search

bottlerocket-os / bottlerocket

An operating system designed for hosting containers
https://bottlerocket.dev
Other
8.78k stars 519 forks source link

kmod-*-nvidia: switch source for Fabric Manager binaries #4015

Closed yeazelm closed 5 months ago

yeazelm commented 5 months ago

Description of changes: The RPMs vended on the developer portal align with Amazon Linux's consumption of Fabric Manager for AL2023. AL2023 is on different driver versions than Bottlerocket at the moment but this at least moves the build to use the same RPM distributions they consume.

Testing done: Build aws-k8s-1.29-nvidia for x86_64 and aarch64 as well as aws-k8s-1.26-nvidia and validated smoke tests pass:

``` ========================================= Running sample UnifiedMemoryPerf ========================================= GPU Device 0: "Turing" with compute capability 7.5 Running ........................................................ Overall Time For matrixMultiplyPerf Printing Average of 20 measurements in (ms) Size_KB UMhint UMhntAs UMeasy 0Copy MemCopy CpAsync CpHpglk CpPglAs 4 0.178 0.252 0.348 0.018 0.034 0.029 0.037 0.028 16 0.238 0.260 0.525 0.043 0.064 0.053 0.070 0.065 64 0.348 0.358 0.838 0.132 0.177 0.161 0.136 0.124 256 0.883 0.831 1.510 0.748 0.619 0.569 0.484 0.473 1024 3.068 3.032 3.785 5.245 2.491 2.321 1.938 1.986 4096 11.928 10.882 14.221 35.107 9.305 9.238 8.692 8.505 16384 57.204 55.393 65.832 275.677 49.163 48.778 46.061 45.991 NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled. ========================================= Running sample deviceQuery ========================================= ./deviceQuery Starting... CUDA Device Query (Runtime API) version (CUDART static linking) Detected 1 CUDA Capable device(s) Device 0: "Tesla T4" CUDA Driver Version / Runtime Version 12.2 / 11.4 CUDA Capability Major/Minor version number: 7.5 Total amount of global memory: 14931 MBytes (15655829504 bytes) (040) Multiprocessors, (064) CUDA Cores/MP: 2560 CUDA Cores GPU Max Clock rate: 1590 MHz (1.59 GHz) Memory Clock rate: 5001 Mhz Memory Bus Width: 256-bit L2 Cache Size: 4194304 bytes Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384) Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers Total amount of constant memory: 65536 bytes Total amount of shared memory per block: 49152 bytes Total shared memory per multiprocessor: 65536 bytes Total number of registers available per block: 65536 Warp size: 32 Maximum number of threads per multiprocessor: 1024 Maximum number of threads per block: 1024 Max dimension size of a thread block (x,y,z): (1024, 1024, 64) Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535) Maximum memory pitch: 2147483647 bytes Texture alignment: 512 bytes Concurrent copy and kernel execution: Yes with 3 copy engine(s) Run time limit on kernels: No Integrated GPU sharing Host Memory: No Support host page-locked memory mapping: Yes Alignment requirement for Surfaces: Yes Device has ECC support: Enabled Device supports Unified Addressing (UVA): Yes Device supports Managed Memory: Yes Device supports Compute Preemption: Yes Supports Cooperative Kernel Launch: Yes Supports MultiDevice Co-op Kernel Launch: Yes Device PCI Domain ID / Bus ID / location ID: 0 / 0 / 30 Compute Mode: < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) > deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 12.2, CUDA Runtime Version = 11.4, NumDevs = 1 Result = PASS ========================================= Running sample globalToShmemAsyncCopy ========================================= [globalToShmemAsyncCopy] - Starting... GPU Device 0: "Turing" with compute capability 7.5 MatrixA(1280,1280), MatrixB(1280,1280) Running kernel = 0 - AsyncCopyMultiStageLargeChunk Computing result using CUDA Kernel... done Performance= 336.59 GFlop/s, Time= 12.461 msec, Size= 4194304000 Ops, WorkgroupSize= 256 threads/block Checking computed result for correctness: Result = PASS Initializing... GPU Device 0: "Turing" with compute capability 7.5 M: 4096 (16 x 256) N: 4096 (16 x 256) K: 4096 (16 x 256) Preparing data for GPU... Required shared memory size: 64 Kb Computing... using high performance kernel compute_gemm_imma Time: 4.223904 ms TOPS: 32.54 ========================================= Running sample reductionMultiBlockCG ========================================= reductionMultiBlockCG Starting... GPU Device 0: "Turing" with compute capability 7.5 33554432 elements numThreads: 1024 numBlocks: 40 Launching SinglePass Multi Block Cooperative Groups kernel Average time: 1.037900 ms Bandwidth: 129.316572 GB/s GPU result = 1.992401361465 CPU result = 1.992401361465 ========================================= Running sample shfl_scan ========================================= Starting shfl_scan GPU Device 0: "Turing" with compute capability 7.5 > Detected Compute SM 7.5 hardware with 40 multi-processors Starting shfl_scan GPU Device 0: "Turing" with compute capability 7.5 > Detected Compute SM 7.5 hardware with 40 multi-processors Computing Simple Sum test --------------------------------------------------- Initialize test data [1, 1, 1...] Scan summation for 65536 elements, 256 partial sums Partial summing 256 elements with 1 blocks of size 256 Test Sum: 65536 Time (ms): 0.026592 65536 elements scanned in 0.026592 ms -> 2464.500732 MegaElements/s CPU verify result diff (GPUvsCPU) = 0 CPU sum (naive) took 0.030940 ms Computing Integral Image Test on size 1920 x 1080 synthetic data --------------------------------------------------- Method: Fast Time (GPU Timer): 0.051200 ms Diff = 0 Method: Vertical Scan Time (GPU Timer): 0.127936 ms CheckSum: 2073600, (expect 1920x1080=2073600) ========================================= Running sample simpleAWBarrier ========================================= ./simpleAWBarrier starting... GPU Device 0: "Turing" with compute capability 7.5 Launching normVecByDotProductAWBarrier kernel with numBlocks = 40 blockSize = 1024 Result = PASSED ./simpleAWBarrier completed, returned OK ========================================= Running sample simpleAtomicIntrinsics ========================================= simpleAtomicIntrinsics starting... GPU Device 0: "Turing" with compute capability 7.5 Processing time: 126.327003 (ms) simpleAtomicIntrinsics completed, returned OK ========================================= Running sample simpleVoteIntrinsics ========================================= [simpleVoteIntrinsics] GPU Device 0: "Turing" with compute capability 7.5 > GPU device has 40 Multi-Processors, SM 7.5 compute capabilities [VOTE Kernel Test 1/3] Running <> kernel1 ... OK [VOTE Kernel Test 2/3] Running <> kernel2 ... OK [VOTE Kernel Test 3/3] Running <> kernel3 ... OK Shutting down... ========================================= Running sample vectorAdd ========================================= [Vector addition of 50000 elements] Copy input data from the host memory to the CUDA device CUDA kernel launch with 196 blocks of 256 threads Copy output data from the CUDA device to the host memory Test PASSED Done ========================================= Running sample warpAggregatedAtomicsCG ========================================= GPU Device 0: "Turing" with compute capability 7.5 CPU max matches GPU max Warp Aggregated Atomics PASSED ```

Terms of contribution:

By submitting this pull request, I agree that this contribution is dual-licensed under the terms of both the Apache License, version 2.0, and the MIT license.