xpu-smi 1.1 hot fix

Changes in this release

• [XPU-SMI] Roll back the memory size to be consistent with v1.0

Supported devices

• Intel(R) Data Center GPU Flex Series

Supported OSes

• XPU-SMI
  -- CentOS 8 Stream
  -- Ubuntu 20.04.3/22.04
  -- RHEL 8.6
  -- SLES 15 SP4

Notice

• For Flex Series GPU GSC firmware update, please make sure to install iGSC v0.8.4 or newer. If you still use the old iGSC version (such as v0.7.1), please disable GPU power management by ""/sys/bus/platform/devices/i915.mei-gscfi.(xxxxx)/power/control" on all installed GPU devices before updating Flex Series GSC firmware

1.1.0 Golden Release

New features in this release

• Integrate XPU-SMI source code into XPU Manager trunk. XPU-SMI will be released with XPU Manager together since this version.
• [XPU Manager/XPU-SMI] Support multi-GPU in the dump command
• [XPU Manager/XPU-SMI] Improve the UX of GPU firmware update, such as reporting the update percentage.
• [XPU Manager/XPU-SMI] Check the media transcoding performance in the GPU diagnostics feature
• [XPU Manager/XPU-SMI] Show the i915 KMD version and kernel version
• [XPU Manager/XPU-SMI] Report the engine-group utilization
• [XPU Manager/XPU-SMI] Improve the topology performance
• [XPU Manager] Support GPU GFX firmware update in Windows Server OS
• [XPU Manager] Support GPU ECC on/off setting in Windows Server OS
• [XPU Manager] Support AMC firmware update on Supermicro 2U server
• [XPU Manager] Report GPU serial number on Supermicro servers
• [XPU Manager] Show AMC OOB sensor readings on Coyote Pass server
• [XPU Manager] Show GPU throttle info in GPU health
• [XPU Manager] Add GPU pre-check into GPU diagnostics feature
• [XPU Manager] New OS supporting
• [XPU Manager] Reduce the CPU usage of XPU manager daemon
• [XPU-SMI] New OS supporting

Supported devices

• Intel(R) Data Center GPU Flex Series

Supported OSes

• XPU Manager
  -- Ubuntu 20.04.3/22.04
  -- RHEL 8.5/8.6
  -- CentOS 8 Stream
  -- CentOS 7.4/7.9
  -- SLES 15 SP3
  -- Windows Server 2022/2019 (limited supporting: GPU device info, GPU telemetries, GPU settings, GPU GFX firmware update)

• XPU-SMI
  -- CentOS 8 Stream
  -- Ubuntu 20.04.3/22.04
  -- RHEL 8.6
  -- SLES 15 SP4

Notice

• For Flex Series GPU GSC firmware update, please make sure to install iGSC v0.8.4 or newer. If you still use the old iGSC version (such as v0.7.1), please disable GPU power management by ""/sys/bus/platform/devices/i915.mei-gscfi.(xxxxx)/power/control" on all installed GPU devices before updating Flex Series GSC firmware
• Since this 1.1.0 sprint version, XPU Manager Windows CLI tool supports Flex Series GPU memory ECC on/off and Flex Series GPU firmware update features. Please install igsc.dll v0.8.4 or newer versions in your Windows OS

xpu-smi 1.1 Beta

New features in this release

• Integrate XPU-SMI into XPU Manager master tree
• Report the engine-group utilization
• Report more GPU-level telemetries in dump feature
• Only show the "OEM Manu data version" in GFX Data Firmware Version field
• Improve topology performance
• Support multi-GPU in the dump command
• Report the GPU utilization with more fine granularity
• Support upgrading from old version in the rpm/deb installer packages.
• Show the time only in the dump command.

Supported devices

• Intel(R) Data Center GPU Flex Series

Supported OSes

-- Ubuntu 20.04.3
-- CentOS 8 Stream

xpu-smi 1.0.0

XPU-SMI is the daemon-less XPU Manager version.

Ready features

• Show GPU device info
• Show GPU real-time telemetries
• Configure GPU settings
• GPU SOC firmware update
• GPU diagnostics

Supported devices

• Intel Data Center GPU Series

Supported OSes

• Ubuntu 20.04.3
• CentOS 8 Stream

Notice

• For ATS-M GSC firmware update, please make sure to install iGSC v0.8.3. If you still use the old iGSC version (such as v0.7.1), please disable GPU power management by ""/sys/bus/platform/devices/i915.mei-gscfi.(xxxxx)/power/control" on all installed GPU devices before updating ATS-M GSC firmware

1.0.0

Ready features

• GPU device info, including device model, stepping, location, frequency, memory capacity, topology and firmware version, etc.
• GPU telemetries, including GPU power, temperature, utilization, memory throughput, etc.
• GPU firmware update
• 3 levels of GPU diagnostics test
• GPU settings, including power limit, frequency range and ECC on/off, etc.
• GPU health status

Supported devices

• Intel(R) Data Center GPU Series

Supported OSes

• Ubuntu 20.04.3
• RHEL 8.5 and CentOS 8 Stream
• CentOS 7.4
• SLES 15 SP3
• Windows Server 2022 (limited supporting: GPU device info, GPU telemetries and GPU settings)

Notice

• For GPU GSC firmware update, please make sure to install iGSC v0.8.3 or newer. If you still use the old iGSC version (such as v0.7.1), please disable GPU power management by ""/sys/bus/platform/devices/i915.mei-gscfi.(xxxxx)/power/control" on all installed GPU devices before updating GSC firmware

1.0.0 beta 4

Ready features

• Make software exclusive checking not blocking diagnostic task
• Power limit is put on GPU level
• XPUM can show all AMC firmware versions
• XPUM can support AMC firmware 4.0.0

Supported devices

• Intel Data Center GPU

Supported OS

• Ubuntu 20.04.3
• RHEL 8.4