Vijay Tatkar's Blog

Sun Studio Compilers + Solaris have 15% advantage over MS Windows/XP
This interesting performance blog at Geek Patrol was being discussed on two aliases internally:
Sun Ultra 20 M2 Performance
It is an interesting comparison between Windows/XP Compiler and Sun Studio running on Solaris using a benchmark called geekbench. Geek Bench 2006 is available here for Solaris x86 (both 32bit and 64bit versions). Overall, the conclusion is that SunStudio does about 15% better. In particular, I like this pronouncement:
Solaris outperformed Windows in almost every benchmark category, even outperforming Windows dramatically in some specific tests (such as some of the floating point benchmarks). If you’re working with processor-intensive tasks, Solaris might be the operating system for you.
This is inspite of the Sun Studio numbers for STREAM being under-represented. John Poole, who wrote the blog, used the -fast flag, while adding -xvector=simd -xprefetch would have helped it enormously more (by about 50% or so, I'm guessing). My own references to STREAM here and here show these numbers to be compellingly higher.
Comments posted here on OSNews are also worth reading.
Posted by tatkar ( Oct 24 2006, 01:34:58 PM PDT ) Permalink Comments [3]

Compilers continue World Record Performance streak on new SunFire Rev F Opteron
Sun has made much of the Datacenter in a Trailer (aka Project Blackbox) concept, lately. It is clearly one of those ideas that are so compelling that they are also simple. The press has been quick to pick up on it. See here, here at IT Jungle  and here at Reuters, for example.
What didnt get as much press at this time was the other announcement from Sun: Sun has refreshed the entire x64 line of servers with the new AMD Opteron RevF (and whats more, at 2.8GHz, dual-core) chips.
Sun's compilers are at the fore-front once again in the SPEC CPU performance race. The current systems and Sun Studio 11 combine to give very impressive performance overall, nicely edging out systems based on Woodcrest, Itanium and Power5+ by some very healthy margins. In particular:

• The Sun Fire X4100 M2 is up to 40+% faster than Intel Xeon 5160-based servers from Dell, IBM and HP
• Sun Fire X4100 M2 4-core, 2p SPEC 2000 FPrate=119 (peak), 102(base)
• Woodcrest servers from IBM, HP, Dell 4-core, 2p SPEC 2000 FPrate= 82 to 85
  
• The Sun Fire X4200 M2 with  SPEC 2000 FP (peak) of 3515 beats the best Intel Itanium2 result  of 3045 by 15+%
• Sun Fire X4200 M2 with SPEC OMP result of 13222 beats the best 4-thread IBM Power5+ result by 60+%
  

Sun is on a roll here and Sun Studio 11 is helping differentiate. Even on the same system, the advantage with Sun Studio11 is about 20-25% over competitive compilers. For instance, the new Sun Fire X4100 with the score(see above) for SPECfp of 119 beats a similarly configured IBM box with SuSE9 and PathScale compilers with a SPECfp of 90.3 (If you dont like SPECrate numbers, how about SPECfp itself: 2857(base) with SunStudio 11 vs. 2218peak, 2029 base with SuSE9/PathScale on the IBM box... hmm, these IBM numbers are curiosly smaller than I'd have expected on this box.). The idea of using disclosed Base numbers is because these dont use autopar (automatic parallelization, so they are single-core base numbers).

Required Disclosure Statements:

SPEC, SPEComp, SPECCfp and SPECfp Rate are Registered Trademarks of Standard Performance Evaluation Sun's results were submitted for review. For SPEC comparisons, socket equates to chip.
Competitive results from www.spec.org as of Oct 15, 2006.

Sun Fire X4100 M2 (AMD Opteron model 2220SE, 4 cores, 2 chips): SPECfp_rate2000: 119, SPECfp_rate_base2000:  102;
IBM System X 3550 (Xeon 5160, 4 cores, 2 chips, Windows 2003 Server EE): SPECfp_rate_base2000 – 82.1
HP ProLiant DL360G5 (Xeon 5160, 4 cores, 2 chips, RHEL 4AS U3): SPECfp_rate2000 – 84.7
Dell PowerEdge 1950 (Xeon 5160, 4 cores, 2 chipsp, RHEL 4AS U3): SPECfp_rate2000 - 83.4
IBM System X 3455 (AMD Opteron model 2220SE, 4 cores, 2 chips, SLES 9 SP3): SPECfp_rate2000: 90.3

Sun Fire X4100 M2 (AMDOpteron model 2220SE, 4 cores, 2 chips): SPECfp2000 – 3515
Sun Fire X4100 M2 (AMD Opteron model 2220SE, 4 cores, 2 chips): SPECfp2000 Base– 2857
Bull NovaScale 3045 (Itanium2 9050, 8 cores, 4 chips, Bull Linux AS4): SPECfp2000 – 3017
IBM System X 3455 (AMD Opteron model 2220SE, 4 cores, 2 chips, SLES 9 SP3): SPECfp2000: 2218 Peak, 2029 Base

Sun Fire X4100 M2 (AMD Opteron model 2220SE, 4 cores, 2 chips, 4 thread): SPECompM2001 – 13222
IBM System p5 520 (POWER5+, 2 cores, 1 chip, 4 threads, AIX5L V5.3): SPECompM2001 - 8174

Posted by tatkar ( Oct 23 2006, 10:04:32 AM PDT ) Permalink Comments [0]

On vacation till August 24th
See ya folks in about 8 days; I dont know if I can post (or read) while on vacation. Will be taking the family to Niagara Falls and area. Cant promise to be on the web in any case during this period. Looking forward to some nice R&R with the family!
Posted by tatkar ( Aug 15 2006, 11:31:08 PM PDT ) Permalink Comments [0]

SunStudio compiles Loki library!
Good news! Sun C++ now compiles the Loki library (you can download Loki sources here)!
Once considered one of the hardest libraries to compile, Simon who had worked on making BOOST work and reported his results regularly at this blog, has now gotten the last two bugs with Loki worked out! Check out his latest entry related to Loki here. You'll also need a patch which is also uploaded at Simon's site here. The upcoming (soon!) SunStudio Express 2 drop should contain the fixes for it.
Great job, C++ team!
Posted by tatkar ( Jul 21 2006, 08:16:24 AM PDT ) Permalink Comments [0]

Customer Testimonial for Sun Try-and-Buy Heres a Sun Success story (validation) among many others with the Sun Try-And-Buy Program that I described earlier in my blog.
DigitTar, a messaging services outsourcing company, which was an early Sun customer and then fell out of the Sun camp, has reiterated its love with the SunFire T2000 systems. DigiTar has now consolidated 10 Opteron boxes from HP, with 2 T2000 boxes (with 50% headroom for growth) and has this to say (glow) about the servers:
"We were kind of an anti-Sun house for a long time because of our earlier experience with Solaris, and in the past Sun was expensive and the price-performance just wasn't there for their servers," he says. "When we put Solaris 10 and the T2000 into production, those issues were resolved."
and this quote particularly:
"The T2000 did require some tuning to meet DigiTar's production requirements, but once it was working "it scaled incredibly compared to the dual-core Opteron servers we had been running."
May this be a precursor to a great new phase for Sun!
Posted by tatkar ( Jul 18 2006, 09:59:45 AM PDT ) Permalink Comments [0]

Contrarian thinking in Sun's Opteron Server offerings
Sun had plenty to glow about in the Big  Opteron server announcement this Tuesday . The boxes are truly remarkable in design and technical specifications and the press has finally been positive in its reception of these announcements, from Mercury News to The Register to ZDnet to News.com  to Internet News. There have been plenty of accolades from customers who have had a preview of these boxes and you'll find their opinion here.
But in this blog, I'd like to emphasize the contrarian aspect of the offering. True to its history, Sun hasnt just introduced a hot box, but it hopes that it will change the game as well, at least in significant, emerging segments. So let me try and enumerate what I believe constitutes contrarian thinking here:

• Innovation still matters: Sun has proved it can add value to industry-standard configurations
• Differentiate on performance, use HPC needs as a leading indicator of  the market (Best SPEC CPU2000 FP on the planet, World Record SPEC OMP2001, Leading SPEC CPU2000 throughput)
  
• Bet on AMD/Opteron chip, at a time when CEOs are busy proclaiming that Woodcrest will take Opteron to the woodshed in an effort to endear to Intel
  
• Multiple Platforms: All new Opteron servers from Sun, offer Solaris, various flavors of Linux and are certified to run Windows.
• The new Blade Server, Sun Blade 8000 Modular System, is
• Optimized for 4 socket (4 CPU, dual-core today, quad-core in future) offering because Sun believes the market is moving up in this segment and will not stay at 2 socket. To quote demi-god Andy B: "It's not that our blade is too large. It's that the others are too small. . . .There are two megatrends in servers: miniaturization and multicore -- quad-core, octo-core, hexadeci-core. You definitely want bigger blades with more memory and more input-output."
  
•  Big blades: 20 times the throughput of traditional blade servers with an 18U chassis. Sun will introduce smaller chassis later in the year.
• Modular Design: high availability, hot-pluggable modular components including I/O, processor, chassis infrastructure or system management.
  
• Built for Reliability, Maintainability and longer life
• The world's first hybrid data-server, Sun Fire X4500,  a combination of 4-way x64 server with 24TB of data storage.

There probably are a few more out there. But these are the ones that occurred to me as I was thinking about this. I'd dearly like to hear from you on what you think about this. Its critical to Sun: in this marketplace, you can make compelling margins by bucking the mainstream.
If this can change the mind Sun's leading critic, Toni Sacconaghi, (read here), I wonder what it can do in the marketplace.
Posted by tatkar ( Jul 14 2006, 11:13:10 AM PDT ) Permalink Comments [0]

SunStudio Portal article compares Solaris and Linux for developers
There is a new article on the Sun Developer's portal that compares Solaris and Linux from a application developer's viewpoint! Highly recommended reading for someone who is planning to be on both platforms; read the full article here. To quote from the article: "This article examines similarities and differences in the development environments of both operating systems. Someone responsible for porting applications from Linux to the Solaris OS, or programmers with prior Linux experience that want to learn development on the Solaris OS, should benefit from this article."
Of course, SunStudio now runs on both, even though the Linux version is still in pre-release (TP) stage . With equivalent features and performance on both, we are hoping that it will be compelling choice for those looking for a common build environment . Posted by tatkar ( Jul 13 2006, 09:14:34 AM PDT ) Permalink Comments [0]

Can we Manage Failure to achieve Success? A new businessweek article...
I am fascinated with how businesses use (or often dont use ) controlled failure as a mechanism for achieving greater success.
In the past, I posted, for example, in my blog, Google VP, Marissa Ann Meyer, touting the value of  Constrained creativity.
Or see here for a more hilariously presented video on the Importance of mistakes by John Cleese.
There is plenty of material on the web on the topic of learning from failures or on the importance of using failure as a creativity tool (just google these phrases and see how many hits you get!). But it still remains a daunting topic, rarely mentioned alongside popular positive attitude books (you know which ones I mean, just walk down your favorite Airport bookstore for a collection!).

So, its good to see when popular and reputable publications embrace the topic from time to time. The most recent cover story Businessweek  titled: Eureka, we failed! caught my eye. It talks about how smart companies learn from their failures and gives some interesting subtext with stories of favorite failures of such luminaries as Jeff Immelt of GE, Pete Carroll of USC, etc. The online article is titled: How failure breeds success!
For fun, see this  Businessweek online extra about some spectacular failures.
Makes for interesting reading. I'd urge you to do so. Reading it online might need a subscription number or registration, but you can always look it up in your favorite library!

My primary interest in this is  in day-to-day application. As  managers of  technology groups, I feel we are constantly walking the thin line between encouraging  engineers to  push the leading edge of technology (and often, as part of that effort, push the edge in terms of their own experience, expertise and horizons) and controlling the path which  leads to unmitigated failure. Failure detection and management is a very useful control tool (IMO) to achieve aggressive product goals . In his hilarious teaching video, John Cleese, eg. makes the point that heat-seeking missiles hit their target by constantly readjusting their trajectory based on updated sensory readings of the target and that learning from failure is no different (in the sense of readjusting frequently based on violations of  parameters or boundaries). Dont we contain  forest fires  through controlled burns to limit the otherwise extensive damage that may be caused by unsuccessfully fighting a larger fire. In everyday lives, we recognize stumbling frequently as a necessary step to learning to walk. If  thats the case, why are organizations so reluctant to manage failure?

Posted by tatkar ( Jul 12 2006, 11:32:20 AM PDT ) Permalink Comments [0]

Three New Opteron Servers from Sun
Big Announcement(s) from Sun today.
Sun introduces the next generation of technology servers today based on three primary guiding principles:

• Innovation matters and is a key differentiator
• HPC market is a leading indicator
• Power: Cooling and efficiency are primary concerns

In true Contrary fashion, Sun has introduced three new systems will change the game (we hope, in our favor):

• Sun Fire X4600 Server: The industry's first 16-way x64 server in a single 4U chassis.
• Sun Fire X4500 Server: The world's first hybrid data server -- a combination of a 4-way x64 server with 24TB of data storage starting as low as $2/GB.
• Sun Blade 8000 Modular System: The world's first no compromise blade platform -- a breakthrough modular architecture that delivers the performance, price, and flexibility of rackmount servers with the serviceability and efficiency of blades.

These systems all run Solaris, various flavors of Linux and Microsoft Windows. If you couldnt listen to the announcement earlier, you can view the keynote and announcement here.

So, why am I excited? On the Sun Studio compiler side, here is where the compiler drove performance metrics on these:
We set these new benchmark records on all three of these systems. Whats more is that we trump everyone in terms of SPEC CPU2000 FP with a score of 3538, beating out not just Woodcrest's impressive numbers, but also IBM Power5+ systems.

But beyond the parochial pride and joy of achieving (and beating) World Records, this is the first time since the late 80s that I have seen such impressively and elegantly simple, plug-and-play, extensible systems from Sun. To be fair, I have been in many presentations where Andy Bechtolsheim has rolled out one or more of these systems to show to customers, but I couldnt blog about it until now, though there were plenty of rumors floating about their existence(like this article in The Register). 

Over the next few days or weeks, I'd like to discuss various aspects of the new offerings from Sun from my (humble?) perspective. I hope you'll join me in the discussion, because I'm truly impressed by these offerings and I dearly hope that they will define the next successful phase for Sun.
Posted by tatkar ( Jul 11 2006, 02:11:10 PM PDT ) Permalink Comments [0]

Sun's Try and Buy Program for Servers and Storage
There are 8 products in Sun's Try and Buy Program.
Here they are (follow the links to follow the program in each):

• Sun Fire T2000 Server
• Sun Fire T1000 Server
• Sun Fire X4200 Server
• Sun Fire X4100 Server
• Sun Fire X2100 Server
• Sun Ultra 40 Workstation
• Sun Ultra 20 Workstation
• Sun StorageTek 5320 NAS Appliance

You can follow this link to get the details on the program.
The page also directs you to a page that will help you qualify for the program.
 Go on, give it a try! I am sure that you'll be pleased with the quality of these offerings.
Posted by tatkar ( Jul 10 2006, 08:45:29 AM PDT ) Permalink Comments [4]

Sun Studio Service Plans Explained
Did you have difficulty navigating through Sun's fairly complex (and complete) support plans matrix before?
If so, Sun has just revamped the service offerings, especially for Sun Studio compilers and tools.
This document is an easy and lucid 1-pager on support offerings.

If this is not clear enough send (me) email and I'll help clarify in this document as well as to you.
Posted by tatkar ( Jul 07 2006, 02:05:21 PM PDT ) Permalink Comments [0]

SunFire X4100 and X4200 Shatter Another World Record: TPC-H
Sun Fire X4100 and X4200 servers  now hold world-records in price-performance for TPC-H (100GB and 300GB categories).

With this new benchmark, Sun Fire X4100 offers:

• Best: TPC-H price-performance among all systems at 100GB
• Best: TPC-H performance among all 2-way systems at 100GB
• Best: data storage ratio
• Best: lowest cost of ownership among all 100GB submissions

Sun Fire X4200 offers:

• Best: TPC-H price-performance among all systems at 300GB
• Best: TPC-H performance among all 2-way systems at 300GB
• Best: lowest cost of ownership among all 300GB submissions

The officially posted results are here.
Posted by tatkar ( Jul 05 2006, 10:35:11 AM PDT ) Permalink Comments [0]

Happy Birthday, America!
America turns a young 230!
A country that has opened the doors to much of the world at a shot at "Life, liberty and the pursuit of happiness". Not without controversy -and not without its detractors, none of whom have generally done any better- but most definitely a nation that has stands out as a bright beacon of the principles on which it was founded.
And, moreover, one that has attracted with open arms and thrived on the backs of immigrants, such as myself!
Happy Independence Day, my fellow citizens, and may we be able to celebrate another 230 with as much pride!
Posted by tatkar ( Jul 04 2006, 12:32:59 PM PDT ) Permalink Comments [0]

Welcome to our Compilers-for-Linux Commander-in-Chief!
I promised out primary instigator and evangelist of our Linux Compilers that I'd welcome him to Sun's blogging community as its Commander-in-Chief when he decides to join us here.
He is now here !
I am glad to see that he has jumped in. He is sure to bring his passion, immense energy, keen intellect and perception and wholly different viewpoint here. Welcome, Roman Shaposhnick!
Roman has a long history at Sun. He joined us at one of the first engineers on the compiler project in St. Petersburg, Russia, long before we started our own Sun Engineering Center in SPB. He has worked in C++ Compiler, and dbx before bringing his passion for Linux to this product!
Hope you enjoy this forum as much as I think you will and you will have a long, and enjoyable experience here!

Posted by tatkar ( Jun 30 2006, 06:42:42 PM PDT ) Permalink Comments [0]

Sun Studio Linux Compiler TP4: Whats New?
Sun Studio for Linux Technology Preview 4 is open for business!
 You can download it and get the details here.

Here is the list of new features implemented as described in this active forum:

• OpenMP support for cc and f95
• -xipo[=1,2], whole-program interprocedural optimization

For Sun C and C++ compilers

• Thread local storage support
• -fast command line macro now includes:
• -dalign -fns -nofstore -fsimple=2 -fsingle -xalias_level=basic -xtarget=native -xO5 -xregs=frameptr
• suncc, sunCC, sunf90, sunf77, sunf95, sunc89, sunc99, sunas compiler driver nicknames
• -m32 and -m64 flags specifying respectively, the ILP32 and LP64 address models
• The -xMD, -xMF, -xMMD flags for controlling makefile dependency information
• #include_next directive
• aligned and visibility __attribute__ support in C
• value returning blocks in C
• __typeof__ support in C
• Zero length arrays in C
• C++ compiler: stlport4 and rw7 libraries

For Fortran compiler:

• -fast command line macro now includes:
• -xO5 -xtarget=native -xprefetch=auto -dalign -fsimple=2 -fns=yes -ftrap=common -nofstore
• interval arithmetic on Solaris Intel
• CAUTION messages issued when using the same symbols that are defined in different modules.
• Faster compilation time for source files with many constant values.
• Intrinsic functions called with UNSIGNED arguments improved.

Keep the feedback coming, folks!
Posted by tatkar ( Jun 30 2006, 09:22:41 AM PDT ) Permalink Comments [0]