new compute kernel design, less mem read delay

This commit is contained in:
Nitload
2026-04-07 01:04:05 +08:00
parent d702b3224c
commit 12a216d404
47 changed files with 225 additions and 145 deletions
+5 -6
View File
@@ -85,8 +85,8 @@
<!-- 把 ARM64 → arm64,x64 保持 x64 -->
<_ArchTag Condition="'$(Platform)'=='x64'">x64</_ArchTag>
<_ArchTag Condition="'$(Platform)'=='ARM64'">arm64</_ArchTag>
<!-- 输出到 build/win-<arch>/ -->
<OutDir>$(SolutionDir)build\win-$(_ArchTag)\</OutDir>
<!-- 输出到 cmake_build/win_<arch>/,与 CMake 目录约定保持一致 -->
<OutDir>$(SolutionDir)cmake_build\win_$(_ArchTag)\</OutDir>
<IntDir>$(SolutionDir)obj\$(ProjectName)\$(Configuration)\$(Platform)\</IntDir>
</PropertyGroup>
<PropertyGroup Condition="'$(Configuration)'=='Debug'">
@@ -110,7 +110,7 @@
<LanguageStandard_C>stdc11</LanguageStandard_C>
<MultiProcessorCompilation>true</MultiProcessorCompilation>
<FavorSizeOrSpeed>Speed</FavorSizeOrSpeed>
<AdditionalIncludeDirectories>$(ProjectDir)lib;$(ProjectDir)sn-head;$(ProjectDir)sn-template;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
<AdditionalIncludeDirectories>$(ProjectDir)lib;$(ProjectDir)sn-head;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
</ClCompile>
<Link>
<!-- DLL 应该用 Windows,不是 Console -->
@@ -172,7 +172,7 @@
<!-- source files -->
<!-- ════════════════════════════════════════════════════════ -->
<ItemGroup>
<ClCompile Include="sn-src\m.transpose.x86.c" />
<ClCompile Include="sn-src\m.basic.x86.c" />
<ClCompile Include="sn-src\dllmain.c" />
<ClCompile Include="sn-src\m.bufop.x86.c" />
<ClCompile Include="sn-src\m.solve.x86.c" />
@@ -185,8 +185,7 @@
<ClCompile Include="sn-src\simdid.x86.c" />
<ClCompile Include="sn-src\simdid_compare.x86.c" />
<ClCompile Include="sn-src\sn_native.c" />
<ClCompile Include="sn-src\thread.linux.c" />
<ClCompile Include="sn-src\thread.windows.c" />
<ClCompile Include="sn-src\t.affinity.c" />
<ClCompile Include="sn-src\vector.c" />
<ClCompile Include="sn-src\vector_dot.c" />
</ItemGroup>