Slashdot Mirror

This is typical of IBM; they have always gravitated towards the docking station instead of building bulky devices on board. They expect the user to dock the device for charging and stationary usage, and mobilize it when making rounds, or whatever. http://www-131.ibm.com/webapp/wcs/stores/servlet/P roductDisplay?catalogId=-840&langId=-1&partNumber= 250610U&storeId=10000001 This is the docking station for the X41, including a cd-rw/dvd-rom drive. Perhaps you guys should investigate available accessories before beating on IBM's choice of how to design a tablet pc ;-).

Also, to all you guys beating on the IBM/Lenovo thing- IBM designed this thing. Lenovo was simply in charge when it made it to market.

Peace!

Your argument is effectively countered by this excellent article on why java memory management is actually faster than the way it is handled in C/C++: http://www-128.ibm.com/developerworks/java/library /j-jtp09275.html?ca=dgr-lnxw07JavaUrbanLegends

In short, memory allocation in Java is faster (measured in cpu cycles used) and it can use both stack and heap based memory allocation (unlike the urban myth that Java can't do that). Additionally, for short lived objects, garbage collection is essentially free (unlike calling free() in C++). Of course you can do clever stuff in C like not actually using malloc and free but most programmers default to not doing except for some very specific cases. In other words, most of the time Java is actually faster for both allocation and deallocation of memory.

People like you run into trouble when they manually call the garbage collector (bad idea, can cause severe performance issues) or when they misconfigure the garbage collector (e.g. if you are running a server with multiple cpus and multiple gigabytes of memory). Also a big issue is C programmers trying to program Java like it is C and making wrong assumptions about the cost of certain code and the impact of certain optimizations (e.g. pooling is actually bad for performance in some cases).

The only real problem that Java has with respect to memory management is that java programmers tend to use lots of memory when writing their software. You don't have to do that but using the default APIs and recommended practices in e.g. the java tutorial and many books on this matter you will probably end up using substantial amounts of memory. Many Java programs continually create and destroy large amounts of small objects. If you'd do that in C/C++ you would really notice this. In Java you hardly notice this at all because it allocates and deallocates in a much smarter way.

I don't know what 'personal tests' you have been performing. Up front I'd say you may have been measuring the wrong things the wrong way and have been drawing the wrong conclusions based on what your results. The JVM is a highly complex piece of software and very difficult to benchmark properly. Doing so requires that you know

Actualy memory allocation/dealocation can be faster in java that in C.
Also, memory fragmentation is a non issue with modern GC wich is not to be said about C.

read this : http://www-128.ibm.com/developerworks/java/library /j-jtp09275.html?ca=dgr-jw22JavaUrbanLegends

Wait, wait, wait. Your telling me in Soviet Russia computers' secure humans while in America humans' secure computers?

Shit, those Soviet Russians must have really smart computers to be able to do this crazy stuff.

While you Americans must have really dumb computers that even really dumb humans' can secure.

X10 Halloween Webcam is online - is it a "Trick" or a "Treat" ... ;-)

Yeah, I liked the keyboard + trackpoint + trackpad on my IBM T41 so much I went out and bought one of these for my regular desktop PC.

It is just like the T41 keyboard only a little bigger, with a numeric keypad added on. I tend to use the trackpoint most since it lets me keep my fingers near the home keys but it has a trackpad too. There is also a wee model without the numpad.

- flip

See Volume 49, Number 2/3, 2005.

ODF Myths

Some of the replies below comment on the operations-per-second estimates of the brain, so I won't. There is, however, a really interesting project at the Brain Mind Institute in Switzerland to simulate, in large detail, a full neocortical column of the brain on several BG/L racks, donated by IBM for this cause. As a consequence, this project is called the "Blue Brain Project", and you can read a bit more about it here:

IBM
http://domino.research.ibm.com/comm/pr.nsf/pages/n ews.20050606_CognitiveIntelligence.html

New Scientist
http://www.newscientist.com/article.ns?id=dn7470

Main site (which doesn't seem to be working at this moment!)
http://bluebrainproject.epfl.ch/

And, finally, from the article:

"We are stronger than ever because we have a research lab in Cambridge, we have one now in China, one in India and that is where the top problems in computer science are going to be solved."

I'm not sure what Mr. Gates is implying here. But if I were on one of the U.S. campuses, I'd be pissed, and a little nervous.

And, finally, from the article:

"We are stronger than ever because we have a research lab in Cambridge, we have one now in China, one in India and that is where the top problems in computer science are going to be solved."

I'm not sure what Mr. Gates is implying here. But if I were on one of the U.S. campuses, I'd be pissed, and a little nervous.

And, finally, from the article:

"We are stronger than ever because we have a research lab in Cambridge, we have one now in China, one in India and that is where the top problems in computer science are going to be solved."

I'm not sure what Mr. Gates is implying here. But if I were on one of the U.S. campuses, I'd be pissed, and a little nervous.

IBM offers similar services with their Deep Computing Capacity On Demand Centers. They've publically announced customers.

http://www-03.ibm.com/servers/deepcomputing/cod.ht ml

http://www-03.ibm.com/servers/deepcomputing/succes s.html

They offer x86, Power4/5, and Blue Gene clusters. They have multiple centers (Poughkeepsie, Houston, Montpellier), which customers can access remotely.

IBM offers similar services with their Deep Computing Capacity On Demand Centers. They've publically announced customers.

http://www-03.ibm.com/servers/deepcomputing/cod.ht ml

http://www-03.ibm.com/servers/deepcomputing/succes s.html

They offer x86, Power4/5, and Blue Gene clusters. They have multiple centers (Poughkeepsie, Houston, Montpellier), which customers can access remotely.

The art of metaprogramming, Part 1: Introduction to metaprogramming
Write programs to generate other programs

One backs up a Petabyte with something like this...

http://www-132.ibm.com/webapp/wcs/stores/servlet/C ategoryDisplay?storeId=1&catalogId=-840&langId=-1& dualCurrId=73&categoryId=2058932&x=6&y=11

Priced anywhere from $150k to $275k depending on the drive types and configuration. We have a total of 70 or so of these frames as they are referred to, spread out in 3 datacenters in delaware alone. But then again, while our budget is small... its an enterprise level system, so it gets an enterprise level budget. A weekend's worth of full backups exceeds 50 TB at my site alone.

It sounds like your requirements are good (scalable) , fast (or large in this case) and cheap. I'm pretty sure that only 2 out of 3 are possible.

How important is the data? More importantly how much will it cost if you lose the data? Without redundancy if you lose one disk you are likely to lose the whole volume... and cheap disks WILL die. If your budget won't allow you to purchase a real enterprise storage system will it allow you to buy an adequate backup solution?

Check this out it might be your best bet http://www-03.ibm.com/servers/storage/disk/ds4000/

I just sat through a demo of one of these units. I was impressed with their goodness, storage capabilities and the ability to use SATA drives brings the cost for large amounts storage cheaper than your typical SCSI enterprise storage solution. These boxes are also fully RAID capable... but you still need a backup... but everyone who reads /. knows that :)

Kailash Nadh writes to tell us ABC News is reporting that IBM is teaming up with several other companies to form a group called Aperi. This group will attempt to "push the open source idea deeper into computing" and "free up the bottlenecks that can occur when a business has bought tape and disk storage systems from a variety of vendors." The partnership is to include companies like Cisco, Sun, Fujitsu, and several others.

If I'm not mistaken, GPFS is free as in beer for Linux. Download it and give it a spin...

GPFS download page...

GPFS
Take it from someone who's messed with nearly every storage product on the market, if you want something that works fairly simply, performs at approaching spindle speed ( meaning the file system is not the bottleneck - if you have 10 GB/sec. storage bandwidth, expect to see near that with proper tuning ), is very stable ( compared to most storage solutions on the market - bear in mind that most storage products are aimed at large-block sequential I/O, and fall down - either performance-wise or stability-wise - when you throw other I/O patterns or combinations of patterns at them ), and is portable across nearly any Linux distribution ( with varying amounts of difficulty, I have had to hack their kernel patches before when using a unsupported kernel ), GPFS is the one. Of course, the problem there is I believe it's pretty expensive to run on non-IBM hardware. But if you have IBM hardware ( even if it's not the hardware you're running the FS on ) or some sort of in with IBM, they'll let you have it for a song and a dance.

Having said that, Lustre is getting there. I'd say it's the equal of GPFS ( as a parallel filesystem - I believe it is even more flexible as a distributed filesystem ) in performance, probably scales roughly the same ( haven't played with it in a large installation, so can't tell you beyond looking at the architecture ), and is going to the be the biggest player on the market in the future. It's also free ( IIRC Cluster File Systems sells support, but the code is freely available ) and not tied to IBM and whatnot, like GPFS is. Of course, HP has a big connection with Lustre, but not ownership thereof.

Those are really the only two that I would consider for a serious high-performance storage project. If you don't need great performance, that's when you can start looking at things like GFS, ADIC's StorNext, Ibrix, etc.

Oh, Gautham Sastri ( of former Maximum Throughput fame ) has a newer company called Terrascale, I recall them putting on a presentation at the 2003 or 2004 ( can't remember ) Supercomputing conference ( SC2005 is coming up in a few weeks, yeah!!! ) which showed pretty good performance ( relative to the small system they were using ), not sure how they're coming along...

Anyways, good luck...and don't forget to use Iozone to benchmark the damn thing!

Transarc was acquired by IBM in 1998, and released OpenAFS in 2000. This used to be IBM's site for Transarc technologies, but it looks like it doesn't exist anymore, and instead just redirects to IBM's software page.

Transarc was acquired by IBM in 1998, and released OpenAFS in 2000. This used to be IBM's site for Transarc technologies, but it looks like it doesn't exist anymore, and instead just redirects to IBM's software page.

May or may not be what you search. Quite expensive but impressive featurelist.

http://www-03.ibm.com/servers/eserver/clusters/sof tware/gpfs.html

Neither PowerPC nor POWER is a subset of the other (check the IBM docs). They do, however, have a common subset. Most POWER code only uses this subset, and GCC can be configured to emit code which only uses this subset. This code will run without modification on both POWER and PowerPC chips. The later POWER-branded chips (POWER4 and later) actually use the PowerPC instruction set, not the POWER instruction set.

AIX runs on Power, not PowerPC

AIX runs on both POWER and PowerPC branded CPUs. IBM even sells PowerPC branded systems that run AIX. The execution core of the PowerPC 970 (G5) and the POWER4 is the same - the only difference is the cache logic and size, and the presence of the vector unit on the 970 (and the fact that the POWER4 series chips were dual-core before the 970 was released).

How can AIX trap an instruction that is already supported by it's processor?

Some legacy code designed for POWER3 and earlier uses the POWER instruction set, not the PowerPC instruction set used by later POWER CPUs. Most of the time, this is not important since the majority (or even the entirety) of the code will contain instructions in the subset supported by both. On the occasions when the CPU encounters a POWER instruction not present in PowerPC, it will call AIX's illegal instruction trap handler and replace the missing instruction with a small sequence that will have the same effect.

See http://www-128.ibm.com/developerworks/views/openso urce/projects.jsp for a list of some of the projects they're actively involved with. This list doesn't seem to include projects that they have released to the community, such as Cloudscape.

The old water-cooled systems that sprawled all over the place and took several very large cabinets have been replaced over the past decade or so with CMOS boxes roughly the size of a fridge or three.

You still might need several cabinets depending on the type of peripherals attached (and automated tape silos are not small), but while the hardware is still related in terms of basic architecture, instruction set, and operating system, basic mainframe technology has come a long way in the past 10-20 years.

The Unisys Clearpath mainframes I've worked on use either MCP (for those descended from the Burroughs A-series line) or OS2200 (for those descended from the Sperry 1100-series), and IBM mainframes typically run Z/OS (descended from OS/390). You can get some idea of the size of these boxes by reading the spec sheets for some of them

Cisco's Internetworking Technology Handbook is a bit dated but a great base resource downloadable in pdf.

Pair the above with IBM's TCP/IP Tutorial and Technical Overview, and round things off by downloading Bable: A Glossary of Computer Oriented Abbreviations and Acronyms since you'll be in acrynom hell.

Probably few /.ers need the above but they've given me a good overview and reference.

For What it's Worth :)

(I'd really like to find a way to surf the 'net and sew at the same time) ZK- http://www-306.ibm.com/software/voice/viavoice/ http://www.footmouse.com/

Just because there's a version of NT that can run on a 128-CPU SMP box doesn't mean it scales that in real-life situations.

If it did, with all Microsoft's billions of dollars, how come there's no NT equivalent to this for Linux, or this for Solaris?

Those two bad boys scale damn near linearly. I know that, I don't have assume that. I can afford a 7-figure house because I can make those things sing. That Sunfire E25K has 72 CPU slots, and each UltraSPARC-IV chip has 2 full CPUs on each die. The IBM 595 has 64 CPU slots, and when I was at SC-04 in Pittsburgh last year, IBM claimed they were working on an 8-way version of their Power CPU. That's 512 CPUs on an SMP box.

There's nothing like that in the NT world that anyone could buy. And you don't have to sign some NDA that would keep you from getting a job in a lot of places to see the source code for either OS.

Keep your damn toy OS, and your self-admitted assumption that "NT knows how to handle more than 2 processors", because there's no commercially-available system to support that assumption.

Well, I guess this article might be about you then...

"Solar flares dump huge amounts of energy into the Earth's magnetic field. The more engergy, the further south the Aurora Borealis appears, as the magnetic field pulls particles from the solar wind to larger circles further away from the poles."

Not exactly. What happens is that with the increased solar wind pressure the magnetosphere becomes distorted, the auroral oval expands equatorwards, and electric currents start flowing. "Magnetic reconnection" events in the magnetotail energize magnetospheric (not solar wind) particles and inject them into the auroral zones resulting in the visible aurora.

Whether or not astronauts are exposed to auroral zone radiation depends on the location of the auroral oval (it expands equatorwards during high geomagnetic activity induced by solar disturbances) and the inclination of the spacecraft orbit. I have satellite (X-ray) images of the auroral oval as far south as Washington DC during one large storm.

For a view of the current "space weather" conditions, see:

http://sohowww.nascom.nasa.gov/spaceweather/

The ACE satellite sits upstream of earth in the solar wind and monitors the solar wind ram pressure and the interplanetary magnetic field Bz component. The NOAA/POES satellite shows images of the current auroral oval (now in its quiet phase).

An excellent article on galactic and solar cosmic rays is here:

http://www.research.ibm.com/journal/rd/421/ziegler .html

If you're wondering why companies like IBM are interested in this, it's because cosmic rays penetrate semiconductor chips and cause "single event upsets" in computers. Or they can knock out an entire satellite. Computer chips need to be "hardened" against radiation before they are space-qualified.

In case you're wondering, I have a PhD in space science lying around here somewhere .....

It may be connected with with IBM Hursley, their UK laboratories in Winchester, England, which did some of the original development of disk drives. IBM has a review and discussionof disk 'file' innovation in the 25 years up to 1981 which describes the Winchester technology in some detail, but doesn't seem to identify where the work was carried out.

It may be connected with with IBM Hursley, their UK laboratories in Winchester, England, which did some of the original development of disk drives. IBM has a review and discussionof disk 'file' innovation in the 25 years up to 1981 which describes the Winchester technology in some detail, but doesn't seem to identify where the work was carried out.

It may be connected with with IBM Hursley, their UK laboratories in Winchester, England, which did some of the original development of disk drives. IBM has a review and discussionof disk 'file' innovation in the 25 years up to 1981 which describes the Winchester technology in some detail, but doesn't seem to identify where the work was carried out.

Oh look, another amazing 4 line code example doing oh so real things to PROVE java is really slow! And on slashdot no less! You're fucking brilliant. Funny enough, IBM had something to say just recently about java's speed too, and it seems they took a slightly different conclusion: http://www-128.ibm.com/developerworks/java/library /j-jtp09275.html?ca=dgr-lnxw01JavaUrbanLegends

And by the way, all those phone games, in that billion dollar industry of phone games, what do you think they run on? Thats right, java. And do you see consumers throwing their phones in disgust of the slow speed? No? Gee.

> I was always under the impression that David Patterson at
> Berkeley and John Hennessy (now President of Stanford) invented
> the RISC architecture and then took it to Sun? The Patterson bio
> linked to above seems to indicate that he did invent the
> RISC architecture. Huh.

Nope. The IBM 801 project began in 1975, and I'm fairly sure they had a machine up and running 2 or 3 years later, perhaps sooner.

The Stanford work on MIPS didn't begin until 1981. I was in John's group at Stanford at that time, though not working on RISC, and I distinctly remember that among the factors that led to the university work on RISC was early information on the 801 that started to come out of IBM. I believe that the Berkeley work was roughly contemporary with the Stanford project, though perhaps a bit ahead. Dave Paterson's bio claims that RISC I was the first VLSI RISC, and I suspect that's true. Hard as it may be to believe now, the IBM 801 was built at a time when even a simple CPU took many chips. I recall the actual box being perhaps 2-3 feet long, and maybe 1.5 feet high.

In any case, the IBM 801 work clearly came years earlier than either the Stanford or Bekerely projects, and I think John H. and Dave P. would be the first to acknowledge the seminal work of John Cocke and the IBM 801 team. My impression is that the respect was mutual, and that all involved agreed that both the Standford and Berkeley teams made very important later contributions.

The history of Fortran is quite interesting. My understanding is that John Backus and the team who built Fortran were so worried that assembler programmers wouldn't trust a compiler to generate code fast enough for the slow machines of the day that they implemented a slew of optimizations that were still viewed as aggressive 10-15 years later. Keep in mind that the compiler itself had to run on these slow machines, with limited memory (tens of KBytes), and mostly punch cards for storing object code, math libraries, etc. By the way, I met Backus once or twice in the late 1970's when I was a very junior member of the programming staff at IBM. He was already something of a legend in the languages community, and I've never met anyone in the field who was kinder, more down to earth, or more interested in having a chat with anyone, regardless of how old or young. The field needs more people like him.

... in Yorktown Heights, NY; Hawthorne, NY; San Jose, CA; Austin, TX; Beijing; Haifa; Tokyo; Zurich; Delhi and Bangalore.

http://www.research.ibm.com/worldwide/

Oh, yes, let's not forget the patents from the product development teams in NY, NC, FL, MA, MN, TX, CA, Ontario, Japan, China, India, Ireland, UK, France, Germany ...

And the patents from the services organization worldwide.

The U.S. doesn't have a monopoly on smart people with good ideas and the ability to work on technology. Research and development with a global reach is a very good thing.

IBM didn't create the economic and political conditions that make outsourcing possible, and IBM can't turn back the clock. Governments probably can't turn back the clock, either.

I was always under the impression that David Patterson at Berkeley and John Hennessy (now President of Stanford) invented the RISC architecture and then took it to Sun? The Patterson bio linked to above seems to indicate that he did invent the RISC architecture. Huh.

While SPARC may have been the first VLSI based RISC architecture, I think the IBM 801 architecture may have preceded it. John Cocke at IBM was a seminal thinker in the area and may have developed the RISC concept and was awarded a Turing award for this work, so he might have a claim for the innovation.

You hardly ever hear about their teleportation research.

What kind of power draw do the blade chassis have? What blades? What version of Red Hat?!?!?!

From my experience, a blade chassis runs with 4 1875 Watt power units (they have more powerful models). Each of the 14 blades is budgeted a maximum of around 210 to 260 watts (dependent on model). Given the CPU speed described, I would imagine that is the 2CPU xeon model, ie the HS20

If they are running IBM cluster managment tools, like CSM (which would be very probable given that it has a module designed to interface directly with the blade chassis management system), that would mean the are limited to the CSM support list.

What kind of power draw do the blade chassis have? What blades? What version of Red Hat?!?!?!

From my experience, a blade chassis runs with 4 1875 Watt power units (they have more powerful models). Each of the 14 blades is budgeted a maximum of around 210 to 260 watts (dependent on model). Given the CPU speed described, I would imagine that is the 2CPU xeon model, ie the HS20

If they are running IBM cluster managment tools, like CSM (which would be very probable given that it has a module designed to interface directly with the blade chassis management system), that would mean the are limited to the CSM support list.

That doesn't work for protocols that don't use URLs, whereas the current system does.

That doesn't work for protocols that don't use URLs, whereas the current system does.

It would be easier to say:

http://www.ibm.com/

to get to the German ibm.com site if you're in .de land, whereas if a US-dweller wanted to get to a German IBM site, he'd say

http://www.ibm.com.de/

You leave off the country identifier to get to sites inside your country, but add it when going international.

That could be extended in a natural way by saying anyone inside, e.g., the ibm.com.us domain only need refer to "http://www" to get to http://www.ibm.com.us/". In other words, the parts of the URL that match your domain need not be supplied.

Or you could just use the system that's currently in place, where http://www.ibm.com/ and http://www.ibm.de/ are different...

de.http://www.ibm.com/ not being the same as us.http://www.ibm.com/

That would be a giant PITA and would be like having the entire World use the NANP system of phone numbers yet each nation assigns them locally -- so 212 in the US would be NYC but not in the UK. It would probably also violate a ton of treaties related to trademarks and copyrights if anybody but IBM controlled that domain in Germany (per your example).

My suggestion (it will never happen) to solve this "problem" (I don't think there is one but the rest of the World seems to...) would be to get rid of the TLDs like .com, .net, .org, etc. and go back to the country code TLDs. Let every nation set their own policy for how they work. Then you could have .com.us, .com.uk, etc.

I've advocated this for years but there is no way in hell it would happen because of the saturation of .com.

de.http://www.ibm.com/ not being the same as us.http://www.ibm.com/

That would be a giant PITA and would be like having the entire World use the NANP system of phone numbers yet each nation assigns them locally -- so 212 in the US would be NYC but not in the UK. It would probably also violate a ton of treaties related to trademarks and copyrights if anybody but IBM controlled that domain in Germany (per your example).

My suggestion (it will never happen) to solve this "problem" (I don't think there is one but the rest of the World seems to...) would be to get rid of the TLDs like .com, .net, .org, etc. and go back to the country code TLDs. Let every nation set their own policy for how they work. Then you could have .com.us, .com.uk, etc.

I've advocated this for years but there is no way in hell it would happen because of the saturation of .com.

Actually, I think the per country root node solution is the right technical solution. The internet and its protocols should learn to deal with this since it is bound to happen eventually.

It would be annoying, but we probably need to allow for something like this:

de.http://www.ibm.com/ not being the same as us.http://www.ibm.com/

Actually, I think the per country root node solution is the right technical solution. The internet and its protocols should learn to deal with this since it is bound to happen eventually.

It would be annoying, but we probably need to allow for something like this:

de.http://www.ibm.com/ not being the same as us.http://www.ibm.com/