Wire delay is not a problem for SMT (in the near future)

Previous papers have shown that the slow scaling of wire delays compared to logic delays will prevent superscalar performance from scaling with technology. In this paper, we show that the optimal pipeline for superscalar becomes shallower with technology, when wire delays are considered, tightening previous results that deeper pipelines perform only as well as shallower pipelines. The key reason for the lack of performance scaling is that superscalar does not have sufficient parallelism to hide the relatively-increased wire delays. However, Simultaneous Multithreading (SMT) provides the much-needed parallelism. We show that an SMT running a multiprogrammed workload with just 4-way issue not only retains the optimal pipeline depth over technology generations, enabling at least 43% increase in clock speed every generation, but also achieves the remainder of the expected speedup of two per generation through IPC. As wire delays become more dominant in future technologies, the number of programs needs to be scaled modestly to maintain the scaling trends, at least till the near-future 50nm technology. While this result ignores bandwidth constraints, using SMT to tolerate latency due to wire delays is not that simple because SMT causes bandwidth problems. Most of the stages of a modern out-of-order-issue pipeline employ RAM and CAM structures. Wire delays in conventional, latency-optimized RAM/CAM structures prevent them from being pipelined in a scaled manner. We show that this limitation prevents scaling of SMT throughput. We use bitline scaling to allow RAM/CAM bandwidth to scale with technology. Bitline scaling enables SMT throughput to scale at the rate of two per technology generation in the near future.

[1]  Yale N. Patt,et al.  On pipelining dynamic instruction scheduling logic , 2000, MICRO 33.

[2]  Rajiv V. Joshi,et al.  A 2-ns cycle, 3.8-ns access 512-kb CMOS ECL SRAM with a fully pipelined architecture , 1991 .

[3]  Todd M. Austin,et al.  Efficient dynamic scheduling through tag elimination , 2002, ISCA.

[4]  Steven K. Reinhardt,et al.  A scalable instruction queue design using dependence chains , 2002, ISCA.

[5]  Vivek De,et al.  Technology and design challenges for low power and high performance [microprocessors] , 1999, Proceedings. 1999 International Symposium on Low Power Electronics and Design (Cat. No.99TH8477).

[6]  M.A. Horowitz,et al.  Speed and power scaling of SRAM's , 2000, IEEE Journal of Solid-State Circuits.

[7]  Dean M. Tullsen,et al.  Handling long-latency loads in a simultaneous multithreading processor , 2001, Proceedings. 34th ACM/IEEE International Symposium on Microarchitecture. MICRO-34.

[8]  Gurindar S. Sohi,et al.  High-bandwidth data memory systems for superscalar processors , 1991, ASPLOS IV.

[9]  Joel S. Emer,et al.  Loose loops sink chips , 2002, Proceedings Eighth International Symposium on High Performance Computer Architecture.

[10]  Dean M. Tullsen,et al.  Exploiting Choice: Instruction Fetch and Issue on an Implementable Simultaneous Multithreading Processor , 1996, 23rd Annual International Symposium on Computer Architecture (ISCA'96).

[11]  K. Ishibashi,et al.  A 2 ns access, 285 MHz, two-port cache macro using double global bit-line pairs , 1997, 1997 IEEE International Solids-State Circuits Conference. Digest of Technical Papers.

[12]  Todd M. Austin,et al.  The SimpleScalar tool set, version 2.0 , 1997, CARN.

[13]  Eric Sprangle,et al.  Increasing processor performance by implementing deeper pipelines , 2002, ISCA.

[14]  Vikas Agarwal,et al.  Clock rate versus IPC: the end of the road for conventional microarchitectures , 2000, Proceedings of 27th International Symposium on Computer Architecture (IEEE Cat. No.RS00201).

[15]  M. Horowitz,et al.  Managing wire scaling: a circuit perspective , 2003, Proceedings of the IEEE 2003 International Interconnect Technology Conference (Cat. No.03TH8695).

[16]  David J. Sager,et al.  The microarchitecture of the Pentium 4 processor , 2001 .

[17]  Kaushik Roy,et al.  Exploring high bandwidth pipelined cache architecture for scaled technology , 2003, 2003 Design, Automation and Test in Europe Conference and Exhibition.

[18]  Sumio Tanaka,et al.  A 9-ns HIT-delay 32-kbyte cache macro for high-speed RISC , 1990 .

[19]  Richard E. Kessler,et al.  The Alpha 21264 microprocessor , 1999, IEEE Micro.

[20]  Kaushik Roy,et al.  Exploring High Bandwidth Pipelined Cache Architecture for Scaled Technology , 2003, Embedded Software for SoC.

[21]  Yale N. Patt,et al.  Using internal redundant representations and limited bypass to support pipelined adders and register files , 2002, Proceedings Eighth International Symposium on High Performance Computer Architecture.

[22]  Kunle Olukotun,et al.  Increasing Cache Port Efficiency for Dynamic Superscalar Microprocessors , 1996, 23rd Annual International Symposium on Computer Architecture (ISCA'96).

[23]  G. Varghese,et al.  A pipelined memory architecture for high throughput network processors , 2003, 30th Annual International Symposium on Computer Architecture, 2003. Proceedings..

[24]  Ken Mai,et al.  The future of wires , 2001, Proc. IEEE.

[25]  裕幸 飯田,et al.  International Technology Roadmap for Semiconductors 2003の要求清浄度について － シリコンウエハ表面と雰囲気環境に要求される清浄度, 分析方法の現状について － , 2004 .

[26]  Dean M. Tullsen,et al.  Simultaneous multithreading: Maximizing on-chip parallelism , 1995, Proceedings 22nd Annual International Symposium on Computer Architecture.

[27]  James E. Smith,et al.  Complexity-Effective Superscalar Processors , 1997, ISCA.

[28]  Norman P. Jouppi,et al.  Cacti 3. 0: an integrated cache timing, power, and area model , 2001 .