Merge branch 'master' into dk/triangular

JuliaGPU · Jul 17, 2023 · 94e1e3d · 94e1e3d
2 parents 5e891d3 + 583329d
commit 94e1e3d
Show file tree

Hide file tree

Showing 103 changed files with 911 additions and 636 deletions.
diff --git a/.buildkite/pipeline.yml b/.buildkite/pipeline.yml
@@ -26,7 +26,8 @@ steps:
               # use the CUDA installation from the CI image
               using CUDA
               CUDA.set_runtime_version!("local")'
-        if: build.message !~ /\[skip tests\]/
+        if: build.message !~ /\[skip tests\]/ &&
+            build.message !~ /\[skip julia\]/
         timeout_in_minutes: 120
         matrix:
           setup:
@@ -35,26 +36,23 @@ steps:
               - "1.7"
               - "1.8"
               - "1.9"
-              - "nightly"
-          adjustments:
-            - with:
-                julia: "nightly"
-              soft_fail: true
+          #     - "nightly"
+          # adjustments:
+          #   - with:
+          #       julia: "nightly"
+          #     soft_fail: true
 
   # then, test supported CUDA toolkits (installed through the artifact system)
   - group: "CUDA"
     key: "cuda"
     depends_on: "julia"
     steps:
-      # NOTE: we support those CUDA versions for which the latest cuDNN is available
-      #       https://developer.nvidia.com/rdp/cudnn-archive
-
       - label: "CUDA {{matrix.cuda}}"
         plugins:
           - JuliaCI/julia#v1:
               version: 1.8
           - JuliaCI/julia-test#v1:
-              test_args: "--thorough"
+              test_args: "core base libraries"
           - JuliaCI/julia-coverage#v1:
               dirs:
                 - src
@@ -63,7 +61,9 @@ steps:
         agents:
           queue: "juliagpu"
           cuda: "*"
-        if: build.message !~ /\[skip tests\]/ && !build.pull_request.draft
+        if: build.message !~ /\[skip tests\]/ &&
+            build.message !~ /\[skip cuda\]/ &&
+            !build.pull_request.draft
         timeout_in_minutes: 120
         matrix:
           setup:
@@ -99,6 +99,9 @@ steps:
             - with:
                 cuda: "12.0"
                 package: "cuDNN"
+            - with:
+                cuda: "12.0"
+                package: "cuTENSOR"
         plugins:
           - JuliaCI/julia#v1:
               version: "1.8"
@@ -110,7 +113,9 @@ steps:
         agents:
           queue: "juliagpu"
           cuda: "*"
-        if: build.message !~ /\[skip tests\]/ && !build.pull_request.draft
+        if: build.message !~ /\[skip tests\]/ &&
+            build.message !~ /\[skip subpackages\]/ &&
+            !build.pull_request.draft
         timeout_in_minutes: 120
         commands: |
           julia -e '
@@ -161,7 +166,9 @@ steps:
         agents:
           queue: "juliagpu"
           cuda: "*"
-        if: build.message !~ /\[skip tests\]/ && !build.pull_request.draft
+        if: build.message !~ /\[skip tests\]/ &&
+            build.message !~ /\[skip downstream\]/ &&
+            !build.pull_request.draft
         timeout_in_minutes: 60
 
   - group: ":eyes: Special"
@@ -212,31 +219,30 @@ steps:
         agents:
           queue: "juliagpu"
           cuda: "*"
-        if: build.message !~ /\[skip docs\]/ && !build.pull_request.draft
+        if: build.message !~ /\[skip docs\]/ &&
+            !build.pull_request.draft
         timeout_in_minutes: 30
 
-      # XXX: fails often, and is very slow
-      #- label: "Compute sanitizer"
-      #  plugins:
-      #    - JuliaCI/julia#v1:
-      #        version: 1.8
-      #    - JuliaCI/julia-test#v1:
-      #        julia_args: "-g2"
-      #        test_args: "--sanitize --quickfail --jobs=2"
-      #    - JuliaCI/julia-coverage#v1:
-      #        dirs:
-      #          - src
-      #          - lib
-      #          - examples
-      #  agents:
-      #    queue: "juliagpu"
-      #    cuda: "*"
-      #  env:
-      #    JULIA_CUDA_USE_COMPAT: 'false'  # NVIDIA bug #3418723: injection tools prevent probing libcuda
-      #  commands: |
-      #    julia --project -e 'using CUDA; CUDA.set_runtime_version!(v"11.6")'
-      #  if: build.message !~ /\[skip tests\]/
-      #  timeout_in_minutes: 240
+      - label: "Compute sanitizer"
+        plugins:
+          - JuliaCI/julia#v1:
+              version: 1.8
+          - JuliaCI/julia-test#v1:
+              test_args: "--sanitize core"
+          - JuliaCI/julia-coverage#v1:
+              dirs:
+                - src
+                - lib
+                - examples
+        agents:
+          queue: "juliagpu"
+          cuda: "*"
+        env:
+          JULIA_CUDA_USE_COMPAT: 'false'  # NVIDIA bug #3418723: injection tools prevent probing libcuda
+        if: build.message !~ /\[skip tests\]/ &&
+            build.message !~ /\[skip sanitizer\]/ &&
+            !build.pull_request.draft
+        timeout_in_minutes: 10
 
   # we want to benchmark every commit on the master branch, even if it failed CI
   - wait: ~

diff --git a/Manifest.toml b/Manifest.toml
@@ -2,9 +2,9 @@
 
 [[AbstractFFTs]]
 deps = ["ChainRulesCore", "LinearAlgebra"]
-git-tree-sha1 = "16b6dbc4cf7caee4e1e75c49485ec67b667098a0"
+git-tree-sha1 = "cad4c758c0038eea30394b1b671526921ca85b21"
 uuid = "621f4979-c628-5d54-868e-fcf4e3e8185c"
-version = "1.3.1"
+version = "1.4.0"
 
 [[Adapt]]
 deps = ["LinearAlgebra", "Requires"]
@@ -63,16 +63,16 @@ uuid = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
 version = "1.16.0"
 
 [[ChangesOfVariables]]
-deps = ["LinearAlgebra", "Test"]
-git-tree-sha1 = "f84967c4497e0e1955f9a582c232b02847c5f589"
+deps = ["InverseFunctions", "LinearAlgebra", "Test"]
+git-tree-sha1 = "2fba81a302a7be671aefe194f0525ef231104e7f"
 uuid = "9e997f8a-9a97-42d5-a9f1-ce6bfc15e2c0"
-version = "0.1.7"
+version = "0.1.8"
 
 [[Compat]]
 deps = ["Dates", "LinearAlgebra", "UUIDs"]
-git-tree-sha1 = "7a60c856b9fa189eb34f5f8a6f6b5529b7942957"
+git-tree-sha1 = "4e88377ae7ebeaf29a047aa1ee40826e0b708a5d"
 uuid = "34da2185-b29b-5c13-b0c7-acf172513d20"
-version = "4.6.1"
+version = "4.7.0"
 
 [[CompilerSupportLibraries_jll]]
 deps = ["Artifacts", "Libdl"]
@@ -99,9 +99,9 @@ version = "0.1.9"
 
 [[GPUArrays]]
 deps = ["Adapt", "GPUArraysCore", "LLVM", "LinearAlgebra", "Printf", "Random", "Reexport", "Serialization", "Statistics"]
-git-tree-sha1 = "a3351bc577a6b49297248aadc23a4add1097c2ac"
+git-tree-sha1 = "2e57b4a4f9cc15e85a24d603256fe08e527f48d1"
 uuid = "0c68f7d7-f131-5f86-a1c3-88cf8149b2d7"
-version = "8.7.1"
+version = "8.8.1"
 
 [[GPUArraysCore]]
 deps = ["Adapt"]
@@ -111,19 +111,19 @@ version = "0.1.5"
 
 [[GPUCompiler]]
 deps = ["ExprTools", "InteractiveUtils", "LLVM", "Libdl", "Logging", "Scratch", "TimerOutputs", "UUIDs"]
-git-tree-sha1 = "cb090aea21c6ca78d59672a7e7d13bd56d09de64"
+git-tree-sha1 = "72b2e3c2ba583d1a7aa35129e56cf92e07c083e3"
 uuid = "61eb1bfa-7361-4325-ad38-22787b887f55"
-version = "0.20.3"
+version = "0.21.4"
 
 [[InteractiveUtils]]
 deps = ["Markdown"]
 uuid = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
 
 [[InverseFunctions]]
 deps = ["Test"]
-git-tree-sha1 = "6667aadd1cdee2c6cd068128b3d226ebc4fb0c67"
+git-tree-sha1 = "eabe3125edba5c9c10b60a160b1779a000dc8b29"
 uuid = "3587e190-3f89-42d0-90ee-14403ec27112"
-version = "0.1.9"
+version = "0.1.11"
 
 [[IrrationalConstants]]
 git-tree-sha1 = "630b497eafcc20001bba38a4651b327dcfc491d2"
@@ -137,22 +137,22 @@ uuid = "692b3bcd-3c85-4b1f-b108-f13ce0eb3210"
 version = "1.4.1"
 
 [[KernelAbstractions]]
-deps = ["Adapt", "Atomix", "InteractiveUtils", "LinearAlgebra", "MacroTools", "PrecompileTools", "SparseArrays", "StaticArrays", "UUIDs", "UnsafeAtomics", "UnsafeAtomicsLLVM"]
-git-tree-sha1 = "47be64f040a7ece575c2b5f53ca6da7b548d69f4"
+deps = ["Adapt", "Atomix", "InteractiveUtils", "LinearAlgebra", "MacroTools", "PrecompileTools", "Requires", "SparseArrays", "StaticArrays", "UUIDs", "UnsafeAtomics", "UnsafeAtomicsLLVM"]
+git-tree-sha1 = "6d08ca80b621635fed9cdfeb9a4280a574020bf3"
 uuid = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
-version = "0.9.4"
+version = "0.9.7"
 
 [[LLVM]]
 deps = ["CEnum", "LLVMExtra_jll", "Libdl", "Printf", "Unicode"]
-git-tree-sha1 = "26a31cdd9f1f4ea74f649a7bf249703c687a953d"
+git-tree-sha1 = "8695a49bfe05a2dc0feeefd06b4ca6361a018729"
 uuid = "929cbde3-209d-540e-8aea-75f648917ca0"
-version = "5.1.0"
+version = "6.1.0"
 
 [[LLVMExtra_jll]]
 deps = ["Artifacts", "JLLWrappers", "LazyArtifacts", "Libdl", "TOML"]
-git-tree-sha1 = "09b7505cc0b1cee87e5d4a26eea61d2e1b0dcd35"
+git-tree-sha1 = "c35203c1e1002747da220ffc3c0762ce7754b08c"
 uuid = "dad2f222-ce93-54a1-a47d-0025e8a3acab"
-version = "0.0.21+0"
+version = "0.0.23+0"
 
 [[LazyArtifacts]]
 deps = ["Artifacts", "Pkg"]
@@ -292,20 +292,20 @@ uuid = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 
 [[SpecialFunctions]]
 deps = ["ChainRulesCore", "IrrationalConstants", "LogExpFunctions", "OpenLibm_jll", "OpenSpecFun_jll"]
-git-tree-sha1 = "ef28127915f4229c971eb43f3fc075dd3fe91880"
+git-tree-sha1 = "7beb031cf8145577fbccacd94b8a8f4ce78428d3"
 uuid = "276daf66-3868-5448-9aa4-cd146d93841b"
-version = "2.2.0"
+version = "2.3.0"
 
 [[StaticArrays]]
 deps = ["LinearAlgebra", "Random", "StaticArraysCore", "Statistics"]
-git-tree-sha1 = "8982b3607a212b070a5e46eea83eb62b4744ae12"
+git-tree-sha1 = "fffc14c695c17bfdbfa92a2a01836cdc542a1e46"
 uuid = "90137ffa-7385-5640-81b9-e52037218182"
-version = "1.5.25"
+version = "1.6.1"
 
 [[StaticArraysCore]]
-git-tree-sha1 = "6b7ba252635a5eff6a0b0664a41ee140a1c9e72a"
+git-tree-sha1 = "1d5708d926c76a505052d0d24a846d5da08bc3a4"
 uuid = "1e83bf80-4336-4d27-bf5d-d5a4f845583c"
-version = "1.4.0"
+version = "1.4.1"
 
 [[Statistics]]
 deps = ["LinearAlgebra", "SparseArrays"]
@@ -343,9 +343,9 @@ version = "0.2.1"
 
 [[UnsafeAtomicsLLVM]]
 deps = ["LLVM", "UnsafeAtomics"]
-git-tree-sha1 = "ea37e6066bf194ab78f4e747f5245261f17a7175"
+git-tree-sha1 = "323e3d0acf5e78a56dfae7bd8928c989b4f3083e"
 uuid = "d80eeb9a-aca5-4d75-85e5-170c8b632249"
-version = "0.1.2"
+version = "0.1.3"
 
 [[Zlib_jll]]
 deps = ["Libdl"]

diff --git a/Project.toml b/Project.toml
@@ -1,6 +1,6 @@
 name = "CUDA"
 uuid = "052768ef-5323-5732-b1bb-66c8b64840ba"
-version = "4.3.2"
+version = "4.4.0"
 
 [deps]
 AbstractFFTs = "621f4979-c628-5d54-868e-fcf4e3e8185c"
@@ -10,7 +10,6 @@ CEnum = "fa961155-64e5-5f13-b03f-caf6b980ea82"
 CUDA_Driver_jll = "4ee394cb-3365-5eb0-8335-949819d2adfc"
 CUDA_Runtime_Discovery = "1af6417a-86b4-443c-805f-a4643ffb695f"
 CUDA_Runtime_jll = "76a88914-d11a-5bdc-97e0-2f5a05c973a2"
-CompilerSupportLibraries_jll = "e66e0078-7015-5450-92f7-15fbd957f2ae"
 ExprTools = "e2ba6199-217a-4e67-a87a-7c52f15ade04"
 GPUArrays = "0c68f7d7-f131-5f86-a1c3-88cf8149b2d7"
 GPUCompiler = "61eb1bfa-7361-4325-ad38-22787b887f55"
@@ -41,9 +40,9 @@ CUDA_Runtime_Discovery = "0.2"
 CUDA_Runtime_jll = "0.6"
 ExprTools = "0.1"
 GPUArrays = "8.6"
-GPUCompiler = "0.20"
+GPUCompiler = "0.21"
 KernelAbstractions = "0.9.2"
-LLVM = "5"
+LLVM = "6"
 Preferences = "1"
 Random123 = "1.2"
 RandomNumbers = "1.5.3"

diff --git a/lib/cublas/CUBLAS.jl b/lib/cublas/CUBLAS.jl
@@ -152,7 +152,7 @@ function xt_handle()
         finalizer(current_task()) do task
             push!(idle_xt_handles, cuda.context, new_handle) do
                 # TODO: which context do we need to destroy this on?
-                cublasXtDestroy(handle)
+                cublasXtDestroy(new_handle)
             end
         end
 

diff --git a/lib/cublas/linalg.jl b/lib/cublas/linalg.jl
@@ -190,7 +190,7 @@ function LinearAlgebra.generic_matvecmul!(Y::CuVector, tA::AbstractChar, A::Stri
     end
 
     T = eltype(Y)
-    alpha, beta = promote(_add.alpha, _add.beta, zero(T))
+    alpha, beta = _add.alpha, _add.beta
     if alpha isa Union{Bool,T} && beta isa Union{Bool,T}
         if T <: CublasFloat && eltype(A) == eltype(B) == T
             if tA in ('N', 'T', 'C')
@@ -281,7 +281,7 @@ end # VERSION
 
 function LinearAlgebra.generic_matmatmul!(C::CuVecOrMat, tA, tB, A::StridedCuVecOrMat, B::StridedCuVecOrMat, _add::MulAddMul)
     T = eltype(C)
-    alpha, beta = promote(_add.alpha, _add.beta, zero(T))
+    alpha, beta = _add.alpha, _add.beta
     mA, nA = size(A, tA == 'N' ? 1 : 2), size(A, tA == 'N' ? 2 : 1)
     mB, nB = size(B, tB == 'N' ? 1 : 2), size(B, tB == 'N' ? 2 : 1)