A flexible simulation framework for graphics architectures

Abstract

In

Analyzing CUDA workloads using a detailed GPU simulator

“…Existing graphics-oriented GPU simulators include Qsilver [43], which does not model programmable shaders, and ATTILLA [10], which focuses on graphics specific features. Ryoo et al [41] use CUDA to speedup a variety of relatively easily parallelizable scientific applications.…”

Applications of Small-Scale Reconfigurability to Graphics Processors

“…The primary advantage of using Chromium is that we ensure that our workloads are not contrived. Although we use Brook and Chromium without modification, we have enhanced the Qsilver graphics architectural simulator [10,11] to model the necessary aspects of the fragment pipeline. A detailed description of our modifications to QSilver and our experimental setup are presented in sections 3 and 4.…”

“…Qsilver is a simulation framework for graphics architectures that can simulate low-level GPU activity for any existing OpenGL application [10]. Qsilver uses Chromium [7] to intercept and transform an OpenGL application's API calls and create an annotated trace that encapsulates geometry, timing, and state information.…”

Studying Thermal Management for Graphics-Processor Architectures

“…In this paper, we use Qsilver [17] to explore a series of thermal management techniques, ranging from classical dynamic voltage scaling (DVS) and clock gating, to techniques like multiple clock domains and temperature-aware floorplans, both of which specifically exploit the parallelism of graphics workloads. Though we also describe two energy efficiency experiments, one varying throughput in the vertex and fragment engines and the other using Multiple Clock Domains or MCD while varying leakage rate, we choose thermal management as a driving problem because we feel it is an area with rich rewards for graphics architecture and an area that can draw a great deal from past work in general purpose processor design.…”