scisharp
/
LLamaSharp
Not watched
1
0
Fork 0
Code
Releases 17 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
Browse Source

Automatic Solution Generator - Work in progress

pull/689/head
Zoli Somogyi 1 year ago
parent
3ded2dd74d
commit
b1f3987fae
5 changed files with 428 additions and 0 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +126
    -0
      CMakeLists.txt
  2. +126
    -0
      LLama.GenerateSolution/CMakeLists.txt.in
  3. +14
    -0
      LLama.GenerateSolution/GenerateSolution.csproj
  4. +25
    -0
      LLama.GenerateSolution/GenerateSolution.sln
  5. +137
    -0
      LLama.GenerateSolution/Program.cs

+ 126
- 0
CMakeLists.txt View File

@@ -0,0 +1,126 @@
#----------------------------------------------------------------------------------------
# INFO:
# - How to use: change the flags in the 'Set one of these ON and all others OFF' section
# - CUDA: it will use automatically the CUDA SDK version installed
#
#----------------------------------------------------------------------------------------
cmake_minimum_required(VERSION 3.8)
project(LLamaSharpCpp VERSION 0.10.0 LANGUAGES CXX CSharp)
if(NOT MSVC)
message(FATAL_ERROR "This CMake file only works with MSVC.")
endif(NOT MSVC)

#--------- Set one of these ON and all others OFF ------------------->
option(LLAMA_CUDA_AVX2 "CUDA + AVX2" ON)
option(LLAMA_AVX2 "AVX2 (no CUDA)" OFF)
option(LLAMA_CUDA "CUDA (no AVX)" OFF)
#etc... add other setups
#<--------- Set one of these ON and all others OFF -------------------

# --------------- Don't change below this line -----------------------

# Variable Settings
if(LLAMA_CUDA_AVX2)
option(LLAMA_AVX "llama: enable AVX" ON)
option(LLAMA_AVX2 "llama: enable AVX2" ON)
option(LLAMA_CUBLAS "llama: use CUDA" ON)
elseif(LLAMA_AVX2)
option(LLAMA_AVX "llama: enable AVX" ON)
option(LLAMA_AVX2 "llama: enable AVX2" ON)
option(LLAMA_CUBLAS "llama: use CUDA" OFF)
elseif(LLAMA_CUDA)
option(LLAMA_AVX "llama: enable AVX" OFF)
option(LLAMA_AVX2 "llama: enable AVX2" OFF)
option(LLAMA_CUBLAS "llama: use CUDA" ON)
elseif(OTHER_SETUPS)
#etc...
endif()

# Fixed Settings
# general
option(BUILD_SHARED_LIBS "build shared libraries" ON)
option(LLAMA_STATIC "llama: static link libraries" OFF)
option(LLAMA_NATIVE "llama: enable -march=native flag" OFF)
option(LLAMA_LTO "llama: enable link time optimization" OFF)
option(LLAMA_CCACHE "llama: use ccache if available" ON)

# debug
option(LLAMA_ALL_WARNINGS "llama: enable all compiler warnings" ON)
option(LLAMA_ALL_WARNINGS_3RD_PARTY "llama: enable all compiler warnings in 3rd party libs" OFF)
option(LLAMA_GPROF "llama: enable gprof" OFF)

# build
option(LLAMA_FATAL_WARNINGS "llama: enable -Werror flag" OFF)

# sanitizers
option(LLAMA_SANITIZE_THREAD "llama: enable thread sanitizer" OFF)
option(LLAMA_SANITIZE_ADDRESS "llama: enable address sanitizer" OFF)
option(LLAMA_SANITIZE_UNDEFINED "llama: enable undefined sanitizer" OFF)

option(LLAMA_AVX512 "llama: enable AVX512" OFF)
option(LLAMA_AVX512_VBMI "llama: enable AVX512-VBMI" OFF)
option(LLAMA_AVX512_VNNI "llama: enable AVX512-VNNI" OFF)
option(LLAMA_FMA "llama: enable FMA" OFF)
# in MSVC F16C is implied with AVX2/AVX512
if (NOT MSVC)
option(LLAMA_F16C "llama: enable F16C" OFF)
endif()

if (WIN32)
set(LLAMA_WIN_VER "0x602" CACHE STRING "llama: Windows Version")
endif()

# 3rd party libs
option(LLAMA_ACCELERATE "llama: enable Accelerate framework" ON)
option(LLAMA_BLAS "llama: use BLAS" OFF)
set(LLAMA_BLAS_VENDOR "Generic" CACHE STRING "llama: BLAS library vendor")
#option(LLAMA_CUDA_CUBLAS "llama: use cuBLAS for prompt processing" OFF)
option(LLAMA_CUDA_FORCE_DMMV "llama: use dmmv instead of mmvq CUDA kernels" OFF)
option(LLAMA_CUDA_FORCE_MMQ "llama: use mmq kernels instead of cuBLAS" OFF)
set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kernels")
set(LLAMA_CUDA_MMV_Y "1" CACHE STRING "llama: y block size for mmv CUDA kernels")
option(LLAMA_CUDA_F16 "llama: use 16 bit floats for some calculations" OFF)
set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K")
set(LLAMA_CUDA_PEER_MAX_BATCH_SIZE "128" CACHE STRING
"llama: max. batch size for using peer access")
option(LLAMA_HIPBLAS "llama: use hipBLAS" OFF)
option(LLAMA_HIP_UMA "llama: use HIP unified memory architecture" OFF)
option(LLAMA_CLBLAST "llama: use CLBlast" OFF)
option(LLAMA_VULKAN "llama: use Vulkan" OFF)
option(LLAMA_VULKAN_CHECK_RESULTS "llama: run Vulkan op checks" OFF)
option(LLAMA_VULKAN_DEBUG "llama: enable Vulkan debug output" OFF)
option(LLAMA_VULKAN_VALIDATE "llama: enable Vulkan validation" OFF)
option(LLAMA_VULKAN_RUN_TESTS "llama: run Vulkan tests" OFF)
option(LLAMA_METAL "llama: use Metal" ${LLAMA_METAL_DEFAULT})
option(LLAMA_METAL_NDEBUG "llama: disable Metal debugging" OFF)
option(LLAMA_METAL_SHADER_DEBUG "llama: compile Metal with -fno-fast-math" OFF)
option(LLAMA_METAL_EMBED_LIBRARY "llama: embed Metal library" OFF)
option(LLAMA_KOMPUTE "llama: use Kompute" OFF)
option(LLAMA_MPI "llama: use MPI" OFF)
option(LLAMA_QKK_64 "llama: use super-block size of 64 for k-quants" OFF)
option(LLAMA_SYCL "llama: use SYCL" OFF)
option(LLAMA_SYCL_F16 "llama: use 16 bit floats for sycl calculations" OFF)
option(LLAMA_CPU_HBM "llama: use memkind for CPU HBM" OFF)

option(LLAMA_BUILD_TESTS "llama: build tests" OFF)
option(LLAMA_BUILD_EXAMPLES "llama: build examples" ON)
option(LLAMA_BUILD_SERVER "llama: build server example" OFF)

# add perf arguments
option(LLAMA_PERF "llama: enable perf" OFF)

include_external_msproject(LLama.Unittest ./LLama.Unittest/LLama.Unittest.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE BAC1CFA9-E6AC-4BD0-A548-A8066D3C467E)

include_external_msproject(LLama.Examples ./LLama.Examples/LLama.Examples.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE BD1909AD-E1F8-476E-BC49-E394FF0470CE)

include_external_msproject(LLamaSharp ./LLama/LLamaSharp.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE 01A12D68-DE95-425E-AEEE-2D099305036D)

include_external_msproject(LLama.WebAPI ./LLama.WebAPI/LLama.WebAPI.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE D3CEC57A-9027-4DA4-AAAC-612A1EB50ADF)

include_external_msproject(LLama.Web ./LLama.Web/LLama.Web.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE C3531DB2-1B2B-433C-8DE6-3541E3620DB1)

include_external_msproject(LLamaSharp.SemanticKernel ./LLama.SemanticKernel/LLamaSharp.SemanticKernel.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE D98F93E3-B344-4F9D-86BB-FDBF6768B587)

include_external_msproject(LLamaSharp.KernelMemory ./LLama.KernelMemory/LLamaSharp.KernelMemory.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE E5589AE7-B86F-4343-A1CC-8E5D34596E52)

add_subdirectory(./llama.cpp)

+ 126
- 0
LLama.GenerateSolution/CMakeLists.txt.in View File

@@ -0,0 +1,126 @@
#----------------------------------------------------------------------------------------
# INFO:
# - How to use: change the flags in the 'Set one of these ON and all others OFF' section
# - CUDA: it will use automatically the CUDA SDK version installed
#
#----------------------------------------------------------------------------------------
cmake_minimum_required(VERSION 3.8)
project(LLamaSharpCpp VERSION 0.10.0 LANGUAGES CXX CSharp)
if(NOT MSVC)
message(FATAL_ERROR "This CMake file only works with MSVC.")
endif(NOT MSVC)

#--------- Set one of these ON and all others OFF ------------------->
option(LLAMA_CUDA_AVX2 "CUDA + AVX2" ON)
option(LLAMA_AVX2 "AVX2 (no CUDA)" OFF)
option(LLAMA_CUDA "CUDA (no AVX)" OFF)
#etc... add other setups
#<--------- Set one of these ON and all others OFF -------------------

# --------------- Don't change below this line -----------------------

# Variable Settings
if(LLAMA_CUDA_AVX2)
option(LLAMA_AVX "llama: enable AVX" ON)
option(LLAMA_AVX2 "llama: enable AVX2" ON)
option(LLAMA_CUBLAS "llama: use CUDA" ON)
elseif(LLAMA_AVX2)
option(LLAMA_AVX "llama: enable AVX" ON)
option(LLAMA_AVX2 "llama: enable AVX2" ON)
option(LLAMA_CUBLAS "llama: use CUDA" OFF)
elseif(LLAMA_CUDA)
option(LLAMA_AVX "llama: enable AVX" OFF)
option(LLAMA_AVX2 "llama: enable AVX2" OFF)
option(LLAMA_CUBLAS "llama: use CUDA" ON)
elseif(OTHER_SETUPS)
#etc...
endif()

# Fixed Settings
# general
option(BUILD_SHARED_LIBS "build shared libraries" ON)
option(LLAMA_STATIC "llama: static link libraries" OFF)
option(LLAMA_NATIVE "llama: enable -march=native flag" OFF)
option(LLAMA_LTO "llama: enable link time optimization" OFF)
option(LLAMA_CCACHE "llama: use ccache if available" ON)

# debug
option(LLAMA_ALL_WARNINGS "llama: enable all compiler warnings" ON)
option(LLAMA_ALL_WARNINGS_3RD_PARTY "llama: enable all compiler warnings in 3rd party libs" OFF)
option(LLAMA_GPROF "llama: enable gprof" OFF)

# build
option(LLAMA_FATAL_WARNINGS "llama: enable -Werror flag" OFF)

# sanitizers
option(LLAMA_SANITIZE_THREAD "llama: enable thread sanitizer" OFF)
option(LLAMA_SANITIZE_ADDRESS "llama: enable address sanitizer" OFF)
option(LLAMA_SANITIZE_UNDEFINED "llama: enable undefined sanitizer" OFF)

option(LLAMA_AVX512 "llama: enable AVX512" OFF)
option(LLAMA_AVX512_VBMI "llama: enable AVX512-VBMI" OFF)
option(LLAMA_AVX512_VNNI "llama: enable AVX512-VNNI" OFF)
option(LLAMA_FMA "llama: enable FMA" OFF)
# in MSVC F16C is implied with AVX2/AVX512
if (NOT MSVC)
option(LLAMA_F16C "llama: enable F16C" OFF)
endif()

if (WIN32)
set(LLAMA_WIN_VER "0x602" CACHE STRING "llama: Windows Version")
endif()

# 3rd party libs
option(LLAMA_ACCELERATE "llama: enable Accelerate framework" ON)
option(LLAMA_BLAS "llama: use BLAS" OFF)
set(LLAMA_BLAS_VENDOR "Generic" CACHE STRING "llama: BLAS library vendor")
#option(LLAMA_CUDA_CUBLAS "llama: use cuBLAS for prompt processing" OFF)
option(LLAMA_CUDA_FORCE_DMMV "llama: use dmmv instead of mmvq CUDA kernels" OFF)
option(LLAMA_CUDA_FORCE_MMQ "llama: use mmq kernels instead of cuBLAS" OFF)
set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kernels")
set(LLAMA_CUDA_MMV_Y "1" CACHE STRING "llama: y block size for mmv CUDA kernels")
option(LLAMA_CUDA_F16 "llama: use 16 bit floats for some calculations" OFF)
set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K")
set(LLAMA_CUDA_PEER_MAX_BATCH_SIZE "128" CACHE STRING
"llama: max. batch size for using peer access")
option(LLAMA_HIPBLAS "llama: use hipBLAS" OFF)
option(LLAMA_HIP_UMA "llama: use HIP unified memory architecture" OFF)
option(LLAMA_CLBLAST "llama: use CLBlast" OFF)
option(LLAMA_VULKAN "llama: use Vulkan" OFF)
option(LLAMA_VULKAN_CHECK_RESULTS "llama: run Vulkan op checks" OFF)
option(LLAMA_VULKAN_DEBUG "llama: enable Vulkan debug output" OFF)
option(LLAMA_VULKAN_VALIDATE "llama: enable Vulkan validation" OFF)
option(LLAMA_VULKAN_RUN_TESTS "llama: run Vulkan tests" OFF)
option(LLAMA_METAL "llama: use Metal" ${LLAMA_METAL_DEFAULT})
option(LLAMA_METAL_NDEBUG "llama: disable Metal debugging" OFF)
option(LLAMA_METAL_SHADER_DEBUG "llama: compile Metal with -fno-fast-math" OFF)
option(LLAMA_METAL_EMBED_LIBRARY "llama: embed Metal library" OFF)
option(LLAMA_KOMPUTE "llama: use Kompute" OFF)
option(LLAMA_MPI "llama: use MPI" OFF)
option(LLAMA_QKK_64 "llama: use super-block size of 64 for k-quants" OFF)
option(LLAMA_SYCL "llama: use SYCL" OFF)
option(LLAMA_SYCL_F16 "llama: use 16 bit floats for sycl calculations" OFF)
option(LLAMA_CPU_HBM "llama: use memkind for CPU HBM" OFF)

option(LLAMA_BUILD_TESTS "llama: build tests" OFF)
option(LLAMA_BUILD_EXAMPLES "llama: build examples" ON)
option(LLAMA_BUILD_SERVER "llama: build server example" OFF)

# add perf arguments
option(LLAMA_PERF "llama: enable perf" OFF)

include_external_msproject(LLama.Unittest ./LLama.Unittest/LLama.Unittest.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE BAC1CFA9-E6AC-4BD0-A548-A8066D3C467E)

include_external_msproject(LLama.Examples ./LLama.Examples/LLama.Examples.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE BD1909AD-E1F8-476E-BC49-E394FF0470CE)

include_external_msproject(LLamaSharp ./LLama/LLamaSharp.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE 01A12D68-DE95-425E-AEEE-2D099305036D)

include_external_msproject(LLama.WebAPI ./LLama.WebAPI/LLama.WebAPI.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE D3CEC57A-9027-4DA4-AAAC-612A1EB50ADF)

include_external_msproject(LLama.Web ./LLama.Web/LLama.Web.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE C3531DB2-1B2B-433C-8DE6-3541E3620DB1)

include_external_msproject(LLamaSharp.SemanticKernel ./LLama.SemanticKernel/LLamaSharp.SemanticKernel.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE D98F93E3-B344-4F9D-86BB-FDBF6768B587)

include_external_msproject(LLamaSharp.KernelMemory ./LLama.KernelMemory/LLamaSharp.KernelMemory.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE E5589AE7-B86F-4343-A1CC-8E5D34596E52)

add_subdirectory(./llama.cpp)

+ 14
- 0
LLama.GenerateSolution/GenerateSolution.csproj View File

@@ -0,0 +1,14 @@
<Project Sdk="Microsoft.NET.Sdk">

<PropertyGroup>
<OutputType>Exe</OutputType>
<TargetFramework>net7.0</TargetFramework>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
</PropertyGroup>

<ItemGroup>
<PackageReference Include="Spectre.Console" Version="0.48.0" />
</ItemGroup>

</Project>

+ 25
- 0
LLama.GenerateSolution/GenerateSolution.sln View File

@@ -0,0 +1,25 @@

Microsoft Visual Studio Solution File, Format Version 12.00
# Visual Studio Version 17
VisualStudioVersion = 17.8.34525.116
MinimumVisualStudioVersion = 10.0.40219.1
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "GenerateSolution", "GenerateSolution.csproj", "{89306FE9-4428-4C70-AF58-0AF871BED56B}"
EndProject
Global
GlobalSection(SolutionConfigurationPlatforms) = preSolution
Debug|Any CPU = Debug|Any CPU
Release|Any CPU = Release|Any CPU
EndGlobalSection
GlobalSection(ProjectConfigurationPlatforms) = postSolution
{89306FE9-4428-4C70-AF58-0AF871BED56B}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
{89306FE9-4428-4C70-AF58-0AF871BED56B}.Debug|Any CPU.Build.0 = Debug|Any CPU
{89306FE9-4428-4C70-AF58-0AF871BED56B}.Release|Any CPU.ActiveCfg = Release|Any CPU
{89306FE9-4428-4C70-AF58-0AF871BED56B}.Release|Any CPU.Build.0 = Release|Any CPU
EndGlobalSection
GlobalSection(SolutionProperties) = preSolution
HideSolutionNode = FALSE
EndGlobalSection
GlobalSection(ExtensibilityGlobals) = postSolution
SolutionGuid = {32128714-30D0-4015-9184-24F609AAE564}
EndGlobalSection
EndGlobal

+ 137
- 0
LLama.GenerateSolution/Program.cs View File

@@ -0,0 +1,137 @@
using Spectre.Console;
using System;
using System.Diagnostics;
using System.Text;
using static System.Runtime.InteropServices.JavaScript.JSType;

namespace GenerateSolution
{
internal class Program
{
static void Main(string[] args)
{
System.Console.InputEncoding = Encoding.Unicode;
System.Console.OutputEncoding = Encoding.Unicode;

// Check if we can accept key strokes
if (!AnsiConsole.Profile.Capabilities.Interactive)
{
AnsiConsole.MarkupLine("[red]Environment does not support interaction.[/]");
return;
}

var options = AskOptions();
var cmakePath = AskCMakePath();
if(string.IsNullOrEmpty(cmakePath) == true)
{
cmakePath = "C:\\Program Files\\CMake\\bin\\cmake.exe";
}
AnsiConsole.MarkupLine("You have selected: [yellow]{0}[/]", cmakePath);

string cmakeListsPath = @"..\..\..\..\CMakeLists.txt";

//cmake [<options>] -B <path-to-build> [-S <path-to-source>]
//TODO: get the chosen arguments from above (hardcoded values below)
//TODO: edit the CMakeList.txt.in template and create the CMakeLists.txt with the chosen options
cmakeListsPath += " -G \"Visual Studio 17 2022\" -A x64 -B ..\\..\\..\\..\\ -S ..\\..\\..\\..\\";

ProcessStartInfo startInfo = new ProcessStartInfo
{
FileName = cmakePath,
Arguments = cmakeListsPath,
RedirectStandardOutput = true,
RedirectStandardError = true,
UseShellExecute = false,
CreateNoWindow = true,
};

try
{
bool bSuccess = false;
string lastError = "";
AnsiConsole.Progress()
.AutoClear(false)
.Columns(new ProgressColumn[]
{
new TaskDescriptionColumn(),
new SpinnerColumn(Spinner.Known.Ascii),
})
.Start(ctx =>
{
var cmakeTask = ctx.AddTask("Generating VS Solution", autoStart: false).IsIndeterminate();
cmakeTask.StartTask();
using (Process process = new Process())
{
process.StartInfo = startInfo;
process.Start();
string output = process.StandardOutput.ReadToEnd();
lastError = process.StandardError.ReadToEnd();
process.WaitForExit();
cmakeTask.StopTask();
if (process.ExitCode == 0)
{
bSuccess = true;
}
}
});

if (bSuccess == true)
{
AnsiConsole.WriteLine("VS solution generated successfully.");
}
else
{
AnsiConsole.WriteLine($"Error running CMake configuration: {lastError}");
}
}
catch (Exception ex)
{
AnsiConsole.WriteLine("[red]ERROR[/] " + ex.Message);
}

Console.ReadLine();
}

public static string AskCMakePath()
{
return AnsiConsole.Prompt(
new TextPrompt<string>("What's your [green]CMake path[/] (default: C:\\Program Files\\CMake\\bin\\cmake.exe)?")
.AllowEmpty());
}

public static List<string> AskOptions()
{
var options = AnsiConsole.Prompt(
new MultiSelectionPrompt<string>()
.PageSize(10)
.Title("Select the preferred [green]options[/]?")
.MoreChoicesText("[grey](Move up and down to reveal more options)[/]")
.InstructionsText("[grey](Press [blue]<space>[/] to toggle an option, [green]<enter>[/] to accept)[/]")
.AddChoiceGroup("Avx", new[]
{
"Avx2", "Avx512"
})
.AddChoiceGroup("Cuda", new[]
{
"Cuda"
})
.AddChoices(new[]
{
"x64",
})
.AddChoiceGroup("Visual Studio", new[]
{
"Visual Studio 16 2019",
"Visual Studio 17 2022"
})
);

if (options.Count > 0)
{
AnsiConsole.MarkupLine("You have selected: [yellow]{0}[/]", string.Join(",",options));
}
return options;
}
}
}

Write Preview
Loading…
Cancel
Save