4,580 Posts served
11,094 Conversations started
- Academic

- Android

- Art, Music, & Animation

- Embedded Computing

- Events

- Game Development

- Graphics & Media

- Intel SW Partner Program

- Intel® AppUp Developer Program

- Manageability & Security

- Mobility

- Open Source

- Parallel Programming

- Performance and Optimization

- Power Efficiency

- Server

- Site News & Announcements

- Software Tools

- Ultrabook

- Association for Computing Machinery TechNews (ACM)
- Go Parallel! (Dr. Dobbs)
- HPCwire (Tabor Communications, Inc.)
- insideHPC (John West)
- Joe Duffy's Weblog (Microsoft)
- Microsoft Parallel Programming Development Center (Microsoft Germany)
- MultiCoreInfo.com
- scalability.org (Scalable Informatics)
- Software Dev Blog (Intel Germany)
- Soft Talk Blog (Intel United Kingdom)
- The Moth (Microsoft)
Intel + Sun + 1 Year = OpenSolaris++
By David Stewart (Intel) (177 posts) on January 22, 2008 at 3:35 pm
Today marks the one year anniversary of Intel and Sun joining together in collaboration to improve the Solaris operating system, among other things. Here's the way I tell it to people: "We're working to make Xeon the best platform to run Solaris."
One year into our collaboration marks a great time to review what we have done together so far in our joint development work to make Xeon/Solaris great. All of these would have been *impossible* without serious collaborative work with Sun. (And incidentally, it's a lot of fun for us to work with the experts in this community).
- Almost at once, we decided to focus our engineering efforts on OpenSolaris, the open source project. We set up the "intel-platform" project, and started consulting on a variety of efforts.
- Right off the bat, we collaborated on a microcode update utility. This basic serviceability feature ensures that Intel processors can get the latest microcode fixes under Solaris.
- We consulted on a number of basic efforts, like proper decoding of CPUID output and the use of the MONITOR and MWAIT instructions, the direct result of which are performance improvements (some of them significant).
- We have supported the OpenSolaris community through participation in the first Developers Summit, CommunityOne, and in various projects and community groups
- We are optimizing various libc and kernel primitives for out current processors. For example in libc, we have optimized memset() and memcpy() routines, in some cases up to 350%. We're also working on strlen(), strcmp(), bzero(), etc. (Some of this work has been completed for some months. I'll put details in a future blog post).
- We together wrote the "iwk" driver, new support for Intel's latest 4965 wireless part, which is part of our shipping "Santa Rosa" mobile platform, better known as "Centrino Pro". We added a project site called http://opensourcewireless.org to improve communication flow about Wireless support in the Solaris and BSD worlds.
- We have collectively increased our efforts in the wired LAN space for Intel LAN silicon. We should very soon seen new support for our roadmap of current and future adaptors.
- Speaking of the world of device support, we have either delivered code or consulted on ICH9 AHCI support, graphics (965), Advanced Management Technology (AMT), I/O Acceleration Technology (IOAT) and it seems like a handful of other things.
- We started enhancing the "Predictive Self-Healing" feature, better known as FMA. Specifically we're helping the FMA team with support for our current Xeon-based server platforms.
- Improving power-performance and power utilization is a major focus for Intel. We implemented PowerTop, an application designed to track down guilty culprits which prevent an otherwise idle system from going into a lower power state (and implemented it using DTrace). We have provided consulting on a number of P- and C- state features of our processors, and set up a lab to measure SPECpower on OpenSolaris.
- We're starting a deeper collaboration on our Virtualization Technology optimized for xVM. More on this later.
- We have teams working on monitoring OpenSolaris performance from week to week, analyzing xVM functionality on Xeon, doing basic validation of our OpenSolaris work.
Well, I'm sure I'm forgetting a lot, but this came off the top of my head today. And of course for this year, I'm staring at a long list of things we're doing to make sure our 2008 processors (Nehalem and friends) are supported right out of the chute on OpenSolaris.
I hope this doesn't sound like bragging or boasting here – I am really extraordinarily proud of the Intel team and the people we have worked with at Sun and other places in the OpenSolaris community to make this project great. Thanks all for a great year and for a fantastic 2008!
Categories: Manageability & Security, Mobility, Open Source, Parallel Programming
For more complete information about compiler optimizations, see our Optimization Notice.
Comments (12)
| January 22, 2008 7:18 PM PST
Lee Hepler |
I hope Intel engineers are also working on support for Intel chip sets. I will be upgrading my hardware this weekend to an X38 based motherboard and a E8400 processor. It will be a bummer to wait for several months to run Solaris on this system if it doesn't run yet. Intel wants open source developers to use their platforms and most of us use the desktop chips and chip sets. I must admit this is only my second Intel build (not counting the Z80 systems) with the first being a DX2-66 machine. I've been running AMD ever since the Pentium first came out but Intel has done such a briliant job on the new CPU's that I just couldn't resist. Too bad they delayed the Quad cores. I wanted the extra cores so I could run 2 or three OS's at once to learn more about virtualization. Lee Hepler |
| January 22, 2008 11:23 PM PST
Jim Grisanzio | cool. congrats. :) |
| January 23, 2008 12:36 AM PST
Glynn Foster | Awesome list - congrats Dave on all the hard work you and your team have done. Very cool to know that you guys are a strong part of the OpenSolaris community. |
| January 23, 2008 1:20 PM PST
David Stewart (Intel)
|
To Lee Hepler - > Hope you have a good experience with your X38 / E8400. > For such new hardware, I would recommend that you go with OpenSolaris rather than Solaris 10. > We're definitely working on enabling new chipset technologies. > I just talked to someone who is running OpenSolaris build 80, and it's running great on a X38 with the E8400 processor. |
| January 24, 2008 11:11 AM PST
Rayson | Very nice!! |
| January 27, 2008 11:35 PM PST
Happy Aether Bunny |
As of last night: E8400 + Intel DQ35JO + 8Gb + b79 = works. b70 suicided in the boot loader, b79 had no such problems, onboard graphics didn't like the 1920x1200 monitor so I ended up using any old nvidia. no sound ^^; I do wish the intel bios would allow you to move that 750mb of junk out of the way, it feels like I'm wasting half a DIMM. |
| January 28, 2008 3:00 AM PST
Rick (Vectorpedia) | Intel and Sun was a great marriage........keep up the good work ! |
| January 29, 2008 8:45 AM PST
Vijay Tatkar |
Great blog, Dave. Working with you in this year has been one of the high points of this relationship. On the compiler side, we've made tremendous strides with Core2 instruction selection and tuning, which resulted in HUGE SPECfp increases. As noted here: http://blogs.sun.com/tatkar/entry/sun_studio_12_patch_performance Similar results with the now tuned Sun Performance Library as well. Onward, to even bigger and better things. Not only is there accomplishments of the past year, but theres wonderful momentum built up as well. Cheers! Awesome work, Intel! |
| January 29, 2008 8:52 AM PST
David Stewart (Intel)
| Congratulations, Vijay! I'm very impressed with the progress you have made on SPECfp and I'm looking forward to more and better things with the premier compiler for Solaris/Xeon. |
| March 12, 2008 3:04 AM PDT
IBRAHIM FAGGE | such joint togetherness is an important mark up point in this IT Era. |
| February 14, 2012 3:46 PM PST
Sergiu | I am lokin for an alternative os for intel ... ca anyone tell me where can i find open solaris |
Trackbacks (1)
- Intel® Software Network Blogs » Making access to memory faster in OpenSolaris (and Core2)
April 4, 2008 4:55 PM PDT



stevel