P
Pedro Lopez
Researcher at Intel
Publications - 28
Citations - 468
Pedro Lopez is an academic researcher from Intel. The author has contributed to research in topics: Cache & Thread (computing). The author has an hindex of 12, co-authored 28 publications receiving 468 citations. Previous affiliations of Pedro Lopez include Polytechnic University of Catalonia & Demos.
Papers
More filters
Patent
Register checkpointing mechanism for multithreading
Pedro Lopez,Carlos Madriles,Alejandro Martinez,Raul Martinez,Josep M. Codina,Enric Gibert Codina,Fernando Latorre,Antonio Gonzalez +7 more
TL;DR: In this paper, a register checkpointing mechanism is proposed to resolve multithreading mis-speculations in a multi-threaded system, where multiple thread units concurrently execute threads to initialize memory to store active checkpoint data including register contents and a checkpoint mask indicating the validity of stored contents.
Patent
Weight-shifting mechanism for convolutional neural networks
Ayose Falcón,Marc Lupon,Herrero Abellanas Enric,Pedro Lopez,Fernando Latorre,Frederico Pratas,Georgios Tournavitis +6 more
TL;DR: A processor includes a processor core and a calculation circuit as discussed by the authors, which includes logic to determine a set of weights for use in a convolutional neural network (CNN) calculation and scale up the weights using a scale value.
Patent
Storage device and method for performing convolution operations
Herrero Abellanas Enric,Georgios Tournavitis,Frederico Pratas,Marc Lupon,Fernando Latorre,Pedro Lopez,Ayose Falcón +6 more
TL;DR: In this paper, a storage device and method for performing convolution operations is described, which comprises a plurality of processing units to execute convolution operation on input data and partial results.
Patent
Method and apparatus for distributed and cooperative computation in artificial neural networks
Frederico Pratas,Ayose Falcón,Marc Lupon,Fernando Latorre,Pedro Lopez,Enric Herrero Abellanas,Georgios Tournavitis +6 more
TL;DR: In this paper, an apparatus and method for distributed and cooperative computation in artificial neural networks is described, which comprises an input/output (I/O) interface, a plurality of processing units communicatively coupled to the I/O interface to receive data for input neurons and synaptic weights associated with each of the input neurons, each unit processing at least a portion of the data for the inputs and weights to generate partial results.
Proceedings ArticleDOI
Boosting single-thread performance in multi-core systems through fine-grain multi-threading
Carlos Madriles,Pedro Lopez,Josep M. Codina,Enric Gibert,Fernando Latorre,Alejandro Martinez,Raul Martinez,Antonio González +7 more
TL;DR: The proposed scheme outperforms previous hardware-only schemes to implement the idea of combining cores for executing single-thread applications in a multi-core design by more than 10% on average on Spec2006 for all configurations, and single- threads performance is improved by 41% when the proposed scheme is used on a Tiny Core, and up to 2.6x for some selected applications.