diff --git a/release/c/wuffs-unsupported-snapshot.c b/release/c/wuffs-unsupported-snapshot.c index 96d49a36..3bf33c4b 100644 --- a/release/c/wuffs-unsupported-snapshot.c +++ b/release/c/wuffs-unsupported-snapshot.c @@ -17620,7 +17620,7 @@ wuffs_adler32__hasher__up_arm_neon( #if defined(WUFFS_BASE__CPU_ARCH__X86_64) static wuffs_base__empty_struct -wuffs_adler32__hasher__up_sse42( +wuffs_adler32__hasher__up_x86_sse42( wuffs_adler32__hasher* self, wuffs_base__slice_u8 a_x); #endif // defined(WUFFS_BASE__CPU_ARCH__X86_64) @@ -17740,7 +17740,7 @@ wuffs_adler32__hasher__update_u32( wuffs_base__cpu_arch__have_arm_neon() ? &wuffs_adler32__hasher__up_arm_neon : #endif #if defined(WUFFS_BASE__CPU_ARCH__X86_64) - wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_adler32__hasher__up_sse42 : + wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_adler32__hasher__up_x86_sse42 : #endif self->private_impl.choosy_up); } @@ -17926,14 +17926,14 @@ wuffs_adler32__hasher__up_arm_neon( } #endif // defined(WUFFS_BASE__CPU_ARCH__ARM_NEON) -// -------- func adler32.hasher.up_sse42 +// -------- func adler32.hasher.up_x86_sse42 #if defined(WUFFS_BASE__CPU_ARCH__X86_64) #if defined(__GNUC__) __attribute__((target("sse4.2"))) #endif static wuffs_base__empty_struct -wuffs_adler32__hasher__up_sse42( +wuffs_adler32__hasher__up_x86_sse42( wuffs_adler32__hasher* self, wuffs_base__slice_u8 a_x) { uint32_t v_s1 = 0; @@ -31374,14 +31374,14 @@ wuffs_png__decoder__filter_4_distance_4_fallback( #if defined(WUFFS_BASE__CPU_ARCH__X86_64) static wuffs_base__empty_struct -wuffs_png__decoder__filter_1_distance_4_sse42( +wuffs_png__decoder__filter_1_distance_4_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr); #endif // defined(WUFFS_BASE__CPU_ARCH__X86_64) #if defined(WUFFS_BASE__CPU_ARCH__X86_64) static wuffs_base__empty_struct -wuffs_png__decoder__filter_3_distance_4_sse42( +wuffs_png__decoder__filter_3_distance_4_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr, wuffs_base__slice_u8 a_prev); @@ -31389,7 +31389,7 @@ wuffs_png__decoder__filter_3_distance_4_sse42( #if defined(WUFFS_BASE__CPU_ARCH__X86_64) static wuffs_base__empty_struct -wuffs_png__decoder__filter_4_distance_3_sse42( +wuffs_png__decoder__filter_4_distance_3_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr, wuffs_base__slice_u8 a_prev); @@ -31397,7 +31397,7 @@ wuffs_png__decoder__filter_4_distance_3_sse42( #if defined(WUFFS_BASE__CPU_ARCH__X86_64) static wuffs_base__empty_struct -wuffs_png__decoder__filter_4_distance_4_sse42( +wuffs_png__decoder__filter_4_distance_4_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr, wuffs_base__slice_u8 a_prev); @@ -32598,14 +32598,14 @@ wuffs_png__decoder__filter_4_distance_4_fallback( return wuffs_base__make_empty_struct(); } -// -------- func png.decoder.filter_1_distance_4_sse42 +// -------- func png.decoder.filter_1_distance_4_x86_sse42 #if defined(WUFFS_BASE__CPU_ARCH__X86_64) #if defined(__GNUC__) __attribute__((target("sse4.2"))) #endif static wuffs_base__empty_struct -wuffs_png__decoder__filter_1_distance_4_sse42( +wuffs_png__decoder__filter_1_distance_4_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr) { wuffs_base__slice_u8 v_c = {0}; @@ -32644,14 +32644,14 @@ wuffs_png__decoder__filter_1_distance_4_sse42( } #endif // defined(WUFFS_BASE__CPU_ARCH__X86_64) -// -------- func png.decoder.filter_3_distance_4_sse42 +// -------- func png.decoder.filter_3_distance_4_x86_sse42 #if defined(WUFFS_BASE__CPU_ARCH__X86_64) #if defined(__GNUC__) __attribute__((target("sse4.2"))) #endif static wuffs_base__empty_struct -wuffs_png__decoder__filter_3_distance_4_sse42( +wuffs_png__decoder__filter_3_distance_4_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr, wuffs_base__slice_u8 a_prev) { @@ -32749,14 +32749,14 @@ wuffs_png__decoder__filter_3_distance_4_sse42( } #endif // defined(WUFFS_BASE__CPU_ARCH__X86_64) -// -------- func png.decoder.filter_4_distance_3_sse42 +// -------- func png.decoder.filter_4_distance_3_x86_sse42 #if defined(WUFFS_BASE__CPU_ARCH__X86_64) #if defined(__GNUC__) __attribute__((target("sse4.2"))) #endif static wuffs_base__empty_struct -wuffs_png__decoder__filter_4_distance_3_sse42( +wuffs_png__decoder__filter_4_distance_3_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr, wuffs_base__slice_u8 a_prev) { @@ -32875,14 +32875,14 @@ wuffs_png__decoder__filter_4_distance_3_sse42( } #endif // defined(WUFFS_BASE__CPU_ARCH__X86_64) -// -------- func png.decoder.filter_4_distance_4_sse42 +// -------- func png.decoder.filter_4_distance_4_x86_sse42 #if defined(WUFFS_BASE__CPU_ARCH__X86_64) #if defined(__GNUC__) __attribute__((target("sse4.2"))) #endif static wuffs_base__empty_struct -wuffs_png__decoder__filter_4_distance_4_sse42( +wuffs_png__decoder__filter_4_distance_4_x86_sse42( wuffs_png__decoder* self, wuffs_base__slice_u8 a_curr, wuffs_base__slice_u8 a_prev) { @@ -33662,7 +33662,7 @@ wuffs_png__decoder__choose_filter_implementations( wuffs_base__cpu_arch__have_arm_neon() ? &wuffs_png__decoder__filter_4_distance_3_arm_neon : #endif #if defined(WUFFS_BASE__CPU_ARCH__X86_64) - wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_4_distance_3_sse42 : + wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_4_distance_3_x86_sse42 : #endif &wuffs_png__decoder__filter_4_distance_3_fallback); } else if (self->private_impl.f_filter_distance == 4) { @@ -33671,7 +33671,7 @@ wuffs_png__decoder__choose_filter_implementations( wuffs_base__cpu_arch__have_arm_neon() ? &wuffs_png__decoder__filter_1_distance_4_arm_neon : #endif #if defined(WUFFS_BASE__CPU_ARCH__X86_64) - wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_1_distance_4_sse42 : + wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_1_distance_4_x86_sse42 : #endif &wuffs_png__decoder__filter_1_distance_4_fallback); self->private_impl.choosy_filter_3 = ( @@ -33679,7 +33679,7 @@ wuffs_png__decoder__choose_filter_implementations( wuffs_base__cpu_arch__have_arm_neon() ? &wuffs_png__decoder__filter_3_distance_4_arm_neon : #endif #if defined(WUFFS_BASE__CPU_ARCH__X86_64) - wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_3_distance_4_sse42 : + wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_3_distance_4_x86_sse42 : #endif &wuffs_png__decoder__filter_3_distance_4_fallback); self->private_impl.choosy_filter_4 = ( @@ -33687,7 +33687,7 @@ wuffs_png__decoder__choose_filter_implementations( wuffs_base__cpu_arch__have_arm_neon() ? &wuffs_png__decoder__filter_4_distance_4_arm_neon : #endif #if defined(WUFFS_BASE__CPU_ARCH__X86_64) - wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_4_distance_4_sse42 : + wuffs_base__cpu_arch__have_x86_sse42() ? &wuffs_png__decoder__filter_4_distance_4_x86_sse42 : #endif &wuffs_png__decoder__filter_4_distance_4_fallback); } diff --git a/std/adler32/common_adler32.wuffs b/std/adler32/common_adler32.wuffs index 6a5f2deb..eb2a93b6 100644 --- a/std/adler32/common_adler32.wuffs +++ b/std/adler32/common_adler32.wuffs @@ -27,7 +27,7 @@ pub func hasher.update_u32!(x: slice base.u8) base.u32 { this.state = 1 choose up = [ up_arm_neon, - up_sse42] + up_x86_sse42] } this.up!(x: args.x) return this.state diff --git a/std/adler32/common_up_x86_sse42.wuffs b/std/adler32/common_up_x86_sse42.wuffs index a9f44c18..06871416 100644 --- a/std/adler32/common_up_x86_sse42.wuffs +++ b/std/adler32/common_up_x86_sse42.wuffs @@ -12,7 +12,7 @@ // See the License for the specific language governing permissions and // limitations under the License. -pri func hasher.up_sse42!(x: slice base.u8), +pri func hasher.up_x86_sse42!(x: slice base.u8), choose cpu_arch >= x86_sse42, { // These variables are the same as the non-SIMD version. diff --git a/std/png/decode_filter_x86_sse42.wuffs b/std/png/decode_filter_x86_sse42.wuffs index 7c19f8d6..7068e80a 100644 --- a/std/png/decode_filter_x86_sse42.wuffs +++ b/std/png/decode_filter_x86_sse42.wuffs @@ -19,7 +19,7 @@ // This (filter = 1, distance = 3) implementation doesn't actually bench faster // than the non-SIMD one. // -// pri func decoder.filter_1_distance_3_sse42!(curr: slice base.u8), +// pri func decoder.filter_1_distance_3_x86_sse42!(curr: slice base.u8), // choose cpu_arch >= x86_sse42, // { // var c : slice base.u8 @@ -42,7 +42,7 @@ // https://github.com/google/wuffs/commit/1660f9268621ed4415b3b363f0a0e1026d4aa83d // "Have std/png filter_1_distance_? use more SIMD" for a pessimizing example. -pri func decoder.filter_1_distance_4_sse42!(curr: slice base.u8), +pri func decoder.filter_1_distance_4_x86_sse42!(curr: slice base.u8), choose cpu_arch >= x86_sse42, { var c : slice base.u8 @@ -61,16 +61,16 @@ pri func decoder.filter_1_distance_4_sse42!(curr: slice base.u8), // Filter 3: Average. -// Similar to filter_1_distance_3_sse42, the SIMD implementation for (filter = -// 3, distance = 3) doesn't actually bench faster than the non-SIMD one. +// Similar to filter_1_distance_3_x86_sse42, the SIMD implementation for +// (filter = 3, distance = 3) doesn't actually bench faster than non-SIMD. // -// pri func decoder.filter_3_distance_3_sse42!(curr: slice base.u8, prev: slice base.u8), +// pri func decoder.filter_3_distance_3_x86_sse42!(curr: slice base.u8, prev: slice base.u8), // choose cpu_arch >= x86_sse42, // { // etc // } -pri func decoder.filter_3_distance_4_sse42!(curr: slice base.u8, prev: slice base.u8), +pri func decoder.filter_3_distance_4_x86_sse42!(curr: slice base.u8, prev: slice base.u8), choose cpu_arch >= x86_sse42, { var c : slice base.u8 @@ -127,13 +127,13 @@ pri func decoder.filter_3_distance_4_sse42!(curr: slice base.u8, prev: slice bas // Filter 4: Paeth. -pri func decoder.filter_4_distance_3_sse42!(curr: slice base.u8, prev: slice base.u8), +pri func decoder.filter_4_distance_3_x86_sse42!(curr: slice base.u8, prev: slice base.u8), choose cpu_arch >= x86_sse42, { - // See the comments in filter_4_distance_4_sse42 for an explanation of how - // this works. That function's single loop is copied twice here, once with - // "length: 4" and once with "length: 3". It's generally faster to load 4 - // bytes at a time instead of 3. + // See the comments in filter_4_distance_4_x86_sse42 for an explanation of + // how this works. That function's single loop is copied twice here, once + // with "length: 4" and once with "length: 3". It's generally faster to + // load 4 bytes at a time instead of 3. // // Differences between that function and this one are marked with a ยง. @@ -205,7 +205,7 @@ pri func decoder.filter_4_distance_3_sse42!(curr: slice base.u8, prev: slice bas } } -pri func decoder.filter_4_distance_4_sse42!(curr: slice base.u8, prev: slice base.u8), +pri func decoder.filter_4_distance_4_x86_sse42!(curr: slice base.u8, prev: slice base.u8), choose cpu_arch >= x86_sse42, { var c : slice base.u8 diff --git a/std/png/decode_png.wuffs b/std/png/decode_png.wuffs index cc6b8a43..97cae7d9 100644 --- a/std/png/decode_png.wuffs +++ b/std/png/decode_png.wuffs @@ -377,20 +377,20 @@ pri func decoder.choose_filter_implementations!() { choose filter_3 = [filter_3_distance_3_fallback] choose filter_4 = [ filter_4_distance_3_arm_neon, - filter_4_distance_3_sse42, + filter_4_distance_3_x86_sse42, filter_4_distance_3_fallback] } else if this.filter_distance == 4 { choose filter_1 = [ filter_1_distance_4_arm_neon, - filter_1_distance_4_sse42, + filter_1_distance_4_x86_sse42, filter_1_distance_4_fallback] choose filter_3 = [ filter_3_distance_4_arm_neon, - filter_3_distance_4_sse42, + filter_3_distance_4_x86_sse42, filter_3_distance_4_fallback] choose filter_4 = [ filter_4_distance_4_arm_neon, - filter_4_distance_4_sse42, + filter_4_distance_4_x86_sse42, filter_4_distance_4_fallback] } }