source/libvpx/vp8/common/x86/variance_impl_mmx.asm - Issue 1154153005: Rename variance files

Side by Side Diff: source/libvpx/vp8/common/x86/variance_impl_mmx.asm

Issue 1154153005: Rename variance files (Closed) Base URL: https://chromium.googlesource.com/chromium/deps/libvpx.git@master

Patch Set: Created 5 years, 6 months ago

Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.

Jump to:

View unified diff | Download patch

OLD	NEW
	(Empty)
1 ;

2 ; Copyright (c) 2010 The WebM project authors. All Rights Reserved.

3 ;

4 ; Use of this source code is governed by a BSD-style license

5 ; that can be found in the LICENSE file in the root of the source

6 ; tree. An additional intellectual property rights grant can be found

7 ; in the file PATENTS. All contributing project authors may

8 ; be found in the AUTHORS file in the root of the source tree.

9 ;

10

11

12 %include "vpx_ports/x86_abi_support.asm"

13

14 %define mmx_filter_shift 7

15

16 ;void vp8_filter_block2d_bil4x4_var_mmx

17 ;(

18 ; unsigned char *ref_ptr,

19 ; int ref_pixels_per_line,

20 ; unsigned char *src_ptr,

21 ; int src_pixels_per_line,

22 ; unsigned short *HFilter,

23 ; unsigned short *VFilter,

24 ; int *sum,

25 ; unsigned int *sumsquared

26 ;)

27 global sym(vp8_filter_block2d_bil4x4_var_mmx) PRIVATE

28 sym(vp8_filter_block2d_bil4x4_var_mmx):

29 push rbp

30 mov rbp, rsp

31 SHADOW_ARGS_TO_STACK 8

32 GET_GOT rbx

33 push rsi

34 push rdi

35 sub rsp, 16

36 ; end prolog

37

38

39 pxor mm6, mm6 ;

40 pxor mm7, mm7 ;

41

42 mov rax, arg(4) ;HFilter ;

43 mov rdx, arg(5) ;VFilter ;

44

45 mov rsi, arg(0) ;ref_ptr ;

46 mov rdi, arg(2) ;src_ptr ;

47

48 mov rcx, 4 ;

49 pxor mm0, mm0 ;

50

51 movd mm1, [rsi] ;

52 movd mm3, [rsi+1] ;

53

54 punpcklbw mm1, mm0 ;

55 pmullw mm1, [rax] ;

56

57 punpcklbw mm3, mm0 ;

58 pmullw mm3, [rax+8] ;

59

60 paddw mm1, mm3 ;

61 paddw mm1, [GLOBAL(mmx_bi_rd)] ;

62

63 psraw mm1, mmx_filter_shift ;

64 movq mm5, mm1

65

66 %if ABI_IS_32BIT

67 add rsi, dword ptr arg(1) ;ref_pixels_per_line ;

68 %else

69 movsxd r8, dword ptr arg(1) ;ref_pixels_per_line ;

70 add rsi, r8

71 %endif

72

73 .filter_block2d_bil4x4_var_mmx_loop:

74

75 movd mm1, [rsi] ;

76 movd mm3, [rsi+1] ;

77

78 punpcklbw mm1, mm0 ;

79 pmullw mm1, [rax] ;

80

81 punpcklbw mm3, mm0 ;

82 pmullw mm3, [rax+8] ;

83

84 paddw mm1, mm3 ;

85 paddw mm1, [GLOBAL(mmx_bi_rd)] ;

86

87 psraw mm1, mmx_filter_shift ;

88 movq mm3, mm5 ;

89

90 movq mm5, mm1 ;

91 pmullw mm3, [rdx] ;

92

93 pmullw mm1, [rdx+8] ;

94 paddw mm1, mm3 ;

95

96

97 paddw mm1, [GLOBAL(mmx_bi_rd)] ;

98 psraw mm1, mmx_filter_shift ;

99

100 movd mm3, [rdi] ;

101 punpcklbw mm3, mm0 ;

102

103 psubw mm1, mm3 ;

104 paddw mm6, mm1 ;

105

106 pmaddwd mm1, mm1 ;

107 paddd mm7, mm1 ;

108

109 %if ABI_IS_32BIT

110 add rsi, dword ptr arg(1) ;ref_pixels_per_line ;

111 add rdi, dword ptr arg(3) ;src_pixels_per_line ;

112 %else

113 movsxd r8, dword ptr arg(1) ;ref_pixels_per_line

114 movsxd r9, dword ptr arg(3) ;src_pixels_per_line

115 add rsi, r8

116 add rdi, r9

117 %endif

118 sub rcx, 1 ;

119 jnz .filter_block2d_bil4x4_var_mmx_loop ;

120

121

122 pxor mm3, mm3 ;

123 pxor mm2, mm2 ;

124

125 punpcklwd mm2, mm6 ;

126 punpckhwd mm3, mm6 ;

127

128 paddd mm2, mm3 ;

129 movq mm6, mm2 ;

130

131 psrlq mm6, 32 ;

132 paddd mm2, mm6 ;

133

134 psrad mm2, 16 ;

135 movq mm4, mm7 ;

136

137 psrlq mm4, 32 ;

138 paddd mm4, mm7 ;

139

140 mov rdi, arg(6) ;sum

141 mov rsi, arg(7) ;sumsquared

142

143 movd dword ptr [rdi], mm2 ;

144 movd dword ptr [rsi], mm4 ;

145

146

147

148 ; begin epilog

149 add rsp, 16

150 pop rdi

151 pop rsi

152 RESTORE_GOT

153 UNSHADOW_ARGS

154 pop rbp

155 ret

156

157

158

159

160 ;void vp8_filter_block2d_bil_var_mmx

161 ;(

162 ; unsigned char *ref_ptr,

163 ; int ref_pixels_per_line,

164 ; unsigned char *src_ptr,

165 ; int src_pixels_per_line,

166 ; unsigned int Height,

167 ; unsigned short *HFilter,

168 ; unsigned short *VFilter,

169 ; int *sum,

170 ; unsigned int *sumsquared

171 ;)

172 global sym(vp8_filter_block2d_bil_var_mmx) PRIVATE

173 sym(vp8_filter_block2d_bil_var_mmx):

174 push rbp

175 mov rbp, rsp

176 SHADOW_ARGS_TO_STACK 9

177 GET_GOT rbx

178 push rsi

179 push rdi

180 sub rsp, 16

181 ; end prolog

182

183 pxor mm6, mm6 ;

184 pxor mm7, mm7 ;

185 mov rax, arg(5) ;HFilter ;

186

187 mov rdx, arg(6) ;VFilter ;

188 mov rsi, arg(0) ;ref_ptr ;

189

190 mov rdi, arg(2) ;src_ptr ;

191 movsxd rcx, dword ptr arg(4) ;Height ;

192

193 pxor mm0, mm0 ;

194 movq mm1, [rsi] ;

195

196 movq mm3, [rsi+1] ;

197 movq mm2, mm1 ;

198

199 movq mm4, mm3 ;

200 punpcklbw mm1, mm0 ;

201

202 punpckhbw mm2, mm0 ;

203 pmullw mm1, [rax] ;

204

205 pmullw mm2, [rax] ;

206 punpcklbw mm3, mm0 ;

207

208 punpckhbw mm4, mm0 ;

209 pmullw mm3, [rax+8] ;

210

211 pmullw mm4, [rax+8] ;

212 paddw mm1, mm3 ;

213

214 paddw mm2, mm4 ;

215 paddw mm1, [GLOBAL(mmx_bi_rd)] ;

216

217 psraw mm1, mmx_filter_shift ;

218 paddw mm2, [GLOBAL(mmx_bi_rd)] ;

219

220 psraw mm2, mmx_filter_shift ;

221 movq mm5, mm1

222

223 packuswb mm5, mm2 ;

224 %if ABI_IS_32BIT

225 add rsi, dword ptr arg(1) ;ref_pixels_per_line

226 %else

227 movsxd r8, dword ptr arg(1) ;ref_pixels_per_line

228 add rsi, r8

229 %endif

230

231 .filter_block2d_bil_var_mmx_loop:

232

233 movq mm1, [rsi] ;

234 movq mm3, [rsi+1] ;

235

236 movq mm2, mm1 ;

237 movq mm4, mm3 ;

238

239 punpcklbw mm1, mm0 ;

240 punpckhbw mm2, mm0 ;

241

242 pmullw mm1, [rax] ;

243 pmullw mm2, [rax] ;

244

245 punpcklbw mm3, mm0 ;

246 punpckhbw mm4, mm0 ;

247

248 pmullw mm3, [rax+8] ;

249 pmullw mm4, [rax+8] ;

250

251 paddw mm1, mm3 ;

252 paddw mm2, mm4 ;

253

254 paddw mm1, [GLOBAL(mmx_bi_rd)] ;

255 psraw mm1, mmx_filter_shift ;

256

257 paddw mm2, [GLOBAL(mmx_bi_rd)] ;

258 psraw mm2, mmx_filter_shift ;

259

260 movq mm3, mm5 ;

261 movq mm4, mm5 ;

262

263 punpcklbw mm3, mm0 ;

264 punpckhbw mm4, mm0 ;

265

266 movq mm5, mm1 ;

267 packuswb mm5, mm2 ;

268

269 pmullw mm3, [rdx] ;

270 pmullw mm4, [rdx] ;

271

272 pmullw mm1, [rdx+8] ;

273 pmullw mm2, [rdx+8] ;

274

275 paddw mm1, mm3 ;

276 paddw mm2, mm4 ;

277

278 paddw mm1, [GLOBAL(mmx_bi_rd)] ;

279 paddw mm2, [GLOBAL(mmx_bi_rd)] ;

280

281 psraw mm1, mmx_filter_shift ;

282 psraw mm2, mmx_filter_shift ;

283

284 movq mm3, [rdi] ;

285 movq mm4, mm3 ;

286

287 punpcklbw mm3, mm0 ;

288 punpckhbw mm4, mm0 ;

289

290 psubw mm1, mm3 ;

291 psubw mm2, mm4 ;

292

293 paddw mm6, mm1 ;

294 pmaddwd mm1, mm1 ;

295

296 paddw mm6, mm2 ;

297 pmaddwd mm2, mm2 ;

298

299 paddd mm7, mm1 ;

300 paddd mm7, mm2 ;

301

302 %if ABI_IS_32BIT

303 add rsi, dword ptr arg(1) ;ref_pixels_per_line ;

304 add rdi, dword ptr arg(3) ;src_pixels_per_line ;

305 %else

306 movsxd r8, dword ptr arg(1) ;ref_pixels_per_line ;

307 movsxd r9, dword ptr arg(3) ;src_pixels_per_line ;

308 add rsi, r8

309 add rdi, r9

310 %endif

311 sub rcx, 1 ;

312 jnz .filter_block2d_bil_var_mmx_loop ;

313

314

315 pxor mm3, mm3 ;

316 pxor mm2, mm2 ;

317

318 punpcklwd mm2, mm6 ;

319 punpckhwd mm3, mm6 ;

320

321 paddd mm2, mm3 ;

322 movq mm6, mm2 ;

323

324 psrlq mm6, 32 ;

325 paddd mm2, mm6 ;

326

327 psrad mm2, 16 ;

328 movq mm4, mm7 ;

329

330 psrlq mm4, 32 ;

331 paddd mm4, mm7 ;

332

333 mov rdi, arg(7) ;sum

334 mov rsi, arg(8) ;sumsquared

335

336 movd dword ptr [rdi], mm2 ;

337 movd dword ptr [rsi], mm4 ;

338

339 ; begin epilog

340 add rsp, 16

341 pop rdi

342 pop rsi

343 RESTORE_GOT

344 UNSHADOW_ARGS

345 pop rbp

346 ret

347

348

349 SECTION_RODATA

350 ;short mmx_bi_rd[4] = { 64, 64, 64, 64};

351 align 16

352 mmx_bi_rd:

353 times 4 dw 64

OLD	NEW

« no previous file with comments | « libvpx_srcs_x86_intrinsics.gypi ('k') | source/libvpx/vp8/common/x86/variance_mmx.c » ('j') | no next file with comments »