Performance considerations.html (1603B)
1 <?xml version="1.0" encoding="UTF-8"?> 2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> 3 <html><head><link rel="stylesheet" href="sitewide.css" /><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/><meta name="exporter-version" content="Evernote Mac 6.13.1 (455785)"/><meta name="altitude" content="-0.1842866390943527"/><meta name="author" content="Alex Balgavy"/><meta name="created" content="2017-12-17 8:45:10 PM +0000"/><meta name="latitude" content="52.37353963015489"/><meta name="longitude" content="4.836234900920008"/><meta name="source" content="desktop.mac"/><meta name="updated" content="2017-12-17 8:50:49 PM +0000"/><title>Performance considerations</title></head><body><div>price/performance ratio — common measure of success</div><div>hit rate — number of hits as a fraction of all attempted accesses</div><div>miss rate — number of misses as fraction of attempted accesses</div><div>miss penalty — total access time seen by processor when miss occurs</div><div><br/></div><div>improvements:</div><div><ul><li>larger cache (increased cost)</li><li>larger block size (only effective up to a certain size, take longer to transfer)</li><li>load-through approach reduces miss penalty</li><li>implement cache on processor chip</li><li>write buffer — temporary storage of write requests, then sent all at once to main memory</li><li>prefetching data into cache</li><li>lockup-free cache — let processor access cache while a miss is being serviced</li></ul></div><div><br/></div></body></html>