r.skoru...@bremultibank.com.pl (R.S.) writes: > WRONG! > I meant it's hard to justfiy the choice: to buy IFL (plus rest of > mainframe) or x64 servers. I meant Linux on IFL is *much* more > expensive than on x64 servers. Things like power, cooling, floor > space, staffing won't change it, but the software licenses could do > it. > BTW: I did not mention RAS, but not everyone will pay for good > RollRoyce, some of us choose Toyota.
re: http://www.garlic.com/~lynn/2012l.html#56 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012l.html#57 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012l.html#59 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012l.html#70 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee z196 with 80 processors is rated at 50BIPS (besides increase from 64 to 80 processors over z10 at 30BIPS ... was the introduction of out-of-order execution ... something that had been used in risc for decades ... and the more recent I86 ... especially after they moved to risc cores with hardware layer that translated i86 to risc micro-ops for execution). At $28M ... z196 at 50BIPS is 624MIPS/processor, $560,000/BIPS and $350,000/processor. z196 80 processor is also 31.7kW ... or 634watts/BIPS. So far ec12 is 101 processors and 1.5times processing of z196 (say 75BIPS) with 26% increase in number of processors. ec12 processors are running 6percent faster than z196 processors ... so much of the rest of the performance improvement may be the statements about improvements to out-of-order execution ... guestimate 75BIPS/101 ... aka 743MIPS/processor (compared to 624MIPS/processor for z196) Analysis from last week is that IBM earns roughly $5.24M in mainframe services, software, and storage for every million in in mainframe processor ... so a $28M z196 would represent total of $175M. IBM has base list price of $1815 for e5-2600 blade. High-end e5-2600 is benchmarked at 527BIPS for two chip, 8processors/chip, 16 processors. (aka single e5-2600 has the processor power equivalent of more than max. configured 10 80processor z196), 33BIPS/processor, $3.44/BIPS, and $113/processor (compared to 624MIPS, $560,000, and $350,000 respectively for z196 ... based on $28M ... not fully loaded $175M ... aka factor 6.25 times). Assumption that disk technology is effectively the same ... except for any additional overhead of CKD layer simulation on regular disks for mainframe (aka CKD disks haven't been manufactured for decades, being simulated on top of industry standard disks). Motherhood that e5-2600 is 50% more energy efficient than previous generation (and "no-brainer when you can replace roughtly 20 old servers with one new E5 2600") http://www.techspot.com/news/48000-xeon-e5-2600-interview-with-intel-its-ajay-chandramouly.html Power consumption and efficiency (e5-2600 at 421watts): http://www.tomshardware.com/reviews/xeon-e5-2687w-benchmark-review,3149-11.html works out to around .8watts/BIPS (compared to 634watts/BIPS for z196). anybody seen published z196 dhrystone and whestone benchmarks compareable to these: http://www.istorya.net/forums/computer-hardware/485176-intel-xeon-e5-2690-and-e5-2660-8-core-sandy-bridge-ep-review.html this mentions kilowatt for ("overclocked" 3.5ghz) e5-4600 (four sockets, 32 processors compared to two sockets, 16 processor for e5-2600) http://www.istorya.net/forums/computer-hardware/459258-turbo-overclocking-high-tdp-cpus-8-core-intel-sandy-bridge-ep-example.html and http://www.istorya.net/forums/computer-hardware/518021-intel-announces-xeon-phi-co-processors.html from above: The Intel Xeon processor E5-2600 product family reached a new supercomputing milestone as the fastest adopted new processing technology to power 44 systems, including 3 Petascale-class supercomputers on the 39th edition of the Top500 list announced today. ... snip ... -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN