[PATCH V8 00/10] AMD XDNA driver

Lizhi Hou lizhi.hou at amd.com
Mon Nov 11 18:15:18 UTC 2024


Sorry, just noticed that I miss merged one line. Please ignore this V8 
set. I will send out V9 instead.


Lizhi

On 11/11/24 09:32, Lizhi Hou wrote:
> This patchset introduces a new Linux Kernel Driver, amdxdna for AMD NPUs.
> The driver is based on Linux accel subsystem.
>
> NPU (Neural Processing Unit) is an AI inference accelerator integrated
> into AMD client CPUs. NPU enables efficient execution of Machine Learning
> applications like CNNs, LLMs, etc.  NPU is based on AMD XDNA
> architecture [1].
>
> AMD NPU consists of the following components:
>
>    - Tiled array of AMD AI Engine processors.
>    - Micro Controller which runs the NPU Firmware responsible for
>      command processing, AIE array configuration, and execution management.
>    - PCI EP for host control of the NPU device.
>    - Interconnect for connecting the NPU components together.
>    - SRAM for use by the NPU Firmware.
>    - Address translation hardware for protected host memory access by the
>      NPU.
>
> NPU supports multiple concurrent fully isolated contexts. Concurrent
> contexts may be bound to AI Engine array spatially and or temporarily.
>
> The driver is licensed under GPL-2.0 except for UAPI header which is
> licensed GPL-2.0 WITH Linux-syscall-note.
>
> User mode driver stack consists of XRT [2] and AMD AIE Plugin for IREE [3].
>
> The firmware for the NPU is distributed as a closed source binary, and has
> already been pushed to the DRM firmware repository [4].
>
> [1] https://www.amd.com/en/technologies/xdna.html
> [2] https://github.com/Xilinx/XRT
> [3] https://github.com/nod-ai/iree-amd-aie
> [4] https://gitlab.freedesktop.org/drm/firmware/-/tree/amd-ipu-staging/amdnpu
>
> Changes since v7:
> - Prealloc dma fence chain before publishing dma-fence
> - Install the job's finished fence in dma-resv rather than driver fence
>
> Changes since v6:
> - Revise command submission flow
>
> Changes since v5:
> - Remove wait_cmd ioctl and use syncobj instead
> - Cleanup spelling errors
> - Add dependencies in Kconfig
>
> Changes since v4:
> - Fix lockdep errors
> - Use __u* structure for struct aie_error
>
> Changes since v3:
> - Remove debug BO patch
> - Changes based on code review comments
>
> Changes since v2:
> - Add document amdnpu.rst
> - Change AIE2_DEVM_SIZE to 64M due to firmware change
> - Changes based on code review comments
>
> Changes since v1:
> - Remove some inline defines
> - Minor changes based on code review comments
>
> Lizhi Hou (10):
>    accel/amdxdna: Add documentation for AMD NPU accelerator driver
>    accel/amdxdna: Add a new driver for AMD AI Engine
>    accel/amdxdna: Support hardware mailbox
>    accel/amdxdna: Add hardware resource solver
>    accel/amdxdna: Add hardware context
>    accel/amdxdna: Add GEM buffer object management
>    accel/amdxdna: Add command execution
>    accel/amdxdna: Add suspend and resume
>    accel/amdxdna: Add error handling
>    accel/amdxdna: Add query functions
>
>   Documentation/accel/amdxdna/amdnpu.rst        | 281 ++++++
>   Documentation/accel/amdxdna/index.rst         |  11 +
>   Documentation/accel/index.rst                 |   1 +
>   MAINTAINERS                                   |  11 +
>   drivers/accel/Kconfig                         |   1 +
>   drivers/accel/Makefile                        |   1 +
>   drivers/accel/amdxdna/Kconfig                 |  18 +
>   drivers/accel/amdxdna/Makefile                |  21 +
>   drivers/accel/amdxdna/TODO                    |   5 +
>   drivers/accel/amdxdna/aie2_ctx.c              | 900 ++++++++++++++++++
>   drivers/accel/amdxdna/aie2_error.c            | 360 +++++++
>   drivers/accel/amdxdna/aie2_message.c          | 791 +++++++++++++++
>   drivers/accel/amdxdna/aie2_msg_priv.h         | 370 +++++++
>   drivers/accel/amdxdna/aie2_pci.c              | 762 +++++++++++++++
>   drivers/accel/amdxdna/aie2_pci.h              | 259 +++++
>   drivers/accel/amdxdna/aie2_psp.c              | 146 +++
>   drivers/accel/amdxdna/aie2_smu.c              | 119 +++
>   drivers/accel/amdxdna/aie2_solver.c           | 330 +++++++
>   drivers/accel/amdxdna/aie2_solver.h           | 154 +++
>   drivers/accel/amdxdna/amdxdna_ctx.c           | 553 +++++++++++
>   drivers/accel/amdxdna/amdxdna_ctx.h           | 162 ++++
>   drivers/accel/amdxdna/amdxdna_gem.c           | 622 ++++++++++++
>   drivers/accel/amdxdna/amdxdna_gem.h           |  65 ++
>   drivers/accel/amdxdna/amdxdna_mailbox.c       | 576 +++++++++++
>   drivers/accel/amdxdna/amdxdna_mailbox.h       | 124 +++
>   .../accel/amdxdna/amdxdna_mailbox_helper.c    |  61 ++
>   .../accel/amdxdna/amdxdna_mailbox_helper.h    |  42 +
>   drivers/accel/amdxdna/amdxdna_pci_drv.c       | 409 ++++++++
>   drivers/accel/amdxdna/amdxdna_pci_drv.h       | 123 +++
>   drivers/accel/amdxdna/amdxdna_sysfs.c         |  67 ++
>   drivers/accel/amdxdna/npu1_regs.c             | 101 ++
>   drivers/accel/amdxdna/npu2_regs.c             | 118 +++
>   drivers/accel/amdxdna/npu4_regs.c             | 118 +++
>   drivers/accel/amdxdna/npu5_regs.c             | 118 +++
>   include/trace/events/amdxdna.h                | 101 ++
>   include/uapi/drm/amdxdna_accel.h              | 436 +++++++++
>   36 files changed, 8337 insertions(+)
>   create mode 100644 Documentation/accel/amdxdna/amdnpu.rst
>   create mode 100644 Documentation/accel/amdxdna/index.rst
>   create mode 100644 drivers/accel/amdxdna/Kconfig
>   create mode 100644 drivers/accel/amdxdna/Makefile
>   create mode 100644 drivers/accel/amdxdna/TODO
>   create mode 100644 drivers/accel/amdxdna/aie2_ctx.c
>   create mode 100644 drivers/accel/amdxdna/aie2_error.c
>   create mode 100644 drivers/accel/amdxdna/aie2_message.c
>   create mode 100644 drivers/accel/amdxdna/aie2_msg_priv.h
>   create mode 100644 drivers/accel/amdxdna/aie2_pci.c
>   create mode 100644 drivers/accel/amdxdna/aie2_pci.h
>   create mode 100644 drivers/accel/amdxdna/aie2_psp.c
>   create mode 100644 drivers/accel/amdxdna/aie2_smu.c
>   create mode 100644 drivers/accel/amdxdna/aie2_solver.c
>   create mode 100644 drivers/accel/amdxdna/aie2_solver.h
>   create mode 100644 drivers/accel/amdxdna/amdxdna_ctx.c
>   create mode 100644 drivers/accel/amdxdna/amdxdna_ctx.h
>   create mode 100644 drivers/accel/amdxdna/amdxdna_gem.c
>   create mode 100644 drivers/accel/amdxdna/amdxdna_gem.h
>   create mode 100644 drivers/accel/amdxdna/amdxdna_mailbox.c
>   create mode 100644 drivers/accel/amdxdna/amdxdna_mailbox.h
>   create mode 100644 drivers/accel/amdxdna/amdxdna_mailbox_helper.c
>   create mode 100644 drivers/accel/amdxdna/amdxdna_mailbox_helper.h
>   create mode 100644 drivers/accel/amdxdna/amdxdna_pci_drv.c
>   create mode 100644 drivers/accel/amdxdna/amdxdna_pci_drv.h
>   create mode 100644 drivers/accel/amdxdna/amdxdna_sysfs.c
>   create mode 100644 drivers/accel/amdxdna/npu1_regs.c
>   create mode 100644 drivers/accel/amdxdna/npu2_regs.c
>   create mode 100644 drivers/accel/amdxdna/npu4_regs.c
>   create mode 100644 drivers/accel/amdxdna/npu5_regs.c
>   create mode 100644 include/trace/events/amdxdna.h
>   create mode 100644 include/uapi/drm/amdxdna_accel.h
>


More information about the dri-devel mailing list