Friday Jun 05, 2009
The last of the 2009 SPECtacular
awards. SPECweb2005 is the
industry standard performance metric for web servers, and today it is
joined by SPECweb2009, the
industry standard performance and energy metric for web servers. The
benchmark includes a banking workload (all SSL), a support workload
(no SSL), and an ecommerce workload (mixed). This is the first
application of the SPECpower
methodology to potentially large system under test
configurations. In the initial
benchmark results you can see one system with and one without
external storage, and the test report lets you see the power
consumption of just the server, of the storage, and of the entire
configuration at various utilization levels. The entire committee did
a fantastic job with this benchmark. As always, I won't list anyone's
name without permission. (But give me the okay and I'll update this
posting!) SPEC recognizes:
Gary
Frost (AMD) who
stepped in to fill a key developer role in an emergency with the
release clock ticking. He took over the control code after a sudden
reassignment, and frankly we handed him quite an undocumented mess.
Gary was up to the challenge and produced the finished code.
Another engineer from AMD
had primary responsibility for the reporting page generator. You
often can't know exactly what information ought to go into a full
disclosure report (FDR) until you see it. Nor how you want it
organized and arranged. Nor what data integrity cross checks need be
present to avoid errors. So the committee changed requirements often
during development. But no matter how many requirements were placed
on him, he turned around with the needed code within a week!
An engineer from Fujitsu
Technology Solutions became the de facto quality assurance
office because of his thorough and methodical testing practices. If
there are a hundred ways software in general can go wrong, then there
are a thousand ways benchmark software can go wrong, as by its nature
it runs on systems stressed to the limit. When SPEC benchmark
software just works that is largely due to people like this engineer
who forsee, test, and diagnose every possible failure unanticipated
by the authors.
And, if you'd like to see all of the
SPECtacular awards, then follow
the tags!
Monday Jun 01, 2009
Another SPECtacular award from the SPEC annual meeting: Klaus Lange (HP) has become a valuable conduit across different levels of the organization and across benchmark subcommittees, by virtue of becoming indispensable in all of them. Though Klaus is an experienced "SPEC hand" he never forgot what he faced as a newcomer, and took it on himself to organize a new member orientation program to help new institutions integrate into SPEC more easily and effectively. As chair of the SPECpower committee Klaus delivered the industry's first energy efficiency benchmark, and leads the committee in aiding other groups as they add energy metrics to a wide range of benchmarks. These groups include many SPEC committees as well as other industry consortia. As HP's representative on the OSG steering committee Klaus has earned respect for his opinions with his diligence and fair mindedness. As a member of the Board of Directors he is often the first to step up to volunteer for important projects, as well as exercising sound judgment in conducting SPEC's business operations.
Friday May 22, 2009
More 2009
SPECtacular awards. The SPECpower committee has been busy. They
released version 1.10 of
the SPECpower_ssj2008 benchmark as a no-cost upgrade to existing
licensees. It adds support for measurement of multi-node (blade)
servers, improves usability, and adds a graphical display of power
data during benchmark execution. Review and publication of benchmark
results continues apace, with a spirited competition for first place,
and with ever more power
analyzers accepted for testing, and more test labs qualified for
independent publication. They have also been assisting several other
benchmark committees inside SPEC, and other industry
standard benchmark organizations, to implement energy measurement for
their benchmarks. SPECpower is more than just a benchmark; it is a
methodology,
and the methodology is modified and expanded as necessary over time
to accommodate energy measurements for all the different workloads
which are relevant to the real world in those market segments. In
alphabetical order SPEC recognizes:
-
Chris
Boire (Sun
Microsystems) – As release manager he coordinated and
integrated development activities to keep the deliverables on
schedule.
-
David
Schmidt (HP)
– He created stand-alone and network integrated tools for
automated results checking to help insure that results submissions
are correct and complete.
-
Greg Darnell (Dell)
– Author of the PTDaemon, he helped many other groups get started
measuring power for their benchmarks. He helps out with whatever
needs to be done, technical or organizational.
-
Hansfried
Block (Fujitsu
Technology Solutions) - He automated the process of determining
power analyzer precision, handled the acceptance of several new
power analyzers, and was instrumental in getting multi-channel
analyzers accepted.
-
Harry
Li (Intel)
– He was primary developer of the Visual Activity Monitor, giving
an unique view of the system's activity.
-
Jeremy
Arnold (IBM)
– If I tried to recount all the accomplishments Jeremy was cited
for I'd probably run into some internal blog size limit. Suffice it
to say he is a primary developer on many parts of the code, who
never turns down a plea for help, and who is never satisfied until
the entire benchmark package is right.
-
Karl
Huppler (IBM)
– As primary author/editor of the Power and Performance
Methodology, he organized the document to capture deep technical
consensus in the committee, and made it readable and understandable
for people new to the field.
-
Matthew
Galloway (HP)
– He designed the control software to drive multiple JVMs,
enabling multi node (blade) testing.
-
An engineer (AMD)
– Who created and maintained much of the web content explaining
the benchmark and methodology to the public.
Wednesday May 13, 2009
More 2009
SPECtacular awards. Sometimes even success doesn't succeed, at first.
SPEC developed a workstation
energy consumption benchmark, and a lot of people worked extra
hard to deliver it in time for EPA to consider using it in the Energy
Star program which is being extended beyond PC's to also include
workstations, servers, thin clients, and storage. Although EPA
decided not to use our test for the workstation program at this time,
the work is still important and I am confident it will be used in
some way. A graphics processor can easily use
more energy than a CPU, especially a high performance accelerated
3D processor. For their exceptional work in producing this benchmark
I thank David Reiner of AMD,
Joerg Grosshennig of Fujitsu
Technology Solutions, Paul Besl of Intel,
and an engineer from NVIDIA.
Monday May 05, 2008
Not NASDAQ - solar power. The ASES annual solar energy conference is in San Diego this week. The top question I get about my solar panels is how long is my return on investment? I did calculate it before we installed them. At current electricity prices and time value of money they will just break even over their useful life. (And we live near the ocean where morning fog obscures the panels on many summer mornings.) Still, if we had another price shock equivalent to the 1973 oil embargo they would pay back about twice the initial investment, in current dollars. And if we had another price shock equivalent to Kenny-Boy Lay's market manipulation, they would pay back about five times the initial investment. Economically, call the panels zero cost insurance.
Now what's the ROI on an SUV? Our solar panels cost about a quarter to a half the price of a big SUV. Will that Escalade have a productive life of 20 years? And over that time how many dollars will it return to your pocket? Or will it perhaps take more money out of your pocket? For the price of the SUV you could instead buy solar panels, zero out your electric bill, buy a Chevy Malibu (which sits in clogged traffic equally well as the SUV), and have enough money left over to pay for over 200,000 miles worth of gas for it, at $4/gallon.
So why aren't there more solar panels in sunny Southern California? Why is Germany, in the cloudy wintery north, so far ahead of the U.S.? Two reasons: (1) money, and (2) money.
(1) Lots of people don't have the luxury of deciding whether to spend discretionary money on a new SUV or on solar panels; they're deciding whether to pay the mortgage, pay the electric bill, or fill up the gas tank. Ditto businesses hard pressed to show a profitable bottom line. Increasingly solar energy entrepreneurs are in effect buying energy "drilling rights" on rooftops. LA's electric utility Edison is building the equivalent of a new generating plant by putting panels on the roofs of commercial and industrial buildings. The building owners pay nothing, and get a good long term locked in electricity rate. Here in San Diego, Hewlett-Packard is converting its campus to solar power. HP stockholders will pay nothing for it, and HP will get substantial energy cost savings in the future. While they're at it, HP is matching the rebate to their employees who want to put solar panels on their homes.
(2) The recent earth shaking discovery that people are more willing to give goods and services in exchange for money, than to give them with nothing in return. (See capitalism.) The biggest barrier to local development of solar energy in San Diego has been a convoluted rate structure that in many cases actually made businesses that installed solar generators pay more money to use less electricity, than before they installed them. Small wonder that northern California is far ahead of sunnier southern California in solar power installations. Now that crazy rate structure is changing, which could bring a boom in locally generated solar power.
For homeowners in San Diego no change is forthcoming. Germany has all those solar installations because of a rate structure that pays for solar electricity at much higher than market rates. In San Diego you see many solar panel installations like ours covering a small portion of the roof. The rate structure here is fair up to the point that you replace your total annual electricity usage with solar power. Produce more than you use, however, and all the excess is just "donated" to the utility without compensation. So you're okay if your solar installation is a bit smaller than you need, but it's economic madness to make it any larger than you need. If not for this rate structure, our solar panel installation could have produced enough electricity for one or two of our neighbors in addition to our own needs.
Saturday Apr 05, 2008
After wrestling with incompatibilities of 64-bit Linux for a while, I finally downgraded my home PC to 32-bit Ubuntu 7.10 (Gutsy). I found some nice and less nice workarounds, like running the Windows version of Firefox under Wine in order to get Flash to work. I hadn't found a workaround for the Java browser plugin, or for Skype, and was considering a 32-bit chroot environment.
Finally one of those automatic updates decided it for me. You know the ones, the messages offering later versions of software, critical security updates and recommended updates. Being able to just click OK to automatically be upgraded to the latest software is part of what makes Ubuntu so friendly. But this time it wasn't so friendly. Something left my PC unable to boot to multi-user, unable to start networking, and unable to start graphics. I don't know what because I didn't keep the disk image around for a post-mortem. It was much much faster simply to blow away my root partition with a complete new OS installation. So while I was at it, I dropped down to 32-bit.
Lots of things started working, but some got worse. I had always had a problem on Gutsy that after suspend/resume the Ethernet driver would get a reversed MAC address, complain that it was invalid, and switch to a new eth instance with a random MAC. Of course this played havoc with my router trying to keep track of where my PC was in order to provide DNS. This problem occurs in the forcedeth driver, reverse engineered for the nForce chipset. Some people worked around the problem with limited success by adding commands in the suspend/resume scripts to stop and restart networking.
But now on 32-bit Gutsy it got worse. Upon resuming the screen stayed black, and since the network seemed to be down I couldn't remotely login to find out what was wrong. I found lots of reports on the web about suspend/resume problems with the same error message in my .xsession-errors
Gtk-WARNING **: This process is currently running setuid or setgid.
This seems related to my NVIDIA GeForce 6150 LE graphics. Like many others who posted their experiences, the problem occurred for me both with the generic open source driver and with the Nvidia proprietary accelerated driver. One person mentioned a workaround by logging out, logging in to a failsafe X-terminal, and suspending manually from there.
Irony: The main reason I'm running Ubuntu instead of Solaris is that Solaris doesn't yet have power management, and for a home PC, suspend and resume are essential. I've been eagerly watching the power mangement project Tesla at opensolaris.org, wondering why it's taking so long. I guess like most things it's easier to do, than it is to do right. By comparison, my PC's when running Windows 98SE often fail to wake up at all, and those running Windows XP tend to wake up by themselves, unbidden. The only systems where suspend/resume always worked were Linspire and, of course, MacOS.
Neither workaround by itself would work for me, but putting them both together I end up with a clumsy workaround that lets me suspend/resume, and may possibly point the way towards a less cumbersome workaround.
- Disable networking via gnome panel
- Logout
- Select failsafe X-terminal session
- Login
- sudo /etc/acpi/sleep.sh
- (system sleeps)
- (normal wakeup by pressing ENTER)
- Logout
- Select normal gnome session
- Login
- Enable networking via gnome panel
Wednesday Mar 26, 2008
More SPECtacular awards given at SPEC's 2008 annual meeting in San Francisco, to members of the power committee who produced the SPECpower_ssj2008 benchmark. This wasn't an easy benchmark to do, taking us into areas of engineering not so familiar to performance analysts. Along the way we picked up some new contributors, and some of us picked up some new knowledge and skills. Energy efficiency is increasingly important, and eventually I expect to see power measurements as part of every performance benchmark. But for now, SPECpower_ssj2008 is a great start that establishes a fair and practical methodology for consistent measurement.
As before I won't cite names without permission, but will add them later if given the okay. SPEC thanks:
- Paul Muehr from AMD
- Greg Darnell, and another engineer from Dell
- Karin Wulf, and another engineer from Fujitsu-Siemens
- Klaus Lange, and another engineer from HP
- Jeremy Arnold, Alan Adamson, and another engineer from IBM
- Anil Kumar, and two other engineers from Intel
- an engineer from Sun
- Michael Armbrust from UC Berkeley RAD Lab
Friday Dec 14, 2007
George Ou wrote the most detailed article I've seen on SPEC's new energy benchmark for Java server workload. You could reasonably think I like the article because he mentions my name. But actually I've been a fan of Ou's column for some time, like when he wrote about the Green Grid consortium, Wifi security, and how to build a 50 watt home PC out of commodity parts.
Tuesday Dec 11, 2007
Today SPEC announced SPECpower_ssj2008, the first industry standard power performance benchmark. It measures electric power used at various load levels from active idle to 100% of possible throughput. The workload tested is a server side Java workload. The methodology is applicable to many workloads, and I hope in the future we will see more standard benchmarks, and application of these methods to measuring power consumption of customers' own workloads. This benchmark is the result of long hard work by dedicated engineers from many companies, universities, and Lawrence
Berkeley National Laboratory. Congratulations!
Wednesday Oct 31, 2007
"Is storage becoming IT's Hummer?" asks The Register. Reporting from SNW Europe they report that as data centers reduce the power cost of computing, storage is poised to become the biggest energy consumer. Well that's just the outcome SNIA hopes to avoid with their Green Storage Initiative. The efficiency race between computing and storage is one where we can cheer for both sides. Besides, as Jonathan points out, the distinction between computers and storage is blurring fast.
The Reg says that virtualization will be primarily responsible for reducing computing power usage through consolidation. Certainly the most effective way to save energy is to follow your mother's command: "Turn that thing off if you're not using it!" But I think the Reg is a bit premature in giving the industry credit for solving the problems of computing power usage. Yes there's a lot of innovation in this area making data centers more efficient in many different ways. But there remains a lot of hard work to do by vendors and users alike.
Thursday Oct 25, 2007
Earlier I wrote that the Camp Pendelton fire didn't threaten the nuclear power plant. But my photo did show the fire burning underneath the power lines, and fire under the lines caused them to shut the plant down. This contributed to the power shortages prompting SDG&E to call for more energy curtailments, and for Sun to take steps to reduce our power load. With so many power lines taken out by the fires, we narrowly avoided blackouts with some help from the Navy.
Here at home I continue better than self sufficient during the peak hours of electricity demand, but by a much smaller margin than usual. With a layer of ash coating my solar panels, efficiency is way down and I'm only generating 1 to 1.3 kW. Soon I'll have to venture up on the roof with a garden hose again to rinse them off, and think some more about my remote solar panel sprinkler project.
All this makes me think that San Diego really could be energy independent, and we ought to be. Start with lots more solar power, and wind power, which have different time-of-day power curves. Add some giant batteries, as developed by E.ON AG. And modernize the natural gas plants to use primarily as peakers in times of high demand. Power lines are great to move energy around for small adjustments in supply and demand, but are just too vulnerable to rely on for a large fraction of the region's power.
Thursday Oct 04, 2007
I measured electricity usage of our
home computers using a cheap clamp-on ammeter from Radio Shack. I
concluded that you need to make your own measurements because many of
my preconceptions did not hold up to actual measurement. Since then
I bought a more accurate and convenient Watts
Up meter for $21 from amazon.com, and re-measured everything.
To my
colleagues at HP: I apologize for saying your PC used 300 watts.
100 watts is more like it. I likewise
overstated the power usage of other PC's. I correct the measurements
here, along with some discoveries about Windows and Ubuntu power
management. You may be asking, why should I believe you now? Good
question. You should not believe some random person you find on the
net. You should get your own meter and measure your own equipment.
Why might you halfway believe me,
enough to read a bit more? Well, Watts Up does seem to be a more
reliable instrument, and unlike my earlier measurements it corrects
for the power factor. It displays the power factor, which can vary
greatly from one device to another. As a check on the new meter I
measured a 3-way light bulb: the 60w setting measured 44w, the 100w
setting measured 89w, and the 150w setting measured 135w. At least
I'm assured it's not measuring high.
A colleague in the SPEC power committee
laughed at my new meter, “You mean you trust a Watts Up meter?”
For the upcoming SPEC power
benchmarks that team of engineers from many companies, universities,
and agencies has written elaborate requirements for the power meter
and measurement protocol to be sure that results tested by one lab
will be comparable to results tested by another lab. But for home, if you can get any information at all about the power your electronic gear uses, that's good.
First I looked at how many watts they
used in normal home, light load operation. I switched the Linux
distro on PC2002 to Ubuntu 7.04 Feisty Fawn, and its power management
software seems to be the equal of Windows.
|
OS |
Computer |
Monitor |
Speakers |
Total |
| PC1999 |
Linux |
39 |
26 |
1 |
66 |
| PC1999 |
Windows 98 |
37 |
26 |
1 |
64 |
| PC2002 |
Linux JDS |
95 |
34 |
- |
129 |
| PC2002 |
Windows XP |
61 |
34 |
- |
95 |
| PC2002 |
Linux Ubuntu |
57 |
34 |
- |
91 |
| PC2006 |
Solaris |
77 |
22 |
- |
99 |
| PC2006 |
Windows XP |
67 |
22 |
- |
89 |
Next I looked at how many watts they
used under a heavy load. These weren't identical controlled loads,
and workload really matters to how much power is used.
|
OS |
Computer |
Monitor |
Speakers |
Total |
| PC1999 |
Linux |
46 |
26 |
1 |
73 |
| PC1999 |
Windows 98 |
55 |
26 |
1 |
82 |
| PC2002 |
Linux JDS |
95 |
34 |
- |
129 |
| PC2002 |
Windows XP |
100 |
34 |
- |
134 |
| PC2002 |
Linux Ubuntu |
118 |
34 |
- |
152 |
| PC2006 |
Solaris |
? |
22 |
- |
? |
| PC2006 |
Windows |
115 |
22 |
- |
137 |
How much of difference does workload
make? Here is the power consumption of PC2002 running Ubuntu:
|
Operation
|
Watts (PC only)
|
|
download OS updates
|
70
|
|
system quiescent
|
57
|
|
100-Base-T file copies
|
80 - 100
|
|
Play Supertux arcade game
|
116
|
|
Scanning document with USB scanner and xsane
|
118
|
|
Afterwards [Note 1]
|
115
|
|
Killed mysqld
|
60
|
[Note 1] I don't know what set off the
MySQL daemon, whether it was doing useful work, or had just run wild.
But without the power meter, had I noticed anything amiss at all, I
would have just been puzzled that the fan was running. When I stopped
the daemon, power usage fell back down to normal levels.
Next I suspended each PC by pressing
the sleep button as the system is configured out of the box.
|
OS |
Computer |
| PC1999 |
Linux |
27 |
| PC1999 |
Windows |
27 |
| PC2002 |
Windows |
2 |
| PC2002 |
Linux Ubuntu |
2 |
| PC2006 |
Windows |
44 |
But wait, what's going on with that new
HP PC taking so much more power suspended than the old HP PC? I
looked at the Windows power management control panel and started
reading the documentation. It turns out that the old PC was set by
default for the sleep button to put it into what Windows calls
“standby” state, while the new PC was set by default for
the sleep button to put it into “away” mode. Away mode
doesn't do much, and is intended for systems that might still be
doing some processing without a person sitting there. So it's kind of
like forcing it immediately into the quiescent state it would
eventually go into if you just left it alone, and the monitor blanks.
I changed it to go into suspend state, and then it used just 6 watts.
Hibernate state used 4 watts. Michael Chu wrote a great
description of how Windows XP power management works.
Here are some details of my
configurations.
|
OS |
Computer |
Monitor |
Speakers |
| PC1999 |
Primary: Sun JDS3 Linux; Secondary Windows 98SE |
PC, Sony Vaio, Pentium III, 500MHz |
NEC, 14” LCD, 1024x768 |
Sony |
| M2000 |
Primary: Linspire Linux; Secondary: Windows 98SE |
Laptop, Sharp Actius, Mobile Pentium II, 300MHz |
integrated |
integrated |
| PC2002 |
Primary: Ubuntu 7.04 Linux; Secondary Windows XP Home |
PC, Compaq S6000Z, Athlon 2600XP, 1.9GHz |
Samsung 170MP, 17” LCD, 1280x1024 |
integrated |
| PC2006 |
Primary: Solaris Nevada; Secondary: Windows XP Media |
PC, HP a1540n, Athlon 64 x2 4200 dual core, 2.2GHz |
Sharp LL193A, 19” LCD, 1280x1024 |
integrated |
| M2007 |
Debian Linux |
Tablet, Nokia N800, TI OMAP 2420, 330MHz |
integrated |
integrated |
Wednesday Sep 19, 2007
George Ou at ZDnet shows how to build your own PC out of commodity parts using only 50 watts. A crucial point for him is the efficiency of the power supply. He does use regular disks. I would have expected one flash memory "disk" and one spinning disk, but maybe the power management software isn't yet able to distinguish and power down the more energy hungry disk first.
Tuesday Sep 04, 2007
Yet another scorching day and another energy curtailment at Sun in San Diego. So in my office the lights and workstation are powered off while I work at home with all the windows open. On Labor Day weekend it hit 110 degrees in the east county and over 90 by the coast. At Legoland the most popular attraction was Soak-n-Sail Pirate Shores that dumps a 300 gallon bucket of water on you.
Yesterday SDG&E hit an all-time record demand of 4,636 MW. Although my solar panels hit their peak production in the early afternoon, already by 8am they were producing twice as much energy as my house was using, and the gap is steadily rising. Peak demand for SDG&E is from 4 to 6 pm. Though that's past the peak of my panels' production, my house remains a net generator of electricity through 6pm when it finally begins drawing power from the grid.
Tuesday Aug 21, 2007
Gene Saunders wonders about the pace of electric utilities implementing rebate programs for efficient computer servers. I don't have any inside information about what utilities are doing what, but it's worth watching the Consortium for Energy Efficiency: " CEE members include utilities,
statewide and regional market transformation administrators,
environmental groups, research organizations and state energy offices
in the U.S. and Canada."
Their Data Centers and Servers initiative seeks to promote power efficient computing just as they already promote energy efficiency in everything from commercial dishwashers to HVAC equipment. The Spring newsletter mentioned the research done at Lawrence Berkeley National Laboratory, the Green Grid, the SPEC power committee, and the PG&E rebate program.
Walter,
Does the engineer from AMD wish to remai...
The AMD engineer hasn't said that it's okay to pos...