Offloading compute-intensive kernels to hardware accelerators relies on the large degree of
parallelism offered by these platforms. However, the effective bandwidth of the memory …

This paper discusses several automated methods of acceleration of program operation. The
acceleration is achieved by parallelization and optimization of memory access. Optimization …

The increasing complexity of transformer models in artificial intelligence expands their
computational costs, memory usage, and energy consumption. Hardware acceleration …

In this work, the perspective of using Optimizing Parallelizing System (http://ops. rsu. ru/en/)
together with Clang compiler is considered. The converters from Clang intermediate …

As processors compute power keeps increasing, so do their demands in memory accesses:
some computations will require a higher bandwidth and exhibit regular memory access …

Dense linear algebra codes are often expressed and coded in terms of BLAS calls. This
approach, however, achieves suboptimal performance due to the overheads associated to …

Cette thèse aborde les difficultés de mise au point de codes multicibles-c'est-à-dire de
codes dont les performances sont portables entre différentes cibles matérielles. Nous avons …

Efficient execution on processors with multiple cores requires the exploitation of parallelism
within the processor. For many dense linear algebra codes this, in turn, requires the efficient …

In this paper, we introduce a supernode amalgamation algorithm which takes into account
the characteristics of a hypermatrix data structure. The resulting frontal tree is then used to …

Представлен метод преобразования размещения матриц между строчным и блочным
представлениями. Строчное размещение используется в языке программирования Си …