Wednesday Jan 16, 2008
IBM is still actively trying to confuse customers with core count.
Remember the internal structure (read "core") is NOT important. What is
important is system-to-system comparison.
People buy systems, put workloads on them, and measure performance. No one cares
how many widgets are inside the system! If the faster system is a lower-cost
fewer chip system - all the better! This is the reason why many customers are switching
from POWER6 systems to buy Sun's US T2 systems. No one cares that IBM has 4 widgets
buried inside that expensive IBM box.
It is system perf, system $/perf, system watt/perf, -- it
is not a system's core-count!
IBM continues to withhold actual POWER6 power measurements on actual
SPEC benchmarks. It is so easy to hook up a power meter to systems
under test on a benchmark. Is IBM scared of the truth?
In the mean time, Sun will continue to do good faith estimates of
32GB and 64GB POWER6 systems. Sun never uses book maximums or
nameplate maximums, we make allowances (which by the way, are correct
whenever we measure on competitive systems). If Sun is so wrong
why doesn't IBM publish actual data?
...And please make it on Power6 systems
with 32GB-64GB and 4.7GHz only, just like the systems IBM uses on
performance benchmarks.
Here's Sun's data compared to IBM on perf, watt/perf:
Wednesday Sep 26, 2007
IBM proudly claims best-in-class 4-core system, and continue with best-in-class x-core system etc. But how dubious are these claims based on core count? I'm speaking to the true nature of making claims that are made to give the impression that the word cores matter and that this sort of comparison has gravity. Let's check it out.
IBM claims:
- "A 4-core IBM System p 570 (4.7 GHz) is the best-in-class 4-core system, 122 SPECint_rate2006"
We should note the configurations, (size & cost) and then compare it against a competitive system. Let us compare the 4-core Power6 system to a different system with 8 cores --
WHAT?!? THAT'S CRAZY an 8-core system has twice as many cores as IBM power6 p570, that is simply not fair... or is it?. I'll compare them anyway
- system_A: 4-core 4RU, 32GB system cost about $80k-200K?, two 4.7GHz dual-core Power6, vendor=IBM
- system_B: 8-core 2RU, 32GB, cost about $40K, two 3GHz quad-core, vendor=???
Wow it seems a system with
Twice the core count can be smaller and cheaper? But didn't IBM imply that knowing core count allowed you to understand the size/cost of a system? On the pricing, this must be because some up-start vendor is undercutting IBM to gain market share....
- System_A=IBM p570 with 4.7GHz Power6 (total of 4 cores, 8 threads)
- System_B=IBM System x3650 with two quad-core 3GHz Intel Xeons (total of 8 cores, 8 threads)
OK, but how do they compare on performance?
- System_A: IBM p570 (two dual-core 4.7GHz Power6) 122 SPECint_rate2006
- System_B: IBM System x3650 (two quad-core 3GHz Xeon X5365) 103 SPECint_rate2006
- System_C: Supermicro X7DB8+ (two quad-core 3GHz Xeon X5365) 116 SPECint_rate2006
IBM can still claim "best-in-class 4 core result", but it doesn't matter. I think any comparison on "core count" means absolutely nothing in the computing industry. "Just like comparing car-engine horsepower based on valve-count," as a friend of mine says
Disclosure Statement
SPEC, SPECint, SPECfp reg tm of Standard Performance Evaluation Corporation. Competitive results from www.spec.org as of 9/01/07. IBM p570, POWER6 4.7GHz (8-thread, 4-core), 122 SPECint_rate2006. IBM System x 3650 (Intel Xeon X5365 quad-core) 8cores total, 103 SPECint_rate2006. Supermicro X7DB8+ (two quad-core Xeon X5365, 3GHz) 116 SPECint_rate2006 1333 MHz system bus, 2x4MB L2 Cache, 16GB DDR2 ECC, SUSE Linux Enterprise Server 10 SP1 (x86_64), Intel C++ Compiler for IA32/EM64T 10.1. For more information see http://www.spec.org/ IBM claims from http://www-03.ibm.com/systems/p/benchmarks/index.html
IBM X3650 pricing from: http://www-132.ibm.com/content/home/store_IBMPublicUSA/en_US/evp/x3650.html. IBM closely hides 40core IBM POWER6 4.7GHz prices for any reasonable memory size of 16GB to 32GB (fast memory only). If you have exact pricing feel free to post it in a comment, and I'll update the entry. Please don't add discounts.
Oracle users cares about cores because of lic...
Any idea's when are we likely to see SPECpower_ssj...
Any ideas when HP will publish results on reasonab...
The TCO is of a high end system configured with mi...
Best not to be vague Alex, a 4-core IBM p570 4.7GH...
48gb of 667mhz, or 192GB of 400mhz.
So your righ...
IBM uses 667MHz on with 4.7GHz performance benchma...