Introducing Runahead Threads for SMT Processors

نویسندگان

  • Tanausú Ramírez
  • Alex Pajuelo
  • Oliverio J. Santana
  • Mateo Valero
چکیده

In this paper, we propose Runahead threads on Simultaneous Multithreading processors as a valuable solution for both exploiting the memory-level parallelism and reducing the resource contention. This approach transforms a memory-bounded eager resource thread into a speculative light thread, alleviating critical resource con icts among multiple threads. Furthermore, it improves the threadlevel parallelism by removing long-latency memory operations from the instruction window, and thus it releases busy resources otherwise. We compare an SMT architecture using Runahead threads (SMTRA) to both state-of-the-art static fetch and dynamic resource control policies. Our results show that the SMTRA combination performs better, in terms of throughput and fairness, than any of the other policies.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

MLP-Aware Runahead Threads in a Simultaneous Multithreading Processor

Threads experiencing long-latency loads on a simultaneous multithreading (SMT) processor may clog shared processor resources without making forward progress, thereby starving other threads and reducing overall system throughput. An elegant solution to the long-latency load problem in SMT processors is to employ runahead execution. Runahead threads do not block commit on a longlatency load but i...

متن کامل

Integrating Multiple Forms of Multithreaded Execution on SMT Processors: A Quantitative Study with Scientific Workloads

Simultaneous multithreaded (SMT) processors have penetrated the mainstream computing market, since they offer a number of cost / performance advantages over conventional superscalar processors at a nominal additional cost. Simultaneous multithreading can be used in the execution engine of a single monolithic microprocessor, or be embedded and replicated in the execution cores of a chip multipro...

متن کامل

Algebraic Models of Simultaneous Multi-Threaded and multi-core Microprocessors

Superscalar microprocessors execute multiple instructions simultaneously by virtue of large amounts of (possibly duplicated) hardware. Much of this hardware is idle at least part of the time. simultaneous multithreaded (SMT) microprocessors utilize this idle hardware by interleaving multiple independent execution threads. In essence, a single physical processor appears to be multiple virtual pr...

متن کامل

Sensitivity Analysis on Hardware Resources in SMT Processors

The microprocessor performance is increased by allowing multiple threads per clock cycle to issue instructions in simultaneous multithreading processors. Shared hardware resources are the key components of SMT processor performance. In SMT processors, impact to the performance will not be same among many hardware resources. To design an optimal hardware configuration for SMT processors, sensiti...

متن کامل

Understanding the Impact of Inter-Thread Cache Interference on ILP in Modern SMT Processors

Simultaneous Multithreading (SMT) has emerged as an effective method of increasing utilization of resources in modern super-scalar processors. SMT processors increase instruction-level parallelism (ILP) and resource utilization by simultaneously executing instructions from multiple independent threads. Although simultaneously sharing resources benefits system throughput, coscheduled threads oft...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2007