std/jpeg: implement Inverse DCT

Updates #42
This commit is contained in:
Nigel Tao committed 2023-05-08 21:51:47 +10:00
1 parent cac4df3bde
commit 8cfe304f3c
7 files changed
+2662

No files matched your search

+76
View File
@@ -30,3 +30,79 @@ pri const UNZIG : roarray[80] base.u8[..= 63] = [
0x3F, 0x3F, 0x3F, 0x3F, 0x3F, 0x3F, 0x3F, 0x3F,
0x3F, 0x3F, 0x3F, 0x3F, 0x3F, 0x3F, 0x3F, 0x3F,
]
// BIAS_AND_CLAMP maps 10-bit signed values (0 centered, in the range -512 ..
// +512) to 8-bit unsigned values (128 centered, in the range 0 .. 256).
pri const BIAS_AND_CLAMP : roarray[1024] base.u8 = [
0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F,
0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D, 0x9E, 0x9F,
0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF,
0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF,
0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7, 0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7, 0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F,
0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x3B, 0x3C, 0x3D, 0x3E, 0x3F,
0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F,
0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x5D, 0x5E, 0x5F,
0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F,
0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F,
]
+1230
View File
@@ -0,0 +1,1230 @@
// Copyright 2023 The Wuffs Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b: base.u32[..= 9], q: base.u32[..= 3]) {
// This method implements the same algorithm as libjpeg-turbo's jidctint.c.
var bq0 : base.u32
var bq2 : base.u32
var bq4 : base.u32
var bq6 : base.u32
var ca : base.u32
var cb2 : base.u32
var cb6 : base.u32
var ccp : base.u32
var ccm : base.u32
var cd0 : base.u32
var cd1 : base.u32
var cd2 : base.u32
var cd3 : base.u32
var bq1 : base.u32
var bq3 : base.u32
var bq5 : base.u32
var bq7 : base.u32
var ci51 : base.u32
var ci53 : base.u32
var ci71 : base.u32
var ci73 : base.u32
var cj : base.u32
var ck1 : base.u32
var ck3 : base.u32
var ck5 : base.u32
var ck7 : base.u32
var cl51 : base.u32
var cl73 : base.u32
var in0 : base.u32
var in2 : base.u32
var in4 : base.u32
var in6 : base.u32
var ra : base.u32
var rb2 : base.u32
var rb6 : base.u32
var rcp : base.u32
var rcm : base.u32
var rd0 : base.u32
var rd1 : base.u32
var rd2 : base.u32
var rd3 : base.u32
var in1 : base.u32
var in3 : base.u32
var in5 : base.u32
var in7 : base.u32
var ri51 : base.u32
var ri53 : base.u32
var ri71 : base.u32
var ri73 : base.u32
var rj : base.u32
var rk1 : base.u32
var rk3 : base.u32
var rk5 : base.u32
var rk7 : base.u32
var rl51 : base.u32
var rl73 : base.u32
var intermediate : array[64] base.u32
if 8 > args.dst_stride {
return nothing
}
// -------- BEGIN generated by script/print-jpeg-idct-code.go
// p0_298631336 = 0x098E = 2446
// p0_390180644 = 0x0C7C = 3196
// p0_541196100 = 0x1151 = 4433
// p0_765366865 = 0x187E = 6270
// p0_899976223 = 0x1CCD = 7373
// p1_175875602 = 0x25A1 = 9633
// p1_501321110 = 0x300B = 12299
// p1_847759065 = 0x3B21 = 15137
// p1_961570560 = 0x3EC5 = 16069
// p2_053119869 = 0x41B3 = 16819
// p2_562915447 = 0x5203 = 20995
// p3_072711026 = 0x6254 = 25172
//
// m0_390180644 = 0xFFFFF384 = 4294964100
// m0_899976223 = 0xFFFFE333 = 4294959923
// m1_961570560 = 0xFFFFC13B = 4294951227
// m2_562915447 = 0xFFFFADFD = 4294946301
// ==== First pass, column 0.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x10]) ~mod* (this.quant_tables[args.q][0x10] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x30]) ~mod* (this.quant_tables[args.q][0x30] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x00]) ~mod* (this.quant_tables[args.q][0x00] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x20]) ~mod* (this.quant_tables[args.q][0x20] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x08]) ~mod* (this.quant_tables[args.q][0x08] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x18]) ~mod* (this.quant_tables[args.q][0x18] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x28]) ~mod* (this.quant_tables[args.q][0x28] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x38]) ~mod* (this.quant_tables[args.q][0x38] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x00] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x38] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x08] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x30] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x10] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x28] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x18] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x20] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 1.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x11]) ~mod* (this.quant_tables[args.q][0x11] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x31]) ~mod* (this.quant_tables[args.q][0x31] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x01]) ~mod* (this.quant_tables[args.q][0x01] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x21]) ~mod* (this.quant_tables[args.q][0x21] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x09]) ~mod* (this.quant_tables[args.q][0x09] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x19]) ~mod* (this.quant_tables[args.q][0x19] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x29]) ~mod* (this.quant_tables[args.q][0x29] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x39]) ~mod* (this.quant_tables[args.q][0x39] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x01] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x39] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x09] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x31] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x11] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x29] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x19] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x21] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 2.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x12]) ~mod* (this.quant_tables[args.q][0x12] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x32]) ~mod* (this.quant_tables[args.q][0x32] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x02]) ~mod* (this.quant_tables[args.q][0x02] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x22]) ~mod* (this.quant_tables[args.q][0x22] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0A]) ~mod* (this.quant_tables[args.q][0x0A] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1A]) ~mod* (this.quant_tables[args.q][0x1A] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2A]) ~mod* (this.quant_tables[args.q][0x2A] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3A]) ~mod* (this.quant_tables[args.q][0x3A] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x02] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x3A] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x0A] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x32] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x12] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x2A] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x1A] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x22] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 3.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x13]) ~mod* (this.quant_tables[args.q][0x13] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x33]) ~mod* (this.quant_tables[args.q][0x33] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x03]) ~mod* (this.quant_tables[args.q][0x03] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x23]) ~mod* (this.quant_tables[args.q][0x23] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0B]) ~mod* (this.quant_tables[args.q][0x0B] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1B]) ~mod* (this.quant_tables[args.q][0x1B] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2B]) ~mod* (this.quant_tables[args.q][0x2B] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3B]) ~mod* (this.quant_tables[args.q][0x3B] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x03] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x3B] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x0B] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x33] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x13] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x2B] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x1B] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x23] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 4.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x14]) ~mod* (this.quant_tables[args.q][0x14] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x34]) ~mod* (this.quant_tables[args.q][0x34] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x04]) ~mod* (this.quant_tables[args.q][0x04] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x24]) ~mod* (this.quant_tables[args.q][0x24] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0C]) ~mod* (this.quant_tables[args.q][0x0C] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1C]) ~mod* (this.quant_tables[args.q][0x1C] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2C]) ~mod* (this.quant_tables[args.q][0x2C] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3C]) ~mod* (this.quant_tables[args.q][0x3C] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x04] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x3C] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x0C] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x34] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x14] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x2C] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x1C] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x24] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 5.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x15]) ~mod* (this.quant_tables[args.q][0x15] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x35]) ~mod* (this.quant_tables[args.q][0x35] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x05]) ~mod* (this.quant_tables[args.q][0x05] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x25]) ~mod* (this.quant_tables[args.q][0x25] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0D]) ~mod* (this.quant_tables[args.q][0x0D] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1D]) ~mod* (this.quant_tables[args.q][0x1D] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2D]) ~mod* (this.quant_tables[args.q][0x2D] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3D]) ~mod* (this.quant_tables[args.q][0x3D] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x05] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x3D] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x0D] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x35] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x15] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x2D] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x1D] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x25] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 6.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x16]) ~mod* (this.quant_tables[args.q][0x16] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x36]) ~mod* (this.quant_tables[args.q][0x36] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x06]) ~mod* (this.quant_tables[args.q][0x06] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x26]) ~mod* (this.quant_tables[args.q][0x26] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0E]) ~mod* (this.quant_tables[args.q][0x0E] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1E]) ~mod* (this.quant_tables[args.q][0x1E] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2E]) ~mod* (this.quant_tables[args.q][0x2E] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3E]) ~mod* (this.quant_tables[args.q][0x3E] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x06] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x3E] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x0E] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x36] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x16] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x2E] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x1E] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x26] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== First pass, column 7.
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x17]) ~mod* (this.quant_tables[args.q][0x17] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x37]) ~mod* (this.quant_tables[args.q][0x37] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x07]) ~mod* (this.quant_tables[args.q][0x07] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x27]) ~mod* (this.quant_tables[args.q][0x27] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
cd0 = ccp ~mod+ cb2
cd1 = ccm ~mod+ cb6
cd2 = ccm ~mod- cb6
cd3 = ccp ~mod- cb2
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0F]) ~mod* (this.quant_tables[args.q][0x0F] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1F]) ~mod* (this.quant_tables[args.q][0x1F] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2F]) ~mod* (this.quant_tables[args.q][0x2F] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3F]) ~mod* (this.quant_tables[args.q][0x3F] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
ci71 = bq7 ~mod+ bq1
ci73 = bq7 ~mod+ bq3
cj = (ci73 ~mod+ ci51) ~mod* 0x25A1
ck1 = bq1 ~mod* 0x300B
ck3 = bq3 ~mod* 0x6254
ck5 = bq5 ~mod* 0x41B3
ck7 = bq7 ~mod* 0x098E
ci51 ~mod*= 0xFFFF_F384
ci53 ~mod*= 0xFFFF_ADFD
ci71 ~mod*= 0xFFFF_E333
ci73 ~mod*= 0xFFFF_C13B
cl51 = ci51 ~mod+ cj
cl73 = ci73 ~mod+ cj
ck1 ~mod+= ci71 ~mod+ cl51
ck3 ~mod+= ci53 ~mod+ cl73
ck5 ~mod+= ci53 ~mod+ cl51
ck7 ~mod+= ci71 ~mod+ cl73
// Combine rows.
intermediate[0x07] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod+ ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x3F] = this.util.sign_extend_rshift_u32(a: (cd0 ~mod- ck1) ~mod+ (1 << 10), n: 11)
intermediate[0x0F] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod+ ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x37] = this.util.sign_extend_rshift_u32(a: (cd1 ~mod- ck3) ~mod+ (1 << 10), n: 11)
intermediate[0x17] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod+ ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x2F] = this.util.sign_extend_rshift_u32(a: (cd2 ~mod- ck5) ~mod+ (1 << 10), n: 11)
intermediate[0x1F] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod+ ck7) ~mod+ (1 << 10), n: 11)
intermediate[0x27] = this.util.sign_extend_rshift_u32(a: (cd3 ~mod- ck7) ~mod+ (1 << 10), n: 11)
// ==== Second pass, row 0.
// Even columns.
in2 = intermediate[0x02]
in6 = intermediate[0x06]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x00]
in4 = intermediate[0x04]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x01]
in3 = intermediate[0x03]
in5 = intermediate[0x05]
in7 = intermediate[0x07]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 1.
// Even columns.
in2 = intermediate[0x0A]
in6 = intermediate[0x0E]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x08]
in4 = intermediate[0x0C]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x09]
in3 = intermediate[0x0B]
in5 = intermediate[0x0D]
in7 = intermediate[0x0F]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 2.
// Even columns.
in2 = intermediate[0x12]
in6 = intermediate[0x16]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x10]
in4 = intermediate[0x14]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x11]
in3 = intermediate[0x13]
in5 = intermediate[0x15]
in7 = intermediate[0x17]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 3.
// Even columns.
in2 = intermediate[0x1A]
in6 = intermediate[0x1E]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x18]
in4 = intermediate[0x1C]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x19]
in3 = intermediate[0x1B]
in5 = intermediate[0x1D]
in7 = intermediate[0x1F]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 4.
// Even columns.
in2 = intermediate[0x22]
in6 = intermediate[0x26]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x20]
in4 = intermediate[0x24]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x21]
in3 = intermediate[0x23]
in5 = intermediate[0x25]
in7 = intermediate[0x27]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 5.
// Even columns.
in2 = intermediate[0x2A]
in6 = intermediate[0x2E]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x28]
in4 = intermediate[0x2C]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x29]
in3 = intermediate[0x2B]
in5 = intermediate[0x2D]
in7 = intermediate[0x2F]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 6.
// Even columns.
in2 = intermediate[0x32]
in6 = intermediate[0x36]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x30]
in4 = intermediate[0x34]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x31]
in3 = intermediate[0x33]
in5 = intermediate[0x35]
in7 = intermediate[0x37]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if args.dst_stride > args.dst_buffer.length() {
return nothing
}
assert 8 <= args.dst_buffer.length() via "a <= b: a <= c; c <= b"(c: args.dst_stride)
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer = args.dst_buffer[args.dst_stride ..]
// ==== Second pass, row 7.
// Even columns.
in2 = intermediate[0x3A]
in6 = intermediate[0x3E]
ra = (in2 ~mod+ in6) ~mod* 0x1151
rb2 = ra ~mod+ (in2 ~mod* 0x187E)
rb6 = ra ~mod- (in6 ~mod* 0x3B21)
in0 = intermediate[0x38]
in4 = intermediate[0x3C]
rcp = (in0 ~mod+ in4) ~mod<< 13
rcm = (in0 ~mod- in4) ~mod<< 13
rd0 = rcp ~mod+ rb2
rd1 = rcm ~mod+ rb6
rd2 = rcm ~mod- rb6
rd3 = rcp ~mod- rb2
// Odd columns.
in1 = intermediate[0x39]
in3 = intermediate[0x3B]
in5 = intermediate[0x3D]
in7 = intermediate[0x3F]
ri51 = in5 ~mod+ in1
ri53 = in5 ~mod+ in3
ri71 = in7 ~mod+ in1
ri73 = in7 ~mod+ in3
rj = (ri73 ~mod+ ri51) ~mod* 0x25A1
rk1 = in1 ~mod* 0x300B
rk3 = in3 ~mod* 0x6254
rk5 = in5 ~mod* 0x41B3
rk7 = in7 ~mod* 0x098E
ri51 ~mod*= 0xFFFF_F384
ri53 ~mod*= 0xFFFF_ADFD
ri71 ~mod*= 0xFFFF_E333
ri73 ~mod*= 0xFFFF_C13B
rl51 = ri51 ~mod+ rj
rl73 = ri73 ~mod+ rj
rk1 ~mod+= ri71 ~mod+ rl51
rk3 ~mod+= ri53 ~mod+ rl73
rk5 ~mod+= ri53 ~mod+ rl51
rk7 ~mod+= ri71 ~mod+ rl73
// Combine columns.
if 8 > args.dst_buffer.length() {
return nothing
}
args.dst_buffer[0] = BIAS_AND_CLAMP[(((rd0 ~mod+ rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[7] = BIAS_AND_CLAMP[(((rd0 ~mod- rk1) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[1] = BIAS_AND_CLAMP[(((rd1 ~mod+ rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[6] = BIAS_AND_CLAMP[(((rd1 ~mod- rk3) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[2] = BIAS_AND_CLAMP[(((rd2 ~mod+ rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[5] = BIAS_AND_CLAMP[(((rd2 ~mod- rk5) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[3] = BIAS_AND_CLAMP[(((rd3 ~mod+ rk7) ~mod+ (1 << 17)) >> 18) & 1023]
args.dst_buffer[4] = BIAS_AND_CLAMP[(((rd3 ~mod- rk7) ~mod+ (1 << 17)) >> 18) & 1023]
// -------- END generated by script/print-jpeg-idct-code.go
}