First public working draft from W3C: WebGPU and WebGPU shading language
Kickstarting AI for Code: Introducing IBM’s Project CodeNet
12-May-2021: IBM announces CodeNet, a collection of 14 million code examples that solve 4053 typical programming problems, a total of 8 GB of archived data. We presume that the problem of improving the quality of source codes will be constantly and highly relevant. Also, we believe that the active introduction of machine learning technologies in this problem space attests to the successful results of such research. Consequently, it will lead to a rapid increase in the share of artificial intelligence in the source codes analysis compared to static methods.
QEMU version 6.0.0 released
30-Apr-2021: The undoubted advantage of QEMU is that it allows emulating any architecture on any machine (or server). The new 6.0 version offers a significant number of changes to support modern platforms. In our opinion, the tool is a sound basis for more in-depth testing of the quality and functionality of the source code, for example, for building CI/CD systems that simultaneously support several hardware architectures. As a result, it will simplify the development of system software, compilers, virtual machines, and other platform-dependent code.
HPVM v1.0 released
9-Apr-2021: Heterogeneous computing refers to systems that use more than one kind of processor or cores. Many chip manufacturers offer their solutions, in particular, for the binding of the CPU and neural processors. The disadvantages of this approach are proprietary source code of compilers and closed interfaces, which do not allow developers to improve the quality of such applications and implement code analyzers and automatic bug fixing tools. The new HPVM compiler is an attempt to find an effective standardized common solution. This solution positioning itself as a more low-level and flexible approach than competitive frameworks such as OpenCL. We presume that the release of the HPVM backend will encourage the development of high-performance specified programming languages related to heterogeneous computing such as graphics processing, neural algorithms, or neural networks.