minusindex), "m"(last_off), - "i"(offsetof(CABACContext, range)), "i"(offsetof(CABACContext, low)), "i"(offsetof(CABACContext, bytestream)) : "%"REG_c, "memory" ); @@ -104,14 +95,9 @@ static int decode_significance_8x8_x86(CABACContext *c, int minusindex= 4-(intptr_t)index; int bit; x86_reg coeff_count; - int low; - int range; x86_reg last=0; x86_reg state; __asm__ volatile( - "movl %a12(%7), %5 \n\t" - "movl %a13(%7), %3 \n\t" - "mov %1, %6 \n\t" "2: \n\t" @@ -120,7 +106,7 @@ static int decode_significance_8x8_x86(CABACContext *c, "add %9, %6 \n\t" BRANCHLESS_GET_CABAC("%4", "%7", "(%6)", "%3", - "%w3", "%5", "%k0", "%b0", "%a14") + "%w3", "%5", "%k0", "%b0", "%a12") "mov %1, %k6 \n\t" "test $1, %4 \n\t" @@ -130,7 +116,7 @@ static int decode_significance_8x8_x86(CABACContext *c, "add %11, %6 \n\t" BRANCHLESS_GET_CABAC("%4", "%7", "(%6)", "%3", - "%w3", "%5", "%k0", "%b0", "%a14") + "%w3", "%5", "%k0", "%b0", "%a12") "mov %2, %0 \n\t" "mov %1, %k6 \n\t" @@ -151,13 +137,9 @@ static int decode_significance_8x8_x86(CABACContext *c, "4: \n\t" "addl %8, %k0 \n\t" "shr $2, %k0 \n\t" - - "movl %5, %a12(%7) \n\t" - "movl %3, %a13(%7) \n\t" - :"=&q"(coeff_count),"+m"(last), "+m"(index), "=&r"(low), "=&r"(bit), - "=&r"(range), "=&r"(state) + :"=&q"(coeff_count),"+m"(last), "+m"(index), "+&r"(c->low), "=&r"(bit), + "+&r"(c->range), "=&r"(state) :"r"(c), "m"(minusindex), "m"(significant_coeff_ctx_base), "m"(sig_off), "m"(last_coeff_ctx_base), - "i"(offsetof(CABACContext, range)), "i"(offsetof(CABACContext, low)), "i"(offsetof(CABACContext, bytestream)) : "%"REG_c, "memory" ); |