- Ecp-aurora-sdk-discuss - mailman3.cels.anl.gov

Re: [Ecp-aurora-sdk-discuss] New oneAPI release and eng-compiler versions for sunspot.
by John Mellor-Crummey 08 May '24

08 May '24

I hate to complain all the time about unexpected changes in the Intel software stack, but I’m tired of reverse engineering what Intel’s software does differently with each release so that I can build tools. The new software stack both changed how the generated code uses level0 command lists AND it changed important characteristics of GPU binaries. I just figured out that the reason HPCToolkit’s binary analyzer is having trouble with GPU binaries generated by the new release is that in prior releases, all text segments for GPU kernels in a binary each had a unique address range that doesn’t overlap with any other text segments. In the new release, all text segments for GPU kernels start with address 0. I hate that we don’t know what the invariants are for anything. In the binaries generated by prior releases of the software stack, kernels were at non-overlapping addresses and we wrote our software accordingly. Should we have relied on that? Apparently not. Did we have any idea that this might change? Absolutely not. Was the change documented in release notes? Not as far as I can tell. What can be done? Intel should document invariants and what may change. -- John Mellor-Crummey Professor Dept of Computer Science Rice University email: johnmc(a)rice.edu phone: 713-348-5179 > On Mar 29, 2024, at 11:27 AM, Chan-nui, Christopher via Ecp-aurora-sdk-discuss <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> wrote: > > This Message Is From an External Sender > This message came from outside your organization. > With the Sunspot update, a new version of the Aurora SDK is now available on Sunspot. > > The module files for this release, which are loaded by default: > > oneapi/eng-compiler/2023.12.15.002 > > This is based on the 2024.0 release of the oneAPI SDK. > > Release notes for the publicly available portions are available at: > https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope… <https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope…> > https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope… <https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope…> > > The engineering build contains bug fixes and performance improvements. > > Colleen will provide caveats, workarounds and other useful information about these releases shortly. > > > -- > Christopher Chan-Nui > christopher.chan-nui(a)intel.com > > -- > Ecp-aurora-sdk-discuss mailing list > Ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov > https://urldefense.us/v3/__https://lists.jlse.anl.gov/mailman/listinfo/ecp-…

1 0

Frameworks 2024.1 'Preview' Release with updated Pytorch feature release available on Sunspot
by Ramer, Rebecca 03 May '24

03 May '24

There is now an updated 2024.1 DL Frameworks release with support for pytorch and tensorflow on the Sunspot deployment. This is a 'preview' release which has an update release of Intel pytorch DL Framework, along with the same Intel tensorflow DL Framework provided in the DL Frameworks deployment announced in April. This deployment uses the 'preview' version oneAPI SDK release, Intel's 2024.1.0 oneapi basekit. This release is available by loading the frameworks modulefile: module use /soft/preview-modulefiles/24.086.0 module load frameworks/2024.04.15.002 Loading this module activates the aurora_nre_models_frameworks-2024.1_preview_u1 conda environment. Along with the pytorch and tensorflow packages, all necessary python packages needed to run the 11 'Aurora NRE' models have been installed into this conda environment. An experimental version of Intel's xpu backend for triton is also provided. More information about this preview release can be found here: https://urldefense.us/v3/__https://apps.cels.anl.gov/confluence/display/int… Please let me know of any issues encountered. Rebecca Ramer rebecca.ramer(a)intel.com<mailto:[email protected]>

1 0

Frameworks 2024.1 preview release modulefile correction!
by Ramer, Rebecca 16 Apr '24

16 Apr '24

Hi, It appears that I mistyped the name of the frameworks modulefile in the announcement I sent yesterday. Here is the correct name: module use /soft/preview-modulefiles/24.086.0 module load frameworks/2024.04.15.001 Rebecca Ramer rebecca.ramer(a)intel.com

1 0

Frameworks 2024.1 'Preview' Release available on Sunspot
by Ramer, Rebecca 15 Apr '24

15 Apr '24

Hi, I would like to announce availability of the 2024.1 DL Frameworks release with support for pytorch and tensorflow on the Sunspot deployment. This is a 'preview' release, as a different SW release deployment process is expected to be introduced soon. The modulefile is located in a different directory from past releases, but the conda environment location has not changed for this preview release. The release is based on the 2024.1 versions of the Deep Learning Frameworks and uses a 'preview' version oneAPI SDK release which uses Intel's 2024.1.0 oneapi basekit. This release is available by loading the frameworks modulefile: module use /soft/preview-modulefiles/24.086.0 module load frameworks/2021.04.15.001 Loading this module activates the aurora_nre_models_frameworks-2024.1_preview conda environment. Along with the pytorch and tensorflow packages, all necessary python packages needed to run the 11 'Aurora NRE' models have been installed into this conda environment. More information about this preview release can be found here: https://urldefense.us/v3/__https://apps.cels.anl.gov/confluence/display/int… Please let me know of any issues encountered. Rebecca Ramer rebecca.ramer(a)intel.com<mailto:[email protected]>

1 0

Re: [Ecp-aurora-sdk-discuss] New oneAPI release and eng-compiler versions for sunspot.
by Bertoni, Colleen 29 Mar '24

29 Mar '24

Thanks Chris! Important note: with the new SDK drop, if you were specifying a std-c++ standard for OpenMP code, you’ll need to update it to -std=c++17 as a workaround or remove it (see below for more details). Comparing this new release to the previous default on Sunspot (oneapi/eng-compiler/2023.10.15.002) we see 44 fixes and 20 regressions in the bug reproducers (below). However, the 20 regressions fall in mainly 2 categories. Most of the regressions are from a compile-time bug in the C++ openmp compiler that complains that the compiler requires “-std=c++17” if the code was compiled with an older standard (-std=C++11 for example). The error message looks like: ``` static assertion failed due to requirement '201103L >= 201703L': DPCPP does not support C++ version earlier than C++17 ``` This can be worked around by changing it to “-std=c++17” for the time being. This is a bug, will not be required in the future, and already is fixed internally at Intel. Let us know if you have any questions, and thanks for all the bug reporting! Colleen Fixes: source/reproducers/dpcpp/ATSsubdevice_regression source/reproducers/dpcpp/VirtualFunction:VirtualFunction_minimal_ath_run [CMPLRLLVM-35295,XDEPS-6157] P0 source/reproducers/dpcpp/fortran_link:fortran_link_AOT [CMPLRLLVM-41268,cmplrllvm-44716] source/reproducers/dpcpp/kokkos_segfault_zebin [CMPLRLLVM-42875] source/reproducers/dpcpp/kokkos_sycl1a_bessel_fail_simd16 [GSD-6244] source/reproducers/dpcpp/kokkos_sycl2a_fp_model_precise_relwithdebinfo [GSD-3120] source/reproducers/dpcpp/syclmemcpy2d:syclmemcpy2d_ath_run [GSD-5515] P0 source/reproducers/dpcpp/work_item_larger_than_uint:work_item_larger_than_uint_max_work_item_no_fit-in-int [CMPLRLLVM-38306,XDEPS-4586] P0 source/reproducers/general/genxir_file source/reproducers/hybrid/oversubscribing:oversubscribing_more_device_memory_ath_run [GSD-786] source/reproducers/icx/ftrapv_error [CMPLRLLVM-44158] source/reproducers/ifx/complex_private:complex_private_WA [CMPLRLLVM-42381,GSD-3550,XDEPS-5672] P0 source/reproducers/ifx/emit_copy_from_ice [CMPLRLLVM-49292] source/reproducers/ifx/fopenmp-target-simd-data:fopenmp-target-simd-data_four source/reproducers/ifx/fopenmp-target-simd-data:fopenmp-target-simd-data_one source/reproducers/ifx/fopenmp-target-simd-data:fopenmp-target-simd-data_three source/reproducers/ifx/fopenmp-target-simd-data:fopenmp-target-simd-data_two source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_ath_compile [MKLD-16295] source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_batched_ath_run [MKLD-16295] source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_nobatched_ath_run [MKLD-16295] source/reproducers/mkl/minimal_zgeqrf_tester:minimal_zgeqrf_tester_bigger_ze00_ath_run [MKLD-15938] source/reproducers/mkl/minimal_zgeqrf_tester:minimal_zgeqrf_tester_ze00_ath_run [MKLD-15938] source/reproducers/mpi/openmp_target_zam:openmp_target_zam_zam1_ath_run [HPCS-8169] source/reproducers/opencl/concurrent_execution source/reproducers/openmp/FMM:FMM_ath_run [CMPLRLLVM-41769,CMPLRLLVM-41769(utils.F90)] source/reproducers/openmp/array_map:array_map_larger [GSD-5020,GSD-5208] source/reproducers/openmp/array_map:array_map_simple [GSD-5020,GSD-5208] source/reproducers/openmp/bigger_fmm_ice_array_reduction [CMPLRLLVM-48992] source/reproducers/openmp/bigger_fmm_ice_array_reduction:bigger_fmm_ice_array_reduction_n65 [CMPLRLLVM-48992] source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j19 [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j21 [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] source/reproducers/openmp/gamess_mini_wrong_answer:gamess_mini_wrong_answer_offload_ath_compile [CMPLRLLVM-45082] source/reproducers/openmp/hanging_gamess:hanging_gamess_AOT_version [CMPLRLLVM-39824,GSD-2102,XDEPS-5609] source/reproducers/openmp/hanging_gamess:hanging_gamess_ath_run [CMPLRLLVM-39824,GSD-2102,XDEPS-5609] source/reproducers/openmp/helpers_thread_is_not_helping:helpers_thread_is_not_helping_2 [XDEPS-3493] P0 source/reproducers/openmp/helpers_thread_is_not_helping:helpers_thread_is_not_helping_threads_2 [XDEPS-3493] P0 source/reproducers/openmp/ifx-O2:ifx-O2_ath_run [CMPLRLLVM-43398] P0 source/reproducers/openmp/mem_transfer_hang:mem_transfer_hang_ath_run source/reproducers/openmp/memcopy_multidevice source/reproducers/openmp/multi_gpu_hanging:multi_gpu_hanging_parallel_for [GSD-2288] P0 source/reproducers/openmp/multi_gpu_ms67:multi_gpu_ms67_ath_run source/reproducers/openmp/qmcpack.cpu.qmcpack.ompsimd.sincos:qmcpack.cpu.qmcpack.ompsimd.sincos_ath_compile [CMPLRLLVM-24010,CMPLRLLVM-28600] source/reproducers/openmp/task_reduction [CMPLRLLVM-23590] source/reproducers/openmp/vla_quda:vla_quda_ath_run [GSD-4147,GSD-4255,GSD-5040,IGC-7270] P0 Regressions: > Related to omp needing std=c++17 bug: source/reproducers/openmp/snap source/reproducers/openmp/snap.collapse source/reproducers/openmp/su3_bench:su3_bench_ath_compile [CMPLRLLVM-10536] source/reproducers/openmp/su3_bench:su3_bench_ath_run [CMPLRLLVM-10536] source/reproducers/openmp/testsnap_ats source/reproducers/openmp/thermo:thermo_ath_compile [CMPLRLIBS-34271,CMPLRLLVM-52687] source/reproducers/openmp/thermo:thermo_ath_run [CMPLRLIBS-34271,CMPLRLLVM-52687] source/reproducers/openmp/thermo:thermo_static_library [CMPLRLIBS-34271,CMPLRLLVM-52687] source/reproducers/openmp/vpo [CMPLRLLVM-19583] source/reproducers/openmp/raja.kernel_lambda_args [CMPLRLLVM-11192] source/reproducers/openmp/miniqmc-smallrun [CMPLRLLVM-27865] P0 > Related to “omp simd” fails: source/reproducers/ifx/CMPLRLLVM-35621 source/reproducers/ifx/XDEPS-5141 source/reproducers/ifx/phasta_target_simd [CMPLRLLVM-35621,CMPLRLLVM-40195,CMPLRLLVM-52024,GSD-6634] P0 source/reproducers/openmp/simd_loops [CMPLRLLVM-38420] P0 > fail from using too much scratch, fixed in UMD 775 or UMD 803.7 source/reproducers/openmp/ddpp_gamess_mini_wrong_answer_fp_precise [CMPLRLLVM-45082] source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j20 [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] source/reproducers/openmp/gamess_minispd_nan:gamess_minispd_nan_ath_compile [CMPLRLLVM-47316,GSD-2325,GSD-3402,GSD-4784,XDEPS-5607,XDEPS-6215] P0 > reported with new JIRAs: source/reproducers/dpcpp/supercontext:supercontext_shared source/reproducers/mkl/zgetrs_batch_slowdown [MKLD-15212,MKLD-15906,MKLD-16680] ________________________________ From: Ecp-aurora-sdk-discuss <ecp-aurora-sdk-discuss-bounces(a)lists.jlse.anl.gov> on behalf of Chan-nui, Christopher via Ecp-aurora-sdk-discuss <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> Sent: Friday, March 29, 2024 11:27 AM To: ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> Subject: [Ecp-aurora-sdk-discuss] New oneAPI release and eng-compiler versions for sunspot. With the Sunspot update, a new version of the Aurora SDK is now available on Sunspot. The module files for this release, which are loaded by default: oneapi/eng-compiler/2023. 12. 15. 002 This is based on the 2024. 0 release of the oneAPI SDK. ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd With the Sunspot update, a new version of the Aurora SDK is now available on Sunspot. The module files for this release, which are loaded by default: oneapi/eng-compiler/2023.12.15.002 This is based on the 2024.0 release of the oneAPI SDK. Release notes for the publicly available portions are available at: https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope… <https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope…> https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope… <https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope…> The engineering build contains bug fixes and performance improvements. Colleen will provide caveats, workarounds and other useful information about these releases shortly. -- Christopher Chan-Nui christopher.chan-nui(a)intel.com

1 0

New oneAPI release and eng-compiler versions for sunspot.
by Chan-nui, Christopher 29 Mar '24

29 Mar '24

With the Sunspot update, a new version of the Aurora SDK is now available on Sunspot. The module files for this release, which are loaded by default: oneapi/eng-compiler/2023.12.15.002 This is based on the 2024.0 release of the oneAPI SDK. Release notes for the publicly available portions are available at: https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope… https://urldefense.us/v3/__https://www.intel.com/content/www/us/en/develope… The engineering build contains bug fixes and performance improvements. Colleen will provide caveats, workarounds and other useful information about these releases shortly. -- Christopher Chan-Nui christopher.chan-nui(a)intel.com

1 0

DL Frameworks update for use with Aurora LustreApps queue
by Ramer, Rebecca 02 Feb '24

02 Feb '24

Hi, Stating the obvious, the NRE model DL frameworks workloads take a long time to initialize and run when the SW stack components, including the Aurora NRE model frameworks conda environment are located on the shared Lustre FS /soft mount. Until a permanent solution for this issue is deployed on Aurora, the SW stack components have been archived in a 'local-frameworks' tarball located on the gecko FS. This tarball can be untarred to compute node's /tmp directory. A script to setup the user's environment is then run. This will minimize the time to access the SW components when framework workloads are run. Please see this wiki page for more information on how to use the tarball: https://apps.cels.anl.gov/confluence/pages/viewpage.action?spaceKey=inteldg… Please let me know if you have any questions or run into any issues. Thanks, Rebecca Ramer rebecca.ramer(a)intel.com

1 0

Frameworks 2024.0 release available
by Ramer, Rebecca 22 Dec '23

22 Dec '23

I would like to announce availability of the 2024.0 DL Frameworks release with support for pytorch and tensorflow on the Aurora and Sunspot deployments. This release is based on the 2024.0 versions of the DeepLearning Frameworks and uses the recently announced oneAPI SDK release which uses Intel's 2024.0.0 oneapi basekit via the oneapi/release/2023.12.15.001 modulefile. This release is available by loading the frameworks modulefile: module use /soft/modulefiles module load frameworks/2023.12.15.001 Loading this module activates the aurora_nre_models_frameworks-2024.0 conda environment. Along with the pytorch and tensorflow packages, all necessary python packages needed to run the 11 'Aurora NRE' models have been installed into this conda environment. This frameworks modulefile replaces the 'hidden' modulefile frameworks/.2023.12.15.001 and is the default frameworks modulefile. NRE model workloads have been validated on Aurora and Sunspot using the frameworks/2023.12.15.001 modulefile. A couple of issues were encountered when validating this release on Aurora: 1. conda environment creation encountered a problem accessing package caches, the root cause is under-investigation, more information about this issue can be found here: https://wiki.jlse.anl.gov/display/inteldga/2024.0+-+Release#id-2024.0Releas… 2. distributed training runs scaling out beyond 8 nodes tend to hang or terminate with node failures, as these issues are identified, worked around, or resolved, information will be shared, and the release wiki pages will be updated here: https://wiki.jlse.anl.gov/display/inteldga/2024.0+-+Release#id-2024.0Releas… More information about the 2024.0 DL Frameworks release can be found here: https://wiki.jlse.anl.gov/display/inteldga/2024.0+-+Release Please let me know of any issues encountered. Rebecca Ramer rebecca.ramer(a)intel.com<mailto:[email protected]>

1 0

Re: [Ecp-aurora-sdk-discuss] New oneAPI release version
by Emil Briggs 13 Dec '23

13 Dec '23

NM. Needed to log out and then log back in. On 12/13/23 16:41, Emil Briggs wrote: > > How does one load this module? It won't load and spidering it doesn't > yield any useful information on what needs to be loaded first. > > > On 12/13/23 15:58, Bertoni, Colleen via Ecp-aurora-sdk-discuss wrote: >> Thanks Chris! >> Comparing this new release to the current latest >> (oneapi/eng-compiler/2023.10.15.002) we see 37 fixes and 5 >> regressions in the bug reproducers (below). >> >> For applications in the test set, we saw some performance >> improvements in the HYPRE and madgraph tests, and several fixes for >> kokkos unit tests (but one regression). The application owners have >> been notified. Two applications, QUDA and FMM saw regressions as >> well. Just a note in case you see it: the compile times on Aurora can >> be much slower than on Sunspot, but the runtimes are comparable or >> faster. >> Just to reiterate from last time, keep in mind that there’s a new >> directory structure. So if you were hardcoding to a library at >> MKLROOT/lib/intel64 for example, you’ll need to change it. And if you >> were using the subdevice() clause in OpenMP to target GPU tiles >> explicitly, that has been deprecated. Instead you can use >> gpu_tile_compact.sh or ONEAPI_DEVICE_SELECTOR to target tiles. >> >> Let us know if you have any questions, and thanks for all the bug >> reporting! >> Colleen >> >> >> Fixes: >> > ./ath.py compare --list --status=fixed --new-file >> result/x4407c0s2b0n0_2023-12-12-1702413670/output.json >> --reference-file standard/2023.10.15.002/reproducers/output.json >> source/reproducers/dpcpp/ATSsubdevice_regression >> source/reproducers/dpcpp/VirtualFunction:VirtualFunction_exw-virtuals-app_ath_run >> [CMPLRLLVM-35295,XDEPS-6157] P0 >> source/reproducers/dpcpp/VirtualFunction:VirtualFunction_minimal_ath_run >> [CMPLRLLVM-35295,XDEPS-6157] P0 >> source/reproducers/dpcpp/fortran_link:fortran_link_AOT >> [CMPLRLLVM-41268,cmplrllvm-44716] >> source/reproducers/dpcpp/kokkos_segfault_zebin [CMPLRLLVM-42875] >> source/reproducers/dpcpp/kokkos_sycl2a_fp_model_precise_relwithdebinfo >> [GSD-3120] >> source/reproducers/dpcpp/multigpu_multiqueue [CMPLRLLVM-49688] >> source/reproducers/dpcpp/vecadd_v0_multidevice [CMPLRLLVM-49688] >> source/reproducers/general/genxir_file >> source/reproducers/hybrid/oversubscribing:oversubscribing_more_device_memory_ath_run >> [GSD-786] >> source/reproducers/ifx/XDEPS-5191:XDEPS-5191_simd >> source/reproducers/ifx/emit_copy_from_ice [CMPLRLLVM-49292] >> source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_ath_compile >> [MKLD-16295] >> source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_batched_ath_run >> [MKLD-16295] >> source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_nobatched_ath_run >> [MKLD-16295] >> source/reproducers/mkl/minimal_zgeqrf_tester:minimal_zgeqrf_tester_bigger_ze00_ath_run >> source/reproducers/mkl/minimal_zgeqrf_tester:minimal_zgeqrf_tester_ze00_ath_run >> source/reproducers/mpi/openmp_target_zam:openmp_target_zam_zam1_ath_run >> [HPCS-8169] >> source/reproducers/opencl/concurrent_execution >> source/reproducers/openmp/array_map:array_map_larger [GSD-5020,GSD-5208] >> source/reproducers/openmp/array_map:array_map_simple [GSD-5020,GSD-5208] >> source/reproducers/openmp/bigger_fmm_ice_array_reduction >> [CMPLRLLVM-48992] >> source/reproducers/openmp/bigger_fmm_ice_array_reduction:bigger_fmm_ice_array_reduction_n65 >> [CMPLRLLVM-48992] >> source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j19 >> [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] >> source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j21 >> [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] >> source/reproducers/openmp/gamess_mini_wrong_answer:gamess_mini_wrong_answer_offload_ath_compile >> [CMPLRLLVM-45082] >> source/reproducers/openmp/gamess_mini_wrong_answer:gamess_mini_wrong_answer_offload_sp_ath_run >> [CMPLRLLVM-45082] >> source/reproducers/openmp/hanging_gamess:hanging_gamess_AOT_version >> [CMPLRLLVM-39824,GSD-2102,XDEPS-5609] >> source/reproducers/openmp/hanging_gamess:hanging_gamess_ath_run >> [CMPLRLLVM-39824,GSD-2102,XDEPS-5609] >> source/reproducers/openmp/helpers_thread_is_not_helping:helpers_thread_is_not_helping_2 >> [XDEPS-3493] P0 >> source/reproducers/openmp/helpers_thread_is_not_helping:helpers_thread_is_not_helping_threads_2 >> [XDEPS-3493] P0 >> source/reproducers/openmp/mem_transfer_hang:mem_transfer_hang_ath_run >> source/reproducers/openmp/memcopy_multidevice >> source/reproducers/openmp/multi_gpu_ms67:multi_gpu_ms67_ath_run >> source/reproducers/openmp/vla_quda:vla_quda_ath_run >> [GSD-4147,GSD-4255,GSD-5040,IGC-7270] P0 >> source/reproducers/tools/advisor_gflop:advisor_gflop.w_an_OpenMP_app >> source/reproducers/tools/advisor_gflop:advisor_gflop.w_an_OpenMP_app_ze_binary >> >> Regressions: >> source/reproducers/openmp/fortran_slice_leak [CMPLRLLVM-50506] >> source/reproducers/mpi/mpi_gatherv:mpi_gatherv_ath_run >> source/reproducers/mkl/zgetrs_batch_slowdown [MKLD-15212,MKLD-15906] >> source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j20_reduction >> [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] >> source/reproducers/tools/vtune_amrex_vector:vtune_amrex_vector.amrex_vector >> source/reproducers/tools/vtune_amrex_vector:vtune_amrex_vector.amrex_vector_vtune >> >> >> ------------------------------------------------------------------------ >> *From:* Ecp-aurora-sdk-discuss >> <ecp-aurora-sdk-discuss-bounces(a)lists.jlse.anl.gov> on behalf of >> Chan-nui, Christopher via Ecp-aurora-sdk-discuss >> <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> >> *Sent:* Wednesday, December 13, 2023 2:40 PM >> *To:* ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov >> <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> >> *Subject:* [Ecp-aurora-sdk-discuss] New oneAPI release version >> >> A new version of the Aurora SDK is now available on the Aurora and >> JLSE systems. >> >> The module files for this release are: >> >> oneapi/release/2023.12.15.001 >> >> This is based on the 2024.0 release of the oneAPI SDK. >> >> Release notes are available at >> >> https://www.intel.com/content/www/us/en/developer/articles/release-notes/in… >> <https://www.intel.com/content/www/us/en/developer/articles/release-notes/in…> >> >> https://www.intel.com/content/www/us/en/developer/articles/release-notes/in… >> <https://www.intel.com/content/www/us/en/developer/articles/release-notes/in…> >> >> Colleen will provide caveats, workarounds and other useful >> information about these releases shortly. >> >> -- >> >> Christopher Chan-Nui >> >> christopher.chan-nui(a)intel.com >> >>

1 0

Re: [Ecp-aurora-sdk-discuss] New oneAPI release version
by Emil Briggs 13 Dec '23

13 Dec '23

How does one load this module? It won't load and spidering it doesn't yield any useful information on what needs to be loaded first. On 12/13/23 15:58, Bertoni, Colleen via Ecp-aurora-sdk-discuss wrote: > Thanks Chris! > Comparing this new release to the current latest > (oneapi/eng-compiler/2023.10.15.002) we see 37 fixes and 5 regressions > in the bug reproducers (below). > > For applications in the test set, we saw some performance improvements > in the HYPRE and madgraph tests, and several fixes for kokkos unit > tests (but one regression). The application owners have been notified. > Two applications, QUDA and FMM saw regressions as well. Just a note in > case you see it: the compile times on Aurora can be much slower than > on Sunspot, but the runtimes are comparable or faster. > Just to reiterate from last time, keep in mind that there’s a new > directory structure. So if you were hardcoding to a library at > MKLROOT/lib/intel64 for example, you’ll need to change it. And if you > were using the subdevice() clause in OpenMP to target GPU tiles > explicitly, that has been deprecated. Instead you can use > gpu_tile_compact.sh or ONEAPI_DEVICE_SELECTOR to target tiles. > > Let us know if you have any questions, and thanks for all the bug > reporting! > Colleen > > > Fixes: > > ./ath.py compare --list --status=fixed --new-file > result/x4407c0s2b0n0_2023-12-12-1702413670/output.json > --reference-file standard/2023.10.15.002/reproducers/output.json > source/reproducers/dpcpp/ATSsubdevice_regression > source/reproducers/dpcpp/VirtualFunction:VirtualFunction_exw-virtuals-app_ath_run > [CMPLRLLVM-35295,XDEPS-6157] P0 > source/reproducers/dpcpp/VirtualFunction:VirtualFunction_minimal_ath_run > [CMPLRLLVM-35295,XDEPS-6157] P0 > source/reproducers/dpcpp/fortran_link:fortran_link_AOT > [CMPLRLLVM-41268,cmplrllvm-44716] > source/reproducers/dpcpp/kokkos_segfault_zebin [CMPLRLLVM-42875] > source/reproducers/dpcpp/kokkos_sycl2a_fp_model_precise_relwithdebinfo > [GSD-3120] > source/reproducers/dpcpp/multigpu_multiqueue [CMPLRLLVM-49688] > source/reproducers/dpcpp/vecadd_v0_multidevice [CMPLRLLVM-49688] > source/reproducers/general/genxir_file > source/reproducers/hybrid/oversubscribing:oversubscribing_more_device_memory_ath_run > [GSD-786] > source/reproducers/ifx/XDEPS-5191:XDEPS-5191_simd > source/reproducers/ifx/emit_copy_from_ice [CMPLRLLVM-49292] > source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_ath_compile > [MKLD-16295] > source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_batched_ath_run > [MKLD-16295] > source/reproducers/mkl/batched_dgemm_incorrect:batched_dgemm_incorrect_nobatched_ath_run > [MKLD-16295] > source/reproducers/mkl/minimal_zgeqrf_tester:minimal_zgeqrf_tester_bigger_ze00_ath_run > source/reproducers/mkl/minimal_zgeqrf_tester:minimal_zgeqrf_tester_ze00_ath_run > source/reproducers/mpi/openmp_target_zam:openmp_target_zam_zam1_ath_run > [HPCS-8169] > source/reproducers/opencl/concurrent_execution > source/reproducers/openmp/array_map:array_map_larger [GSD-5020,GSD-5208] > source/reproducers/openmp/array_map:array_map_simple [GSD-5020,GSD-5208] > source/reproducers/openmp/bigger_fmm_ice_array_reduction [CMPLRLLVM-48992] > source/reproducers/openmp/bigger_fmm_ice_array_reduction:bigger_fmm_ice_array_reduction_n65 > [CMPLRLLVM-48992] > source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j19 > [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] > source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j21 > [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] > source/reproducers/openmp/gamess_mini_wrong_answer:gamess_mini_wrong_answer_offload_ath_compile > [CMPLRLLVM-45082] > source/reproducers/openmp/gamess_mini_wrong_answer:gamess_mini_wrong_answer_offload_sp_ath_run > [CMPLRLLVM-45082] > source/reproducers/openmp/hanging_gamess:hanging_gamess_AOT_version > [CMPLRLLVM-39824,GSD-2102,XDEPS-5609] > source/reproducers/openmp/hanging_gamess:hanging_gamess_ath_run > [CMPLRLLVM-39824,GSD-2102,XDEPS-5609] > source/reproducers/openmp/helpers_thread_is_not_helping:helpers_thread_is_not_helping_2 > [XDEPS-3493] P0 > source/reproducers/openmp/helpers_thread_is_not_helping:helpers_thread_is_not_helping_threads_2 > [XDEPS-3493] P0 > source/reproducers/openmp/mem_transfer_hang:mem_transfer_hang_ath_run > source/reproducers/openmp/memcopy_multidevice > source/reproducers/openmp/multi_gpu_ms67:multi_gpu_ms67_ath_run > source/reproducers/openmp/vla_quda:vla_quda_ath_run > [GSD-4147,GSD-4255,GSD-5040,IGC-7270] P0 > source/reproducers/tools/advisor_gflop:advisor_gflop.w_an_OpenMP_app > source/reproducers/tools/advisor_gflop:advisor_gflop.w_an_OpenMP_app_ze_binary > > Regressions: > source/reproducers/openmp/fortran_slice_leak [CMPLRLLVM-50506] > source/reproducers/mpi/mpi_gatherv:mpi_gatherv_ath_run > source/reproducers/mkl/zgetrs_batch_slowdown [MKLD-15212,MKLD-15906] > source/reproducers/openmp/gamess_mini_all_kernels:gamess_mini_all_kernels_j20_reduction > [CMPLRLLVM-45082,CMPLRLLVM-45890,CMPLRTST-21020,IGC-6246,XDEPS-5609,GSD-5514] > source/reproducers/tools/vtune_amrex_vector:vtune_amrex_vector.amrex_vector > source/reproducers/tools/vtune_amrex_vector:vtune_amrex_vector.amrex_vector_vtune > > > ------------------------------------------------------------------------ > *From:* Ecp-aurora-sdk-discuss > <ecp-aurora-sdk-discuss-bounces(a)lists.jlse.anl.gov> on behalf of > Chan-nui, Christopher via Ecp-aurora-sdk-discuss > <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> > *Sent:* Wednesday, December 13, 2023 2:40 PM > *To:* ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov > <ecp-aurora-sdk-discuss(a)lists.jlse.anl.gov> > *Subject:* [Ecp-aurora-sdk-discuss] New oneAPI release version > > A new version of the Aurora SDK is now available on the Aurora and > JLSE systems. > > The module files for this release are: > > oneapi/release/2023.12.15.001 > > This is based on the 2024.0 release of the oneAPI SDK. > > Release notes are available at > > https://www.intel.com/content/www/us/en/developer/articles/release-notes/in… > <https://www.intel.com/content/www/us/en/developer/articles/release-notes/in…> > > https://www.intel.com/content/www/us/en/developer/articles/release-notes/in… > <https://www.intel.com/content/www/us/en/developer/articles/release-notes/in…> > > Colleen will provide caveats, workarounds and other useful information > about these releases shortly. > > -- > > Christopher Chan-Nui > > christopher.chan-nui(a)intel.com > >

1 0