Slashdot is powered by your submissions, so send in your scoop


Forgot your password?

Comment Lack of a use case (Score 3, Informative) 128

From the article:

Programs like "Mail" or "Messages" could be implemented in reprogrammable silicon.

You need how much compute power to read mail?

Most users just don't need that much power. Once everybody could play streaming HDTV, the couch potato market was covered. Rendering in gaming could still improve, and NPC behavior could get smarter, but really, GTA V pretty much has that nailed and it runs on last-generation consoles.

There are people who need more power, but they're running fluid dynamics simulations or rendering movies or simulating new ICs or something like that. I've run Autodesk Inventor on 24-CPU workstations. That's one of the few interactive programs that can usefully use a 24-CPU workstation. It's not a mass market product.

The applications that need vast amounts of additional compute power are there, but they're not high-volume applications. Nor are they "enthusiast" applications. There's not enough volume there to justify heavy investment in faster CPUs.

This may change as we have better robots or something like that. But speeding up existing desktop apps, no. (Program load times are still ridiculous long, but mostly because of stupidity like phoning home for updates, waiting for the license server, fetching ads, or using virtual memory in a world where memory is cheap.)

Comment Re:No point pussy-footing around (Score 4, Interesting) 128

An interesting scenario just came to mind...

1) RSA intentionally weakens their crypto at the behest of the NSA (this is fairly certain)
2) Chinese hack RSA - the only question is just how thoroughly (a known fact)

Now comes the speculation.

3) China analyzes what they got from RSA and discover the crypto is weaker than expected.
4) Quietly, China also begins to take advantage of this breakable crypto the NSA foisted on US companies and citizens.
5) China deduces why it was done and starts looking for weaknesses in other US crypto products - possibly succeeding, given they have a decent idea what to look for.

Followed by

6) China successfully and quietly penetrates most US defense contractors and financial institutions.

Comment Re:I'll tell you what it means ... (Score 3, Informative) 286

"The NSA/CSS Memorial Wall lists the names of 171 cryptologists who have died in the line of duty since the Agency's inception in 1952," according to the letter.

This refers to members of the US military doing cryptographic duty who died in the line of duty. Here's the list. Most died during the Cold War or in Vietnam. In recent years, in Afghanistan or Iraq. Only one civilian, Alan M. Blue, who was on the USS Liberty when the Israelis attacked it.

Comment Re:old news (Score 1) 110

GI in Japan after surrender of Japan in WWII, picked up gut bacteria and whenever he ate carbs he got drunk.

That may be the one my parents told me about, back in the '60s or so (but as a war story which probably puts it in WWII.

In the one I heard about the GI was thrown in the brig and put on bread and water - which of course made him even more intoxicated. Then they mounted an investigation to see how he was getting the booze smuggled in. That finally showed it was a medical problem.

Turns out he had diverticulosis - one or more failures in the intestinall muscle wall where the gut membrane bubbles out into a little appendix-like pocket and is prone to infections - and one of these became home to a culture of brewer's yeast.

Comment Old and kludgy makes it harder to port. (Score 2) 157

Not only does it cost a LOT to port this stuff and risk errors in doing so, but the cruftier it is the harder (and more expensive and error-prone) it is to port it.

If, instead, you can get the new machines to run the old code, why port it? Decades of Moore's Law made the performance improve by orders of magnitude, and the behavior is otherwise unchanged.

If you have an application where most of the work is done in a library that is largely parallelizable, and with a few tiny tweaks you can plug in a modern multiprocessor-capable library and run it on a cluster, you get another factor of almost as-many-processors-as-I-decide-to-throw-at-it, with small effort and negligible chance of breaking the legacy code.

What a deal!

And it's one less reason to touch the tarbaby of the rest of the working legacy code.

Let the COMPUTER do the work. People are for setting it up - with as little effort as practical - and moving on to something else that is important and can't yet be automated.

Eventually somebody will teach the computers to convert the Fortran to a readable and easily understandable modern language - while both keeping the behavior identical and highlighting likely bugs and opportunities for refactoring. Until then, keeping such applications in the legacy language (unless there's a really good reason to pay to port them) is often the better approach - both for economy and reliability.

Comment Re:Don't mess with America (Score 0, Troll) 162

Submersible hunter-killer drones lie in wait to defend America's freedom cable and orbital defense platforms defend the space above from communist tyranny. Long live freedom's reign.

Says the guy living in the country with the highest incarceration rate on Earth source

We're defending something, sure, but I don't think it's freedom.

Comment That's too bad (Score 1) 23

That's too bad. I've met many of their people. They were doing good work.

"Suitable Technologies" is just another company producing those annoying "remote presence" robots with a video phone on top. There are four or five manufacturers of those things. They can't do anything; they have no manipulation capability. They just talk.

"Remote presence" is only useful if the person running it is someone who gets sucked up to. Like doctors. A number of health care vendors are trying the things.

Comment Bandwidth of tape is terrible (Score 1) 208

A few years ago, I was involved in the conversion of the Stanford AI lab tape archive to modern media. This involved reading thousands of reels of 1/2" magnetic tape. It was a slow process. Volunteers were loading a tape onto a tape drive every 15 minutes for weeks. After each tape was loaded, its contents were sent over an Internet connection in under a minute. It took much longer to wind through the tapes than to transmit the data.

The data went to a server farm at IBM Almaden Research, where the file systems were reassembled (these were incremental dump tapes) and text files were converted from the Stanford AI lab's unique character set to Unicode.

The result was the SAIL DART archive. See the source code for EMACS, the early years.

Comment Easy answer... (Score -1, Flamebait) 162

The only country that has thus far been able to check the United States' unmitigated love of bombing people has been Russia. 'Murica wanted to bomb Syria. Russia came in and said "We'll take those chemical weapons off your hands, then 'Murica has no reason to bomb you." ... Suddenly, Syria looks rational, Russia looks peaceful, and America looks like the playground bully kicking sand in everyone's faces. Naturally, this didn't go over very well with the war hawks in Congress... but to date, no bombs have dropped.

China needs someone in their corner with nuclear weapons. Either that, or develop their own. America does not play well with others; They only back down and act reasonable when there's a risk of total and immediate thermonuclear destruction of the planet... anything conventional and it's bombs-away! I only wish I were joking. China has already taken the first steps -- realizing that America can't be handled conventionally. They've started developing their economy and cyberwarfare resources at a pace that exceeds America's, and the disparity is growing measurably every few months. It will only be a decade at most before they're left eating the dust of China as it rises to become a global economic superpower.

America is looking at undersea cables and going; If we can delay this a bit somehow... it'll slow 'em down. Their policies towards china have become very much about delaying and frustrating them, because stopping them isn't an option anymore. There's billions of chinese, and only millions of Americans. But stubborn nationalistic pride is keeping both sides from finding a mutually-acceptable middle ground. Unfortunately for America... they're rapidly losing their position at the bargaining table. They may not have a chair in a few more years at the rate China is developing.

And I think that, more than anything, is what is driving behavior like this. Worrying about the Chinese spying on everyone and putting backdoors in telecom equipment is a pretty pitiful excuse when America has been pants'd internationally over the exact same thing recently, and new examples are being made public weekly. And China isn't running around hunting down its ex-pats in Russian airports when its citizens come forward and say what its government is up to. They just stare blankly into the camera and then say "We make you iphone! iPhone good! You want more iphones? Shut up." ... and that's the end of it. -_-

Comment Re:Really? (Score 1) 552

I found spinning rust to at least give some clues prior to a crash and burn.

You know, I find this attitude to be both prevalent, and strange for supposed IT experts. Most of your computer doesn't run on "spinning rust". CPUs, memory, motherboards, power supplies... nobody says the lack of noise they make when they die (unless you count the screams of the souls that are released with the smoke) is a problem... but somehow, when it comes to SSDs, the "I can hear it dying" argument comes up. A lot.

I suspect this is a psychological attachment, with a healthy helping of overvaluation of personal experience instead of objective data. The weird part? When you point it out, geeks tend to dismiss it as "Well, they just aren't as good" as though 'goodness' was some kind of objective measure. I find this all the time amongst otherwise perfectly rational IT people: The belief that because the solution isn't perfect, it is therefore wrong, while ignoring the fact that the current solution they're supporting is also not perfect.

But the fact is, SSDs are many multiples faster than regular old "spinning rust" and more reliable. Ask any major manufacturer what their average warranty RMA rate is on their SSDs versus any other manufacturer's RMA rate on regular old "spinning rust". You'll find that SSD manufacturers regularly offer 3 and 5 year warranties. You're lucky to get a 90 day return policy on spinning rust bought off Amazon.

Now, all that said, dig into the data and you will find some new failure conditions that spinning rust doesn't have. For example, sudden power loss can cause a temporary loss of capacity, which will show up as bad sectors, in many SSDs. Very few IT professionals are aware of this; Or the fix: Physically disconnecting it for at least an hour, then wiping it (SATA command, not OS) and restoring the data. Many will RMA a drive claiming 'bad sectors' when there's nothing physically wrong with the drive... it's just buggy firmware.

Everyone points to write-exhaustion; The overly-focused on issue of repeated writes eventually 'wearing out' the drive. But guys... the average cycles here are 3,000 to 5,000 per cell. If you are writing 10GB a day to your drive, then a tiny 80 GB SSD will take 18.7 years before it gives up the ghost; Or about 68.5 TB of data written to the disk. If you opted for a 160GB drive, kick that out to 37.5 years. And that's for it to start showing physical loss of storage capacity.

The problems of SSDs is not electrical. It is not physical. It is entirely software. The firmware on many of these drives is buggy and this is covered up by the SATA / AHCI interfaces, which were designed for spinning rust, and thus have no direct way to signal the myriad of weird firmware glitches.

The electrical/physical part of SSDs is proven tech. It doesn't go bad, not under the usage conditions that the average computer user will put them into. And yes, I know, you don't think of yourself as average... but you are, ok? Even you, Mr. Programmer, Mr. Video Editor, and Mr. Super Linux Power User ZOMFG I Built My Own Raid In Mom's Basement. All of you are the 'average' case. The only time I've heard of mechanical drives being preferred is in usage conditions where data is being constantly written out -- such as a monitoring system like the Large Hadron Supercollider that collects terabytes upon terabytes of data, which is then processed and flushed, many times a day. SSDs would be bad in that environment. But unless you're building your own LHC in the garage... SSDs will work just fine.

That said... I have considered writing to OCZ and Intel and asking them if they could make their SSDs make the same noises as mechanical drives. There's a proven psychological value in this; Just like how your cell phone camera is programmed to emit a shutter snap sound... despite shutters not being around since the 80s. Because there are a lot of people that apparently need reassurances that their computer be making noise in the corner for them to feel good about it's performance and reliability. It may be too soon for geeks to live with silent computers.

Comment Re:Give consumers more privacy? (Score 4, Informative) 147

Or take away their ability to block tracking as they can currently do with cookies?

That's the basic idea. CNET covered this a few days ago. "The AdID would be transmitted to advertisers and ad networks that have agreed to basic guidelines, giving consumers more privacy and control over how they browse the Web,"

Expect meaningless, easy to evade "basic guidelines", like TrustE.

Comment Re:Not really notable at all (Score 1) 214

That presumably helped getting (relatively NDA free, no less!) access to the part. Again, I have no idea what Intel says if you want to buy less than a tray worth of anything (much less get detailed assistance/permission, they certainly weren't shy about squishing Nvidia on QPI chipsets...); but the position on the rPI is that (while it is sincerely cheap, and the vendor has no obvious conflicts of interest in selling as many of the things as they can, as uncrippled as they can), it barely matters whether the Gerbers are open or not when the core SoC is practically a special favor from Broadcom.

Intel's availability situation is clearly better for socketed parts (since, even if they won't talk in units of less than a thousand, resellers abound who do whatever they want); but I don't know what the story is for things like chipsets and BGA parts. My personal interest is less in the freedom of the board (since the economics of DIY vs. some pacific-rim slave factory will always be dubious); but the firmware. Intel have, in contrast to their generally good position on graphics, been kind of dicks compared to AMD in terms of Coreboot vs. UEFI+whatever proprietary features Intel has dreamed up these days. An Intel-supported, enough-UEFI-to-actually-be-useful, board would be real news(yeah, Intel has 'Tianocore', which is the OSS implementation of all the boring parts of UEFI; but the good shit, of the sort actually required to do a coreboot port or a fully-open UEFI build, has been lacking for all but the most antique of their gear).

Slashdot Top Deals

You can tell the ideals of a nation by its advertisements. -- Norman Douglas