From a094eed596891f4feba0e2b15628ca6abd3a3c77 Mon Sep 17 00:00:00 2001
From: Connor Baker <ConnorBaker01@gmail.com>
Date: Tue, 23 Sep 2025 17:11:20 +0000
Subject: [PATCH] cudaPackages.cublasmp: init at 0.6.0

Signed-off-by: Connor Baker <ConnorBaker01@gmail.com>
---
 .../_cuda/manifests/cublasmp/README.md        |  5 +++
 .../manifests/cublasmp/redistrib_0.6.0.json   | 43 +++++++++++++++++++
 .../cuda-modules/packages/libcublasmp.nix     | 38 ++++++++++++++++
 pkgs/top-level/cuda-packages.nix              |  4 ++
 4 files changed, 90 insertions(+)
 create mode 100644 pkgs/development/cuda-modules/_cuda/manifests/cublasmp/README.md
 create mode 100644 pkgs/development/cuda-modules/_cuda/manifests/cublasmp/redistrib_0.6.0.json
 create mode 100644 pkgs/development/cuda-modules/packages/libcublasmp.nix

diff --git a/pkgs/development/cuda-modules/_cuda/manifests/cublasmp/README.md b/pkgs/development/cuda-modules/_cuda/manifests/cublasmp/README.md
new file mode 100644
index 000000000000..de9d7fcd41c2
--- /dev/null
+++ b/pkgs/development/cuda-modules/_cuda/manifests/cublasmp/README.md
@@ -0,0 +1,5 @@
+# cublasmp
+
+Link: <https://developer.download.nvidia.com/compute/cublasmp/redist/>
+
+Requirements: <https://docs.nvidia.com/cuda/cublasmp/getting_started/index.html#hardware-and-software-requirements>
diff --git a/pkgs/development/cuda-modules/_cuda/manifests/cublasmp/redistrib_0.6.0.json b/pkgs/development/cuda-modules/_cuda/manifests/cublasmp/redistrib_0.6.0.json
new file mode 100644
index 000000000000..f3c90fe91791
--- /dev/null
+++ b/pkgs/development/cuda-modules/_cuda/manifests/cublasmp/redistrib_0.6.0.json
@@ -0,0 +1,43 @@
+{
+    "release_date": "2025-09-09",
+    "release_label": "0.6.0",
+    "release_product": "cublasmp",
+    "libcublasmp": {
+        "name": "NVIDIA cuBLASMp library",
+        "license": "cuBLASMp library",
+        "license_path": "libcublasmp/LICENSE.txt",
+        "version": "0.6.0.84",
+        "linux-x86_64": {
+            "cuda12": {
+                "relative_path": "libcublasmp/linux-x86_64/libcublasmp-linux-x86_64-0.6.0.84_cuda12-archive.tar.xz",
+                "sha256": "214a439031cc53be7d02961651e5e6ee520d80ab09b772d5a470e678477a6c57",
+                "md5": "2a6a91fd58b90a16a1c2b3c3e4d2bdce",
+                "size": "4324732"
+            },
+            "cuda13": {
+                "relative_path": "libcublasmp/linux-x86_64/libcublasmp-linux-x86_64-0.6.0.84_cuda13-archive.tar.xz",
+                "sha256": "f3892486ac72649ab5e140fd1466421e5638ce23a56a5360a42f32450fcfbf83",
+                "md5": "b3f96dce5e52f432a36e0a6a006f6b27",
+                "size": "4815848"
+            }
+        },
+        "cuda_variant": [
+            "12",
+            "13"
+        ],
+        "linux-sbsa": {
+            "cuda12": {
+                "relative_path": "libcublasmp/linux-sbsa/libcublasmp-linux-sbsa-0.6.0.84_cuda12-archive.tar.xz",
+                "sha256": "6af07f02ed01eee761509ad5c733a7196520f09ce036d5d047f38a1768287080",
+                "md5": "f6efeba7b2e1ae8b164a69f208c5a53b",
+                "size": "4273376"
+            },
+            "cuda13": {
+                "relative_path": "libcublasmp/linux-sbsa/libcublasmp-linux-sbsa-0.6.0.84_cuda13-archive.tar.xz",
+                "sha256": "9c3ea75c2a2705cb415d37316a6f540dbeb021ac3dc7bf0404dac314eb098aa0",
+                "md5": "35bac35e00eb29a86e54bb4fb703d258",
+                "size": "4751836"
+            }
+        }
+    }
+}
diff --git a/pkgs/development/cuda-modules/packages/libcublasmp.nix b/pkgs/development/cuda-modules/packages/libcublasmp.nix
new file mode 100644
index 000000000000..92130c135a06
--- /dev/null
+++ b/pkgs/development/cuda-modules/packages/libcublasmp.nix
@@ -0,0 +1,38 @@
+{
+  _cuda,
+  buildRedist,
+  libcublas,
+  nvshmem ? null, # TODO(@connorbaker): package this
+}:
+buildRedist {
+  redistName = "cublasmp";
+  pname = "libcublasmp";
+
+  outputs = [
+    "out"
+    "dev"
+    "include"
+    "lib"
+  ];
+
+  # TODO: Looks like the minimum supported capability is 7.0 as of the latest:
+  # https://docs.nvidia.com/cuda/cublasmp/getting_started/index.html
+  buildInputs = [
+    libcublas
+  ];
+
+  platformAssertions = _cuda.lib._mkMissingPackagesAssertions { inherit nvshmem; };
+
+  meta = {
+    description = "High-performance, multi-process, GPU-accelerated library for distributed basic dense linear algebra";
+    longDescription = ''
+      NVIDIA cuBLASMp is a high-performance, multi-process, GPU-accelerated library for distributed basic dense linear
+      algebra.
+
+      cuBLASMp is compatible with 2D block-cyclic data layout and provides PBLAS-like C APIs.
+    '';
+    homepage = "https://docs.nvidia.com/cuda/cublasmp";
+    changelog = "https://docs.nvidia.com/cuda/cublasmp/release_notes";
+    license = _cuda.lib.licenses.math_sdk_sla;
+  };
+}
diff --git a/pkgs/top-level/cuda-packages.nix b/pkgs/top-level/cuda-packages.nix
index 8ca8bbca41c3..7c18919d703e 100644
--- a/pkgs/top-level/cuda-packages.nix
+++ b/pkgs/top-level/cuda-packages.nix
@@ -25,6 +25,7 @@ in
 {
   cudaPackages_12_6 = callPackage ../development/cuda-modules {
     manifests = selectManifests {
+      cublasmp = "0.6.0";
       cuda = "12.6.3";
       cudnn = "9.13.0";
       cusparselt = "0.6.3";
@@ -35,6 +36,7 @@ in
 
   cudaPackages_12_8 = callPackage ../development/cuda-modules {
     manifests = selectManifests {
+      cublasmp = "0.6.0";
       cuda = "12.8.1";
       cudnn = "9.13.0";
       cusparselt = "0.8.1";
@@ -45,6 +47,7 @@ in
 
   cudaPackages_12_9 = callPackage ../development/cuda-modules {
     manifests = selectManifests {
+      cublasmp = "0.6.0";
       cuda = "12.9.1";
       cudnn = "9.13.0";
       cusparselt = "0.8.1";
@@ -55,6 +58,7 @@ in
 
   cudaPackages_13_0 = callPackage ../development/cuda-modules {
     manifests = selectManifests {
+      cublasmp = "0.6.0";
       cuda = "13.0.2";
       cudnn = "9.13.0";
       cusparselt = "0.8.1";