| Skadron, Kevin, Martonosi, Margaret, and Clark, Douglas W. Alloying Global and Local Branch History: Taxonomy, Performance, and Analysis. Tech. rep., Princeton Dept. of Computer Science, 1999. |
....rate achieved by MAs. must be used. Finding the best composition of PHT index bits was done using brute force, simulating all possible combinations of global, local, and address bits for the desired branch predictor size (plots of this design space for gcc and m88ksim can be found in [16]. Finding equal area configurations must also account for the BHT s size. We explored all possible BHT configurations for the chosen size, ranging from wide and short (many local history bits and few BHT entries) to narrow and tall. The GAs, PAs, and MAs configurations chosen appear in Table 2. ....
K. Skadron, M. Martonosi, and D. W. Clark. Alloying global and local branch history: Taxonomy, performance, and analysis. Tech. Report TR-594-99, Princeton Dept. of Computer Science, Jan. 1999.
....also have a large static branch footprint, while most SPEC programs do not [11] Other fundamental research has focused on understanding and improving branch prediction accuracy via both hardware and software means. Lee, Chen, and Mudge [32] Sechrest, Lee, and Mudge [49] and Skadron et al. [53], among others, have described and measured the different causes of mispredictions, and Evers et al. 13] recently measured the benefits of branch correlation. Work by Yeh and Patt [65] Pan et al. 40] McFarling [36] Sprangle et al. 57] and Eden and Mudge [10] among others, have proposed ....
K. Skadron, M. Martonosi, and D. W. Clark. Alloying global and local branch history: Taxonomy, performance, and analysis. Technical Report TR-594-99, Princeton Dept. of Computer Science, Jan. 1999.
....and local history available, Section 4.5 compares alloying to two varieties of hybrid prediction and Section 4.6 briefly compares small sizes of all the predictor organizations. Section 4.7 presents related work, and Section 4.8 reviews the chapter s results. Note that this work also appears as [88]. 4.2 Taxonomy As mentioned, many studies have measured the importance of PHT interference as a source of mispredictions [10, 57, 64, 66, 82, 94, 96, 116] and other work has shown the importance of wrong history mispredictions [21] This section unifies the di#erent misprediction types in a ....
K. Skadron, M. Martonosi, and D. W. Clark. Alloying global and local branch history: Taxonomy, performance, and analysis. Tech. Report TR-594-99, Princeton University Dept. of Computer Science, Jan. 1999.
No context found.
Skadron, Kevin, Martonosi, Margaret, and Clark, Douglas W. Alloying Global and Local Branch History: Taxonomy, Performance, and Analysis. Tech. rep., Princeton Dept. of Computer Science, 1999.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC