Merge branch 'main' into raft-public-npy-helpers

Author: cjnolet

.devcontainer/Dockerfile

@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2023-2025, NVIDIA CORPORATION. All rights reserved.
+# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 # syntax=docker/dockerfile:1.5
@@ -42,7 +42,8 @@ ENV HISTFILE="/home/coder/.cache/._bash_history"
 ###
 # sccache configuration
 ###
-ENV AWS_ROLE_ARN="arn:aws:iam::279114543810:role/nv-gha-token-sccache-devs"
+ENV AWS_IDP_URL="https://token.rapids.nvidia.com"
+ENV AWS_ROLE_ARN="arn:aws:iam::279114543810:role/rapids-token-sccache-devs"
 ENV SCCACHE_REGION="us-east-2"
 ENV SCCACHE_BUCKET="rapids-sccache-devs"
 ENV SCCACHE_S3_USE_PREPROCESSOR_CACHE_MODE=true

cpp/cmake/modules/ConfigureCUDA.cmake

@@ -37,6 +37,10 @@ if(CUDA_LOG_COMPILE_TIME)
   list(APPEND CUVS_CUDA_FLAGS "--time=nvcc_compile_log.csv")
 endif()
 
+# disable warpspeed scan / workaround for CCCL bug https://github.com/NVIDIA/cccl/issues/8838
+list(APPEND CUVS_CXX_FLAGS "-DCCCL_DISABLE_WARPSPEED_SCAN")
+list(APPEND CUVS_CUDA_FLAGS "-DCCCL_DISABLE_WARPSPEED_SCAN")
+
 list(APPEND CUVS_CUDA_FLAGS --expt-extended-lambda --expt-relaxed-constexpr)
 list(APPEND CUVS_CXX_FLAGS "-DCUDA_API_PER_THREAD_DEFAULT_STREAM")
 list(APPEND CUVS_CUDA_FLAGS "-DCUDA_API_PER_THREAD_DEFAULT_STREAM")