Changes:
- Updates to amdsmi_asic_info_t structure to include:
target_graphics_version, kfd_id, node_id, partition_id
- Updates to amd-smi static --asic to display new
samdsmi_asic_info_t fields
- Updates to gpu enumeration during amdsmi_init()
to discover all logical GPUs when in a non-SPX mode
(ex. DPX, TPX, QPX, or CPX)
- Updates to amdsmi_get_gpu_bdf_id(..) to include
partition_id details when in BDF or optional bits.
- bits [63:32] = domain
- bits [31:28] or bits [2:0] = partition id
- bits [27:16] = reserved
- bits [15:8] = Bus
- bits [7:3] = Device
- bits [2:0] = Function (partition id maybe in bits [2:0]) <-- Fallback for non SPX modes
- C++/Python tests updated to reflect these outputs
Change-Id: I4be0ea35bb98f3109ae2ca9e82f6b21baa38de29
Signed-off-by: Charis Poag <Charis.Poag@amd.com>
[ROCm/amdsmi commit: a33e4c9e14]
1. Add a API interface amdsmi_topo_get_p2p_status to retrieve
connection type and P2P capabilities between 2 GPUs.
2. Add getting p2p status test in hw_topology_read
to print P2P capability information.
3. Add below tables for cli topology sub commands:
- CACHE COHERANCY TABLE
- ATOMICS TABLE
- DMA TABLE
- BI-DIRECTIONAL TABLE
Change-Id: I199173030d4170115cea27c472958a4826e4e1bf
Signed-off-by: Tim Huang <tim.huang@amd.com>
[ROCm/amdsmi commit: 260edaa752]
I ran a test that exercised this code in dev mode and ASAN found a memory access issue due to the iterator returned by lower_bound being dereferenced unconditionally. I believe the right fix is to check if the iterator is within the map and if not go to the else branch
Change-Id: I34fdce634791a09a89eee76c8b2b64a9607d57f9
Signed-off-by: Galantsev, Dmitrii <dmitrii.galantsev@amd.com>
[ROCm/amdsmi commit: 2066872297]
number of compute units `amdgpu_gpu_info.num_of_compute_units` is exposed through amdsmi_get_gpu_asic_info().
Code changes related to the following:
* API
* CLI
* Unit tests
* Examples
Change-Id: Ibeb612d079ed87437a0e56124b8504098fc2dcfd
Signed-off-by: Oliveira, Daniel <daniel.oliveira@amd.com>
[ROCm/amdsmi commit: b05849dad0]
GPU Metrics info `gpu_metrics.vcn_activity` is exposed through amdsmi_get_utilization_count().
Code changes related to the following:
* API
* CLI
* Unit tests
Change-Id: I831b2a81bdc0e090a6698dcb689d10f91ed87dd9
Signed-off-by: Oliveira, Daniel <daniel.oliveira@amd.com>
[ROCm/amdsmi commit: af3670d758]
When set PM policy in sysfs, the driver expect a return character.
Change-Id: I83cddb3cdb14c226e6e856776176000eea33b251
[ROCm/amdsmi commit: 4441249ffa]
The function is to clean the local data in LDS/GPRs. The clear sram
is misleading.
Change-Id: I0385e6d6348602fe0f347d17e48ed8983f7ceb87
[ROCm/amdsmi commit: 4cf59c4edb]
The sysfs is changed to use the pm_policy folder with multiple
dpm_policy files.
Change-Id: I40fac8de2d0cb127950d238b8196f6d2416778d0
[ROCm/amdsmi commit: e5d1ba4621]
When discover the amdgpu, if the assigned numbers are not consecutive,
not all GPU can be discovered. The code is change to discover the
GPU based on max card number.
Change-Id: Icf4c1df4a1651093b5de3cd7a25a9bd69a299075
[ROCm/amdsmi commit: 437cb07db6]
The process isolation sysfs format is changed. This fix will
adapt to the new sysfs format.
Change-Id: Id6fd7eeb3e25525047dccab248fd9cfb206cbf62
[ROCm/amdsmi commit: a0d0210761]
A few APIs and command line options are added to support process
isolation and clean shader.
Change-Id: I98ad3fc9fc7429799a21798b7fca1c307de7f403
[ROCm/amdsmi commit: 7d2ab7970d]
Drops checks that are invalid with the new pp_od_clk_voltage format
Code changes related to the following:
* get_od_clk_volt_info()
* get_od_clk_volt_curve_regions()
Change-Id: I534c920e00fa3dacdb980f431db5eef260ac93f5
Signed-off-by: Oliveira, Daniel <daniel.oliveira@amd.com>
[ROCm/amdsmi commit: 1ae3a5b6cb]
Update the API and CLI to support XGMI Per-Link Power Down Policy.
Change-Id: Iaf04a771eb8bb0829a5b3088d803a7355a8dfd0b
[ROCm/amdsmi commit: e4085c6414]
Fixes asserts in unit tests, and 'pp_dpm_pcie' condition
Code changes related to the following:
* rsmi_dev_pci_bandwidth_set()
* Functional tests
Change-Id: Id5e6851393fa3b51bb8cad87daca1efaf500a7e0
Signed-off-by: Oliveira, Daniel <daniel.oliveira@amd.com>
[ROCm/amdsmi commit: 475424525e]
Checks and forces rereading gpu metrics unconditionally
Code changes related to the following:
* Device::dev_log_gpu_metrics()
* amdsmi_get_gpu_metrics_header_info()
Removed unintentionally during work on 'header cleanup Remove non-unified headers'
* Examples
* Unit tests
Change-Id: I83710e173c0f7102d0b7f865c18474c979a95cd8
Signed-off-by: Oliveira, Daniel <daniel.oliveira@amd.com>
[ROCm/amdsmi commit: 78074d7d77]
The current code assume err_count sysfs only have 2 lines, which is
changed for umc_err_count by adding extra line for defer errors.
The code is changed to relax such check.
Change-Id: I7129bfcf9ef47346bca3c5a9535bc8c16c670f88
[ROCm/amdsmi commit: b125f3ee8f]
- CLI: Added average_power to display if current_power is empty
- CLI: fixed PCIe current_speed not displaying GT/s
- ROCm API: 1.3 & 1.4
-> commented out setting avg clocks to current clock value
(leave as max uint value, not re-assign; these are not same values)
-> commented out setting current_socket_power = average_power
(leave as max uint value, not re-assign; these are not same values)
-> For all non-array clocks, placed value in first
array[0] to keep outputs consistent
(helps xcd calc)
- ROCm API: rsmi_dev_metrics_curr_gfxclk_get fixed to count
XCDs using backwards compatible rsmi_dev_gpu_metrics_info_get.
- ^ Fixes XCD count overall + assigning clock[0] in 1.3 to curr
freq
- AMD SMI API: amdsmi_get_gpu_metrics_info() initialized all new
1.5 metric values for all lower metric tables
- AMD SMI API: wrapper -> fix is here + returns correct AMD SMI return
- AMD SMI API: wrapper -> now displays amdsmi return status as
string in logs
- gpu_metrics_read.cc -> now has better overview of backwards
compatible output
- gpu_metrics_read.cc -> Cleaned up output, added units, and
display all array output
Signed-off-by: Charis Poag <Charis.Poag@amd.com>
Change-Id: Id5b60ded5b0ed2cdf0f96ca72c79e356f0410960
[ROCm/amdsmi commit: 5ff5af0b5a]
Add above fields for cache info. Remove driver_date in CLI and
Remove the disable properties of cache.
Change-Id: I80672490908d9e32a149076cc37459fa56b8b0bf
[ROCm/amdsmi commit: 59b510de2b]
Eisuke Kawashima
Maisam Arif
danzimm
Charis Poag
Tim Huang
Xiaodong Wang
Oliveira, Daniel
Bill(Shuzhou) Liu
Maisam Arif
Galantsev, Dmitrii