HIP supports compiling kernels from assembly into exec.
The device assembly needs to be compiled into a fat binary
object. This device object is embedded into a host object using
llvm-mc directives. Then, any host linker may link the host and
device objects together into an executable. A README is added.
Change-Id: I59d3a8b5363073810ffc3aa0d57f21b0df272369
1.Make directed_test apps linked against static libs
of hip, rocclr, rocr, roct and amd_comgr.
2.Remove custom_target amdhip64_static_combiner.
3.Support EXCLUDE_HIP_LIB_TYPE <static|shared>.
4.Simplify argument list parsing.
5.Install rocclr when rocm is installed.
6.Fix some original small bugs.
Revert "Revert "Make directed_test support static libs""
This reverts commit 144a6fb100.
Change-Id: I918eeae94487e5e2ff5bfde083667ac65fb6e702
1.Make directed_test apps linked against static libs
of hip, rocclr, rocr, roct and amd_comgr.
2.Remove custom_target amdhip64_static_combiner.
3.Support EXCLUDE_HIP_LIB_TYPE <static|shared>.
4.Simplify argument list parsing.
5.Fix hip_pch.o missing issue in samples' build.
6.Fix some original small bugs.
Change-Id: I30f5e65367613152ce86b80b13e1c9f2a28da803
Only cmake can support static lib of hip rt.
Thus samples will support static lib of hip
rt when this is done.
Change-Id: I70e8d06e85084369a035b42c5d1d56287c874ac9
- The driver code should not re-define `tex` again as it's already
defined in the kernel code. Eventually, the driver code should be as
regular C++ code instad of HIP code.
Change-Id: I8c7cab204b98990619d6e7109b990d7089ea9261
1.Combine libamdhip64_static_base.a and libamdvdi_static.a into libamdhip64_static.a.
2.Let hipcc use -use-staticlib to link libamdhip64_static.a.
3.Add some samples for static lib.
4.Fix compiling failure of code object.
Change-Id: Ic8c95228eb139058da8b5d66ba8439486154ca6f
This reverts commit ed3b0eb391.
Reason for revert: It is causing dkms-no-npi-hipclang broken.
It is top priority to maintain dkms-no-npi-hipclang build, otherwise we lose track of regression analysis.
So revert the change for now and recommit it after fixing it.
Change-Id: Ia5136e888baecb6148c6c18eedbf37066fcb1eaa
1.Combine libamdhip64_static_base.a and libamdvdi_static.a into libamdhip64_static.a.
2.Let hipcc use -use-staticlib to link libamdhip64_static.a.
3.Add some samples for static lib.
4.Fix compiling failure of code object.
Change-Id: Ia2333622a8d05639b90974c4c5d3d85654ba0138
The maxSharedMemoryPerMultiProcessor attribute is meant to describe
the number of bytes of shared memory (LDS space in AMD terminology)
in each SM (CU in AMD terminology). For instance, on AMD GPUs this
is often 64KB per CU, and some Nvidia GPUs it's 96KB per SM.
This shared memory is a different address space from the normal
global memory. However, the current HIP-HCC properties fill this
in with a size that matches the totalGlboalMem property. This gives
a drastically too-high calculation for the amount of LDS space that
each CU has -- tens of GBs vs. 10s of KBs.
This patch fixes this by pulling the maxSharedMemoryPerMultiProcessor
property from the HSA pool that describes how much workgroup-local
space is available on each CU. The HSA runtime eventually pulls
this from the topology information about LDSSizeInKB, defined as
"Size of Local Data Store in Kilobytes per SIMD".
Previously, this HSA query was used to fill in the value of the
sharedMemPerBlock property. On today's AMD GPUs, we know that
the amount of LDS avaialble to the workgroup is identical to the
amount of LDS space in the CU. However, in the future this may
differ. As such, this patch changes around the order and fills
in the "PerMultiProcessor" property from the HSA query (since
what's what the query is defined to return), and then separately
fills in the "PerBlock" property as we know it.
Temporarily comment out Hcc-specific template functions
hipExtLaunchKernelGGL and hipOccupancyMaxPotentialBlockSize for CLang
compiler so that all test cases under hip/samples can be built
successfully for Clang + Hip/Hcc runtime.
Change-Id: Iafc761257be4a7b34eafa6759a01f369570cd6ce
* Use hipExtLaunchKernelGGL in dispatchlatency sample
* Let it run on NVCC path too
* Refactoring
* Add test_kernel source
* Remove ResultDB
* Remove error checks
Fixes SWDEV-203394.
Currently in runTest() returns true, even if the texture reference copy does not happen. Using the existing testResult Flag to return from runTest().
Aaron En Ye Shi
Tao Sang
Sarbojit Sarkar
Vladislav Sytchenko
Julia Jiang
Satyanvesh Dittakavi
Rahul Garg
Jason Tang
agodavar
Aaron En Ye Shi
Alex Xie
Michael LIAO
Maneesh Gupta
Evgeny Mankov
Rahul Garg
Sarbojit2019
Joseph Greathouse
Lad, Aditya
satyanveshd
Anusha Godavarthy Surya
kjayapra-amd
Yaxun (Sam) Liu