6 ms·
Effects of the x86 ISA on the Front End: Where have all the cycles gone? (2001) [pdf]
- jws 12y agoI am unclear when this was published. It does not reference an Intel microarchitecture past P6 (Pentium Pro) and contains no citations later than 2000. TLDR Abstract: We look at 8 aspects of front-end that can contribute to performance loss; then based on this information, we introduce an improvement that yields 17 % speedup in overall execution time.
- _delirium 12y agoIt's from 2001. See: http://www.eecs.umich.edu/eecs/research/techreports/cse_tr/database/reports.cgi?01 http://www.eecs.umich.edu/eecs/research/techreports/cse_tr/d...
- dang 12y agoThank you. I searched earlier and that information is surprisingly hard to come by.
- _delirium 12y agoYeah, it's weird that the document doesn't "self-identify" its publication information anywhere. The only reason I found the metadata quickly is that I glanced at the URL and realized it was a tech report, so Googled for Umich CSE's tech report archive.
- nkurz 12y agoI tried to search for it so I could include it in the title myself, but eventually gave up and posted without. Then I was confused when the usual 'automated' date tagging system (dang) didn't work as expected!
- userbinator 12y agoThe creation date of the PDF is 2001, so this is an extremely old paper and the details may not have all that much relevance anymore since x86 CPU development has progressed very quickly; I'd guess that AMD and Intel have already made use of the techniques mentioned here. It's still worth reading though - the uop decoder mechanism of microarchitectures today remains very similar to that introduced with the P5, just wider and deeper.
- wsxcde 12y agoThe paper's main problem is that they have to second guess all of Intel and AMD's design decisions. And it looks like they got some important things wrong, which makes their conclusions suspect. For instance they claim an ADD mem, reg instruction would be translated into three 3 uops. This is definitely not the case, x86 chips can do a load+arith_op or a store+arith_op in the same uop. The reason they need to do this is because the old x86 (in the pre-AMD64 days) had only 8 registers so spills were very common and so loads and stores were very frequently executed, much more so than in RISC code. This meant that getting maximum performance out of the memory subsystem was very critical for x86. I remember an AMD chief architect telling me that the hardest part of building a performant x86 core is getting the memory subsystem right. Anyway, back to the paper, I'm not terribly impressed. The uops/instruction number looks bogus. I remember those numbers being very close 1. The cycles per uop of 0.37 seems wildly optimistic. I remember numbers closer to 1. My only caveat is that my experience working on x86 is from about 3-4 years ago. Maybe things were very different in 2000, but I doubt it.
- Ono-Sendai 12y ago18 cycle latency for memory access is way too low as well.
- sliverstorm 12y agoSome of the figures seem hard to believe, but on the other hand this was 2001. The first 1GHz chip hit shelves in 2000 (?). So 18 cycles then was (very roughly) 100 cycles now. That puts us more in the ballpark. (At 3.3GHz, it's around 200 cycles) Then add in, if my memory serves DDR was lower latency than DDR3.
- userbinator 12y agoIt's in the right order of magnitude, considering that at the time the most popular type of PC memory would be PC100/133 SDRAM with a latency of 2-3 cycles at 100/133MHz FSB, and a typical CPU being a Pentium II/III at 400-500MHz - 4-5x FSB frequency meaning 8-15 cycles at the core.