Skip to content

Commit 2d63652

Browse files
yangpanMSyangpanMS
authored
Rollback driver change for MLPerf (#326)
* Revert "Fixing Nvidia Drivers to latest (#312)" This reverts commit 20b6c8d. * Revert "Fix unit test break for mlperf (#321)" This reverts commit 152a360.
1 parent f7f224a commit 2d63652

File tree

11 files changed

+23
-28
lines changed

11 files changed

+23
-28
lines changed

‎.pipelines/azure-pipelines-linux.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -16,7 +16,7 @@ resources:
1616
options: --entrypoint=""
1717

1818
variables:
19-
VcVersion : 1.14.33
19+
VcVersion : 1.14.34
2020
ROOT: $(Build.SourcesDirectory)
2121
CDP_DEFINITION_BUILD_COUNT: $[counter('', 0)] # needed for onebranch.pipeline.version task https://aka.ms/obpipelines/versioning
2222
ENABLE_PRS_DELAYSIGN: 1

‎.pipelines/azure-pipelines.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -18,7 +18,7 @@ pool:
1818
vmImage: windows-latest
1919

2020
variables:
21-
VcVersion : 1.14.33
21+
VcVersion : 1.14.34
2222
ROOT: $(Build.SourcesDirectory)
2323
CDP_DEFINITION_BUILD_COUNT: $[counter('', 0)] # needed for onebranch.pipeline.version task https://aka.ms/obpipelines/versioning
2424
ENABLE_PRS_DELAYSIGN: 1

‎src/VirtualClient/VirtualClient.Actions.FunctionalTests/MLPerfProfileTests.cs

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -46,8 +46,8 @@ public async Task MLPerfWorkloadProfileExecutesTheExpectedDependenciesAndReboot(
4646
{
4747
$"sudo apt update",
4848
$"sudo apt install build-essential -yq",
49-
$"sudo wget https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run",
50-
$"sudo sh cuda_12.4.1_550.54.15_linux.run --silent",
49+
$"sudo wget https://developer.download.nvidia.com/compute/cuda/12.0.0/local_installers/cuda_12.0.0_525.60.13_linux.run",
50+
$"sudo sh cuda_12.0.0_525.60.13_linux.run --silent",
5151
$"sudo bash -c \"echo 'export PATH=/usr/local/cuda-11.6/bin${{PATH:+:${{PATH}}}}' | sudo tee -a /home/[a-z]+/.bashrc\"",
5252
$"bash -c \"echo 'export LD_LIBRARY_PATH=/usr/local/cuda-11.6/lib64${{LD_LIBRARY_PATH:+:${{LD_LIBRARY_PATH}}}}' " +
5353
"| sudo tee -a /home/[a-z]+/.bashrc\""

‎src/VirtualClient/VirtualClient.Actions.FunctionalTests/SuperBenchmarkProfileTests.cs

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -46,8 +46,8 @@ public async Task SuperBenchmarkWorkloadProfileExecutesTheExpectedDependenciesAn
4646
{
4747
$"sudo apt update",
4848
$"sudo apt install build-essential -yq",
49-
$"sudo wget https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run",
50-
$"sudo sh cuda_12.4.1_550.54.15_linux.run --silent",
49+
$"sudo wget https://developer.download.nvidia.com/compute/cuda/12.0.0/local_installers/cuda_12.0.0_525.60.13_linux.run",
50+
$"sudo sh cuda_12.0.0_525.60.13_linux.run --silent",
5151
$"sudo bash -c \"echo 'export PATH=/usr/local/cuda-12.0/bin${{PATH:+:${{PATH}}}}' | sudo tee -a /home/[a-z]+/.bashrc\"",
5252
$"bash -c \"echo 'export LD_LIBRARY_PATH=/usr/local/cuda-12.0/lib64${{LD_LIBRARY_PATH:+:${{LD_LIBRARY_PATH}}}}' | " +
5353
$"sudo tee -a /home/[a-z]+/.bashrc\""

‎src/VirtualClient/VirtualClient.Dependencies.UnitTests/CUDAAndNvidiaGPUDriverInstallationTests.cs

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -37,7 +37,7 @@ public class CUDAAndNvidiaGPUDriverInstallationTests
3737
$"sudo tee -a /home/anyuser/.bashrc\"";
3838

3939
private const string UpgradeCommand = "apt upgrade -y";
40-
private const string InstallDriverCommand = "apt install nvidia-driver-510-server nvidia-dkms-510-server -y";
40+
private const string InstallDriverCommand = "apt install nvidia-driver-510 nvidia-dkms-510 -y";
4141
private const string InstallFabricManagerCommand = "apt install cuda-drivers-fabricmanager-510 -y";
4242

4343
private MockFixture fixture;

‎src/VirtualClient/VirtualClient.Dependencies/CUDAAndNvidiaGPUDriverInstallation.cs

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -343,7 +343,7 @@ private List<string> VersionSpecificInstallationCommands(LinuxDistribution linux
343343
case LinuxDistribution.Ubuntu:
344344
commands.Add("apt update");
345345
commands.Add("apt upgrade -y");
346-
commands.Add($"apt install nvidia-driver-{this.LinuxDriverVersion}-server nvidia-dkms-{this.LinuxDriverVersion}-server -y");
346+
commands.Add($"apt install nvidia-driver-{this.LinuxDriverVersion} nvidia-dkms-{this.LinuxDriverVersion} -y");
347347
commands.Add($"apt install cuda-drivers-fabricmanager-{this.LinuxDriverVersion} -y");
348348

349349
break;

‎src/VirtualClient/VirtualClient.Main/profiles/PERF-GPU-MLPERF-NVIDIA.json

Lines changed: 3 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -4,16 +4,15 @@
44
"RecommendedMinimumExecutionTime": "10:00:00",
55
"SupportedPlatforms": "linux-x64",
66
"SupportedOperatingSystems": "Ubuntu",
7-
"RecommendOperationSystem": "Ubunutu-22.04, Default LinusCudaVersion,LinuxDriverVersion,LinuxLocalRunFile support this OS only.",
87
"SupportedRequireCustomSystemSupport": "A100_PCIe_80GBx4",
98
"SpecialRequirements": "This is an Nvidia GPU-specialized workload. It depends upon the system having an Nvidia GPU card/chip."
109
},
1110
"Parameters": {
1211
"Username": "",
1312
"DiskFilter": "SizeGreaterThan:1000gb",
14-
"LinuxCudaVersion": "12.4",
15-
"LinuxDriverVersion": "550",
16-
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run",
13+
"LinuxCudaVersion": "12.0",
14+
"LinuxDriverVersion": "525",
15+
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.0.0/local_installers/cuda_12.0.0_525.60.13_linux.run",
1716
"RequireCustomSystemSupport": true
1817
},
1918
"Actions": [

‎src/VirtualClient/VirtualClient.Main/profiles/PERF-GPU-MLPERF-TRAINING-NVIDIA.json

Lines changed: 3 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -4,17 +4,16 @@
44
"RecommendedMinimumExecutionTime": "10:00:00",
55
"SupportedPlatforms": "linux-x64",
66
"SupportedOperatingSystems": "Ubuntu",
7-
"RecommendOperationSystem": "Ubunutu-22.04, Default LinusCudaVersion,LinuxDriverVersion,LinuxLocalRunFile support this OS only.",
87
"SpecialRequirements": "This is an Nvidia GPU-specialized workload. It depends upon the system having an Nvidia GPU card/chip. Requires an OSDisk of >=256GB size and Data disk of >=8TB size",
98
"SupportedImplementations": "Visit the link https://github.com/mlcommons/training_results_v2.1/tree/main/NVIDIA/benchmarks enter the implementation folder to see supported implementations.",
109
"SupportedConfigFiles": "Visit the link https://github.com/mlcommons/training_results_v2.1/tree/main/NVIDIA/benchmarks/bert/implementations/pytorch-22.09 for each benchmark and implementation you can the supported "
1110
},
1211
"Parameters": {
1312
"Username": null,
1413
"DiskFilter": "BiggestSize",
15-
"LinuxCudaVersion": "12.4",
16-
"LinuxDriverVersion": "550",
17-
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run"
14+
"LinuxCudaVersion": "12.0",
15+
"LinuxDriverVersion": "525",
16+
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.0.0/local_installers/cuda_12.0.0_525.60.13_linux.run"
1817
},
1918
"Actions": [
2019
{

‎src/VirtualClient/VirtualClient.Main/profiles/PERF-GPU-MLPERF.json

Lines changed: 3 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -4,16 +4,15 @@
44
"RecommendedMinimumExecutionTime": "10:00:00",
55
"SupportedPlatforms": "linux-x64",
66
"SupportedOperatingSystems": "Ubuntu",
7-
"RecommendOperationSystem": "Ubunutu-22.04, Default drivers support this OS only.",
87
"SupportedRequireCustomSystemSupport": "A100_PCIe_80GBx4",
98
"SpecialRequirements": "This is an Nvidia GPU-specialized workload. It depends upon the system having an Nvidia GPU card/chip."
109
},
1110
"Parameters": {
1211
"Username": "",
1312
"DiskFilter": "SizeGreaterThan:1000gb",
14-
"LinuxCudaVersion": "12.4",
15-
"LinuxDriverVersion": "550",
16-
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run",
13+
"LinuxCudaVersion": "12.0",
14+
"LinuxDriverVersion": "525",
15+
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.0.0/local_installers/cuda_12.0.0_525.60.13_linux.run",
1716
"RequireCustomSystemSupport": true
1817
},
1918
"Actions": [

‎src/VirtualClient/VirtualClient.Main/profiles/PERF-GPU-SUPERBENCH-NVIDIA.json

Lines changed: 3 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -5,15 +5,14 @@
55
"RecommendedMinimumExecutionTime": "08:00:00",
66
"SupportedPlatforms": "linux-x64",
77
"SupportedOperatingSystems": "Ubuntu",
8-
"RecommendOperationSystem": "Ubunutu-22.04, Default LinusCudaVersion,LinuxDriverVersion,LinuxLocalRunFile support this OS only.",
98
"SpecialRequirements": "This is an Nvidia GPU-specialized workload. It depends upon the system having an Nvidia GPU card/chip."
109
},
1110
"Parameters": {
1211
"ConfigurationFile": "default.yaml",
1312
"Username": "",
14-
"LinuxCudaVersion": "12.4",
15-
"LinuxDriverVersion": "550",
16-
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run"
13+
"LinuxCudaVersion": "12.0",
14+
"LinuxDriverVersion": "525",
15+
"LinuxLocalRunFile": "https://developer.download.nvidia.com/compute/cuda/12.0.0/local_installers/cuda_12.0.0_525.60.13_linux.run"
1716
},
1817
"Actions": [
1918
{

0 commit comments

Comments
 (0)