Oneapi6 fix #1251

yhmtsai · 2023-01-03T14:11:46Z

oneapi change their structure internally after major 6 and start to drop subgroup 8 for some GPU.
Note. the version after compiler is just date, which does not indicate any version information.
the original DPCPP version we catch is the date of building package, so I can not use it.
oneapi 6 change:

sycl is not under cl namespace
they put using sycl = ::sycl in namespace cl to provide some back compatibility, but we have forward declaration and specialization. It will lead compiler error, so we need to adjust namespace.
onemkl gemm works only works on no-empty matrix like other vendor

tcojean

Minor nit, I didn't look at the code.

As I feared originally with these magic numbers, we might need a hardware SIMD table somewhere to keep 7 for some hardware while updating for newer ones?

tcojean · 2023-01-26T16:26:37Z

include/ginkgo/core/matrix/csr.hpp

@@ -393,7 +393,7 @@ class Csr : public EnableLinOp<Csr<ValueType, IndexType>>,
 * number of threads in a SIMD unit is 7


Suggested change

* number of threads in a SIMD unit is 7

* number of threads in a SIMD unit is 8

tcojean · 2023-01-26T16:27:46Z

include/ginkgo/core/matrix/csr.hpp

@@ -583,7 +583,7 @@ class Csr : public EnableLinOp<Csr<ValueType, IndexType>>,
 * number of threads in a SIMD unit is 7


Suggested change

* number of threads in a SIMD unit is 7

* number of threads in a SIMD unit is 8

upsj · 2023-02-08T13:58:53Z

include/ginkgo/core/base/executor.hpp

+
+// after intel/llvm September'22 release, which uses major version 6, they
+// introduce another inline namespace _V1.
+#if GINKGO_DPCPP_MAJOR_VERSION >= 6


Do we need to keep supporting the old version? It sounds like their ecosystem is slowly stabilizing, maybe we can follow that?

For the next release I would consider it, though this _V1 namespace thing is concerning, it's not stabilizing that much...

Yes, the _V1 is annoying to keep on our side. They do this way to keep the non-breaking interface after this changes. (_V1->_V2->...). Thus, it will not be shown on the release documentation, I think. I have tried something to avoid _V1 but failed. maybe I miss something

include/ginkgo/core/base/executor.hpp

dpcpp/matrix/dense_kernels.dp.cpp

dpcpp/matrix/csr_kernels.dp.cpp

upsj · 2023-02-08T14:01:46Z

dpcpp/base/executor.dp.cpp

@@ -274,15 +274,38 @@ void DpcppExecutor::set_device_property(dpcpp_queue_property property)
 }
 this->get_exec_info().max_workgroup_size = static_cast<int>(
 device.get_info<sycl::info::device::max_work_group_size>());
+// They change the max_work_item_size with template parameter Dimension after
+// major version 6 and adding the default = 3 is not in the same release.


would 3 be a sensible default at all? It sounds like that is a pretty CUDA-specific view? maybe just state that this is the dimensionality we use.

I think so. At least, it's the default dimension template choice from the spec.
They just forgot to add the default in some version when updating the code with updated spec

dpcpp/components/cooperative_groups.dp.hpp

yhmtsai · 2023-02-11T13:23:53Z

rebase!

sonarcloud · 2023-02-11T22:06:39Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
1 Code Smell

0.0% Coverage
0.0% Duplication

pratikvn

Is something holding this PR back ? Maybe we can push this ?

include/ginkgo/core/base/executor.hpp

pratikvn

LGTM!

tcojean

LGTM

include/ginkgo/config.hpp.in

dpcpp/CMakeLists.txt

dpcpp/base/executor.dp.cpp

dpcpp/matrix/csr_kernels.dp.cpp

include/ginkgo/core/base/executor.hpp

tcojean · 2023-03-14T09:38:00Z

include/ginkgo/core/base/executor.hpp

+
+// after intel/llvm September'22 release, which uses major version 6, they
+// introduce another inline namespace _V1.
+#if GINKGO_DPCPP_MAJOR_VERSION >= 6


For the next release I would consider it, though this _V1 namespace thing is concerning, it's not stabilizing that much...

include/ginkgo/core/base/executor.hpp

The failure is NULL pointer argument in memory fill operation. -30 (PI_ERROR_INVALID_VALUE)

If the device does not support it, set it as 1 currently Co-authored-by: Terry Cojean <[email protected]>

…module

ginkgo-bot · 2023-03-14T14:47:43Z

Note: This PR changes the Ginkgo ABI:

Functions changes summary: 1 Removed, 0 Changed, 1 Added functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

For details check the full ABI diff under Artifacts here

sonarcloud · 2023-03-15T02:51:44Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
1 Code Smell

0.0% Coverage
6.9% Duplication

upsj

LGTM!

codecov · 2023-03-15T15:11:18Z

Codecov Report

Patch coverage has no change and project coverage change: -0.06 ⚠️

Comparison is base (f050721) 90.83% compared to head (ec8981a) 90.77%.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1251      +/-   ##
===========================================
- Coverage    90.83%   90.77%   -0.06%     
===========================================
  Files          570      570              
  Lines        48621    48557      -64     
===========================================
- Hits         44164    44079      -85     
- Misses        4457     4478      +21

Impacted Files	Coverage Δ
include/ginkgo/core/base/executor.hpp	`75.23% <0.00%> (-0.36%)`	⬇️
include/ginkgo/core/base/math.hpp	`100.00% <ø> (ø)`
include/ginkgo/core/matrix/csr.hpp	`44.33% <0.00%> (+0.28%)`	⬆️

... and 17 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

Release 1.6.0 of Ginkgo. The Ginkgo team is proud to announce the new Ginkgo minor release 1.6.0. This release brings new features such as: - Several building blocks for GPU-resident sparse direct solvers like symbolic and numerical LU and Cholesky factorization, ..., - A distributed Schwarz preconditioner, - New FGMRES and GCR solvers, - Distributed benchmarks for the SpMV operation, solvers, ... - Support for non-default streams in the CUDA and HIP backends, - Mixed precision support for the CSR SpMV, - A new profiling logger which integrates with NVTX, ROCTX, TAU and VTune to provide internal Ginkgo knowledge to most HPC profilers! and much more. If you face an issue, please first check our [known issues page](https://github.com/ginkgo-project/ginkgo/wiki/Known-Issues) and the [open issues list](https://github.com/ginkgo-project/ginkgo/issues) and if you do not find a solution, feel free to [open a new issue](https://github.com/ginkgo-project/ginkgo/issues/new/choose) or ask a question using the [github discussions](https://github.com/ginkgo-project/ginkgo/discussions). Supported systems and requirements: + For all platforms, CMake 3.13+ + C++14 compliant compiler + Linux and macOS + GCC: 5.5+ + clang: 3.9+ + Intel compiler: 2018+ + Apple Clang: 14.0 is tested. Earlier versions might also work. + NVHPC: 22.7+ + Cray Compiler: 14.0.1+ + CUDA module: CUDA 9.2+ or NVHPC 22.7+ + HIP module: ROCm 4.5+ + DPC++ module: Intel OneAPI 2021.3+ with oneMKL and oneDPL. Set the CXX compiler to `dpcpp`. + Windows + MinGW: GCC 5.5+ + Microsoft Visual Studio: VS 2019+ + CUDA module: CUDA 9.2+, Microsoft Visual Studio + OpenMP module: MinGW. ### Version Support Changes + ROCm 4.0+ -> 4.5+ after [#1303](#1303) + Removed Cygwin pipeline and support [#1283](#1283) ### Interface Changes + Due to internal changes, `ConcreteExecutor::run` will now always throw if the corresponding module for the `ConcreteExecutor` is not build [#1234](#1234) + The constructor of `experimental::distributed::Vector` was changed to only accept local vectors as `std::unique_ptr` [#1284](#1284) + The default parameters for the `solver::MultiGrid` were improved. In particular, the smoother defaults to one iteration of `Ir` with `Jacobi` preconditioner, and the coarse grid solver uses the new direct solver with LU factorization. [#1291](#1291) [#1327](#1327) + The `iteration_complete` event gained a more expressive overload with additional parameters, the old overloads were deprecated. [#1288](#1288) [#1327](#1327) ### Deprecations + Deprecated less expressive `iteration_complete` event. Users are advised to now implement the function `void iteration_complete(const LinOp* solver, const LinOp* b, const LinOp* x, const size_type& it, const LinOp* r, const LinOp* tau, const LinOp* implicit_tau_sq, const array<stopping_status>* status, bool stopped)` [#1288](#1288) ### Added Features + A distributed Schwarz preconditioner. [#1248](#1248) + A GCR solver [#1239](#1239) + Flexible Gmres solver [#1244](#1244) + Enable Gmres solver for distributed matrices and vectors [#1201](#1201) + An example that uses Kokkos to assemble the system matrix [#1216](#1216) + A symbolic LU factorization allowing the `gko::experimental::factorization::Lu` and `gko::experimental::solver::Direct` classes to be used for matrices with non-symmetric sparsity pattern [#1210](#1210) + A numerical Cholesky factorization [#1215](#1215) + Symbolic factorizations in host-side operations are now wrapped in a host-side `Operation` to make their execution visible to loggers. This means that profiling loggers and benchmarks are no longer missing a separate entry for their runtime [#1232](#1232) + Symbolic factorization benchmark [#1302](#1302) + The `ProfilerHook` logger allows annotating the Ginkgo execution (apply, operations, ...) for profiling frameworks like NVTX, ROCTX and TAU. [#1055](#1055) + `ProfilerHook::created_(nested_)summary` allows the generation of a lightweight runtime profile over all Ginkgo functions written to a user-defined stream [#1270](#1270) for both host and device timing functionality [#1313](#1313) + It is now possible to enable host buffers for MPI communications at runtime even if the compile option `GINKGO_FORCE_GPU_AWARE_MPI` is set. [#1228](#1228) + A stencil matrices generator (5-pt, 7-pt, 9-pt, and 27-pt) for benchmarks [#1204](#1204) + Distributed benchmarks (multi-vector blas, SpMV, solver) [#1204](#1204) + Benchmarks for CSR sorting and lookup [#1219](#1219) + A timer for MPI benchmarks that reports the longest time [#1217](#1217) + A `timer_method=min|max|average|median` flag for benchmark timing summary [#1294](#1294) + Support for non-default streams in CUDA and HIP executors [#1236](#1236) + METIS integration for nested dissection reordering [#1296](#1296) + SuiteSparse AMD integration for fillin-reducing reordering [#1328](#1328) + Csr mixed-precision SpMV support [#1319](#1319) + A `with_loggers` function for all `Factory` parameters [#1337](#1337) ### Improvements + Improve naming of kernel operations for loggers [#1277](#1277) + Annotate solver iterations in `ProfilerHook` [#1290](#1290) + Allow using the profiler hooks and inline input strings in benchmarks [#1342](#1342) + Allow passing smart pointers in place of raw pointers to most matrix functions. This means that things like `vec->compute_norm2(x.get())` or `vec->compute_norm2(lend(x))` can be simplified to `vec->compute_norm2(x)` [#1279](#1279) [#1261](#1261) + Catch overflows in prefix sum operations, which makes Ginkgo's operations much less likely to crash. This also improves the performance of the prefix sum kernel [#1303](#1303) + Make the installed GinkgoConfig.cmake file relocatable and follow more best practices [#1325](#1325) ### Fixes + Fix OpenMPI version check [#1200](#1200) + Fix the mpi cxx type binding by c binding [#1306](#1306) + Fix runtime failures for one-sided MPI wrapper functions observed on some OpenMPI versions [#1249](#1249) + Disable thread pinning with GPU executors due to poor performance [#1230](#1230) + Fix hwloc version detection [#1266](#1266) + Fix PAPI detection in non-implicit include directories [#1268](#1268) + Fix PAPI support for newer PAPI versions: [#1321](#1321) + Fix pkg-config file generation for library paths outside prefix [#1271](#1271) + Fix various build failures with ROCm 5.4, CUDA 12, and OneAPI 6 [#1214](#1214), [#1235](#1235), [#1251](#1251) + Fix incorrect read for skew-symmetric MatrixMarket files with explicit diagonal entries [#1272](#1272) + Fix handling of missing diagonal entries in symbolic factorizations [#1263](#1263) + Fix segmentation fault in benchmark matrix construction [#1299](#1299) + Fix the stencil matrix creation for benchmarking [#1305](#1305) + Fix the additional residual check in IR [#1307](#1307) + Fix the cuSPARSE CSR SpMM issue on single strided vector when cuda >= 11.6 [#1322](#1322) [#1331](#1331) + Fix Isai generation for large sparsity powers [#1327](#1327) + Fix Ginkgo compilation and test with NVHPC >= 22.7 [#1331](#1331) + Fix Ginkgo compilation of 32 bit binaries with MSVC [#1349](#1349)

yhmtsai requested a review from a team January 3, 2023 14:11

yhmtsai self-assigned this Jan 3, 2023

ginkgo-bot added mod:core This is related to the core module. mod:dpcpp This is related to the DPC++ module. reg:build This is related to the build system. type:factorization This is related to the Factorizations type:matrix-format This is related to the Matrix formats labels Jan 3, 2023

yhmtsai mentioned this pull request Jan 12, 2023

Half precision support #1257

Open

12 tasks

tcojean reviewed Jan 26, 2023

View reviewed changes

upsj self-requested a review February 6, 2023 13:31

yhmtsai force-pushed the oneapi6_fix branch 3 times, most recently from 9e631fe to 1ed4027 Compare February 8, 2023 07:55

yhmtsai added the 1:ST:ready-for-review This PR is ready for review label Feb 8, 2023

yhmtsai requested review from tcojean and a team February 8, 2023 10:03

upsj reviewed Feb 8, 2023

View reviewed changes

yhmtsai force-pushed the oneapi6_fix branch from 1ed4027 to 542aa4f Compare February 9, 2023 22:47

ginkgo-bot force-pushed the oneapi6_fix branch from 542aa4f to e863e9d Compare February 11, 2023 13:24

yhmtsai force-pushed the oneapi6_fix branch from e863e9d to 447798f Compare February 24, 2023 12:18

yhmtsai force-pushed the oneapi6_fix branch from 447798f to d78ef86 Compare March 6, 2023 10:24

pratikvn reviewed Mar 9, 2023

View reviewed changes

include/ginkgo/core/base/executor.hpp Outdated Show resolved Hide resolved

pratikvn approved these changes Mar 9, 2023

View reviewed changes

tcojean approved these changes Mar 14, 2023

View reviewed changes

yhmtsai added 4 commits March 14, 2023 11:06

remove the cl namespace

6f05c61

remove the subgroup 8

8220d4f

add the macro to fit intel's structure changes

a886c4b

fix the zero size matrix in onemkl

20da270

The failure is NULL pointer argument in memory fill operation. -30 (PI_ERROR_INVALID_VALUE)

yhmtsai and others added 4 commits March 14, 2023 11:06

fix the namespace on old/new compiler

c4a5439

only change the threads_per_cu

af601ff

Use the intel extension to get the hardware thread

1b5d4ec

If the device does not support it, set it as 1 currently Co-authored-by: Terry Cojean <[email protected]>

fix mismatch sycl from 6.0.0 when using dpcpp compiler without dpcpp …

9d75dc3

…module

yhmtsai force-pushed the oneapi6_fix branch from d78ef86 to 9d75dc3 Compare March 14, 2023 10:06

yhmtsai mentioned this pull request Mar 14, 2023

add a770 action #1297

Merged

2 tasks

update doc, variable, function name

ec8981a

yhmtsai added 1:ST:ready-to-merge This PR is ready to merge. and removed 1:ST:ready-for-review This PR is ready for review labels Mar 14, 2023

upsj approved these changes Mar 15, 2023

View reviewed changes

yhmtsai merged commit c97d33e into develop Mar 15, 2023

yhmtsai deleted the oneapi6_fix branch March 15, 2023 15:39

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Oneapi6 fix #1251

Oneapi6 fix #1251

yhmtsai commented Jan 3, 2023

tcojean left a comment

tcojean Jan 26, 2023

tcojean Jan 26, 2023

upsj Feb 8, 2023

tcojean Mar 14, 2023

yhmtsai Mar 14, 2023

upsj Feb 8, 2023

yhmtsai Feb 8, 2023

yhmtsai commented Feb 11, 2023

sonarcloud bot commented Feb 11, 2023

pratikvn left a comment

pratikvn left a comment

tcojean left a comment

tcojean Mar 14, 2023

ginkgo-bot commented Mar 14, 2023

sonarcloud bot commented Mar 15, 2023

upsj left a comment

codecov bot commented Mar 15, 2023

		@@ -393,7 +393,7 @@ class Csr : public EnableLinOp<Csr<ValueType, IndexType>>,
		* number of threads in a SIMD unit is 7

	* number of threads in a SIMD unit is 7
	* number of threads in a SIMD unit is 8

		@@ -583,7 +583,7 @@ class Csr : public EnableLinOp<Csr<ValueType, IndexType>>,
		* number of threads in a SIMD unit is 7

Oneapi6 fix #1251

Oneapi6 fix #1251

Conversation

yhmtsai commented Jan 3, 2023

tcojean left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yhmtsai commented Feb 11, 2023

sonarcloud bot commented Feb 11, 2023

pratikvn left a comment

Choose a reason for hiding this comment

pratikvn left a comment

Choose a reason for hiding this comment

tcojean left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ginkgo-bot commented Mar 14, 2023

sonarcloud bot commented Mar 15, 2023

upsj left a comment

Choose a reason for hiding this comment

codecov bot commented Mar 15, 2023

Codecov Report