鏈榪戝湪鍐?4浣嶆眹緙栵紝璁劇疆緙栬瘧鍛戒護鏄歡綣佺悙鑰屽張閲嶅鐨勫姵鍔ㄣ傜綉涓婁篃鏈変笉灝戞枃绔犱粙緇嶆庝箞鏍風紪璇?4浣嶆眹緙栫殑錛屼絾閮借寰楀鏉傦紝鏈夌殑榪橀渶瑕佷笅杞藉叾浠栫殑姹囩紪杞歡錛屾渶鍥х殑鏄細鏈夌瘒鏂囩珷緇欏嚭鐨勪笅杞藉湴鍧涓嬭澆涓嬫潵鐨勬枃浠跺共鑴嗗氨鐩存帴鏄湪椹紒
鏃㈢劧鎴戜滑浣跨敤VS緋誨垪錛屽氨涓嶈鑸嶈繎姹傝繙浜嗐備嬌鐢∕S鎻愪緵鐨刴l64錛屽畨鍏ㄦ斁蹇冿紝涔熸棤鍏煎鎬ч棶棰樸備笅闈㈠紑濮嬭鏄庢庝箞鏍瘋緗細
涓銆佸湪vc鐨勫畨瑁呯洰褰曚笅錛屾壘鍒?#8220;VCProjectDefaults/masm.rules”錛屽鍒跺嚭鏉ワ紝鏀逛釜鍚嶅瓧錛岀敤璁頒簨鏈瓑綰枃鏈紪杈戝櫒鎵撳紑
浜屻乵asm.rules鐨勫墠闈㈠嚑琛岀殑鍐呭濡備笅錛?/p>
- 锘?/span><?xml version="1.0" encoding="utf-8"?>
- <VisualStudioToolFile
- Name="Microsoft Macro Assembler"
- Version="8.00"
- >
- <Rules>
- <CustomBuildRule
- Name="MASM"
- DisplayName="Microsoft Macro Assembler"
- CommandLine="ml.exe /c [AllOptions] [AdditionalOptions] /Ta[inputs]"
- Outputs="[$ObjectFileName]"
- FileExtensions="*.asm"
- ExecutionDescription="Assembling..."
- >
鏇挎崲ml.exe涓簃l64.exe錛岀劧鍚庢妸鍚嶇О錛屾墿灞曞悕淇敼鎴愪綘鍠滄鐨勶紝涓嬮潰鏄垜鐨勪慨鏀癸細
- 锘?/span><?xml version="1.0" encoding="utf-8"?>
- <VisualStudioToolFile
- Name="Macro Assembler x64"
- Version="8.00"
- >
- <Rules>
- <CustomBuildRule
- Name="MASM x64"
- DisplayName="Macro Assembler x64"
- CommandLine="ml64.exe /c [AllOptions] [AdditionalOptions] /Ta[inputs]"
- Outputs="[$ObjectFileName]"
- FileExtensions="*.asm64"
- ExecutionDescription="Assembling..."
- >
淇濆瓨錛屽叧闂枃浠躲?/span>
涓夈佸湪瑙ｅ喅鏂規璧勬簮綆＄悊鍣ㄩ噷錛屽彸閿偣涓綘鐨勯」鐩紝鍑烘潵涓涓彍鍗曪紱閫夋嫨“鑷畾涔夌敓鎴愯鍒?..”錛屽湪寮瑰嚭鐨勫璇濇涓婏紝鐐?#8220;鏌ユ壘鐜版湁鐨?..”錛屾墦寮浣犲垰鎵嶇紪杈戠殑.rules鏂囦歡銆傜劧鍚庝綘灝卞簲璇ュ湪“鍙敤鐨勮鍒欐枃浠?#8221;鍒楄〃閲岀湅鍒頒綘鍒氭墠淇敼鐨勫悕瀛楀拰鎵╁睍鍚嶏紙鎴戠殑鍙仛asm64錛夛紝鍦ㄥ叾宸﹁竟鐨勬鏌ユ閲屽嬀涓婏紝鍏抽棴榪欎釜紿楀彛銆?/span>
鍥涖佹柊寤轟竴涓枃鏈枃浠訛紝鏀瑰悕鎴?#8220;鏂囦歡鍚嶇О.浣犲枩嬈㈢殑鎵╁睍鍚?#8221;錛屽鎴戠殑灝卞彨鍋?#8220;1.asm64”錛岀劧鍚庢妸榪欎釜鏂囦歡鍔犲叆鍒板伐紼嬮噷銆備笉鍑烘剰澶栵紝VS灝辮嚜鍔ㄦ牴鎹墿灞曞悕閫夋嫨浜嗕綘鍒氭墠淇敼鐨勮鍒欙紱濡傛灉鍑轟簡鎰忓錛屽垯鍗曠嫭瀵?#8220;1.asm64”榪涜璁劇疆錛氬湪瑙ｅ喅鏂規璧勬簮綆＄悊鍣ㄩ噷錛屽彸閿偣涓繖涓枃浠訛紝閫夋嫨灞炴э紱鐒跺悗鍦ㄥ璇濇閲屽乏杈歸夋嫨“閰嶅埗灞炴?甯歌”錛岀劧鍚庡湪宸ュ叿閲岄夋嫨浣犲垰鎵嶅懡鍚嶇殑瑙勫垯銆傚鎴戠殑鍙仛“Macro Assembler x64”銆傜偣“紜”鎸夐挳鍏抽棴榪欎釜瀵硅瘽妗嗐?/span>
浜斻佸湪“1.asm64”鏂囦歡閲屽啓涓婃紜殑姹囩紪錛岀紪璇戜箣銆備笉鍑烘剰澶栵紝64浣嶆眹緙栧凡緇忔紜繛鎺ュ埌浣犵殑紼嬪簭閲屼簡銆?/span>
鍏佹墦寮“1.asm64”鏂囦歡鐨勫睘鎬у璇濇錛屽簲璇ヨ兘鐪嬪埌涓嶅皯璁劇疆閫夐」錛屾參鎱㈢悽紓ㄥ惂銆?/span>
show涓孌?4浣嶆眹緙栦唬鐮?
- INCLUDE listing.inc
- pdata SEGMENT
- ___real_value_1 DD 03f800000r
- DD 03f800000r
- DD 03f800000r
- DD 03f800000r
- ___real_value_negation DD 080000000H
- DD 080000000H
- DD 080000000H
- DD 080000000H
- pdata ENDS
- ;void MulAdd(D3DMATRIX & Out,const D3DMATRIX & m1,FLOAT f,const D3DMATRIX & m2)
- ;{
- ; register FLOAT * pOut = (FLOAT *)&Out;
- ; register const FLOAT * pIn1 = (FLOAT *)&m1;
- ; register const FLOAT * pIn2 = (FLOAT *)&m2;
- ; for(register int i=15;i>=0;--i)
- ; pOut[i] = pIn1[i] * f + pIn2[i];
- ;}
- PUBLIC ?MulAdd@@YAXAEAU_D3DMATRIX@@AEBU1@M1@Z
- _TEXT SEGMENT
- ?MulAdd@@YAXAEAU_D3DMATRIX@@AEBU1@M1@Z PROC
- shufps xmm2,xmm2,0
- movups xmm0,[rdx]
- movups xmm1,[r9]
- mulps xmm0,xmm2
- addps xmm0,xmm1
- movups [rcx],xmm0
- movups xmm0,[rdx+16]
- movups xmm1,[r9+16]
- mulps xmm0,xmm2
- addps xmm0,xmm1
- movups [rcx+16],xmm0
- movups xmm0,[rdx+32]
- movups xmm1,[r9+32]
- mulps xmm0,xmm2
- addps xmm0,xmm1
- movups [rcx+32],xmm0
- movups xmm0,[rdx+48]
- movups xmm1,[r9+48]
- mulps xmm0,xmm2
- addps xmm0,xmm1
- movups [rcx+48],xmm0
- ret 0
- ?MulAdd@@YAXAEAU_D3DMATRIX@@AEBU1@M1@Z ENDP
- _TEXT ENDS
紲濆悇浣嶅悓浠乢WIN64鏃呯▼鎰夊揩錛?/p>
]]>
1銆丷elease涓嶈兘緙栬瘧鎴愬姛錛屾彁紺烘棤娉曡В鏋愮殑澶栭儴鍙橀噺錛屾垨鑰呮棤娉曟壘鍒?*.lib閾炬帴搴?/p>
瑙ｅ喅鏂規硶鏄湪欏圭洰-->灞炴?->榪炴帴鍣?->甯歌-->闄勫姞搴撶洰褰曪紝灝嗗唴瀹逛慨鏀逛負錛?.\..\..\common\$(PlatformName)\$(ConfigurationName);$(HMPP_PATH)\lib
2銆佸ぇ瀹舵湁娌℃湁瑙夊緱鏃墮棿鍙婃椂涓嶅ぇ瀵瑰憖錛?/p>
鍛靛懙錛岃繖涓槸鐪熺殑銆傚湪lab涓紝灝?/p>
fprintf( stdout, "CPU time : %.2f ms(best)\n", best_measure/10e3); 鏀逛負
fprintf( stdout, "CPU time : %.2f ms(best)\n", best_measure/1e3);錛屽疄闄呬笂灝辨槸涓涓暟閲忕駭鐨勯棶棰樸?
涓轟簡涓嶆敼鍙楥APS HMPP labs鐨勫師鏈夌粨鏋勶紝鍚庨潰鐨勮瑙ｄ腑灝嗛噸鏂版惌寤洪」鐩傚笇鏈涘ぇ瀹跺叧娉ㄥゥ銆傘?img alt="寰瑧" src="http://static.blog.csdn.net/xheditor/xheditor_emot/default/smile.gif" />
]]>