[ mikikg @ 09.12.2019. 12:14 ] @
Ovo izgleda vrlo vrlo zanimljivo ...

Citat:

DPC++ shorthand for Data Parallel C++, it’s the new direct programming language of oneAPI—an Intel-led initiative to unify and simplify application development across diverse computing architectures.

DPC++ is based on familiar (and industry-standard) C++, incorporates SYCL* specification 1.2.1 from The Khronos Group*, and includes language extensions developed using an open community process. Purposely designed as an open, cross-industry alternative to single-architecture, proprietary languages, DPC++ enables developers to more easily port their code across CPUs, GPUs, and FPGAs, and also tune performance for a specific accelerator.

The oneAPI Specification
The oneAPI specification extends existing developer programming models to enable a diverse set of hardware through language, a set of library APIs, and a low level hardware interface to support cross-architecture programming. To promote compatibility and enable developer productivity and innovation, the oneAPI specification builds upon industry standards and provides an open, cross-platform developer stack.

The Language
At the core of the oneAPI specification is DPC++, an open, cross-architecture language built upon the ISO C++ and Khronos SYCL standards. DPC++ extends these standards and provides explicit parallel constructs and offload interfaces to support a broad range of computing architectures and processors, including CPUs and accelerator architectures.

The Libraries
The set of APIs spans several domains that benefit from acceleration, including an interface for deep learning; general libraries for linear algebra math, video, and media processing; and others.

The Hardware Abstraction Layer
The low-level hardware interface defines a set of capabilities and services that a hardware accelerator needs to interface with the broad set of languages in support of consumer to Deep/Machine Learning and HPC class solutions.


https://techdecoded.intel.io/e...&erpm_id=8844009#gs.ldjhsp

https://techdecoded.intel.io/topics/oneapi/

https://www.oneapi.com
[ Branimir Maksimovic @ 09.12.2019. 13:48 ] @
Izgleda i Intel je ulozio neke pare u reklamu ocigledno....
[ nkrgovic @ 09.12.2019. 21:18 ] @
U cemu je tacno prednost u odnosu na TensorFlow ili CUDA? Osim sto oni rade za brdo jezika i ne zahteva intel kompajler.... Ovo lici na nesto izmedju TensorFlow i CUDA, gde oba lepo rade sa mnogo vise jezika, zasto bi neko furao ovo?
[ Branimir Maksimovic @ 09.12.2019. 22:04 ] @
Pa kazu oni, open source i te fore. Unifikacija. Ali tu je OpenCL koji se svuda koristi. Ova Intelova inicijativa svakako da se nece svideti Nvidiji vec sam video kako CEO Nvidije kritikuje ovo.