1# SPDX-License-Identifier: MIT 2# 3# Heterogeneous system architecture configuration 4# 5 6config HSA_AMD 7 bool "HSA kernel driver for AMD GPU devices" 8 depends on DRM_AMDGPU && (X86_64 || ARM64 || PPC64) 9 select HMM_MIRROR 10 select MMU_NOTIFIER 11 select DRM_AMDGPU_USERPTR 12 help 13 Enable this if you want to use HSA features on AMD GPU devices. 14 15config HSA_AMD_SVM 16 bool "Enable HMM-based shared virtual memory manager" 17 depends on HSA_AMD && DEVICE_PRIVATE 18 default y 19 select HMM_MIRROR 20 select MMU_NOTIFIER 21 help 22 Enable this to use unified memory and managed memory in HIP. This 23 memory manager supports two modes of operation. One based on 24 preemptions and one based on page faults. To enable page fault 25 based memory management on most GFXv9 GPUs, set the module 26 parameter amdgpu.noretry=0. 27 28config HSA_AMD_P2P 29 bool "HSA kernel driver support for peer-to-peer for AMD GPU devices" 30 depends on HSA_AMD && PCI_P2PDMA && DMABUF_MOVE_NOTIFY 31 help 32 Enable peer-to-peer (P2P) communication between AMD GPUs over 33 the PCIe bus. This can improve performance of multi-GPU compute 34 applications and libraries by enabling GPUs to access data directly 35 in peer GPUs' memory without intermediate copies in system memory. 36 37 This P2P feature is only enabled on compatible chipsets, and between 38 GPUs with large memory BARs that expose the entire VRAM in PCIe bus 39 address space within the physical address limits of the GPUs. 40 41