Slashdot Mirror

No! As mentioned in several of the replies to that comment, that's not a solid state drive. Google the part number and go to the second page and you'll find the stats, which clearly states it's a 5400 RPM hard drive.

According to this month-old article about Samsung announcing (probably not coincidentally) a 32GB flash hard drive gives a price of it over $900.

Uhm, hello? This isn't a solid state drive. This is a way overpriced 32GB, 5400RPM drive. A 32GB solid state drive would be very, very expensive.

Look it up: http://www-3.ibm.com/pc/support/site.wss/document. do?lndocid=MIGR-44390

You're quite right, the URL for the hard disk at ibm (search: ibm 08K9509 site:ibm.com) is http://www-307.ibm.com/pc/support/site.wss/documen t.do?lndocid=MIGR-44390. It's got 8 heads on 4 platters... SSDs have 0 heads, and 0 platters :)

Java is already fragmented. The result of open sourcing Java will actually be consolidation, i.e. killing of competing VMs. And a huge open source test suite will greatly benefit all surviving JVMs, which is a good thing.

How can you not see this?

Javas problem is not that it might get fragmented, the problem is that it IS fragmented. Do something about it! Let Java free!

I'll bite

Java performance is often better than C when allocating and deallocating memory

Heck, you don't even have to do a real reboot with linux any more to upgrade the kernel. Just kexec into the new kernel http://www-128.ibm.com/developerworks/linux/librar y/l-kexec.html. No cold boot. Not even a warm boot. No POST, no hardware re-initialization.

I seriously doubt the current implementation of Java would fit inside a 400 MHz x86 computer that's designed more to be energy-efficient and small than to be fast. If Java is to run on the OLPC computer, it needs a smaller, simpler implementation.

Check out http://java.sun.com/products/cdc/, it's for smaller computers than their regular J2SE. Here is similar stuff from IBM: http://www-128.ibm.com/developerworks/websphere/zo nes/wireless/weme_eval_runtimes.html.

I run this on a 400 MHz MIPS (Au1500 from AMD) and it works just fine. 400 MHZ x86 should actually be faster and, depending on available memory, may be better off running J2SE.

Basically, here are JVMs for cellphones, server farms and everything in between.

Or get something that has a little more power, like an HP server with iLO (Integrated Lights Out) Advanced, or an IBM server with the Remote Supervisor Adapter II. You can have complete console control of the remote system, including virtual drives.

I've used Thinkpads since 1997 or so. They are well designed tanks. If you do a lot of text input, you will want the joystick mouse control. Touch pads, drive me bats now. Over the years, they have gotten a little less sturdy but they are still very good. My favorite is still a 600 for it's small size and reliability. My current model is a poorly kept T23, which I did not buy from Certified Used. Power management works flawlessly on all models, with some tweaking - usually as simple as turning off ACPI and using APM for sleep.

The only strenuous advice I have is to avoid "desktop replacement" pigs. All computers look "obsolete" in a few years. The small difference in performance between small, cute laptops does not justify the extra weight. You might think it does today, but two or three years from now, when clock speeds have doubled again, you won't. As an extreme example consider two 10 year old laptops, a 560 and a 380 thinkpad. Today, the 560, is still cute but a technically superior 380 is an ugly brick. At the time, the 380 was 50% faster and had twice the memory and a much better screen. The screen is still better, but the fan is loud, the case is huge, the 16MB of RAM is laughable and it's just too heavy. Unless your hands are unusually large, consider an X series.

Avoid high school castoffs and other poorly handled and maintained notebooks. Screws should be replaced every time because they depend on a nylon coating to work. When you take them out, you mess that coat up and things get loose. Really badly maintained models will have missing screws and broken structural parts. They are not reliable and you might have to boot them daily like a Windoze machine. Yes, that's the worst I've ever seen in a Thinkpad. Lesser computers might not boot at all after such bad treatment.

Sony's achievement used helical scan -- IBM's is with linear tape, which is much more commonly used for data storage. Also, Sony used evaporated metal tape. IBM used the economical "dual-coat" tape, which is a newly created formulation of Fuji Photo Film's NANOCUBIC tape.

For more details, and magnetic force microscope images of the bits themselves, see: http://domino.research.ibm.com/comm/pr.nsf/pages/n ews.20060516_magnetic.html

Check out the details, including comparisons to LTO-3, and magnetic force microscope images of the bits themselves here: http://domino.research.ibm.com/comm/pr.nsf/pages/n ews.20060516_magnetic.html

Unfortunatley, the Library of Congress doesn't say give any TB or PB equivalent for their collections. Over the years, I've seen statements of the size of their text only equivalent -- the LoC has lots of images, maps and photos that wouldn't be counted in this metric -- rise from 2 to 10 TB.

If it's cheap, then Netapp might not qualify... :)

What about technologies - NAS? Host-attached? Gateway/NAS? Grids?

Other companies/products to consider:

EMC (The Celerra is a nice product)

Onstor Bobcat

HP

IBM

Hitachi

Panasas

Exanet

Yotta Yotta

StoreAge

If you want basic raid devices look at Infortrend/Transtec. Their S-ATA offerings now support RAID-6 and are dirt cheap.

I agree with most of what you said, but don't understand why the point of adding the bardcode bit at the end was for.

With barcodes, you can scale up your system with a tape storage robot. The barcode is mostly there for error checking to make sure a tape is in the correct spot. When I worked at the PSC, they had two fairly large tape robots. I had to rewrite the barcode generator program, so that it no longer depended on a library with a strict beerware license (No, I did not make this up).

Wonder why no one posted that actual release from the Almaden research site for a better review of what was accomplished. IBM Researchers Set World Record in Magnetic Tape Data Density

Thank you for that link! It contains this one, which I also didn't know about and makes important distinctions.

I don't know if anybody uses Lotus Smart Suite any more, but... shouldn't THAT be the place where they add ODF?

How did this get modded down to 0?

Penguin Power on PS3? Probably. has some useful information.

With 7 SPUs on the PS3 Cell, that's 160-175 single-precision GFLOPS on Linpack. Even on double precision, at 8.5 GFLOPS it's still faster than a P4. Cell Broadband Engine Architecture and its first implementation-A performance view /insert obligatory "imagine a Beowulf..." here

This is going to be used for real work.

That means that every Java app out there, and there are a lot, should be able to run much faster

Compiling has trade-offs. You must target the end environment (CPU, OS), and also try to optimize code (for the target CPU). But you can only do static optimization.

Modern JVMs optimize on the fly. So the more you use a particular path through the code, the more it will be optimized (obviously only so far...). See this article: "The dynamic nature of the Java language provides opportunities for better optimization based on runtime profile information, and this is a significant advantage of a Java dynamic compiler over a traditional static compiler."

So having a compiled executable may not yield faster run times. It may have faster load times, which is where most of the perception of slowness comes from.

I use a rather large Java IDE (Eclipse) for development. It takes 10-15 seconds to load up the IDE from a cold start (2.1 GHz laptop). After that it is just as fast as I can type, which includes on-the-fly error/syntax checking, code assist, and so on.

...than a Dell Optiplex GX520 does today.

So, what exactly is the point?

In 1962 an IBM 7094 cost $3,134,500.

Does that mean that $19,356,198.10 is a reasonable price to pay for a Dell Optiplex GX520 today?

Ahem.

"1992 - ThinkPad 700C
IBM launched the first edition of its ThinkPad series, the ThinkPad 700C. Right from the start IBM set the standard: the first notebook with a 10.4" color TFT screen and a new TrackPoint pointing device. This notebook weighed 5.7 lb And had a memory of 120 MB--respectable for the era. "

http://www.pc.ibm.com/us/thinkpad/anniversary/hist ory.html#1992

And the color of the case? Darth vader black.

Leave it to Apple to set the trends again. I bet all the other companies are gonna copy them and come out with black laptops now... ; )

Another example of the influence of the original PowerBook on the rest of the laptop industry is the placement of the keyboard and pointing devices: prior to the release of the PowerBook pointing devices were either non-existant on PC laptops (purchased separately and hung off the side of the laptop by a hook or bracket) or were situated above the keyboard or on the display and keyboards were placed in a "key-forward" position without a palm or wrist rest. After the release of the powerbook, however, almost all laptops placed pointing devices below the keyboard or used the IBM TrackPoint eraser-head mechanism in the middle of the keyboard itself and moved the keyboard back to allow the user's wrists to rest on the blank space at the front of the laptop case.

I would suggest you look at the Cell BE SDK from IBM. Cell isn't completely impossible to code for. From what I understand, you just write a small app, you get your arguments from main(), and there are a few functions to interface back with the main program. There are a few caviots about instructions, but for the most part it's just like writing in C.

I'm into Bioinformatics, and I would love to see if I could get each of the individual SPE doing dynamic programming on short sequences. A database search could be sped up 8 fold (or 7 fold on the PS3).

Hopefully they'll open it up for the genius's of the world to take a stab at coding in the cell environment

I'm no coder, but perhaps you meant something like this ?

But you don't have to be a Sci-Fi author to imagine crazy abuses of this data.

Windows CE may be an attractive option for you. You'll have the quintessential Windows-like interface, a ridiculously low memory interface (~400KB), and the OS itself is built into ROM - restoring the OS can be as simple as flipping a switch. Also, Windows CE is (for the most part) invulerable to viruses that plague traditional Win32 systems.

Lenovo/Neoware Thin Clients for example (not affiliated, just found them through a google search) low-end models cost >$400ea., have a 400MHz VIA processor, 64MB of RAM, 2 USB2.0 Ports, and no fans. I'm not a systems administrator, but these systems sound ideal for poorly-ventilated/cooled areas (face it, universities are full of them) where students can work on documents from their thumbdrives and browse the internet without risking network or system integrity.

As with hundreds of foreign-owned companies that did business in Germany at that time, Dehomag came under the control of Nazi authorities prior to and during World War II.

The same is true of, say, Ford. Did Ford also have a role in the Holocaust because its (seized by the Nazis) German subsidiary produced vehicles which probably transported SS officers to/from death camps? I'd say not, but perhaps you have another opinion.

I understand your point, having come from a mainframe background

Or, even better, googled "SL8500" and checked the datasheets. IBM specs the z9 at over 1Tb/s. You think customers who'll pay for that are going to let the library use 2,048 tape drives to play keepaway with their petabytes?

If you're a Linux programmer (or Java programmer, for that matter) you're already a mainframe programmer.

IBM has a program called the Academic Initiative, which provides free of charge, course material, hands-on systems and software, to colleges and universities who want to start teaching their students about mainframes. There was also a Mainframe Programming Contest that ran just this last year, where over 700 students from the US and Canada logged into a system and performed various tasks to win prizes. The top five performers were flown out to IBM Poughkeepsie, NY for a tour of the facilities and to meet with some of the higher-ups.

Mainframes aren't going away. There's simply nothing else out there that can handle the amount of data that a mainframe can process. That's simply the way it was built. It doesn't matter if someone comes out with a 25ghz core cpu with fiber interconnects. That doesn't drive business. IBM's mainframes are built for business. Read up on GDPS, Parallel Sysplex, WLM and CICS. Comparing a "omg fast 15ghz dell server" or whatever to a mainframe is like comparing a Ferarri to a freight train. Sure, the Ferarri will smoke the train in the quarter mile, but there's no way you'd want to use one to move coal from one side of the country to the other.

If you're interested in learning the stuff, just ask around a company that uses them. While it is an entirely different world, it is fun to learn, and people are really trying to hire young mainframe system programmers now.

z/OS hasn't changed in 30 years? I beg to differ

According to this, Sun's next JVM (1.6 'Mustang') performs escape analysis. For example, if it can see that a StringBuffer is only used locally in one thread and never 'escapes' to the heap, then it knows no synchronisation will ever be needed, and it can bypass all that, effectively reducing it to a StringBuilder anyway. Mustang can also change lock granularity, move stuff from the heap to the stack, and make many other optimisations.

Tony Hoare said (and Donald Knuth quoted) that premature optimisation is the root of all evil; 'premature' is starting to look more and more like 'by the programmer', as decent compilers and runtimes start to do a better job of it than we can.

Sera

indeed. How much for the book? IBM's giving the information away for free: http://www-128.ibm.com/developerworks/java/library /wa-ajaxintro1.html

IBM Workplace Documents runs on Windows and supports ODF - uses it as its native format in fact.

Take a look at java.util.concurrent (as another response has suggested). Specifically, look at:

"Concurrent Programming with J2SE 5.0"
http://java.sun.com/developer/technicalArticles/J2 SE/concurrency/index.html

and

"Concurrency Utilities Overview"
http://java.sun.com/j2se/1.5.0/docs/guide/concurre ncy/overview.html

There's much more to concurrency in Java (especially Java 5) than simple target-object synchronization. (For example, you can set up queues to pass safely from one thread to another.)

You might also want to look at some other frameworks that have been created in Java, such as CSP, covered in a series of DeveloperWorks articles beginning with:

"CSP for Java programmers, Part 1"
http://www-128.ibm.com/developerworks/java/library /j-csp1.html

There's a lot out there to let you think in much higher-level concepts than simple locks (although you can do that to, for the learning exercise ;-).

Talks about the gigalapse and CmdrTaco! Bunch of seemingly random stuff tied together pretty well

Talks about the gigalapse and CmdrTaco! Bunch of seemingly random stuff tied together pretty well

Well, yeah, you could use Java I suppose. But it's still using a shared-state concurrency model, which is inherently complex no matter what you do. Instead of that, you could use Erlang, a functional language using message-passing concurrency, and make all that complexity go away. Ericsson uses it to run telephone switches. Someone else wrote a first-person shooter with it. It can handle thousands of lightweight threads at once, and does distributed apps transparently too. For an intro to Erlang from a Java guy's perspective, check out this article by Bruce Tate.

Again, for quick and dirty jobs you use scripting languages. For massive projects, you use Java or .NET.

BTW, you do realise that Java is on almost all occations faster than any scripting language, and using JIT compiling you can increase your Java applications speed far beyond any scripting language.

You're right! It would be better to get a technology designed for multi-threaded programming. But then you go on to push Java? I'm confused.

Now, granted, Java does provide builtin support for concurrency. Although that support is an implementation of the 30-year old monitor concurrency primitive, which even one of its principal inventors (Tony Hoare) long ago abandoned in favor of better things. Not to mention that Java's implementation of monitors is known to be broken (or at least was - maybe they've fixed it now?).

Doug Lea's util.concurrent package (now part of Java 1.5) certainly helps things on the Java front, but it's hardly a panacaea. Depending on the application, you may want to take a look at JCSP instead - it provides for an almost direct translation of ideas from Tony Hoare's CSP into Java. For languages with good concurrency support out of the box, you might try Erlang, Ada, Mozart/Oz, E, or Eiffel.

http://www.usenix.org/publications/library/proceed ings/javavm02/full_papers/jacob/jacob_html/node12. html

http://www.shudo.net/jit/perf/

IBM's VM looks very favorable in these tests. You can get it for Linux (also available for windows and other platforms) here:

http://www-128.ibm.com/developerworks/java/jdk/lin ux/download.html

Enjoy!

Taft

I don't think he meant to say that Eclipse helps with the exception "problem" (which really isn't a big problem: good Java programs use exceptions only when necessary). For that he referred to other programming languages.

It would be spiffy-keen to have a decent JVM on my PPC box running linux...

Now one might explain why IBM suddenly is selling these systems with very similar specs to these. Yes, the 185 is a bit neutered (memory, undocumented AIX only 3d graphics, PCI-X versus PCI-E), but they'd make for a nice system to use recycled 970's.

Yes - the ability to take corruption into account is what differs mainframes (and also high-end IBM UNIX servers like p595) from PCs.

I'm going to go out on a limb here, but I have a distinct feeling that Apple moved more units of PPC Macs than IBM did of RS/6000s. Those things aren't exactly a dime a dozen; you can't just walk down to CompUSA and pick one up to put under the Christmas tree.

I can't find any hard sales figures recently, but back in 2000, at the height of the Bubble, IBM was touting that it had sold "over 1,000,000 RS/6000" servers total to date, while Apple shipped more machines than that (1.12M) in the fourth quarter of 2000 alone.

I've been told on good authority that the biggest users of Power chips have always been embedded systems, and that this is where IBM was putting the majority of its development focus (particularly on lowering power consumption and making them run cooler).

Specifically related to the issues I mentioned: If you are interested in some of the challenges around flatness, you can learn more about dummy fill that must be added to metal layers, by looking at the layman's version or a technical description.

With regard to reflection, you can check out a rather old background article or how anti-reflection layers must be used in modern semiconductor manufacturing to reduce problems.

More specific articles on 3D fabrication can probably be found in recent journals (most likely not available online), or if you're not concerned about reading patents, by reading patents from the USPTO (for reasons of US law which you're probably familiar with, I'm not going to search that and provide you any links). There may also be more by searching for Matrix Semiconductor (which I didn't realize at the time of my first posting has been acquired by SanDisk).

Having said that, there is also 3D packaging, which takes various forms. Semiconductor Cubing (as it's apparently called) can stack lots of semiconductor devices, but note that these are originally fabricated as single layer chips and then they are bonded together to form a larger block.

More recently (and in real production), 3D packaging is being performed through a System in Package (SiP) methodology (you may also see this referred to as a 'chip stack' technology). This is distinct from a multi-chip module (MCM), where the chips are aligned horizontally on the packaging substrate. Today, a SiP is generally a memory module bonded upside down onto a non-memory device (though it can also be used to bond an RF device onto a non-RF device). This form of packaging is receiving attention from SEMATECH as well. Further information from SEMI is also available if you Google for "SEMI Forum: Mapping progress in 3D IC integration".

Beyond that, it's again hard, due to the password protected nature of conference materials and journals... but hopefully that's a good set of links to explore.

Specifically related to the issues I mentioned: If you are interested in some of the challenges around flatness, you can learn more about dummy fill that must be added to metal layers, by looking at the layman's version or a technical description.

With regard to reflection, you can check out a rather old background article or how anti-reflection layers must be used in modern semiconductor manufacturing to reduce problems.

More specific articles on 3D fabrication can probably be found in recent journals (most likely not available online), or if you're not concerned about reading patents, by reading patents from the USPTO (for reasons of US law which you're probably familiar with, I'm not going to search that and provide you any links). There may also be more by searching for Matrix Semiconductor (which I didn't realize at the time of my first posting has been acquired by SanDisk).

Having said that, there is also 3D packaging, which takes various forms. Semiconductor Cubing (as it's apparently called) can stack lots of semiconductor devices, but note that these are originally fabricated as single layer chips and then they are bonded together to form a larger block.

More recently (and in real production), 3D packaging is being performed through a System in Package (SiP) methodology (you may also see this referred to as a 'chip stack' technology). This is distinct from a multi-chip module (MCM), where the chips are aligned horizontally on the packaging substrate. Today, a SiP is generally a memory module bonded upside down onto a non-memory device (though it can also be used to bond an RF device onto a non-RF device). This form of packaging is receiving attention from SEMATECH as well. Further information from SEMI is also available if you Google for "SEMI Forum: Mapping progress in 3D IC integration".

Beyond that, it's again hard, due to the password protected nature of conference materials and journals... but hopefully that's a good set of links to explore.

IBM carefully limits the use of its logos. No other company may use IBM logos unless it has the express written permission of IBM, or is licensed by IBM to do so.

To obtain permission to use any IBM logo, contact your IBM representative or the IBM Call Center at 1-800-IBM4YOU (1-800-426-4968) and ask for Corporate Branding.

Here is the link for 3G laptops with EVDO (faster than GPRS/EDGE): http://www-131.ibm.com/webapp/wcs/stores/servlet/P roductDisplay?productId=4611686018425151954&storeI d=10000001&langId=-1&categoryId=2049168&dualCurrId =1000073&catalogId=-840