How about Vulcan OS which ran on Harris minicomputers in the late 70s and 80s. I cut my teeth on Vulcan on a Harris H500 on my junior college. Even learned Harris assembly language.
Grew up in and around Eureka in Humboldt County from ages 5-20. It definitely is a different place. (325 miles north of San Fransisco on HWY101 for those who don't know.) People in Humboldt tend to believe and act like they live behind the Redwood Curtain. And it is true.
One of the problems with running Linux on the compute nodes is that the normal mechanism of loading applications using demand paging doesn't fit well. I worked in the MPP supercomputing business 15 years ago and we tried running a full Unix kernel (Linux was just coming into being.) and the users ended having to run they apps twice. The first time to get all the memory warmed up with the pages loaded.
We were working to fix it, but then end of the cold war killed the business.
Have you ever even been to Eureka ?? Your statement is total hogwash regarding the flooding. I know, I lived in a around Eureka, Loleta, Fortuna and the other parts of the county for almost 20 years and my parents are still there. Actually the locals would love to have some good flooding like we had in the early 1960s. That how the farmland around Eel River stays fertile.
I grew up 10 miles south of Eureka and 10 miles north of Scotia where Pacific Lumber is based. These tree sitters are invading private property and breaking the law. Pacific Lumber has a legal right to harvest those trees. Why we tolerate this is beyond me. If the target was a high tech office building in Santa Clara instead of remote Humboldt County, Calif. it would not have been allowed to continue. There is only around 100,000 people in the entire county and the major economic activities are Tourism, Agriculture, Fishing and Timber. The only one that pays worth spit is timber.
Having worked for Intel for 15 years (no longer however) and based on my observations during the Microsoft anti-trust effort, I'd say that Intel as a company is vastly more mature and sophisticated that Microsoft in protecting itself against anti-trust allegations. We had training down to the individual contributor level and the are active in making sure they are not put at risk.
Locus Computing (which doesn't exist anymore.) had a technology called Locus TNC. When I was still at Intel doing supercomputers, we used this in the Paragon MPP system to do full SSI. (We did it across hundreds of nodes.)
If you notice on the Compaq SSI web page, one of the principles is Bruce Walker who was one of the architects at Locus when I was involved with them. (I never dealt with him personally.)
There is a fairly old book describing Locus TNC which was published through MIT Press called The LOCUS Distributed System Architecture by Popek and Walker. It worked fairly well but we never implemented any of the availability or reliability ideas that were kicked around.
I assume that since Apple is using the Mach 3 kernel that they are running with a user mode server ?? Does anyone know ?? We used Mach 3 and the OSF/1 single server on the Paragon supercomputer and could never get around the performance hit that was caused by the user mode server and all the protection domains you had to cross to do anything. (networking in particular.)
I left Intel after 15 years. (I've been out 10 months.) I think the main reason for disasters like Rambus and many of the other execution problems is that the traditional Intel culture has been allowed to slip away. Believe it or not, the intenral culture revolved around responsibility and accountability. Around 6 years ago that started to change. Disenting opinions where not welcomed. (Shoot the messager.) and too many decisions are being made too high in the chain. (Specifically technical decisions.)
The only thing new here is that the processor manuals are being released prior to release of the chip and they are not under NDA restrictions. With previous chips, the release of the manuals would have waited until the chip was released. (Full Disclosure: I worked for Intel for 15 years and quit last December.)
The real reason that supercomputing appears to be dieing is that there is no market. There is only one actual customer - the U.S. government. Many of what use to be supercomuting applications can now be served with standard commercial systems. Does anyone know if IBM really makes any money on their supercomputer systems ?? I bet not. Great vehicles for doing R&D however.
Intel has been out of supercomputing for sometime. Their TFLOPS machine was the last and it was not a "product". (A one time prototype.) Their last product was the Paragon XP/S.
This is somewhat ironic. The first week in December, I ended 15 years of employment at Intel Corp. One of the major frustrations over the years was that their entire I.T. organization was set up to support someone who runs MS Office and that's it. It was very difficult for engineers (particually software engineers.) to get decent machines. So it seems funny that they will be giving employees new machines for home but can't seem to give their engineers decent systems for their desk at work.
This just doesn't make sense. What about the companies such as SmartHome.com which sell hidden video cameras ? (http://www.smarthome.com/7569.html is a good example.) There is a number of companies which sell these. Are they going to be shut down too ??
You've hit it on the head. This is the real reason for the decline of the supercomputing market. The end of the cold war and the rise in performance of general purpose systems has made the traditional supercomputer obsolete except in some extreme niche markets. I predict that in the future, the only participants in this market will be companies like IBM which can afford to use their supercomputer business as a place to develop new technology but not necessarily make any money. (I spent 6 years in Intel's Supercomputer Systems Division).
Given the apparent endorsement and possibly connections with the DOE (given the quote in the ZDNET article, this is another sign of the end of the supercomputing business. Look at all the roadkill: Thinking Machines KSR Ncube - still exists does video servers I guess. Cray Computer Intel Supercomputer Systems Division (now defunct.) Convex - Bought by HP Cray Research - Bought by SGI Seems IBM is the only really viable player anymore. When I was with Intel SSD, it was obvious that the government was making it really hard to make a profit in that business.
This is all so deja vu. Does anyone know if they are talking about a single system image style cluster (something like the old Locus TCF/TNC) where the cluster just looks like a big system as far as users and apps go ?? In a past life I worked on such as system which supported hundreds of nodes. Instead of Linux, we used the Mach 3.0 microkernel from CMU and a user mode Unix server. Scalability and availability will be big challenges.
Slashdot Mirror
How about Vulcan OS which ran on Harris minicomputers in the late 70s and 80s. I cut my teeth on Vulcan on a Harris H500 on my junior college. Even learned Harris assembly language.
Grew up in and around Eureka in Humboldt County from ages 5-20. It definitely is a different place. (325 miles north of San Fransisco on HWY101 for those who don't know.) People in Humboldt tend to believe and act like they live behind the Redwood Curtain. And it is true.
The i/o nodes and service nodes run Linux. Compute nodes run a custom kernel called CNK.- Utah-2005/BGL-SSW03-CNK-CIOD.pdf)
(See http://www-unix.mcs.anl.gov/~beckman/bluegene/SSW
One of the problems with running Linux on the compute nodes is that the normal mechanism of loading applications using demand paging doesn't fit well. I worked in the MPP supercomputing business 15 years ago and we tried running a full Unix kernel (Linux was just coming into being.) and the users ended having to run they apps twice. The first time to get all the memory warmed up with the pages loaded.
We were working to fix it, but then end of the cold war killed the business.
Have you ever even been to Eureka ?? Your statement is total hogwash regarding the flooding. I know, I lived in a around Eureka, Loleta, Fortuna and the other parts of the county for almost 20 years and my parents are still there. Actually the locals would love to have some good flooding like we had in the early 1960s. That how the farmland around Eel River stays fertile.
I grew up 10 miles south of Eureka and 10 miles north of Scotia where Pacific Lumber is based. These tree sitters are invading private property and breaking the law. Pacific Lumber has a legal right to harvest those trees. Why we tolerate this is beyond me. If the target was a high tech office building in Santa Clara instead of remote Humboldt County, Calif. it would not have been allowed to continue. There is only around 100,000 people in the entire county and the major economic activities are Tourism, Agriculture, Fishing and Timber. The only one that pays worth spit is timber.
Having worked for Intel for 15 years (no longer however) and based on my observations during the Microsoft anti-trust effort, I'd say that Intel as a company is vastly more mature and sophisticated that Microsoft in protecting itself against anti-trust allegations. We had training down to the individual contributor level and the are active in making sure they are not put at risk.
Locus Computing (which doesn't exist anymore.) had a technology called Locus TNC. When I was still at Intel doing supercomputers, we used this in the Paragon MPP system to do full SSI. (We did it across hundreds of nodes.) If you notice on the Compaq SSI web page, one of the principles is Bruce Walker who was one of the architects at Locus when I was involved with them. (I never dealt with him personally.) There is a fairly old book describing Locus TNC which was published through MIT Press called The LOCUS Distributed System Architecture by Popek and Walker. It worked fairly well but we never implemented any of the availability or reliability ideas that were kicked around.
I assume that since Apple is using the Mach 3 kernel that they are running with a user mode server ?? Does anyone know ?? We used Mach 3 and the OSF/1 single server on the Paragon supercomputer and could never get around the performance hit that was caused by the user mode server and all the protection domains you had to cross to do anything. (networking in particular.)
I left Intel after 15 years. (I've been out 10 months.) I think the main reason for disasters like Rambus and many of the other execution problems is that the traditional Intel culture has been allowed to slip away. Believe it or not, the intenral culture revolved around responsibility and accountability. Around 6 years ago that started to change. Disenting opinions where not welcomed. (Shoot the messager.) and too many decisions are being made too high in the chain. (Specifically technical decisions.)
Always does it for me!
The only thing new here is that the processor manuals are being released prior to release of the chip and they are not under NDA restrictions. With previous chips, the release of the manuals would have waited until the chip was released. (Full Disclosure: I worked for Intel for 15 years and quit last December.)
The real reason that supercomputing appears to be dieing is that there is no market. There is only one actual customer - the U.S. government. Many of what use to be supercomuting applications can now be served with standard commercial systems. Does anyone know if IBM really makes any money on their supercomputer systems ?? I bet not. Great vehicles for doing R&D however.
Intel has been out of supercomputing for sometime. Their TFLOPS machine was the last and it was not a "product". (A one time prototype.) Their last product was the Paragon XP/S.
This is somewhat ironic. The first week in December, I ended 15 years of employment at Intel Corp. One of the major frustrations over the years was that their entire I.T. organization was set up to support someone who runs MS Office and that's it. It was very difficult for engineers (particually software engineers.) to get decent machines. So it seems funny that they will be giving employees new machines for home but can't seem to give their engineers decent systems for their desk at work.
This just doesn't make sense. What about the companies such as SmartHome.com which sell hidden video cameras ? (http://www.smarthome.com/7569.html is a good example.) There is a number of companies which sell these. Are they going to be shut down too ??
You've hit it on the head. This is the real reason for the decline of the supercomputing market. The end of the cold war and the rise in performance of general purpose systems has made the traditional supercomputer obsolete except in some extreme niche markets. I predict that in the future, the only participants in this market will be companies like IBM which can afford to use their supercomputer business as a place to develop new technology but not necessarily make any money. (I spent 6 years in Intel's Supercomputer Systems Division).
Sorry, I didn't realize that it would compact my list down to one line.
Given the apparent endorsement and possibly connections with the DOE (given the quote in the ZDNET article, this is another sign of the end of the supercomputing business. Look at all the roadkill: Thinking Machines KSR Ncube - still exists does video servers I guess. Cray Computer Intel Supercomputer Systems Division (now defunct.) Convex - Bought by HP Cray Research - Bought by SGI Seems IBM is the only really viable player anymore. When I was with Intel SSD, it was obvious that the government was making it really hard to make a profit in that business.
This is all so deja vu. Does anyone know if they are talking about a single system image style cluster (something like the old Locus TCF/TNC) where the cluster just looks like a big system as far as users and apps go ?? In a past life I worked on such as system which supported hundreds of nodes. Instead of Linux, we used the Mach 3.0 microkernel from CMU and a user mode Unix server. Scalability and availability will be big challenges.
Any speculation on how IBM is going to position Sequent's (once the merger is completed.) against their RS/6000 line ?? Seams like the same market ?