- Added Physical State Query API calls
- Added Clock, Power and Performance Query API calls
- README for both additions
Change-Id: Icf412364a13c9e51b9630f19f29a7cdfbe46f7fe
Signed-off-by: Dalibor Stanisavljevic <Dalibor.Stanisavljevic@amd.com>
- Add up to date README file for python tool
Change-Id: I7a02f79469e990870398b3741b033ea447998fdd
Signed-off-by: Dalibor Stanisavljevic <Dalibor.Stanisavljevic@amd.com>
- Implement context manager event class for easier usage instead of having 4 different
APIs.
- Add support for with statement for easier cleanup.
Change-Id: I396947a93d3f04efc068e17804eab12eefa97cf0
Signed-off-by: Dalibor Stanisavljevic <Dalibor.Stanisavljevic@amd.com>
- Added APIs for "Supported Function Checks"
- Added first half of the APIs for "Unsupported Functions In Virtual Environment"
Change-Id: Ia83303e6b7ae65aada608bd9ca5f56fd33e4ef34
Signed-off-by: Dalibor Stanisavljevic <Dalibor.Stanisavljevic@amd.com>
Init the ams_smi using the rocm-smi, which makes the GPU discovery
consistent with or without libdrm.
Change-Id: Ic714781f8ce791451b0c057621525926edb7f5ee
Return value from ReadSysfsStr function that reads cu_occupancy file
was not handled correctly. Modified the script to handle any fail conditions.
Change-Id: I3c71e0f6f288f196ed1f833e8709255c2b6e78ee
Get the device name from rocm-smi which is not displayed properly
in some cards. Set the vibos version using the rocm-smi.
Change-Id: I138f1760cde94007cb93cad02c6d8cccbb4afa28
Those two APIs are changed to let the user get the handles count,
allocate memory, and then return handles to the allocated memory.
Change-Id: Ibe28a89ad188c99da6af3af1740b2b25ff22ba06
Move rocm_smi related function to rocm_smi folder. Move amd_smi to
top level include/ and src/ folder. Remove obsolte oam folder.
Change the CMakeLists.txt to update folder locations.
Change-Id: I52e6be739e49f3b0545865f25364787f5985e9c3
A few fields are added to board_info and asic_info for rocm-smi
device information.
Implement rocm-smi related fw block in amdsmi_get_fw_info().
Change-Id: I825d3e5c7feaa07a6e05386d4f1a59ebf528dfc0
Install interface should provide /opt/rocm-ver/include as the include path
Path /opt/rocm-ver/rocm_smi/include should be used only as part of backward compatibility support
Change-Id: Idc1f663069356c6b1fbd492f45ef4637fc90e4eb
Invalid peer links are labeled as N/A during topology report creation.
This invalid link type could be triggered by having a configuration
with CPU XGMI iolinks and disable XGMI peer to peer access. This can
be done by setting the driver parameter 'use_xgmi_p2p = 0'.
Signed-off-by: Alex Sierra <Alex.Sierra@amd.com>
Change-Id: Ifb09a8f3266a3f07686615dfb45781d6cfe55e83
Devices with CPU XGMI iolink do not support PCIe peer access. Therefore,
they should not be reported as accessible links in the topology.
Signed-off-by: Alex Sierra <Alex.Sierra@amd.com>
Change-Id: I3ee51796945dc0966200dee03886510e8f1846b7
The purpose of this patch is to modify the column header of the default
'./rocm-smi' command from 'Temp' to 'Temp (DieEdge)' for clarity.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I127a9214be97a1185c3db010f1c9176d1f412ec9
--showdeviceid
Fix for false-positive "FRU is corrupted" messages,
since str(sn).isalphanum() triggers on empty struct.
--showproductname
fix script termination on non-alphanum product name
Change-Id: I78d4998e156f9b0d9f45338bed2a0d30b789e220
I6520b51aac34060b5b90f94a016cec1827a4973f happens after uninstall, which
leaves a dangling directory under /opt/rocm/libexec/rocm_smi.
Removing __pycache__ before uninstall fixes the issue.
Signed-off-by: Galantsev, Dmitrii <dmitrii.galantsev@amd.com>
Change-Id: I695bd085d4a43b678b563b4c35f6d2e8ddfa7d7c
Any default value if required should be controlled from outside.
For ROCM, build script is setting the value to "lib"
Change-Id: I12a2951307fe64e46a4e608476bfceb678bdc97d
This fixes the 'unknown' value being displayed
for Perf Level because of a missing mapping of
RSMI_DEV_PERF_LEVEL_DETERMINISM to its string
value.
Signed-off-by: Divya Shikre <DivyaUday.Shikre@amd.com>
Change-Id: I479c2baea450f0ff61640ad81cbd4d08ad56ff8e
The purpose of this patch is to set RETCODE equal to 0 by default
unless an appropriate '--loglevel LEVEL' has been set.
To allow a non-zero RETCODE value, you must use any loglevel that
is not 'warning' or 'None' (default).
You can set the loglevel in the CLI with:
--loglevel <debug/info/warning/error/critical>
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I9484a750206a3f464c59952304e72c59c3d12465
Changes made on rsmi_perf_determinism_mode_set function documentation
as well for styling consistency.
Change-Id: I09ce8139eb9cbda94352ac7725c4c9b9bb06bd59
Remove the compile time dependency on libdrm. Load it at the run
time instead.
Add the headers missed from smi-lib
Change-Id: Ie1ecf293b51425b6a61c502d11a42809dc099f70
If the FRU has been corrupted, then the serial number will come in with
any manner of random bytes, which will cause decode() to fail
spectacularily. Check that the serial returned by the kernel is
alphanumeric, and print to the error log if not (then continue to the
next device).
Change-Id: If4f35b140b6089e02729b1490ed6b48d614a122a
Use GNUInstallDirs variables to determine the location of LIBDIR, BINDIR, INCLUDEDIR, DOCDIR
Note that CMAKE_INSTALL_LIBDIR is overriden, since the default for RHEL
is lib64, but ROCm packaging wants it to be lib always. Distros or users
can easily override this.
Change-Id: I616152ccd2bc1f5a60bffa940312b38ca6e88c04
This patch changes the error handling for setClockRange.
When a device does not support modifying a clock type (sclk/mclk),
an error message is printed through the python CLI.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I37d9ea4189b1ca81e5deaab5efa6cfa4901b89b3
The init version includes the amd_smi.h header, an example uses the
amd_smi, folder structure and CMake files.
Add the support to libdrm.
Change-Id: I779e55e4cf9491c61dc226a30d24e96be9bc6016
showpidgpus prints 'none' when no GPU devices are
being used by the running process. Adding a fix
to print a relevant message.
Signed-off-by: Divya Shikre <DivyaUday.Shikre@amd.com>
Change-Id: I165a6644a76c8e1c3c3cad676dcfd41eb1c4724f
This patch fixes a --showvoltagerange bug, which attempts to check
the voltage curve on a device that does not have any voltage
regions in its OverDrive voltage frequency data (odvf).
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I647c30c978ffb13f6819ac3d069ee340710a7f99
Fixes bug in the 'setPowerOverdrive' function which mishandles
GPUs with secondary dies. Secondary dies have a default power cap
of 0W and cannot be changed, so they are now skipped.
Fixes bug in the 'resetPowerOverdrive' function which incorrectly
resets the wattage to the current value.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I483fa3f58b1fa44a3bf7bae3b52c59ce523ae152
1. Memory allocated for handle was not deleted
when no variant, subvariant or supported function
was found
2. handle->func_id_iter address was set to 0
before delete[]
Signed-off-by: Divya Shikre <DivyaUday.Shikre@amd.com>
Change-Id: Iab50fdfbe03eec8e6fd0e84e03bd2c47e645b3d8
Show an optional debug log (RSMI_DEBUG_BITFIELD=2) to
the user in the following scenarios:
1. If more than one current frequency is found
2. If frequencies are not read in increasing order of
their value
If current frequency is not available, index for it is
set to -1, values will not have * next to it in the
output. This will also be handled in rocm_smi.py.
Signed-off-by: Divya Shikre <DivyaUday.Shikre@amd.com>
Change-Id: I477ec065f7513c8045d6392f12ef6cb835a6b8f6
Dalibor Stanisavljevic
Bill(Shuzhou) Liu
Galantsev, Dmitrii
Sreekant Somasekharan
Dejan Andjelkovic
Ranjith Ramakrishnan
Alex Sierra
Ori Messinger
Elena Sakhnovitch
Alexsandar Nedeljkovic
Divya Shikre
Kent Russell