re: http://www.garlic.com/~lynn/2013g.html#5 SAS Deserting the MF? http://www.garlic.com/~lynn/2013g.html#7 SAS Deserting the MF? http://www.garlic.com/~lynn/2013g.html#9 SAS Deserting the MF?
one of the claims about x86 performance increase in BIPS over the past decade has been attributed to competition between multiple vendors producing x86 chips. the other claim is that for the past several x86 chip generations they've gone to RISC cores with hardware layer that translate x86 instructions into RISC micro-ops (for decades, RISC performance has been significantly better than x86, but x86 move to RISC core appears to be negating that difference). base-line of dhrystone is 370/158-3 1 processor, 1MIPS, (1MIPS/proc), 1972 other numbers approx 370/168-3 1 processor 3MIPS, (3MIPS/proc), 3033 1 processor 4.5MIPS, (4.5MIPS/proc), 3081D 2 processor 10MIPS, (5MIPS/proc), 3081K 2 processor 14MIPS, (7MIPS/proc), this is recent post discussing reasons for 3081 throughput being much slower than claimed (& 3090 being first "real" new IBM 370 after 168) http://www.garlic.com/~lynn/2013g.html#1 http://www.garlic.com/~lynn/2013g.html#8 This source has MIPS vendor claims http://www.roylongbottom.org.uk/mips.htm 3090-600E 6 processor, 89MIPS (17MIPS/proc) ES/9000-982 8 processor, 408MIPS (51MIPS/proc) 9672 G6 ZZ7 12 processor, 1644MIPS (137MIPS/proc) the numbers appear to show increase consistent through the generations from the 370/158-3 base ... which is also used as baseline for the industry standard dhrystones benchmark. older post here in ibm-main looking at last decade or so, from ibm publications & website http://www.garlic.com/~lynn/2013b.html#45 z900, 16 processors, 2.5BIPS (156MIPS/proc), Dec2000 z990, 32 processors, 9BIPS, (281MIPS/proc), 2003 z9, 54 processors, 18BIPS (333MIPS/proc), July2005 z10, 64 processors, 30BIPS (469MIPS/proc), Feb2008 z196, 80 processors, 50BIPS (625MIPS/proc), Jul2010 EC12, 101 processors, 75BIPS (743MIPS/proc), Aug2012 this post http://www.garlic.com/~lynn/2013g.html#1 also discusses killing of acs-360 effort in late 60s because company was worried that it would advance computing technology too fast, threatening the company's control of the market. http://people.cs.clemson.edu/~mark/acs_end.html the bottom of the above page discusses acs-360 (from late 60s) features finally showing up in ES/9000 in the early 90s. -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
