Fig 1 - uploaded by Artem Kramov
Content may be subject to copyright.
Source publication
Запропоновано метод розроблення та верифікації застосувань для систем з масовим паралелізмом на основі відеоадаптерів від компанії NVIDIA, який дозволяє створювати абстракції різних рівнів за допомогою апарата розмічених транзиційних систем. Композиції таких систем трансформуються в мережі Петрі, які далі аналізуються відповідними засобами. Метод т...
Contexts in source publication
Context 1
... model in the NVIDIA CUDA architecture, based on LTS and Petri nets was presented in [9]. The main emphasis was put on obtaining a high-level model without a detailed examination of the labels semantics of each transition system. The following briefly recalls the main details: three LTS were emerged as a result of CUDA application decomposition: (Fig. 1, 2, 3). The main activities modelled at this level of abstractions include planning, choosing warp and instruction for it and providing exclusive access to computational resources for the execution time of each warp-instruction tuple. This paper shows creation of a model of the lower level of abstraction, which will allow us to analyze the ...
Context 2
... model in the NVIDIA CUDA architecture, based on LTS and Petri nets was presented in [9]. The main emphasis was put on obtaining a high-level model without a detailed examination of the labels semantics of each transition system. The following briefly recalls the main details: three LTS were emerged as a result of CUDA application decomposition: (Fig. 1, 2, 3). The main activities modelled at this level of abstractions include planning, choosing warp and instruction for it and providing exclusive access to computational resources for the execution time of each warp-instruction tuple. This paper shows creation of a model of the lower level of abstraction, which will allow us to analyze the ...
