IBM Launches Z13 Mainframe • Hacker News Board

Re: IBM Launches Z13 Mainframe

#101

Post by qznc » Thu, Jan 15, 2015, 10:08 AM UTC

TheCondor wrote: post #97

listic wrote: post #95

Earlier quoted context omitted.

I guess this this market of ultra-conservative customers are mostly US companies? I have hard time imagining companies in other countries getting into the habit of buying such kind if hardware?

I traveled all over the world while at IBM in the 1990s. They were doing a very nice business with the biggest banks in Europe as well as many of auto companies in Europe. Major ibm customers in Japan as well. They wouldn't build these if they couldn't sell them. With the io and data flow rates these thing can handle, there really aren't alterntives on the market. the debate I keep spinning through in my head is will…

1. So far no cloud provider tried to provide reliability at the level of a mainframe.

2. Those conservative customers want to keep their data in-house. So a cloud is probably impossible to sell here.

They can either build their own "in-house cloud" (paradox?!) or buy a mainframe. The later is simpler and proven. Cost savings of the roll-your-own-in-house-cloud are questionable.

You could try to sell an ultra-reliable (maybe distributed) system for in-house use. However, you cannot call it a "mainframe".

Re: IBM Launches Z13 Mainframe

#102

Post by aries1980 » Thu, Jan 15, 2015, 11:38 AM UTC

gojock wrote: post #79

aries1980 wrote: post #60

Earlier quoted context omitted.

110 GIPS is not too much for 141 CPU. My Core i7 4700HQ CPU dos 3 GIPS/core at 2.9 GHz.

yes but buying 3 of them gets you 3 machines that can do 3GIPS, not one machine that can do 9GIPS

True, but that is rather motherboard-design and core-stacking question. x86 CPUs are made for consumer use-cases, but that doesn't mean it can't be stacked almost linearly, like the Knights Landing CPUs.

Re: IBM Launches Z13 Mainframe

#103

Post by aries1980 » Thu, Jan 15, 2015, 11:56 AM UTC

dfox wrote: post #76

aries1980 wrote: post #60

Earlier quoted context omitted.

110 GIPS is not too much for 141 CPU. My Core i7 4700HQ CPU dos 3 GIPS/core at 2.9 GHz.

It depends on what kinds of instructions you count. I strongly suspect that your 3GIps number is only register-to-register integer arithmetic, while the 110GIps is some IBM-ish "mixed workload" that uses most of the System/z instruction set. (which includes things like "get CRC32 of this buffer" as one instruction)

Average of Dhrystone and Whetstone, which is not a proper benchmark indeed.

Those special instructions such as CRC32 or SHA256 are rarely needed in scientific simulations such as Monte Carlo simulations or real-time working with Bayesian Network, what this system is aimed for. In these scenarios raw power and big on-board caches what matter and on paper the POWER8 has these, but 110 GIPS is not the best bang for bucks.

Re: IBM Launches Z13 Mainframe

#104

Post by dfox » Thu, Jan 15, 2015, 2:44 PM UTC

aries1980 wrote: post #103

dfox wrote: post #76

Earlier quoted context omitted.

It depends on what kinds of instructions you count. I strongly suspect that your 3GIps number is only register-to-register integer arithmetic, while the 110GIps is some IBM-ish "mixed workload" that uses most of the System/z instruction set. (which includes things like "get CRC32 of this buffer" as one instruction)

Average of Dhrystone and Whetstone, which is not a proper benchmark indeed. Those special instructions such as CRC32 or SHA256 are rarely needed in scientific simulations such as Monte Carlo simulations or real-time working with Bayesian Network, what this system is aimed for. In these scenarios raw power and big on-board caches what matter and on paper the POWER8 has these, but 110 GIPS is not the best bang for buck…

System z is completely different thing from what is currently called Power Systems. Target market for System z is pretty much limited to OLTP and most of the weirdness of the platform comes from the fact that it is heavily optimized for exactly that and not for raw speed in two's-complement or FP arithmetics.