From 1b32aea2910787492269ad65cb03e01477139bb3 Mon Sep 17 00:00:00 2001 From: Shwetha-Selma Date: Sun, 21 Jan 2024 23:56:46 -0800 Subject: [PATCH 1/2] updated README with single precision option --- Libraries/oneMKL/batched_linear_solver/README.md | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/Libraries/oneMKL/batched_linear_solver/README.md b/Libraries/oneMKL/batched_linear_solver/README.md index 7d3fc63521..b419c4fd10 100644 --- a/Libraries/oneMKL/batched_linear_solver/README.md +++ b/Libraries/oneMKL/batched_linear_solver/README.md @@ -52,6 +52,15 @@ Run `make` to build and run the sample. Three programs are generated: Note that the makefile only runs small tests to verify that the executables are working correctly. The problem sizes are too small to justify accelerator offload. Use the following command-line options to run the tests shown in [Solving Linear Systems Using oneMKL and OpenMP Target Offloading](https://www.intel.com/content/www/us/en/developer/articles/technical/solve-linear-systems-onemkl-openmp-target-offload.html): `-n 16000 -b 8 -r 1 -c 5`. +> **Note**: By default this makefile will be executable on devices with double precision support (fp64). +> To execute on devices with only single precision support(e.g., gen11, gen12) include -DSP option as follows in makefile +> +> ``` +> IFX_OPTS_OFFLOAD = -DSP -DMKL_ILP64 -qopenmp -fopenmp-targets=spir64 -fsycl -L${MKLROOT}/lib/intel64 -lmkl_sycl - +>lmkl_intel_ilp64 -lmkl_intel_thread -lmkl_core -liomp5 -lpthread -ldl +> ``` +> + You can remove all generated files with `make clean`. ### Example of Output From 31128144b1442bc6a9aa2f14f73ce2b06a8ce06e Mon Sep 17 00:00:00 2001 From: Shwetha-Selma <134358522+Shwetha-Selma@users.noreply.github.com> Date: Mon, 22 Jan 2024 13:31:53 +0530 Subject: [PATCH 2/2] updated README with single precision option --- Libraries/oneMKL/batched_linear_solver/README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Libraries/oneMKL/batched_linear_solver/README.md b/Libraries/oneMKL/batched_linear_solver/README.md index b419c4fd10..8de10736b9 100644 --- a/Libraries/oneMKL/batched_linear_solver/README.md +++ b/Libraries/oneMKL/batched_linear_solver/README.md @@ -42,7 +42,7 @@ Third party program Licenses can be found here: [third-party-programs.txt](https >For more information on environment variables, see [Use the setvars Script for Linux or macOS](https://www.intel.com/content/www/us/en/develop/documentation/oneapi-programming-guide/top/oneapi-development-environment-setup/use-the-setvars-script-with-linux-or-macos.html). ### Running Samples on the DevCloud -When running a sample in the Intel DevCloud, remember that you must specify the compute node (CPU, GPU, FPGA) as well whether to run in batch or interactive mode. For more information see the Intel® oneAPI Base Toolkit Get Started Guide (https://devcloud.intel.com/oneapi/get-started/base-toolkit/). +When running a sample in the Intel DevCloud, remember that you must specify the compute node (CPU, GPU, FPGA) as well as whether to run in batch or interactive mode. For more information see the Intel® oneAPI Base Toolkit Get Started Guide (https://devcloud.intel.com/oneapi/get-started/base-toolkit/). Run `make` to build and run the sample. Three programs are generated: @@ -53,7 +53,7 @@ Run `make` to build and run the sample. Three programs are generated: Note that the makefile only runs small tests to verify that the executables are working correctly. The problem sizes are too small to justify accelerator offload. Use the following command-line options to run the tests shown in [Solving Linear Systems Using oneMKL and OpenMP Target Offloading](https://www.intel.com/content/www/us/en/developer/articles/technical/solve-linear-systems-onemkl-openmp-target-offload.html): `-n 16000 -b 8 -r 1 -c 5`. > **Note**: By default this makefile will be executable on devices with double precision support (fp64). -> To execute on devices with only single precision support(e.g., gen11, gen12) include -DSP option as follows in makefile +> To execute on devices with only single precision support(e.g., gen11, gen12) include `-DSP` option as follows in makefile > > ``` > IFX_OPTS_OFFLOAD = -DSP -DMKL_ILP64 -qopenmp -fopenmp-targets=spir64 -fsycl -L${MKLROOT}/lib/intel64 -lmkl_sycl -