Intel®, in collaboration with Analytics India Journal, lately concluded its Intel® oneAPI workshop on superior SYCL ideas for heterogeneous computing, on June 24, 2022, at IST 5:00 PM. The session witnessed near 200+ members.
The workshop coated superior ideas and options of the most recent SYCL specs, together with simplified code implementation for heterogeneous computing, an summary of reminiscence fashions involving buffers/accessors and unified shared reminiscence (USM), and an understanding of simplified discount with reside hands-on pattern codes on Intel® DevCloud.
The session was led by Jyotsna Khemka, software program engineering supervisor at Intel Company – Asia Pacific & Japan – South, alongside Subarnarekha Ghosal, a software program know-how consulting engineer at Intel Company.
Key highlights
Opening the workshop, Khemka threw gentle on the oneAPI use circumstances, showcasing one of many case research executed by the College in Berlin, the place they developed a Tsunami software referred to as the Straightforward Wave Simulation, which was developed for the Nvidia GPU. The crew used oneAPI as a programming mannequin, transformed that code into SYCL, and was capable of apply it to completely different sorts of {hardware} structure.
Additional, she spoke about varied oneAPI toolkits. She stated that a few of these instruments enable customers to transform their CUDA code to SYCL and use this programming language to run on a number of {hardware} and cases.
Following this, Ghosal launched SYCL 2020 language specification. Highlighting the most recent options, she stated that it enhances productiveness, and greater than something, it reduces the verbosity of writing code and runs the code sooner throughout laptops. She additionally spoke about SYCL Commonplace, an open collaboration platform. “SYCL is rising daily, and it’s rising taking the suggestions,” she added.
Merely put, Intel®’s SYCL* customary helps in enhancing productiveness, efficiency, and fast-tracking open collaboration.
Evaluating the most recent model of SYCL with conventional coding, she gave an summary of varied parameters that go into writing the code, together with the SYCL buffers methodology, explaining host code, accelerator system code and host code. As well as, she additionally showcased the way it simplifies language simplification and extra.
This was adopted by a hands-on information to studying pointer-based reminiscence administration for heterogeneous computing utilizing unified shared reminiscence, alongside understanding the implicit and express methods of shifting reminiscence utilizing unified shared reminiscence and dealing with knowledge dependency between kernel executions.
As well as, Ghosal showcased some great benefits of utilizing Subgroups in SYCL programming, deciphering varied ideas, and defined how Subgroup Shuffle operations may assist keep away from express reminiscence operations. She additionally showcased a hands-on rationalization of utilizing SYCL discount to simplify discount with parallel kernels and urged the customers to take benefit and cut back features to scale back sub_group and work-group ranges.
Click on right here to obtain Intel® oneAPI Toolkits to get began.
Click on right here to create an Intel® DevCloud account.
On the finish of the work, Analytics India Journal ran a fortunate draw, whereby fortunate members received an Amazon Voucher value INR 2000/- every. The winners had been shortlisted based mostly on their engagement with Discord all through the workshop.
- Nikhitha Avula
- Prasanta Kundu
- Anirban Malla
- Anirban Dasgupta
- Aswin Vijayakumar
- Vidyasagar M
- Rakesh Roy
- Karthik Koundinya S R
- Rahul Raj
- Shobana LakshmiNarsimhan
Right here’s a Github hyperlink to discover the lab session. Join Intel DevCloud.
