[email protected] (Dana Mitchell) writes:
> So you are saying that a sub $2K blade has roughly 10 times raw
> compute power as an 80 way z196?
>
> I'd be interested in references to support that.

re:
http://www.garlic.com/~lynn/2013b.html#5 mainframe "selling" points

IBM publishes 50BIPS for 80-way z196

past post in ibm-main from last fall
http://www.garlic.com/~lynn/2012l.html#59 

quoting benchmarks numbers from this web page
http://www.istorya.net/forums/computer-hardware/485176-intel-xeon-e5-2690-and-e5-2660-8-core-sandy-bridge-ep-review.html

(Dhrystone)
E5-2690 @2.9GHZ 527.55BIPS
E5-2660 @2.2GHZ 428.15BIPS
X5690 @3.45GHZ 307.49BIPS
i7-3690 @4.78GHZ 288BIPS
AMD 6274 @2.4GHZ 272.73BIPS

(Whetstone)
e5-2690 @2.9GHZ 315GFLOPS
E5-2660 @2.2GHZ 263.7GFLOPS
X5690 @3.4GHZ 227GFLOPS
i7-3690 @4.78GHZ 176GFLOPS
AMD 6274 @2.4GHZ 168.11GFLOPS

... aka the e5-2600 at 527.55BIPS is also capable of 315GFLOPS 

and from this recent post 
http://www.garlic.com/~lynn/2012j.html#34 
in "Mainframe Experts" discussion
http://lnkd.in/QhQ73A

including ibm results for lots of platforms (at the time as of
3july2012 ... now current as of 29Jan2003 ... IBM currently
has 873 bencmarks listed in the following ... none of them for
mainframe)
http://www.spec.org/cpu2006/results/cint2006.html

risc chip/processors have had numerous performance features for the past
several decades that provided them with significant throughput
advantages over i86 chips. however, for the past several generations of
i86 chips, they have moved to risc cores with hardware layer that
translates i86 instructions into risc microops ... significantly closing
that throughput gap.

z10 was rated at 30BIPS with 64processors or 469MIPS/processor.  z196 is
rated at 50BIPS with 80processors or 625MIPS/processor.  The description
is that much of the improvement in z196 per processor performance was
introduction of some of the features that have been in risc processors
for decades (and are part of all the latest i86 processors). zEC12 is
rated at 75BIPS with 101processors or 743MIPS/processor ... with claims
that there have been some additonal risc-like features in the way the
zEC12 processors operate.

Part of the claim for the rapid advancement in i86 chip throughput is
there being competition between multiple vendors of i86 chips.

for other drift ... this ibm-main post references peak i/o throughput
benchmark for z196 using 104 FICON channels and 14 system assist
processors that achieves 2M 4k IOPS (although peak SAP all running at
100% is 2.2M and recommendations is to keep SAP utilization below 70% or
1.5M)
http://www.garlic.com/~lynn/2012m.html#4 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee

FICON is an enormously expensive protocol layer on top of FCS, that is
significantly slower than the the throughput of the underlying
fibre-channel standard technology.  Above post reference announce of a
*single* FCS channel (aka technology used in FICON) for e5-2600 capable
of over 1M IOPS (two such FCS channels could give an e5-2600 the same
throughput as z196 peak i/o throughput with 104 FICON channels).

other recent posts here in ibm-main mentioning e5-2600 comparisons
http://www.garlic.com/~lynn/2012l.html#20 X86 server
http://www.garlic.com/~lynn/2012l.html#28 X86 server
http://www.garlic.com/~lynn/2012l.html#30 X86 server
http://www.garlic.com/~lynn/2012l.html#34 X86 server
http://www.garlic.com/~lynn/2012l.html#51 Turn Off Another Light - Univ. of 
Tennessee
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#81 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012l.html#87 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012l.html#88 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012l.html#90 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012l.html#100 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012m.html#3 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012m.html#5 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012m.html#43 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee
http://www.garlic.com/~lynn/2012o.html#25 Blades versus z was Re: Turn Off 
Another Light - Univ. of Tennessee

-- 
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

Reply via email to