Architecture guy on Why TAGE is the best Branch Prediction gu… on Why TAGE is the best a computer architect on Why TAGE is the best
Author Archives: rustam
Graduate students like me read lots of research papers. Not all research papers are available for free download online—yet—so we have to access them through our university’s EZProxy gateway. I made a simple bookmark that makes reading papers with EZProxy … Continue reading
Visualization is a key part of microarchitectural analysis. Visualization is necessary because even the simplest in-order processor is incredibly complex; its performance may be affected by cache misses, branch mispredictions, pipeline stalls, DRAM contention, inaccurate prefetching, and so on. In … Continue reading
The TAGE branch predictor by André Seznec and Pierre Michaud is the best branch predictor today, winning the last two branch predictor competitions (CBP2 and CBP3). It was introduced in a 2006 paper: A case for (partially) tagged Geometric History … Continue reading
I am a computer architecture Ph.D. student at The University of Texas at Austin. This is my blog about computer architecture. I also have a regular academic website with my publications and other academic info.
Asymptotic complexity of algorithms (the “big-O” notation) is a staple of computer science and engineering. We usually focus on asymptotic worst-case time complexity. For example, we know that quicksort can take up to n2 steps to sort a sequence of … Continue reading
While working on a recent paper, I stumbled on an easy way to find errors in a computer architecture simulator. I simulated a processor at various frequencies and plotted the resulting performance. I expected to see a nice “smooth” set of points, perhaps … Continue reading