Build native Windows packages for both RDNA4 targets
Add explicit gfx1200/gfx1201 builds, runtime selection and per-target code objects. Guard staging, packaging and installation against mixed architectures; retain gfx1201 compatibility and strict K/M arithmetic. Adapt realdody's target configuration and MSVC binary16/BOOL portability fixes from ae530b0. Do not import its forced gfx12 layout on RDNA3. Add exhaustive binary16 and C++/PowerShell target contract regressions. Validation: 20 CTest gates on RX 9070 XT; four software-only gfx1200 gates; MSVC portability compile checks; exact 4K K async and M control outputs; isolated package install/restore and mixed-target rejection tests. No physical gfx1200 validation is claimed. Co-authored-by: realdody <[email protected]>
This commit is contained in:
xdfnx-devandrealdody committed
1 parent
8f7058d2ba
commit
98de8e3a3e
51 files changed
+636
-127
No files matched your search
@@ -5,12 +5,13 @@ param(
|
||||
[string]$HipRoot = 'C:\Program Files\AMD\ROCm\7.2',
|
||||
[string]$ZludaRoot,
|
||||
[ValidateSet('cu','wgp')][string]$ShaderMode='cu',
|
||||
[ValidateSet('gfx1200','gfx1201')][string]$GpuArch='gfx1201',
|
||||
[string]$OutputDirectory
|
||||
)
|
||||
$ErrorActionPreference='Stop'
|
||||
$repo=[IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$ZludaRoot) { $ZludaRoot=Join-Path $repo 'dist/zluda-windows-final' }
|
||||
if (!$OutputDirectory) { $OutputDirectory=Join-Path $repo 'build/private-textures-gfx1201' }
|
||||
if (!$OutputDirectory) { $OutputDirectory=Join-Path $repo "build/private-textures-$GpuArch" }
|
||||
& python (Join-Path $repo 'tools/windows/build_native_texture.py') --dlss-dll $DlssDll --hip-root $HipRoot `
|
||||
--zluda-root $ZludaRoot --kernel $Kernel --output-directory $OutputDirectory --shader-mode $ShaderMode
|
||||
--zluda-root $ZludaRoot --kernel $Kernel --output-directory $OutputDirectory --shader-mode $ShaderMode --gpu-arch $GpuArch
|
||||
if ($LASTEXITCODE) { throw "Native texture build failed ($LASTEXITCODE); inspect the output work directory." }
|
||||
@@ -2,6 +2,7 @@
|
||||
param(
|
||||
[string]$HipRoot = $env:HIP_PATH,
|
||||
[ValidateSet('stable', 'therock')][string]$RuntimeProfile = 'stable',
|
||||
[ValidateSet('gfx1200','gfx1201')][string]$GpuArch = 'gfx1201',
|
||||
[string]$ZludaRoot,
|
||||
[string]$ToolchainRoot,
|
||||
[string]$BuildDirectory,
|
||||
@@ -17,6 +18,10 @@ if ($RuntimeProfile -eq 'therock') {
|
||||
if (!$HipRoot) { $HipRoot = 'C:\Program Files\AMD\ROCm\7.2' }
|
||||
if (!$BuildDirectory) { $BuildDirectory = Join-Path $repo 'build/windows-rdna4' }
|
||||
if (!$InstallDirectory) { $InstallDirectory = Join-Path $repo 'dist/windows-rdna4-diagnostics' }
|
||||
if ($GpuArch -ne 'gfx1201') {
|
||||
if (!$PSBoundParameters.ContainsKey('BuildDirectory')) { $BuildDirectory += "-$GpuArch" }
|
||||
if (!$PSBoundParameters.ContainsKey('InstallDirectory')) { $InstallDirectory += "-$GpuArch" }
|
||||
}
|
||||
$HipRoot = [IO.Path]::GetFullPath($HipRoot).TrimEnd('\', '/')
|
||||
$BuildDirectory = [IO.Path]::GetFullPath($BuildDirectory).TrimEnd('\', '/')
|
||||
$InstallDirectory = [IO.Path]::GetFullPath($InstallDirectory).TrimEnd('\', '/')
|
||||
@@ -47,7 +52,7 @@ if ($minhookCommit -ne 'c3fcafdc10146beb5919319d0683e44e3c30d537') {
|
||||
$originalPath = $env:PATH
|
||||
try {
|
||||
$configure = @('--fresh', '-S', $repo, '-B', $BuildDirectory, '-G', 'Ninja', '-DCMAKE_BUILD_TYPE=RelWithDebInfo',
|
||||
"-DCMAKE_MAKE_PROGRAM=$ninja", "-DD4R_HIP_ROOT=$HipRoot", "-DCMAKE_INSTALL_PREFIX=$InstallDirectory")
|
||||
"-DCMAKE_MAKE_PROGRAM=$ninja", "-DD4R_HIP_ROOT=$HipRoot", "-DD4R_GPU_ARCH=$GpuArch", "-DCMAKE_INSTALL_PREFIX=$InstallDirectory")
|
||||
if ($ToolchainRoot) {
|
||||
$compiler = Join-Path $ToolchainRoot 'bin/x86_64-w64-mingw32-clang++.exe'
|
||||
if (!(Test-Path $compiler)) { throw "Compiler missing: $compiler" }
|
||||
|
||||
@@ -0,0 +1,26 @@
|
||||
# Shared by the build, diagnostics, staging and game package scripts.
|
||||
function Get-D4RGpuTarget([string]$Root, [string]$Requested) {
|
||||
$configured=$null
|
||||
$targetFile=Join-Path $Root 'gpu-target.json'
|
||||
if (Test-Path -LiteralPath $targetFile) {
|
||||
$configured=(Get-Content -LiteralPath $targetFile -Raw | ConvertFrom-Json).architecture
|
||||
if ($configured -notin @('gfx1200','gfx1201')) { throw "Unsupported package GPU target: $configured" }
|
||||
}
|
||||
$target=if ($Requested) { $Requested } elseif ($configured) { $configured } else { 'gfx1201' }
|
||||
if ($target -notin @('gfx1200','gfx1201')) { throw "Unsupported GPU target: $target" }
|
||||
if (!$configured -and $target -ne 'gfx1201') { throw 'A gfx1200 build requires gpu-target.json; rebuild with -GpuArch gfx1200.' }
|
||||
if ($configured -and $target -ne $configured) { throw "Requested $target differs from package target $configured" }
|
||||
return $target
|
||||
}
|
||||
function Assert-D4RCodeObjectTarget([string]$Path, [string]$Target) {
|
||||
if ($Target -notin @('gfx1200','gfx1201')) { throw "Unsupported GPU target: $Target" }
|
||||
$reader=[IO.BinaryReader]::new([IO.File]::OpenRead($Path))
|
||||
try { $header=$reader.ReadBytes(64) } finally { $reader.Dispose() }
|
||||
if ($header.Length -ne 64 -or $header[0] -ne 127 -or $header[1] -ne 69 -or $header[2] -ne 76 -or $header[3] -ne 70 -or
|
||||
$header[4] -ne 2 -or $header[5] -ne 1 -or $header[18] -ne 224 -or $header[19] -ne 0) {
|
||||
throw "Invalid AMDGPU ELF64 code object: $Path"
|
||||
}
|
||||
# LLVM AMDGPU e_flags machine IDs; inspect, never patch or spoof them.
|
||||
$expected=if ($Target -eq 'gfx1200') { 0x48 } else { 0x4e }
|
||||
if ($header[48] -ne $expected) { throw "Code object does not target ${Target}: $Path (ELF machine=$($header[48]))" }
|
||||
}
|
||||
@@ -2,18 +2,32 @@
|
||||
param(
|
||||
[string]$PackageRoot, [string]$DiagnosticRoot, [string]$OptiScalerRoot,
|
||||
[string]$ZludaRoot, [string]$HipRoot, [string]$KernelKRoot, [string]$KernelMRoot,
|
||||
[ValidateSet('gfx1200','gfx1201')][string]$GpuArch,
|
||||
[string]$ArchivePath
|
||||
)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
$defaults = @{PackageRoot='dist/windows-rdna4-game'; DiagnosticRoot='dist/windows-rdna4-command-list';
|
||||
OptiScalerRoot='dist/optiscaler-windows-d4r'; ZludaRoot='dist/zluda-windows-final';
|
||||
HipRoot='.tools/therock-10.2.0a20260929/_rocm_sdk_core';
|
||||
KernelKRoot='build/native-k-gfx1201'; KernelMRoot='build/native-m-gfx1201'}
|
||||
HipRoot='.tools/therock-10.2.0a20260929/_rocm_sdk_core'}
|
||||
foreach ($name in $defaults.Keys) {
|
||||
if (!(Get-Variable -Name $name -ValueOnly)) { Set-Variable -Name $name -Value (Join-Path $repo $defaults[$name]) }
|
||||
}
|
||||
$GpuArch=Get-D4RGpuTarget $DiagnosticRoot $GpuArch
|
||||
if ($GpuArch -eq 'gfx1200' -and !$PSBoundParameters.ContainsKey('PackageRoot')) { $PackageRoot += '-gfx1200' }
|
||||
if (!$KernelKRoot) { $KernelKRoot=Join-Path $repo "build/native-k-$GpuArch" }
|
||||
if (!$KernelMRoot) { $KernelMRoot=Join-Path $repo "build/native-m-$GpuArch" }
|
||||
foreach ($directory in @($KernelKRoot,$KernelMRoot)) {
|
||||
$info=Join-Path $directory 'build-info.json'
|
||||
if ((Test-Path -LiteralPath $info) -and (Get-Content -LiteralPath $info -Raw | ConvertFrom-Json).architecture -ne $GpuArch) {
|
||||
throw "Native kernel directory does not target ${GpuArch}: $directory"
|
||||
}
|
||||
}
|
||||
$PackageRoot = [IO.Path]::GetFullPath($PackageRoot)
|
||||
if (Test-Path -LiteralPath (Join-Path $PackageRoot 'gpu-target.json')) {
|
||||
[void](Get-D4RGpuTarget $PackageRoot $GpuArch)
|
||||
}
|
||||
$sourceCommit=(& git -C $repo rev-parse HEAD).Trim()
|
||||
$sourceDirty=[bool](& git -C $repo status --porcelain)
|
||||
if ($ArchivePath -and $sourceDirty) { throw 'Commit source changes before creating the distributable archive.' }
|
||||
@@ -21,6 +35,7 @@ New-Item -ItemType Directory -Force $PackageRoot | Out-Null
|
||||
$stagedFiles = [Collections.Generic.List[string]]::new()
|
||||
function Stage([string]$source, [string]$relative) {
|
||||
if (!(Test-Path -LiteralPath $source -PathType Leaf)) { throw "Required file missing: $source" }
|
||||
if ($source.EndsWith('.hsaco')) { Assert-D4RCodeObjectTarget $source $GpuArch }
|
||||
$destination = Join-Path $PackageRoot $relative
|
||||
New-Item -ItemType Directory -Force (Split-Path $destination) | Out-Null
|
||||
Copy-Item -LiteralPath $source -Destination $destination -Force
|
||||
@@ -28,7 +43,7 @@ function Stage([string]$source, [string]$relative) {
|
||||
}
|
||||
Stage (Join-Path $OptiScalerRoot 'OptiScaler.dll') 'OptiScaler.dll'
|
||||
Stage (Join-Path $DiagnosticRoot 'bin/d4r_nvngx.dll') 'd4r/_nvngx.dll'
|
||||
Stage (Join-Path $DiagnosticRoot 'bin/pixel_convert_gfx1201.hsaco') 'd4r/pixel_convert_gfx1201.hsaco'
|
||||
Stage (Join-Path $DiagnosticRoot "bin/pixel_convert_${GpuArch}.hsaco") "d4r/pixel_convert_${GpuArch}.hsaco"
|
||||
Stage (Join-Path $DiagnosticRoot 'nvapi-compat/nvapi64.dll') 'd4r/nvapi/nvapi64.dll'
|
||||
Stage (Join-Path $DiagnosticRoot 'bin/d4r_debug_launcher.exe') 'd4r/d4r_debug_launcher.exe'
|
||||
foreach ($file in @('nvcuda.dll','nvapi64.dll','libc++.dll','libunwind.dll')) { Stage (Join-Path $ZludaRoot $file) "d4r/zluda/$file" }
|
||||
@@ -46,14 +61,26 @@ foreach ($directory in @($KernelKRoot, $KernelMRoot)) {
|
||||
}
|
||||
}
|
||||
if ($entries.Count -ne 16) { throw "Expected 11 K and 5 M native objects; found $($entries.Count)" }
|
||||
@('# Audited DLSS 310.9.1: K + M gfx1201 native kernels', '# Generated by package-windows-game.ps1') + @($entries.Values | Sort-Object) |
|
||||
@("# Audited DLSS 310.9.1: K + M $GpuArch native kernels", '# Generated by package-windows-game.ps1') + @($entries.Values | Sort-Object) |
|
||||
Set-Content -LiteralPath (Join-Path $PackageRoot 'd4r/native/d4r-kernels.txt') -Encoding ASCII
|
||||
$stagedFiles.Add('d4r/native/d4r-kernels.txt')
|
||||
Stage (Join-Path $repo 'scripts/windows/windows-game.ps1') 'windows-game.ps1'
|
||||
Stage (Join-Path $repo 'scripts/windows/gpu-target.ps1') 'gpu-target.ps1'
|
||||
if (Test-Path -LiteralPath (Join-Path $DiagnosticRoot 'gpu-target.json')) {
|
||||
Stage (Join-Path $DiagnosticRoot 'gpu-target.json') 'gpu-target.json'
|
||||
} else {
|
||||
# Builds predating target metadata were gfx1201 only. The resolver rejects
|
||||
# any attempt to label such a build gfx1200.
|
||||
@{schema=1; architecture=$GpuArch; wavefrontSize=32; validation='Legacy gfx1201 build; run the GPU diagnostics.'} |
|
||||
ConvertTo-Json | Set-Content -LiteralPath (Join-Path $PackageRoot 'gpu-target.json') -Encoding UTF8
|
||||
$stagedFiles.Add('gpu-target.json')
|
||||
}
|
||||
Stage (Join-Path $repo 'scripts/windows/capture-game-window.ps1') 'capture-game-window.ps1'
|
||||
Stage (Join-Path $repo 'scripts/windows/send-game-key.ps1') 'send-game-key.ps1'
|
||||
Stage (Join-Path $repo 'docs/windows-game.md') 'README.md'
|
||||
Stage (Join-Path $repo 'docs/windows-performance.md') 'windows-performance.md'
|
||||
Stage (Join-Path $repo 'docs/windows-gpu-support.md') 'windows-gpu-support.md'
|
||||
Stage (Join-Path $repo 'docs/windows-rdna4-port.md') 'windows-rdna4-port.md'
|
||||
Stage (Join-Path $repo 'LICENSE') 'licenses/d4r.txt'
|
||||
Stage (Join-Path $OptiScalerRoot 'GPL-3.0.txt') 'licenses/OptiScaler.txt'
|
||||
Stage (Join-Path $repo 'external/ZLUDA/LICENSE-MIT') 'licenses/ZLUDA-MIT.txt'
|
||||
@@ -70,7 +97,7 @@ New-Item -ItemType Directory -Force (Split-Path $sourceArchive) | Out-Null
|
||||
if ($LASTEXITCODE) { throw 'Cannot archive the committed d4r source.' }
|
||||
Stage $sourceArchive 'source/d4r-windows-rdna4-source.zip'
|
||||
$metadata = [ordered]@{
|
||||
architecture='gfx1201'; nativeKernels=16; dlssSha256='3975567b8943c53acce397f2b72380092f84f162d00b0d2c7d08a1025c563983';
|
||||
architecture=$GpuArch; hardwareValidation=$(if ($GpuArch -eq 'gfx1201') { 'RX9070XT tested; other SKUs unverified' } else { 'compile-only; hardware unverified' }); nativeKernels=16; dlssSha256='3975567b8943c53acce397f2b72380092f84f162d00b0d2c7d08a1025c563983';
|
||||
d4rCommit=$sourceCommit;
|
||||
d4rWorkingTreeDirty=$sourceDirty;
|
||||
optiScaler=(Get-Content -LiteralPath (Join-Path $OptiScalerRoot 'build-info.json') -Raw | ConvertFrom-Json);
|
||||
|
||||
@@ -6,13 +6,15 @@ param(
|
||||
[switch]$Translated
|
||||
)
|
||||
$ErrorActionPreference='Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo=[IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$ZludaRoot) { $ZludaRoot=Join-Path $repo 'dist/zluda-windows-final' }
|
||||
if (!$PackageRoot) { $PackageRoot=Join-Path $repo 'dist/windows-rdna4-command-list' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$OutputDirectory) { $OutputDirectory=Join-Path $repo ('test-results/profile-' + $Preset + '-' + (Get-Date -Format 'yyyyMMdd-HHmmss-fff')) }
|
||||
$settings=@{D4R_ZLUDA_PROFILE='1'; D4R_PROFILE_STAGES='1'; D4R_VALIDATE_OUTPUT='1';
|
||||
D4R_ZLUDA_PROFILE_DEFERRED=$null; D4R_ZLUDA_PROFILE_ALLOW_LEGACY=$null; D4R_ZLUDA_PROFILE_EVERY=$null;
|
||||
D4R_ZLUDA_NATIVE_DIR=$(if ($Translated) { $null } elseif ($Preset -eq 11) { Join-Path $repo 'build/native-k-gfx1201' } else { Join-Path $repo 'build/native-m-gfx1201' })}
|
||||
D4R_ZLUDA_NATIVE_DIR=$(if ($Translated) { $null } elseif ($Preset -eq 11) { Join-Path $repo "build/native-k-$GpuArch" } else { Join-Path $repo "build/native-m-$GpuArch" })}
|
||||
$old=@{}
|
||||
try {
|
||||
foreach ($key in $settings.Keys) { $old[$key]=[Environment]::GetEnvironmentVariable($key,'Process'); [Environment]::SetEnvironmentVariable($key,$settings[$key],'Process') }
|
||||
|
||||
@@ -1,24 +1,30 @@
|
||||
[CmdletBinding()]
|
||||
param([Parameter(Mandatory=$true)][string]$DlssDll, [string]$PackageRoot, [string]$OutputDirectory)
|
||||
param([Parameter(Mandatory=$true)][string]$DlssDll, [string]$PackageRoot, [string]$OutputDirectory, [ValidateSet('gfx1200','gfx1201')][string]$GpuArch)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo 'build/native-k-gfx1201' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $GpuArch
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo "build/native-k-$GpuArch" }
|
||||
if (![IO.Path]::IsPathRooted($DlssDll) -or !(Test-Path -LiteralPath $DlssDll)) {
|
||||
throw 'Supply the absolute path to your local nvngx_dlss.dll.'
|
||||
}
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
foreach ($object in Get-ChildItem -LiteralPath $OutputDirectory -Filter '*.hsaco') {
|
||||
Assert-D4RCodeObjectTarget $object.FullName $GpuArch
|
||||
}
|
||||
$layers = @('enc0','enc1','enc2','enc3','enc4','dec5','dec4','dec3','dec2','dec1','dec0')
|
||||
foreach ($layer in $layers) {
|
||||
$name = "dltss_pwin_${layer}_layer"
|
||||
$source = Join-Path $PackageRoot "experimental/k/${name}_gfx1201.hsaco"
|
||||
$source = Join-Path $PackageRoot "experimental/k/${name}_${GpuArch}.hsaco"
|
||||
if (!(Test-Path -LiteralPath $source)) { throw "Native K module missing: $source" }
|
||||
Assert-D4RCodeObjectTarget $source $GpuArch
|
||||
Copy-Item -LiteralPath $source -Destination (Join-Path $OutputDirectory "$name.hsaco") -Force
|
||||
}
|
||||
& python (Join-Path $repo 'kernels/tools/kernel_manifest.py') $OutputDirectory $DlssDll
|
||||
if ($LASTEXITCODE) { throw "Native manifest generation failed ($LASTEXITCODE)" }
|
||||
@{architecture='gfx1201'; family='K'; dlssSha256=(Get-FileHash -LiteralPath $DlssDll -Algorithm SHA256).Hash;
|
||||
validation='synthetic layers passed; actual DLSS capture/reference validation required';
|
||||
@{architecture=$GpuArch; family='K'; dlssSha256=(Get-FileHash -LiteralPath $DlssDll -Algorithm SHA256).Hash;
|
||||
validation='staged only; GPU execution and DLSS capture/reference validation required';
|
||||
modules=@(Get-ChildItem -LiteralPath $OutputDirectory -Filter '*.hsaco' | Get-FileHash -Algorithm SHA256)} |
|
||||
ConvertTo-Json -Depth 5 | Set-Content -LiteralPath (Join-Path $OutputDirectory 'build-info.json') -Encoding UTF8
|
||||
Write-Host "Experimental native K directory: $OutputDirectory"
|
||||
@@ -1,24 +1,30 @@
|
||||
[CmdletBinding()]
|
||||
param([Parameter(Mandatory=$true)][string]$DlssDll, [string]$PackageRoot, [string]$OutputDirectory)
|
||||
param([Parameter(Mandatory=$true)][string]$DlssDll, [string]$PackageRoot, [string]$OutputDirectory, [ValidateSet('gfx1200','gfx1201')][string]$GpuArch)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo 'build/native-m-gfx1201' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $GpuArch
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo "build/native-m-$GpuArch" }
|
||||
if (![IO.Path]::IsPathRooted($DlssDll) -or !(Test-Path -LiteralPath $DlssDll)) {
|
||||
throw 'Supply the absolute path to your local nvngx_dlss.dll.'
|
||||
}
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
foreach ($object in Get-ChildItem -LiteralPath $OutputDirectory -Filter '*.hsaco') {
|
||||
Assert-D4RCodeObjectTarget $object.FullName $GpuArch
|
||||
}
|
||||
foreach ($layer in @('enc1','enc2','enc3_tube','dec2','dec1')) {
|
||||
$name = "rrlite_${layer}_4x4"
|
||||
$source = Join-Path $PackageRoot "experimental/m/${name}_gfx1201.hsaco"
|
||||
$source = Join-Path $PackageRoot "experimental/m/${name}_${GpuArch}.hsaco"
|
||||
if (!(Test-Path -LiteralPath $source)) { throw "Native M module missing: $source" }
|
||||
Assert-D4RCodeObjectTarget $source $GpuArch
|
||||
Copy-Item -LiteralPath $source -Destination (Join-Path $OutputDirectory "$name.hsaco") -Force
|
||||
}
|
||||
& python (Join-Path $repo 'kernels/tools/kernel_manifest.py') $OutputDirectory $DlssDll
|
||||
if ($LASTEXITCODE) { throw "Native manifest generation failed ($LASTEXITCODE)" }
|
||||
@{architecture='gfx1201'; family='M'; numerics='SWIN_EXACT; SWIN_EXACT_PV; FP16 widening; native FP8 disabled';
|
||||
@{architecture=$GpuArch; family='M'; numerics='SWIN_EXACT; SWIN_EXACT_PV; FP16 widening; native FP8 disabled';
|
||||
dlssSha256=(Get-FileHash -LiteralPath $DlssDll -Algorithm SHA256).Hash;
|
||||
validation='synthetic layers passed; actual DLSS capture/reference validation required';
|
||||
validation='staged only; GPU execution and DLSS capture/reference validation required';
|
||||
modules=@(Get-ChildItem -LiteralPath $OutputDirectory -Filter '*.hsaco' | Get-FileHash -Algorithm SHA256)} |
|
||||
ConvertTo-Json -Depth 5 | Set-Content -LiteralPath (Join-Path $OutputDirectory 'build-info.json') -Encoding UTF8
|
||||
Write-Host "Experimental native M directory: $OutputDirectory"
|
||||
@@ -11,12 +11,14 @@ param(
|
||||
[string]$OutputDirectory
|
||||
)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$NgxCore) { $NgxCore = Join-Path $repo '_nvngx.dll' }
|
||||
if (!$DlssDll) { $DlssDll = Join-Path $repo 'nvngx_dlss.dll' }
|
||||
if (!$NativeRoot) { $NativeRoot = Join-Path $repo 'build/native-k-gfx1201' }
|
||||
if (!$ZludaRoot) { $ZludaRoot = Join-Path $repo 'dist/zluda-windows-api-profile' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-command-list' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$NativeRoot) { $NativeRoot = Join-Path $repo "build/native-k-$GpuArch" }
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/async-k-' + (Get-Date -Format 'yyyyMMdd-HHmmss-fff')) }
|
||||
if ($Resolution -notmatch '^([0-9]+)x([0-9]+)$') { throw 'Resolution must be WIDTHxHEIGHT' }
|
||||
$width=$Matches[1]; $height=$Matches[2]
|
||||
@@ -71,7 +73,7 @@ try {
|
||||
Tee-Object -FilePath (Join-Path $OutputDirectory "$mode-comparison.log")
|
||||
if ($LASTEXITCODE) { throw "Async $mode output mismatch" }
|
||||
}
|
||||
@{passed=$true; architecture='gfx1201'; preset=11; resolution=$Resolution;
|
||||
@{passed=$true; architecture=$GpuArch; preset=11; resolution=$Resolution;
|
||||
inputResolution=$InputResolution;
|
||||
modes=@('burst','recreate'); framesPerMode=3; exactRgb=$true; gpuBoundaryTiming=[bool]$ProfileGpuBoundary;
|
||||
deferredKernelTiming=[bool]$ProfileKernelsDeferred; batchInputs=[bool]$BatchInputCopies; pixelProfile=$PixelProfile} |
|
||||
|
||||
@@ -1,9 +1,11 @@
|
||||
[CmdletBinding()]
|
||||
param([string]$ZludaRoot, [string]$PackageRoot, [string]$HipRoot, [string]$OutputDirectory)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$ZludaRoot) { $ZludaRoot = Join-Path $repo 'dist/zluda-windows-deferred-profile' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-command-list' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$HipRoot) { $HipRoot = Join-Path $repo '.tools/therock-10.2.0a20260929/_rocm_sdk_core' }
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/deferred-profile-' + (Get-Date -Format 'yyyyMMdd-HHmmss-fff')) }
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
@@ -64,7 +66,7 @@ try {
|
||||
if ($exitCode) { throw "$name failed ($exitCode); logs: $OutputDirectory" }
|
||||
$out=Get-Content -LiteralPath $stdout -Raw
|
||||
$err=Get-Content -LiteralPath $stderr -Raw
|
||||
if ($out -notmatch 'PASS CUDA architecture=gfx1201' -or $out -notmatch 'guard_verified=1') { throw "$name did not verify PTX output" }
|
||||
if ($out -notmatch "PASS CUDA architecture=$GpuArch" -or $out -notmatch 'guard_verified=1') { throw "$name did not verify PTX output" }
|
||||
$records=[regex]::Matches($err, '(?m)^D4R_KERNEL_PROFILE kernel="d4r_ptx_pattern" [^\r\n]*serializing=0 deferred=1 [^\r\n]+')
|
||||
$invalid=[regex]::Matches($err, '(?m)^D4R_KERNEL_PROFILE_INVALID kernel="d4r_ptx_pattern" [^\r\n]*reason=invalid_elapsed_time serializing=0 deferred=1 [^\r\n]+')
|
||||
if ($records.Count + $invalid.Count -ne $iterations -or $err -match 'serializing=1|D4R_KERNEL_PROFILE_SKIPPED') {
|
||||
@@ -89,11 +91,11 @@ try {
|
||||
$arguments=@('--hip-root', ('"' + $HipRoot + '"'), '--cuda-dll', ('"' + (Join-Path $ZludaRoot 'nvcuda.dll') + '"'), '--iterations', '3')
|
||||
$exitCode=RunProbe $arguments $stdout $stderr
|
||||
$out=Get-Content -LiteralPath $stdout -Raw; $err=Get-Content -LiteralPath $stderr -Raw
|
||||
if ($exitCode -or $out -notmatch 'PASS CUDA architecture=gfx1201' -or
|
||||
if ($exitCode -or $out -notmatch "PASS CUDA architecture=$GpuArch" -or
|
||||
$err -notmatch 'reason=legacy_default_stream' -or $err -match '(?m)^D4R_KERNEL_PROFILE ') {
|
||||
throw 'Default-stream diagnostics did not safely skip timing events.'
|
||||
}
|
||||
@{passed=$true; architecture='gfx1201'; cases=$cases; defaultStreamSkipped=$true; addedCompletionWaits=0} |
|
||||
@{passed=$true; architecture=$GpuArch; cases=$cases; defaultStreamSkipped=$true; addedCompletionWaits=0} |
|
||||
ConvertTo-Json -Depth 5 | Set-Content -LiteralPath (Join-Path $OutputDirectory 'validation.json') -Encoding UTF8
|
||||
Write-Host "PASS deferred kernel profiling: $OutputDirectory"
|
||||
} finally {
|
||||
|
||||
@@ -0,0 +1,23 @@
|
||||
[CmdletBinding()]
|
||||
param([Parameter(Mandatory=$true)][string]$PackageRoot, [Parameter(Mandatory=$true)][string]$CodeObject)
|
||||
$ErrorActionPreference='Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$target=Get-D4RGpuTarget $PackageRoot $null
|
||||
$other=if ($target -eq 'gfx1201') { 'gfx1200' } else { 'gfx1201' }
|
||||
function Expect-Rejection([scriptblock]$Action) {
|
||||
$rejected=$false
|
||||
try { & $Action | Out-Null } catch { $rejected=$true }
|
||||
if (!$rejected) { throw 'GPU target contract accepted incompatible input.' }
|
||||
}
|
||||
if ((Get-D4RGpuTarget $PackageRoot $target) -ne $target) { throw 'Incorrect package target.' }
|
||||
Assert-D4RCodeObjectTarget $CodeObject $target
|
||||
Expect-Rejection { Get-D4RGpuTarget $PackageRoot $other }
|
||||
Expect-Rejection { Get-D4RGpuTarget $PackageRoot 'gfx1250' }
|
||||
Expect-Rejection { Assert-D4RCodeObjectTarget $CodeObject $other }
|
||||
Expect-Rejection { Assert-D4RCodeObjectTarget (Join-Path $PackageRoot 'gpu-target.json') $target }
|
||||
# Old packages had no target metadata and were gfx1201 only.
|
||||
$legacy=Join-Path $PackageRoot '__no-target-metadata__'
|
||||
if (Test-Path -LiteralPath (Join-Path $legacy 'gpu-target.json')) { throw 'Unexpected legacy test fixture.' }
|
||||
if ((Get-D4RGpuTarget $legacy $null) -ne 'gfx1201') { throw 'Legacy target changed.' }
|
||||
Expect-Rejection { Get-D4RGpuTarget $legacy 'gfx1200' }
|
||||
Write-Host "PASS GPU_TARGET_SCRIPT target=$target rejected=5 hardware_executed=0"
|
||||
@@ -3,9 +3,11 @@ param([Parameter(Mandatory=$true)][string]$CaptureDirectory, [string]$HipRoot,
|
||||
[string]$PackageRoot, [string]$OutputDirectory, [int]$Windows = 12,
|
||||
[Parameter(Mandatory=$true)][string]$DlssDll, [int]$PtxWindows = 2)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$HipRoot) { $HipRoot = Join-Path $repo '.tools/therock-10.2.0a20260929/_rocm_sdk_core' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/k-real-reference-' + (Get-Date -Format 'yyyyMMdd-HHmmss')) }
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
$previousPython = $env:PYTHONPATH
|
||||
@@ -23,7 +25,7 @@ try {
|
||||
New-Item -ItemType Directory -Force $folder | Out-Null
|
||||
$env:D4R_DIAG_DIR = $folder
|
||||
& (Join-Path $PackageRoot 'bin/d4r_native_replay_probe.exe') --hip-root $HipRoot `
|
||||
--module (Join-Path $PackageRoot "experimental/k/${name}_gfx1201.hsaco") `
|
||||
--module (Join-Path $PackageRoot "experimental/k/${name}_${GpuArch}.hsaco") `
|
||||
--fixture-dir $capture.FullName --output-dir $output --iterations 1 `
|
||||
> (Join-Path $folder 'gpu.stdout.log') 2> (Join-Path $folder 'gpu.stderr.log')
|
||||
if ($LASTEXITCODE) { throw "K $layer actual GPU replay failed: $folder" }
|
||||
|
||||
@@ -7,15 +7,17 @@ param(
|
||||
[int]$Iterations = 32
|
||||
)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$HipRoot) { $HipRoot = Join-Path $repo '.tools/therock-10.2.0a20260929/_rocm_sdk_core' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/k-replay-' + (Get-Date -Format 'yyyyMMdd-HHmmss')) }
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
$previousPython = $env:PYTHONPATH
|
||||
$previousDiag = $env:D4R_DIAG_DIR
|
||||
$env:PYTHONPATH = "$(Join-Path $repo '.tools/python/vendor');$previousPython"
|
||||
$summary = [ordered]@{utc=[DateTime]::UtcNow.ToString('o'); hipRoot=$HipRoot; layers=@(); passed=$false}
|
||||
$summary = [ordered]@{utc=[DateTime]::UtcNow.ToString('o'); hipRoot=$HipRoot; architecture=$GpuArch; layers=@(); passed=$false}
|
||||
try {
|
||||
if ($Iterations -lt 1 -or $Iterations -gt 10000) { throw 'Iterations must be 1..10000' }
|
||||
foreach ($layer in $Layers) {
|
||||
@@ -29,7 +31,7 @@ try {
|
||||
> (Join-Path $folder 'fixture.stdout.log') 2> (Join-Path $folder 'fixture.stderr.log')
|
||||
if ($LASTEXITCODE) { throw "K $layer fixture generation failed; see $folder" }
|
||||
& (Join-Path $PackageRoot 'bin/d4r_native_replay_probe.exe') --hip-root $HipRoot `
|
||||
--module (Join-Path $PackageRoot "experimental/k/dltss_pwin_${layer}_layer_gfx1201.hsaco") `
|
||||
--module (Join-Path $PackageRoot "experimental/k/dltss_pwin_${layer}_layer_${GpuArch}.hsaco") `
|
||||
--fixture-dir $fixture --output-dir $output --iterations $Iterations `
|
||||
> (Join-Path $folder 'gpu.stdout.log') 2> (Join-Path $folder 'gpu.stderr.log')
|
||||
if ($LASTEXITCODE) { throw "K $layer GPU replay failed; see $folder" }
|
||||
|
||||
@@ -3,9 +3,11 @@ param([Parameter(Mandatory=$true)][string]$CaptureDirectory, [string]$HipRoot,
|
||||
[string]$PackageRoot, [string]$OutputDirectory, [int]$Blocks = 12,
|
||||
[Parameter(Mandatory=$true)][string]$DlssDll, [int]$PtxBlocks = 2)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$HipRoot) { $HipRoot = Join-Path $repo '.tools/therock-10.2.0a20260929/_rocm_sdk_core' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/m-real-reference-' + (Get-Date -Format 'yyyyMMdd-HHmmss')) }
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
$previousPython=$env:PYTHONPATH; $previousDiag=$env:D4R_DIAG_DIR
|
||||
@@ -22,7 +24,7 @@ try {
|
||||
New-Item -ItemType Directory -Force $folder | Out-Null
|
||||
$env:D4R_DIAG_DIR = $folder
|
||||
& (Join-Path $PackageRoot 'bin/d4r_native_replay_probe.exe') --hip-root $HipRoot `
|
||||
--module (Join-Path $PackageRoot "experimental/m/${name}_gfx1201.hsaco") `
|
||||
--module (Join-Path $PackageRoot "experimental/m/${name}_${GpuArch}.hsaco") `
|
||||
--fixture-dir $capture.FullName --output-dir $output --iterations 1 `
|
||||
> (Join-Path $folder 'gpu.stdout.log') 2> (Join-Path $folder 'gpu.stderr.log')
|
||||
if ($LASTEXITCODE) { throw "M $layer actual GPU replay failed: $folder" }
|
||||
|
||||
@@ -3,15 +3,17 @@ param([string]$HipRoot, [string]$PackageRoot, [string]$OutputDirectory,
|
||||
[string[]]$Layers = @('enc1','enc2','enc3_tube','dec2','dec1'),
|
||||
[ValidateRange(1,10000)][int]$Iterations = 32)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$HipRoot) { $HipRoot = Join-Path $repo '.tools/therock-10.2.0a20260929/_rocm_sdk_core' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/m-replay-' + (Get-Date -Format 'yyyyMMdd-HHmmss')) }
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
$previousPython = $env:PYTHONPATH
|
||||
$previousDiag = $env:D4R_DIAG_DIR
|
||||
$env:PYTHONPATH = "$(Join-Path $repo '.tools/python/vendor');$previousPython"
|
||||
$summary = [ordered]@{utc=[DateTime]::UtcNow.ToString('o'); hipRoot=$HipRoot; layers=@(); passed=$false}
|
||||
$summary = [ordered]@{utc=[DateTime]::UtcNow.ToString('o'); hipRoot=$HipRoot; architecture=$GpuArch; layers=@(); passed=$false}
|
||||
try {
|
||||
foreach ($layer in $Layers) {
|
||||
if ($layer -notmatch '^(enc[12]|enc3_tube|dec[12])$') { throw "Invalid M layer $layer" }
|
||||
@@ -24,7 +26,7 @@ try {
|
||||
> (Join-Path $folder 'fixture.stdout.log') 2> (Join-Path $folder 'fixture.stderr.log')
|
||||
if ($LASTEXITCODE) { throw "M $layer fixture generation failed: $folder" }
|
||||
& (Join-Path $PackageRoot 'bin/d4r_native_replay_probe.exe') --hip-root $HipRoot `
|
||||
--module (Join-Path $PackageRoot "experimental/m/rrlite_${layer}_4x4_gfx1201.hsaco") `
|
||||
--module (Join-Path $PackageRoot "experimental/m/rrlite_${layer}_4x4_${GpuArch}.hsaco") `
|
||||
--fixture-dir $fixture --output-dir $output --iterations $Iterations `
|
||||
> (Join-Path $folder 'gpu.stdout.log') 2> (Join-Path $folder 'gpu.stderr.log')
|
||||
if ($LASTEXITCODE) { throw "M $layer GPU replay failed: $folder" }
|
||||
|
||||
@@ -1,9 +1,11 @@
|
||||
[CmdletBinding()]
|
||||
param([string]$HipRoot, [string]$PackageRoot, [string]$ZludaRoot, [string]$OutputDirectory)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$HipRoot) { $HipRoot = Join-Path $repo '.tools/therock-10.2.0a20260929/_rocm_sdk_core' }
|
||||
if (!$PackageRoot) { $PackageRoot = Join-Path $repo 'dist/windows-rdna4-therock' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$ZludaRoot) { $ZludaRoot = Join-Path $repo 'dist/zluda-windows-final' }
|
||||
if (!$OutputDirectory) { $OutputDirectory = Join-Path $repo ('test-results/native-identity-' + (Get-Date -Format 'yyyyMMdd-HHmmss')) }
|
||||
New-Item -ItemType Directory -Force $OutputDirectory | Out-Null
|
||||
@@ -29,7 +31,7 @@ try {
|
||||
$good = Join-Path $OutputDirectory 'matching-identity'
|
||||
New-Item -ItemType Directory -Force $bad,$good | Out-Null
|
||||
foreach ($directory in @($bad,$good)) {
|
||||
Copy-Item -LiteralPath (Join-Path $PackageRoot 'bin/probe_gfx1201.hsaco') -Destination (Join-Path $directory 'd4r_ptx_pattern.hsaco') -Force
|
||||
Copy-Item -LiteralPath (Join-Path $PackageRoot "bin/probe_${GpuArch}.hsaco") -Destination (Join-Path $directory 'd4r_ptx_pattern.hsaco') -Force
|
||||
}
|
||||
'd4r_ptx_pattern 0000000000000000' | Set-Content -LiteralPath (Join-Path $bad 'd4r-kernels.txt') -Encoding ASCII
|
||||
$env:D4R_ZLUDA_NATIVE_DIR = $bad
|
||||
|
||||
@@ -1,27 +1,30 @@
|
||||
[CmdletBinding()]
|
||||
param([string]$TextureRoot, [string]$ControlTextureRoot, [string]$CommonTextureRoot, [string]$ZludaRoot,
|
||||
param([string]$TextureRoot, [string]$ControlTextureRoot, [string]$CommonTextureRoot, [string]$ZludaRoot, [string]$PackageRoot,
|
||||
[string]$OutputDirectory, [int]$Iterations=8, [string]$OutputResolution='512x288', [string]$InputResolution,
|
||||
[switch]$CandidateFirst,
|
||||
[ValidateSet('output','input')][string]$KernelStage='output')
|
||||
$ErrorActionPreference='Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repo=[IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if (!$TextureRoot) { $TextureRoot=Join-Path $repo 'build/private-textures-gfx1201' }
|
||||
if (!$PackageRoot) { $PackageRoot=Join-Path $repo 'dist/windows-rdna4-command-list' }
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $null
|
||||
if (!$TextureRoot) { $TextureRoot=Join-Path $repo "build/private-textures-$GpuArch" }
|
||||
if (!$ZludaRoot) { $ZludaRoot=Join-Path $repo 'dist/zluda-windows-final' }
|
||||
if (!$OutputDirectory) { $OutputDirectory=Join-Path $repo 'test-results/k-native-texture-reference' }
|
||||
if ($OutputResolution -notmatch '^([0-9]+)x([0-9]+)$') { throw 'OutputResolution must be WIDTHxHEIGHT.' }
|
||||
$outputWidth=[int]$Matches[1]; $outputHeight=[int]$Matches[2]
|
||||
$combined=Join-Path $OutputDirectory 'private-native'
|
||||
New-Item -ItemType Directory -Force $combined | Out-Null
|
||||
foreach ($directory in @((Join-Path $repo 'build/native-k-gfx1201'),$CommonTextureRoot,$TextureRoot) | Where-Object { $_ }) {
|
||||
foreach ($directory in @((Join-Path $repo "build/native-k-$GpuArch"),$CommonTextureRoot,$TextureRoot) | Where-Object { $_ }) {
|
||||
Get-ChildItem -LiteralPath $directory -Filter '*.hsaco' | Copy-Item -Destination $combined -Force
|
||||
}
|
||||
& python (Join-Path $repo 'kernels/tools/kernel_manifest.py') $combined (Join-Path $repo 'nvngx_dlss.dll')
|
||||
if ($LASTEXITCODE) { throw 'Combined private manifest failed.' }
|
||||
$controlNative=Join-Path $repo 'build/native-k-gfx1201'
|
||||
$controlNative=Join-Path $repo "build/native-k-$GpuArch"
|
||||
if ($ControlTextureRoot -or $CommonTextureRoot) {
|
||||
$controlNative=Join-Path $OutputDirectory 'control-private-native'
|
||||
New-Item -ItemType Directory -Force $controlNative | Out-Null
|
||||
foreach ($directory in @((Join-Path $repo 'build/native-k-gfx1201'),$CommonTextureRoot,$ControlTextureRoot) | Where-Object { $_ }) {
|
||||
foreach ($directory in @((Join-Path $repo "build/native-k-$GpuArch"),$CommonTextureRoot,$ControlTextureRoot) | Where-Object { $_ }) {
|
||||
Get-ChildItem -LiteralPath $directory -Filter '*.hsaco' | Copy-Item -Destination $controlNative -Force
|
||||
}
|
||||
& python (Join-Path $repo 'kernels/tools/kernel_manifest.py') $controlNative (Join-Path $repo 'nvngx_dlss.dll')
|
||||
@@ -37,6 +40,7 @@ $stagedObjects=@{}
|
||||
foreach ($stage in @(@{name='control'; root=$controlNative},@{name='candidate'; root=$combined})) {
|
||||
$stagedObjects[$stage.name]=@{}
|
||||
foreach ($object in Get-ChildItem -LiteralPath $stage.root -Filter '*.hsaco') {
|
||||
Assert-D4RCodeObjectTarget $object.FullName $GpuArch
|
||||
$stagedObjects[$stage.name][$object.Name]=(Get-FileHash -LiteralPath $object.FullName -Algorithm SHA256).Hash.ToLowerInvariant()
|
||||
}
|
||||
}
|
||||
@@ -52,7 +56,7 @@ try {
|
||||
$expectedBackend=if (Test-Path -LiteralPath (Join-Path $env:D4R_ZLUDA_NATIVE_DIR ($kernel + '.hsaco'))) { 'native' } else { 'translated' }
|
||||
if ($mode -eq 'candidate' -and $expectedBackend -ne 'native') { throw "Missing candidate object: $kernel" }
|
||||
$arguments=@{RuntimeProfile='therock'; ZludaRoot=$ZludaRoot;
|
||||
PackageRoot=(Join-Path $repo 'dist/windows-rdna4-command-list'); NgxCore=(Join-Path $repo '_nvngx.dll');
|
||||
PackageRoot=$PackageRoot; NgxCore=(Join-Path $repo '_nvngx.dll');
|
||||
DlssDll=(Join-Path $repo 'nvngx_dlss.dll'); NgxMode='d3d12'; NgxAbi='project-legacy'; Preset=11;
|
||||
NgxCreateFlags=$flags; NgxOutputResolution=$OutputResolution; NgxOnly=$true; RequireNativeNetwork=$true; CommandListBackend=$true;
|
||||
NgxInputResolution=$InputResolution;
|
||||
@@ -76,7 +80,7 @@ try {
|
||||
& python (Join-Path $repo 'tools/windows/frame_compare.py') --reference $control --actual $candidate --width $outputWidth --height $outputHeight --exact
|
||||
if ($LASTEXITCODE) { throw "Texture flags=$flags full-frame RGB mismatch; preserve baseline." }
|
||||
}
|
||||
@{passed=$true; architecture='gfx1201'; framesPerVariant=$Iterations; outputResolution=$OutputResolution; createFlags=@(0,11); strictRgb=$true;
|
||||
@{passed=$true; architecture=$GpuArch; framesPerVariant=$Iterations; outputResolution=$OutputResolution; createFlags=@(0,11); strictRgb=$true;
|
||||
inputResolution=$InputResolution;
|
||||
serializingProfile=$true; candidateFirst=[bool]$CandidateFirst; kernelStage=$KernelStage;
|
||||
controlTextureRoot=$ControlTextureRoot; commonTextureRoot=$CommonTextureRoot;
|
||||
|
||||
@@ -4,6 +4,7 @@ param(
|
||||
[ValidateSet('stable', 'therock')][string]$RuntimeProfile = 'stable',
|
||||
[string]$ZludaRoot,
|
||||
[string]$PackageRoot,
|
||||
[ValidateSet('gfx1200','gfx1201')][string]$GpuArch,
|
||||
[string]$OutputDirectory,
|
||||
[string]$NgxCore = $env:D4R_NGX_CORE,
|
||||
[string]$DlssDll = $env:D4R_DLSS_DLL,
|
||||
@@ -26,6 +27,7 @@ param(
|
||||
[int]$TimeoutSeconds = 180
|
||||
)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
$repoRoot = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../..'))
|
||||
if ($RuntimeProfile -eq 'therock') {
|
||||
# Works both from scripts/windows and the installed dist/<profile> directory.
|
||||
@@ -40,6 +42,7 @@ if (!$PackageRoot) {
|
||||
else { $PackageRoot = [IO.Path]::GetFullPath((Join-Path $PSScriptRoot '../../dist/windows-rdna4-diagnostics')) }
|
||||
}
|
||||
$PackageRoot = [IO.Path]::GetFullPath($PackageRoot)
|
||||
$GpuArch=Get-D4RGpuTarget $PackageRoot $GpuArch
|
||||
if (!$HipRoot) { $HipRoot = 'C:\Program Files\AMD\ROCm\7.2' }
|
||||
$HipRoot = [IO.Path]::GetFullPath($HipRoot)
|
||||
if (!$ZludaRoot) {
|
||||
@@ -68,7 +71,7 @@ foreach ($setting in @('D4R_ZLUDA_WMMA','D4R_ZLUDA_WMMA_FP8','D4R_ZLUDA_WMMA_FP8
|
||||
}
|
||||
$exitStatus = 1
|
||||
$ngxRuntimeDirectory = $null
|
||||
$summary = [ordered]@{utc=[DateTime]::UtcNow.ToString('o'); profile=$RuntimeProfile; hipRoot=$HipRoot; zludaRoot=$ZludaRoot; tests=@()}
|
||||
$summary = [ordered]@{utc=[DateTime]::UtcNow.ToString('o'); profile=$RuntimeProfile; architecture=$GpuArch; hipRoot=$HipRoot; zludaRoot=$ZludaRoot; tests=@()}
|
||||
function Quote-Argument([string]$Value) {
|
||||
# All generated paths are absolute file/directory paths, with no trailing backslash.
|
||||
if ($Value.Contains('"')) { throw 'Double quotes are not allowed in arguments' }
|
||||
@@ -162,10 +165,10 @@ try {
|
||||
$comgr = Join-Path $HipRoot 'bin/amd_comgr_3.dll'
|
||||
if (!(Test-Path -LiteralPath $comgr)) { $comgr = Join-Path $HipRoot 'bin/amd_comgr.dll' }
|
||||
$files = @((Join-Path $HipRoot 'bin/amdhip64_7.dll'), $comgr,
|
||||
(Join-Path $ZludaRoot 'nvcuda.dll'), (Join-Path $PackageRoot 'bin/probe_gfx1201.hsaco'),
|
||||
(Join-Path $PackageRoot 'bin/wmma_gfx1201.hsaco'),
|
||||
(Join-Path $PackageRoot 'experimental/k/dltss_pwin_enc1_layer_gfx1201.hsaco'),
|
||||
(Join-Path $PackageRoot 'experimental/k/dltss_pwin_enc2_layer_gfx1201.hsaco'))
|
||||
(Join-Path $ZludaRoot 'nvcuda.dll'), (Join-Path $PackageRoot "bin/probe_${GpuArch}.hsaco"),
|
||||
(Join-Path $PackageRoot "bin/wmma_${GpuArch}.hsaco"),
|
||||
(Join-Path $PackageRoot "experimental/k/dltss_pwin_enc1_layer_${GpuArch}.hsaco"),
|
||||
(Join-Path $PackageRoot "experimental/k/dltss_pwin_enc2_layer_${GpuArch}.hsaco"))
|
||||
if ($OptiScalerDll) {
|
||||
if ($NgxMode -ne 'd3d12' -or !$CommandListBackend) { throw '-OptiScalerDll requires D3D12 and -CommandListBackend' }
|
||||
if (![IO.Path]::IsPathRooted($OptiScalerDll)) { throw '-OptiScalerDll must be absolute' }
|
||||
@@ -192,12 +195,12 @@ try {
|
||||
$bin = Join-Path $PackageRoot 'bin'
|
||||
$hipOk = $true
|
||||
if (!$NgxOnly) { $hipOk = Invoke-Probe 'hip' (Join-Path $bin 'd4r_hip_gfx1201_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin 'probe_gfx1201.hsaco'), '--iterations', "$Iterations")
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin "probe_${GpuArch}.hsaco"), '--iterations', "$Iterations")
|
||||
}
|
||||
if ($hipOk) {
|
||||
if (!$NgxOnly) {
|
||||
$wmmaOk = Invoke-Probe 'gfx12-wmma' (Join-Path $bin 'd4r_gfx12_wmma_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin 'wmma_gfx1201.hsaco'), '--iterations', "$Iterations")
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin "wmma_${GpuArch}.hsaco"), '--iterations', "$Iterations")
|
||||
if (!$wmmaOk) { throw 'gfx12 WMMA layout validation failed; refusing to mark K/M readiness.' }
|
||||
$pythonCommand = Get-Command python -ErrorAction SilentlyContinue
|
||||
if (!$pythonCommand) { throw 'Python 3.11+ with NumPy 2.4.6 is needed for K reference checks; run setup-windows-tools.ps1.' }
|
||||
@@ -205,7 +208,7 @@ try {
|
||||
$kFixture = Join-Path $OutputDirectory "k-$layer-fixture"
|
||||
$kModuleOk = Invoke-Probe "k-$layer-gpu" (Join-Path $bin 'd4r_k_module_probe.exe') @(
|
||||
'--hip-root', $HipRoot,
|
||||
'--module', (Join-Path $PackageRoot "experimental/k/dltss_pwin_${layer}_layer_gfx1201.hsaco"),
|
||||
'--module', (Join-Path $PackageRoot "experimental/k/dltss_pwin_${layer}_layer_${GpuArch}.hsaco"),
|
||||
'--kernel-name', $layer, '--fixture-dir', $kFixture, '--iterations', "$Iterations")
|
||||
if (!$kModuleOk) { throw "K $layer GPU execution failed; see the k-$layer-gpu logs." }
|
||||
$kReferenceOk = Invoke-Probe "k-$layer-reference" $pythonCommand.Source @(
|
||||
@@ -247,19 +250,19 @@ try {
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin 'd4r_nvngx.dll'), '--interop-mode', 'indirect', '--iterations', "$Iterations")
|
||||
if (!$commandOk) { throw 'D3D12 command-list order/state/lifetime validation failed.' }
|
||||
$pixelsOk = Invoke-Probe 'd3d12-gpu-pixel-formats' (Join-Path $bin 'd4r_d3d12_pixel_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin 'pixel_convert_gfx1201.hsaco'), '--iterations', '2')
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin "pixel_convert_${GpuArch}.hsaco"), '--iterations', '2')
|
||||
if (!$pixelsOk) { throw 'D3D12 typed SRV/UAV format conversion validation failed.' }
|
||||
$mapOk = Invoke-Probe 'interop-map-lifetime' (Join-Path $bin 'd4r_d3d12_hip_interop_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--interop-mode', 'map', '--iterations', '64')
|
||||
$interopOk = Invoke-Probe 'interop-roundtrip' (Join-Path $bin 'd4r_d3d12_hip_interop_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin 'probe_gfx1201.hsaco'), '--iterations', "$Iterations")
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin "probe_${GpuArch}.hsaco"), '--iterations', "$Iterations")
|
||||
if (!$mapOk -or !$interopOk) {
|
||||
foreach ($mode in @('resource', 'import')) {
|
||||
Invoke-Probe "interop-$mode-lifetime" (Join-Path $bin 'd4r_d3d12_hip_interop_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--interop-mode', $mode, '--iterations', '32') | Out-Null
|
||||
}
|
||||
Invoke-Probe 'hip-stream-lifetime' (Join-Path $bin 'd4r_hip_stream_lifecycle_probe.exe') @(
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin 'probe_gfx1201.hsaco'), '--iterations', '32') | Out-Null
|
||||
'--hip-root', $HipRoot, '--module', (Join-Path $bin "probe_${GpuArch}.hsaco"), '--iterations', '32') | Out-Null
|
||||
}
|
||||
}
|
||||
if ($mapOk -and $interopOk) { $exitStatus = 0 }
|
||||
@@ -288,12 +291,12 @@ try {
|
||||
# alongside the private user-supplied DLLs for this test.
|
||||
$localShim = Join-Path $ngxRuntimeDirectory 'd4r_nvngx.dll'
|
||||
Copy-Item -LiteralPath (Join-Path $bin 'd4r_nvngx.dll') -Destination $localShim
|
||||
Copy-Item -LiteralPath (Join-Path $bin 'pixel_convert_gfx1201.hsaco') -Destination $ngxRuntimeDirectory
|
||||
Copy-Item -LiteralPath (Join-Path $bin "pixel_convert_${GpuArch}.hsaco") -Destination $ngxRuntimeDirectory
|
||||
if ($OptiScalerDll) {
|
||||
$bridgeDirectory = Join-Path $ngxRuntimeDirectory 'd4r'
|
||||
New-Item -ItemType Directory -Path $bridgeDirectory | Out-Null
|
||||
Copy-Item -LiteralPath $localShim -Destination (Join-Path $bridgeDirectory '_nvngx.dll')
|
||||
Copy-Item -LiteralPath (Join-Path $bin 'pixel_convert_gfx1201.hsaco') -Destination $bridgeDirectory
|
||||
Copy-Item -LiteralPath (Join-Path $bin "pixel_convert_${GpuArch}.hsaco") -Destination $bridgeDirectory
|
||||
$localShim = Join-Path $ngxRuntimeDirectory 'OptiScaler.dll'
|
||||
Copy-Item -LiteralPath $OptiScalerDll -Destination $localShim
|
||||
$ini = @"
|
||||
|
||||
@@ -26,6 +26,7 @@ param(
|
||||
[string]$OutputDirectory
|
||||
)
|
||||
$ErrorActionPreference = 'Stop'
|
||||
. (Join-Path $PSScriptRoot 'gpu-target.ps1')
|
||||
if ($ProfileGpuBoundary -and !$AsyncInterop) { throw '-ProfileGpuBoundary requires -AsyncInterop.' }
|
||||
if ($ProfileKernels -and $ProfileKernelsDeferred) { throw 'Choose either serializing or deferred kernel profiling.' }
|
||||
if ($ProfileLegacyStream -and !$ProfileKernelsDeferred) { throw '-ProfileLegacyStream requires -ProfileKernelsDeferred.' }
|
||||
@@ -67,7 +68,9 @@ if (!$NgxCore -or !$DlssDll) { throw 'Pass both local NVIDIA DLL paths with -Ngx
|
||||
$NgxCore = (Get-Item -LiteralPath $NgxCore).FullName
|
||||
$DlssDll = (Get-Item -LiteralPath $DlssDll).FullName
|
||||
$metadata = Get-Content -LiteralPath (Join-Path $package 'package.json') -Raw | ConvertFrom-Json
|
||||
$GpuArch=Get-D4RGpuTarget $package $metadata.architecture
|
||||
foreach ($file in $metadata.files) {
|
||||
if ($file.path.EndsWith('.hsaco')) { Assert-D4RCodeObjectTarget (Join-Path $package $file.path) $GpuArch }
|
||||
if ((Get-FileHash -LiteralPath (Join-Path $package $file.path) -Algorithm SHA256).Hash -ne $file.sha256) { throw "Package file changed; rebuild package: $($file.path)" }
|
||||
}
|
||||
if ((Get-FileHash -LiteralPath $DlssDll -Algorithm SHA256).Hash -ne $metadata.dlssSha256) { throw 'DLSS DLL does not match the validated 310.9.1 native manifest. Rebuild and validate native manifests for this DLL first.' }
|
||||
@@ -75,7 +78,7 @@ $textureFiles=@(); $textureManifest=@()
|
||||
if ($LocalTextureKernels) {
|
||||
$LocalTextureKernels=(Get-Item -LiteralPath $LocalTextureKernels -ErrorAction Stop).FullName
|
||||
$validation=Get-Content -LiteralPath (Join-Path $LocalTextureKernels 'validation.json') -Raw | ConvertFrom-Json
|
||||
if (!$validation.passed -or !$validation.strictRgb -or $validation.architecture -ne 'gfx1201' -or
|
||||
if (!$validation.passed -or !$validation.strictRgb -or $validation.architecture -ne $GpuArch -or
|
||||
$validation.source.dlssSha256 -ne $metadata.dlssSha256 -or !$validation.source.accuracy) {
|
||||
throw 'Local texture kernels must pass test-native-texture.ps1 with the exact packaged DLSS identity.'
|
||||
}
|
||||
@@ -85,6 +88,7 @@ if ($LocalTextureKernels) {
|
||||
}
|
||||
$path=Join-Path $LocalTextureKernels $object.Name
|
||||
if ((Get-FileHash -LiteralPath $path -Algorithm SHA256).Hash.ToLowerInvariant() -ne $object.Value) { throw 'Local texture object changed after validation.' }
|
||||
Assert-D4RCodeObjectTarget $path $GpuArch
|
||||
$textureFiles+=Get-Item -LiteralPath $path
|
||||
}
|
||||
if ($textureFiles.Count -ne 2) { throw 'Expected both validated local K output variants.' }
|
||||
@@ -123,7 +127,12 @@ function InstallFile([string]$source, [string]$relative) {
|
||||
$manifest | ConvertTo-Json -Depth 6 | Set-Content -LiteralPath $manifestPath -Encoding UTF8
|
||||
}
|
||||
InstallFile (Join-Path $package 'OptiScaler.dll') 'dxgi.dll'
|
||||
foreach ($file in Get-ChildItem -LiteralPath (Join-Path $package 'd4r') -Recurse -File) { InstallFile $file.FullName $file.FullName.Substring($package.Length+1) }
|
||||
foreach ($file in $metadata.files) {
|
||||
$relative=$file.path.Replace('/','\')
|
||||
if ($relative.StartsWith('d4r\',[StringComparison]::OrdinalIgnoreCase)) {
|
||||
InstallFile (Join-Path $package $relative) $relative
|
||||
}
|
||||
}
|
||||
if ($LocalTextureKernels) {
|
||||
foreach ($file in $textureFiles) { InstallFile $file.FullName ('d4r/native/' + $file.Name) }
|
||||
$combinedManifest=Join-Path $backup 'generated-native-manifest.txt'
|
||||
@@ -177,7 +186,7 @@ $settings = @{
|
||||
D4R_NVCUDA_DLL=(GamePath 'd4r/zluda/nvcuda.dll'); ZLUDA_CUDA_LIB=(GamePath 'd4r/zluda/nvcuda.dll');
|
||||
D4R_NVAPI_DLL=(GamePath 'd4r/nvapi/nvapi64.dll'); D4R_NVAPI_BACKEND=(GamePath 'd4r/zluda/nvapi64.dll');
|
||||
D4R_NGX_CORE=(GamePath 'd4r/vendor/_nvngx.dll'); D4R_DLSS_DLL=(GamePath 'd4r/vendor/nvngx_dlss.dll');
|
||||
D4R_FORMAT_MODULE=(GamePath 'd4r/pixel_convert_gfx1201.hsaco');
|
||||
D4R_FORMAT_MODULE=(GamePath "d4r/pixel_convert_${GpuArch}.hsaco");
|
||||
D4R_ZLUDA_NATIVE_DIR=(GamePath 'd4r/native'); D4R_D3D12_COMMAND_BACKEND='1'; D4R_ZLUDA_VERBOSE='1';
|
||||
D4R_ZLUDA_WMMA='1'; D4R_ZLUDA_WMMA_FP8='1'; D4R_ZLUDA_WMMA_FP8_NATIVE='0'; D4R_ZLUDA_WMMA_F16_REFERENCE='1';
|
||||
# Do not inherit relaxation flags from unrelated experiments. Unset and
|
||||
|
||||
Reference in new issue
Block a user