Thursday Jan 03, 2008
Luiz André Barroso and Urs Hölzle of Google have written an article called:
The Case for Energy-Proportional Computing. You can find the IEEE Computing article here.
They correctly point out that memory (and disk) are a key factor in power used. Too bad that new SPECpower benchmark only favors tiny-memory configurations, and vendors do not publish measured watts on real-size memory configurations.
They miss the fact that low utilization wastes more power than energy-proportional designs will get them,
as we've pointed out
here
Disclosure statement
SPECpower_ssj2008. SPEC, SPECpower reg tm of Standard Performance Evaluation Corporation. See www.spec.org.
Thursday Jul 12, 2007
Sun doesn't sleep in the summer (other vendors are quiet, even
those that have brand new products, huh?), Sun continues to set a variety
of world records, and more to come this month and next month.
Here is a review of 4 very recent HPC benchmarks.
A World Record
Another World Record
Another World Record
Another World Record
Also a couple of commercial ones
note: Sun talks about delivered system performance not
... "use 'per-core' quick hide the fact that
these are super expensive cores or 'look at my peaks'" used by others.
Wednesday Dec 13, 2006
Sun Fire E6900 UltraSPARC IV+ beats the IBM p5 570 POWER5+ on SPECjbb2005. Why haven't we seen any
more POWER5+ performance benchmarks on workloads like this? Might be
the weaknesses we've been talking about when looking at other benchmarks?
The Sun Fire E6900 with US-IV+ maintained price/performance leadership on the SPECjbb2005 server-side Java benchmark. The 16-way Sun Fire E6900 (32 Procs Avail to Java) with 1.8GHz US-IV+ processors outperformed the 16-way IBM P5 570 (32 Procs Avail to Java) with 2.2 GHz POWER5+ processors by 5%.
SPECjbb2005 Performance Chart (ordered by performance,
bops: SPECjbb2005 Business Operations per Second, where bigger is better)
Ch/Co/Thr: Chips, Cores, Threads
|
System
|
Processors
|
Performance
|
|
Ch/Co/Thr
|
GHz
|
Type
|
bops
|
JVMs
|
bops/JVM
|
|
SGI Altix 4700
|
16/32/32
|
1.6
|
Itanium 2
|
471,030
|
8
|
58,879
|
|
SGI Altix 3700
|
32/32/32
|
1.6
|
Itanium 2
|
457,598
|
16
|
28,600
|
|
Fujitsu PP1500
|
32/32/32
|
2.16
|
SPARC64 V
|
361,920
|
8
|
45,161
|
|
Sun Fire E6900
|
16/32/32
|
1.8
|
US-IV+
|
343,512
|
16
|
21,470
|
|
Sun Fire E6900
|
24/48/48
|
1.5
|
US-IV+
|
342,578
|
12
|
28,548
|
|
IBM p5-570
|
8/16/32
|
2.2
|
POWER5+
|
326,651
|
8
|
40,831
|
|
Fujitsu PQ480
|
32/32/32
|
1.6
|
Itanium 2
|
322,719
|
8
|
40,340
|
|
Fujitsu PQ480
|
32/32/32
|
1.6
|
Itanium 2
|
295,161
|
1
|
295,161
|
|
Sun Fire E6900
|
16/32/32
|
1.5
|
US-IV+
|
248,075
|
8
|
31,009
|
|
IBM p5-570
|
8/16/32
|
1.9
|
POWER5
|
244,361
|
8
|
30,545
|
|
Sun Fire E6900
|
16/32/32
|
1.5
|
US-IV+
|
241,560
|
8
|
30,195
|
|
IBM p5-560Q
|
8/16/32
|
1.5
|
POWER5+
|
226,291
|
8
|
28,286
|
|
IBM p5-570
|
8/16/32
|
1.9
|
POWER5
|
224,200
|
1
|
224,200
|
|
|
Complete benchmark results may be found at the SPEC benchmark
website http://www.spec.org.
SPECjbb2005 (Java Business Benchmark) measures the
performance of a Java implemented application tier
(server-side Java). The benchmark is based on the order
processing in a wholesale supplier application. The
performance of the user tier and the database tier are
not measured in this test. The metrics given are number
of SPECjbb2005 bops (Business Operations per Second) and
SPECjbb2005 bops/JVM (bops per JVM instance).
Results Summary
|
Certified Results
|
|
|
Performance:
|
|
343,512 SPECjbb2005 bops
|
|
|
|
|
24,470 SPECjbb2005 bops/JVM
|
|
|
Reference Date:
|
|
December 6, 2006
|
|
Systems:
|
|
Sun Fire E6900
|
|
Total # Processors:
|
|
16
|
|
Processor/GHz:
|
|
US-IV+ 1.8 GHz
|
|
Operating System:
|
|
Solaris 10
|
|
JVM:
|
|
Java HotSpot(TM) 32-Bit, Version 1.6.0
|
Disclosure Statement:
SPECjbb2005 Sun Fire E6900 (16-way, 16 chips, 32 cores, 1.8 GHz) 343,512
SPECjbb2005 bops, 21,470 SPECjbb2005 bops/JVM submitted for review, IBM
eServer p5 570 (8 chips, 16 cores, 16-way, 2.2 GHz) 326,651 SPECjbb2005 bops,
20,415 SPECjbb2005 bops/JVM. SPEC, SPECjbb reg tm of Standard Performance
Evaluation Corporation.
Results as of 12/06/06 on www.spec.org. Prices based on publicly documented list prices. SPECjbb2005 Benchmark Reports
|
|
|
|
