Sun Fire X4440 Best Opteron Multi-JVM Performance SPECjbb2005 at 606watts

Thursday May 15, 2008

The Sun Fire X4440 (4 2.3GHz Opteron QC) (system 606watts measured) running Sun Java SE 6 Update 6-p achieved a result of 372467 SPECjbb2005 bops, 46558 SPECjbb2005 bops/JVM for the best score for all Opteron-based servers on the SPECjbb2005 benchmark.

The Sun Fire X4440 demonstrated better performance over the HP DL585 G5 result of 368,543 SPECjbb2005 bops, 92136 SPECjbb2005 bops/JVM which used 4 2.3GHz Opteron quad-core and the BEA JRocket JDK 1.6.0_03.

The Sun Fire X4440 consumed 606 Watts on average during the execution of this benchmark while occupying 2 RU of space.

The Sun Fire X4440 demonstrated 2.7X better SWaP than the HP DL 585 G5 which requires 4 RU of space and an estimated power rating of 804watts.

The Sun Fire X4440 demonstrated 7% better performance over the IBM p 570 result of 346,742 SPECjbb2005 bops, 86686 SPECjbb2005 bops/JVM which used four 4.7GHz POWER6 (8-cores). The watts on that configuration are estimated power rating of 2,040watts.

The Sun Fire X4440 used Solaris 10 5/08 to obtain this leading result.

The SWaP metric is a measure of server efficiency ratio that includes system performance, power and space consumption on a specific benchmark. (SWaP = Perf /[ Space (RU) x Watts ] )

SPECjbb2005 results comparing HP Opteron quad-core, Sun Opteron quad-core and IBM POWER6 4-Chip systems

System		CPU		Perf
name	sys watts	Chip, Core, Thrd	GHz CPU	SPECjbb 2005 bops	JVMs	SPECjbb 2005 bops/JVM
Sun Fire X4440	606w measured	4,16,16	2.3 Opteron 8356	372,467	8	46558
HP DL585 G5	804w pow est	4,16,16	2.3 Opteron 8356	368,543	4	92136
IBM p 570	1,960w pow est	4,8,16	4.7 POWER6	346,742	4	86686

SPECjbb2005 Performance Chart (ordered by performance)
bops: SPECjbb2005 Business Operations per Second (bigger is better)

Complete benchmark results may be found at the SPEC benchmark website http://www.spec.org.

Benchmark Description

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).

Disclosure Statement: SPEC, SPECjbb reg tm of Standard Performance Evaluation Corporation. Results as of 5/12/08 on www.spec.org. Sun Fire X4440(4 chips, 16 cores) 372467 SPECjbb2005 bops, 46558 SPECjbb2005 bops/JVM submitted for review. HP DL 585 G5(4 chips, 16 cores) 368543 SPECjbb2005 bops, 92136 SPECjbb2005 bops/JVM. IBM p570 (4 chips, 8 cores) 346742 SPECjbb2005 bops, 86686 SPECjbb2005 bops/JVM.

Power References:

HP ProLiant DL585 G5 power rating of 804W is estimated as 70% of maximum PSU rating published on 5/4/2008 at: http://h18004.www1.hp.com/products/quickspecs/13016_na/13016_na.html

IBM p6 570 power specifications from 70% of maximum report power consumption published here, 06/07/07, posted at ftp://ftp.software.ibm.com/common/ssi/rep_sp/n/PSB01628USEN/PSB01628USEN.PDF 4-core at 980w, 8-core at 1960w estimates.

Results Summary

Reference Date:			May 13, 2008
Results			372467 SPECjbb2005 bops, 46558 SPECjbb2005 bops/JVM
System:			Sun Fire X4440
Processor:			4 x AMD Opteron 8356 2.3 GHz
Operating System:			Solaris 10 5/08
JVM:			Java HotSpot(TM) 32-Bit Server, Version 1.6.0_06-p

Dell games with little info, tiny mem, and low GHz

Tuesday Jan 22, 2008

Dell's comparison of power-performance done by Principled Technologies simply isn't very good. They leave out the critical configuration details to make the results look good.

Games Dell played:

• They used tiny 4GB of memory configurations, Read here to see huge power implications of memory size games.
• only filling half DIMM slots, and used small 1GB DIMMS
• They used non-standard BIOS hacks to improve performance (ex: Disabled hardware Prefetcher and Adjacent Cache Line Prefetch!) This helps some Java benchmarks but really HURTS real performance on workloads. We talked about this trick before: Hacking non-standard BIOS
• They only used 2.3GHz processors, hmmmm... Dell always talks about high-GHz on every other benchmark?
• Dell doesn't talk about ways to drive up the system utilisation, which saves more power than any games they played on the benchmark. They talk about small percentage differences in watts/performance - utilisations save factors in terms of watts/performance.
• older 10K RPM drives.
• No 10G Gigabit Ethernet - only promised for the future!. In the power tests they only use one 1GbE.
• showing perf/watt with 5 significant digits, to make this only appear accurate. I'm 99.97462452342534 sure about this.
• Dell System is limited to 350W per blade, given their current supplies, fans, etc - that will continue to mean older CPUs and small memory configurations.
  

(P.S. IBM bloggers couldn't even find the memory configuration details of the test, all they had to do was look beyond the press release It was in the system's packaging report: http://www.principledtechnologies.com/Clients/Reports/Dell/DellHPIBMbladeserverOOB1207.pdf). I agree with those bloggers, GHz & memory details should be in the press release.

Blog info about other power-performance benchmarks with same of the same issues.

Register weighs in on this info-free announcement: http://www.theregister.co.uk/2008/01/21/new_blades_dell_m1000e/

Comparables and IBM's blatant attempt to confuse customers with cores

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:

Siebel CRM 8.0 PSPP UltraSPARC T2 beats POWER6 and sets World Record

Thursday Jan 10, 2008

arrgghhh... I've been asked to show only Sun's results. You must now do your own math with the information posted on Oracle's website: http://www.oracle.com/apps_benchmark/doc/Sun_Siebel8_10000_PSPP_On_Solaris.pdf
http://www.oracle.com/apps_benchmark/doc/IBM_Siebel8_7000_PSPP_On_AIX_POWER6%20Final.pdf

IBM now longer holds the world record and really needs to post a correction on:
http://www-03.ibm.com/systems/p/hardware/benchmarks/erp.html

---

Four Sun SPARC Enterprise T5120 and T5220 servers (UltraSPARC T2 processors) set a new World Record using Siebel's standard Platform Sizing and Performance Program (PSPP) benchmark suite with Siebel CRM 8.0 Industry Applications and Oracle 10g R2 DB running on Solaris 10.

The Sun results using the UltraSPARC T2 supported 30% higher Siebel benchmark concurrent users compared to other results on the Siebel CRM Applications Release 8.0.

Sun again shows the UltraSPARC T2 servers are ideally suited for Oracle database applications. The database server ran Oracle 10g R2 on this Siebel benchmark.

{ Stuff deleted }

Sun's Solaris and Coolthreads based servers proves once again to be the best combination for scalability and resource utilization in the datacenter, giving users a consistent response time on critical applications as shown 10,000 users benchmark on Siebel CRM 8.0.

The 10,000 Siebel benchmark users performance results on 4 Sun SPARC Enterprise T5120/T5220 servers running Solaris 10 delivers a scalable and cost-effective platform for deploying Siebel CRM Application and Oracle 10g R2 deployment.

The result of 10,000 active concurrent Siebel user benchmark was run end to end on the new generation of Sun SPARC Enterprise servers using coolthreads technology with the highest level of space and energy efficiency.

See Also: http://www.oracle.com/apps_benchmark/html/white-papers-siebel.html

Siebel CRM 8.0 PSPP Performance Chart as of 01/04/2008 (bigger is better)

Vendor	Users	Web Server	Application Servers	Database Server
Sun	10,000	1 x Sun SPARC Enterprise T5120 4 cores, 1 chip @1.2 GHz US-T2     8 GB RAM Siebel CRM 8.0 SIA [20204] ENU Sun Java System Web     Server 6.1 SP8 Solaris 10 8/07	1 x Sun SPARC Enterprise T5220 8 cores, 1 chip @1.4 GHz US-T2     32 GB RAM 1 x Sun SPARC Enterprise T5220 8 cores, 1 chip @1.2 GHz US-T2     32 GB RAM Siebel CRM 8.0 SIA [20204] ENU Solaris 10 8/07	1 x Sun SPARC Enterprise T5120 8 cores, 1 chip @1.2 GHz US-T2     32 GB RAM Oracle 10gR2 Database     Server v10.2.0.1.0 Solaris 10 8/07
.	.	.	.	.

As noted on the official benchmark report: "Siebel CRM Release 8.0 Industry Application Platform Sizing and Performance benchmarks are based on Siebel CRM Release 8.0 customized industry applications and reflect a heavier scenario mix and more-aggressive think times than earlier version. Results of this benchmark are not comparable with those of prior Siebel CRM Release 7 benchmarks."

Benchmark Description

Siebel CRM 8.0 Platform Sizing and Performance Program (PSPP) is a multi-tier benchmark designed to stress the Siebel CRM Release 8.0 architecture and to demonstrate that large customers can successfully deploy many thousands of concurrent users. Among the Siebel CRM Release 8.0 architecture features exercised are the following:

• Smart Web Architecture: Takes advantage of the newest Web browser technology to deliver a highly interactive experience. The interaction model, which is similar to Windows-based applications, also improves productivity. Utilization rates on the web server are low, allowing customers to retain existing Web server infrastructure.
• Smart Network Architecture: Allows Siebel CRM Release 8.0 customers to leverage their existing network infrastructure by compressing and caching user interface components, so that browser/Web server interaction occurs only when the application requests data. This allows customers to avoid expensive network upgrades that can be necessary with competing products.

• Server Connection Broker: The Siebel Connection Broker (SCBroker) is a server component that provides intraserver loadbalancing. SCBroker distributes server requests across multiple instances of Application Object Managers (AOMs) running on a Siebel server.
• Smart Database Connection Pooling and Multiplexing: Allows customers to scale their database without intrducing expensive and complex transaction-processing monitors.
• Server Request Broker: Server Request Broker (SRBroker) processes synchronous server requests - reuqests that must be run immediately, and for which the calling process waits for completion.
• Enterprise Application Integration: Allows customers to integrate their existing systems with Siebel CRM applications.
• eScript: eScript is a scripting or programming language that application developers use to write simple scripts to extend Siebel applications. Javascript, a popular scripting language used primarily on Web sites, is its core language.

The test simulated real-world requirements of a large organization, consisting of 10,000 concurrent, active users from multiple departments accessing a call center. Test conditions simulated service representatives running Siebel Financial Services Call Center and partner organizations running Siebel Partner Relationship Management (Web sales and Web service). Siebel Workflow and the Siebel Scripting Engine were used to incorporate business-process-management customizations. The application also simulated integration with Web systems, using the Siebel Enterprise Application Integration component and Siebel Web Services.

Disclosure Statement:

Siebel CRM 8.0 Platform Sizing and Performance Program (PSPP) benchmark as of 01/04/08. Sun Microsystems: 10,000 users, 1 x Sun SPARC Enterprise T5120 web server (4 cores, 1 chip @1.2 GHz US-T2, 8 GB RAM), Siebel CRM 8.0 SIA [20204] ENU, Sun Java System Web Server 6.1 SP8, Solaris 10 8/07, 1 x Sun SPARC Enterprise T5220 application server (8 cores, 1 chip @1.4 GHz US-T2, 32 GB RAM), 1 x Sun SPARC Enterprise T5220 application server (8 cores, 1 chip @1.2 GHz US-T2, 32 GB RAM) Siebel CRM 8.0 SIA [20204] ENU, Solaris 10 8/07, 1 x Sun SPARC Enterprise T5120 database server (8 cores, 1 chip @1.2 GHz US-T2, 32 GB RAM), Oracle 10gR2 Database Server v10.2.0.1.0, Solaris 10 8/07 Oracle, Siebel, registered trademarks of Oracle Corporation and/or its affiliates. More info www.oracle.com/apps_benchmark/html/white-papers-siebel.html

Power Reference:

Sun measured: Database Server (1.2 GHz T5120, 8 core, 32G memory): 291W, Gateway/Application Server #1 (1.4 GHz T5220, 8 core, 32G memory): 323W, Application Server #2 (1.2 GHz T5220, 8 core, 32G memory): 376W, Web Server (1.2 GHz T5120, 4 core, 8G memory): 212W.

IBM power calculation based on the following: The p570 is supplied in building blocks with 2 chips, 4 cores per chassis called a CEC. Up to 4 CECs can be connected together to create a single 16 chip, 32 core SMP system. Each CEC is 4 RU, and each CE is estimatedC to consume 1,040 watts when configured with 2 processors, based on the following: IBM p6 570 power specifications from 80% of maximum report power consumption published here, 06/07/07, posted at ftp://ftp.software.ibm.com/common/ssi/rep_sp/n/PSB01628USEN/PSB01628USEN.PDF

System Configuration

Certified Results			10,000 Users
Reference Date:			January 4, 2008
Systems:			1 x Sun SPARC Enterprise T5120, web server (one 1.2GHz UltraSPARC T2)
			1 x Sun SPARC Enterprise T5220, gateway/application server (one 1.4GHz UltraSPARC T2)
			1 x Sun SPARC Enterprise T5220, application server(one 1.2GHz UltraSPARC T2)
			1 x Sun SPARC Enterprise T5120, database server (one 1.2GHz UltraSPARC T2)
Operating System:			Solaris 10 8/07
Software:			Sun Java System Web Server 6.1 SP8
			Siebel CRM 8.0 SIA [20204] ENU
			Oracle 10gR2 Database Server v10.2.0.1.0

Chopped configs and SPECpowerssj2008

Wednesday Jan 02, 2008

Timothy Prickett Morgan in the IT Jungle missed the point a bit when writing about SPECpower_ssj2008. It is not that you can't use power supply ratings to determine power usage. This difference really proves that these are chopped configs that are being used just for making numbers on SPECpower_ssj2008. Vendors who are trying to build power-efficient systems don't make mistakes like using small configs on power supplies that are too big. Power supplies are most efficient when the watts drawn are near the power supply ratings. If HP thought these small-memory chopped configs were typical they would have put smaller power supplies in them.

Have power supplies that are ONLY at 22% load (269/1200) is smoking gun aimed right at showing this is not typical config...

"it is no surprise at all that the best metrics for this are coming from streamlined two-socket machines so far. An HP ProLiant DL160 G5 with two quad-core Xeon E5450 processors running at 3 GHz with 16 GB of main memory and a single 80 GB 7.2K SATA drive was able to max out at 308,022 ops/sec on the test; the average idle power draw on this machine was 160 watts and when running full speed (99.2 percent of CPU capacity), it drew 269 watts. This machine had a 1200 watt power supply, which tells you how useless these maximum draw ratings on power supplies can be in determining power usage."

http://www.itjungle.com/bns/bns121807-story01.html

Disclosure statement

SPECpower_ssj2008: HP Proliant DL160 G5 (2-chip QC Xeon E5450 3GHz), 698 overall ssj_ops/watt. SPECpower_ssj2008:HP Proliant DL580 G5 (4-chip QC Xeon L7345 1.86GHz), 546 overall ssj_ops/watt, 359523 ssj_ops and 387 watt at 100% target load, 255512 ssj_ops and 359 watt at 70% target load, and 71409 ssj_ops and 294 watt at 20% target load. Results from www.spec.org as of 12/11/07.

SPECpower_ssj2008 power-benchmark needs work

Tuesday Dec 18, 2007

A new SPEC benchmark called SPECpower_ssj2008 was announced last week.. I've taken a little time to look at the new results and benchmark rules.

Quite simply, I don't know why SPEC didn't add power measurement to every benchmark. It is quite easy to add rules that puts one SUT (system under test) on a power meter and a few rules to accurately measure power. That should have been done months ago, it is so much easier than creating a new benchmark.

So why didn't SPEC do this? I have no idea. I can only guess some vendors to show off their power-management software at different utilisation levels. As we've talked about here for a very long time, running servers at low-utilisation (below 50% utilisation) is the worst way in the world to waste power... Even with the most extreme power-saving HW/SW.

With a different benchmark only for power-benchmarking it is really possible to game results. Did anyone?

• HP submitted a SPECpower_ssj2008 on the DL 580 G5 QC Xeon. Cool now we can see what the DL580 G5 does. But HP PICKED 1.86GHz Xeon L7345 for SPECpower_ssj and a very DIFFERENT processor (2.9GHz Xeon) for performance on SPECweb2005, well at least they are not confusing things...
• HP's DL 580 G5 SPECpower_ssj2008 documentation does NOT EVEN mention processor GHz or type! Why?
• The only SPECpower_ssj2008 results were on 4GB, 8GB, 16GB configs. Memory is a huge power draw. So why configure such small systems for power? Why do these vendors use 32GB and 64GB configs on performance benchmarks?
• The HP DL580 G5 used one 60GB 5400 RPM SATA drive. My laptop has better
• HP's DL580 G5 Power supplies used are 2x1200 watts. The 1.86GHz Xeon DL580 G5 draws 387 watts at 100% on SPECpower_ssj2008. Since all vendors know that you use power supplies near their rating for best energy efficiency, why does HP not sell properly-size power supplies? Maybe because HP typically expects to sell configurations that are different than SPECpower_ssj2008 DL580 configs?
• Is running at low utilisation efficient? SIMPLE answer NO! Let's look at SPECpower-ssj2008 HP DL580 G5. Comparing 70% utilisation with 20% utilisation we see that one wastes 3 TIMES more watts for the same amount of work at 20% utilization than at 70% utilization!!! 3 = round(2.93) = (294w/71409ssj_ops) / (359w/255512ssj_ops)
• SPECpower_ssj allows default BIOS settings to be changed to make better results by turning off prefetch (also see SPECjbb & prefetch) - yeah, right, every customer does that get real.

So what serves the industry? Using the same configurations that you benchmark for performance and power-performance.

Sun does this and has been doing this for years:
2007: UltraSPARC T2
2005: UltraSPARC T1 & T2000 blogs with power-performance

The only thing I really can take away from SPECpower_ssj2008 is that running at low utilization is silly, read carefully, and avoid silly configurations.

Disclosure statement

SPECpower_ssj2008: HP Proliant DL160 G5 (2-chip QC Xeon E5450 3GHz), 698 overall ssj_ops/watt. SPECpower_ssj2008:HP Proliant DL580 G5 (4-chip QC Xeon L7345 1.86GHz), 546 overall ssj_ops/watt, 359523 ssj_ops and 387 watt at 100% target load, 255512 ssj_ops and 359 watt at 70% target load, and 71409 ssj_ops and 294 watt at 20% target load. SPECpower_ssj2008: Colfax CX2266-N2 (2-chip AMD Opteron DC 2216HE 2.4GHz), 203 overall ssj_ops/watt. SPEC, SPECpower reg tm of Standard Performance Evaluation Corporation. Results from www.spec.org as of 12/11/07. HP ProLiant DL580 G5 (16 cores, 4 chips) 30261 SPECweb2005. SPEC, SPECweb reg tm of Standard Performance Evaluation Corporation. Results from www.spec.org as of Oct 8, 2007. HP DL580 power consumption from HP Power Calculator system configured with 4 x2.93GHz processors, redundant PSU, 16 x 4GB DIMMs, 8 x 36GB SAS drives,1 x PCI card, 80% utilisation on 9/10/07: http://h30099.www3.hp.com/configurator/powercalcs.asp HP DL385G2 power consumption from HP Power Calculator for system configured with 2 x AMD 2220 2.8GHz processors, redundant PSU, 8 x 4GB DIMMs, 2 x HBAs and 2 x 146GB SAS drives, 80% utilisation on 6/4/07: http://h30099.www3.hp.com/configurator/powercalcs.asp

Interesting hydrogren generation method using Microbial Fuel Cells

Friday Dec 07, 2007

There is a new method combines microbial action and fuel cell ideas in a unique way to generate hydrogen.

You can download an article titled "Microbial Fuel Cells: Methodology and Technology," that describes this work. The lead author is Bruce Logan (professor Penn State)... I now envy the Penn State coffee mug of a co-worker .

This article is the #1 download on the Environmental Science and Technology Journal's web site. This journal is a publication of the American Chemical Society. http://pubs.acs.org/cgi-bin/abstract.cgi/esthag/2006/40/i17/abs/es0605016.html

eWeek article about Economic green with green computing

Thursday Sep 20, 2007

More good advice on datacenters was given by Mark Monroe(Sun) in at eWEEK.com article entitled, "Finding the Economic Green in Green Computing" By Chris Preimesberger September 18, 2007. You can read about it at:

http://www.eweek.com/article2/0,1895,2184747,00.asp

In this blog we covered other important factors:
http://blogs.sun.com/bmseer/entry/the_register_needs_to_ask

and remember measure it in watts/performance, just like $/perf.

Quick way to save datacenter energy costs

Thursday Jun 14, 2007

While this strikes me as very cool, it also bothers me a bit because we can all take action now, that could make a bigger difference

"Climate Savers Computing Initiative (CSCI), has been joined by most of the biggest names in the computing industry, including Microsoft, IBM, HP, AMD, Dell and Sun, among others. The expressed aim is to make all computers being produced 90% efficient by 2010." ITwire: http://www.itwire.com.au/content/view/12869/1023/

Datacenter managers can save money even before all of this technology, and it is because they can change their policies.

New Mantra:
"I'm going to run my servers at 10% higher utilisation and turn off some old servers" (or even a higher % goal)

or...
"I'm going to consolidate and turn off some old servers"

See the math...
http://blogs.sun.com/bmseer/entry/the_total_tyranny_of_low
http://blogs.sun.com/bmseer/entry/total_tyranny_of_low_utilization

Then go home and buy some CFLs, and turn off your lights when not in use...

Like this post? | | | |

Posted at 04:50PM Jun 14, 2007 by BM Seer in Wattage-Power
Tags: power watt/performance wattage watts