Matthew Ernest: Pushing rectangles...

Popularity doesn't scale

Sun, 12 Sep 2004 17:42:15 GMT

I could have quoted from any of a numbers of posters about the RSS non-scaling non-controversy, but Dare offers a round-up of the bleating masses

The RSS Sky is Falling...Again.This is becoming a broken record. Every couple of months some web site that hasn't properly prepared for the amount of bandwidth consumed by having a popular RSS feed loudly complains and the usual suspects complain that RSS is broken. [Dare Obasanjo aka Carnage4Life]

Sure, RSS doesn't scale. It doesn't not scale either. All of the problems with scaling have thus far been shown to depend on nothing particular to RSS.

Scoble attemps to show RSS not scaling a) generating a gedankenexperiment where hits on the RSS are a fixed factor greater than HTML hits, and b) by quoting figures that the bandwidth consumed by RSS requests (for what site I'm not sure) per month is growing at a much faster rate that bandwidth for HTML. The former shows exact parity in scaling between RSS and HTML, while the latter fails to factor in the relative popularity of RSS versus HTML readers. If new traffic for RSS is N times new traffic for HTML, but new requestors of RSS are 10 times new requestors for HTML , this is also parity in scaling between RSS and HTML.

There are attempts to blame the automated nature of RSS requests, but HTML can be and is requested by machine as well. Either way, requests at regular intervals may affect the absolute value of the traffic but bear no relation to the scaling of the traffic growth.

Certainly Amdahl's law would tell you that if RSS is a significatly larger portion of your traffic that it may be worth while applying some RSS specific fixes. However, the discussion is not really about a scaling problem in the sense that the proposed solutions do not affect scaling. If requests are still growing exponentially, reducing response size may delay the point at which you hit any particular traffic level but it cannot prevent it. This is independant of what the response is, and also independant of the absolute size of the response.

My greatest fear

Mon, 09 Aug 2004 01:39:54 GMT

I do not know what idea is more frighening: that regular people might think we use such overwrought terms as "silicon nanosurgery", or that someone at work might read this and think it would be a good idea if we started.

Intel Technicians Use Delicate Silicon Surgery to Fine-Tune Microchips. A technique known as silicon nanosurgery, routinely used at nine Intel chip factories around the world, has transformed the way modern computer chips are developed. By By JOHN MARKOFF. [The New York Times > Technology]

Dumbest. Marketing. Evar.

Sat, 19 Jun 2004 16:41:26 GMT

Some days I feel like my company's products are marketed by morons. Other days I have proof.

Intel's Tablet PC Centrino Surfboard.
I don't know what to say. Intel has gone and taken the ur-joke of the internet and made it reality by incorporating a tablet PC into a surfboard, allowing pro surfer Duncan Scott to literally, yes, surf the web. The laptop is integrated into a Jools Matthews board and communicates via WiFi with a hotspot on the beach. I fully expect to go outside now and find a laptop built into the BQE with a huge arrow pointing to the 'Information Highway.'
Read [Pocket-Lint]
[Gizmodo]

Dualies

Tue, 15 Jun 2004 14:18:11 GMT

AMD taped out the dual-core Opteron. I wouldn't mind swapping out my current Opteron for a 3GHz dual-core version. [Hack the Planet]

No pressure, eh?

Home on the page

Mon, 14 Jun 2004 14:53:42 GMT

K10 Architect Leaves AMD. AMD Zone Jun 13 2004 5:41PM GMT [Moreover - moreover...]

All I can say is that's it's kind of sad that Andy Glew has his CV posted on a geocities site.

Not only does it taste bad, but the portions are so small

Mon, 05 Apr 2004 02:23:14 GMT

Intel's AMD64 Sucks?. AMD Zone Apr 4 2004 5:55PM GMT [Moreover - moreover...]

The fanboys that complained about how Intel's 64-bit was identical to AMD's 64 bit now say that it performs badly and go off on some tangent about "emulation". What they fail to mention is that EVERY x86 processor on the market emulates x86 anyway. Accusing Intel's 64-bit of being emulated is like accusing water of being wet.

I can't read German, so I don't actually know if the original c't article quotes actual perfromance metrics. Certainly I haven't seen anyone who points to the article quote any numbers, so no one can comment on the numbers.

Let them eat red/black trees

Wed, 03 Mar 2004 14:35:51 GMT

Tim Bray deleted his commentary on his own site, leaving only the quote, so you'll have to go off what's here. Anyway... Tim, up in the this air where people make new languages like XML, is shocked (shocked, I tell you!) to read that there are function of programming that are "mind-numbing". Not only are they mind-numbing, but most of the low end of the salary spectrum does nothing but those kinds of task all day long. Welcome to the quite desperation of someone who works for a living.

At best, I could read that as Tim misinterpreting the quote as saying that all code-writing is mind-numbing.

Coding Makes You Dumb. I quote from an article in this weekâ€™s Economist (read it here if youâ€™re a subscriber) arguing that the negative impact of â€œOffshoringâ€� is exaggerated. The reasons we need not worry include ... the bulk of these exports will not be the high-flying jobs of IT consultants, but the mind-numbing functions of code-writing. There you have it; I guess Dennis Ritchie and Bill Joy and Dave Cutler and Charles Simonyi and Linus Torvalds and Larry Wall somehow managed to conceal the mental numbing. [ongoing]

Litigation as a business case

Wed, 18 Feb 2004 15:06:53 GMT

Judge Rules Throws Out Rambus Antitrust Case. Boston Globe Feb 18 2004 3:45AM GMT [Moreover - moreover...]

Oh, how I long for the days when a business had to make something or provide a service. However, this is just a preliminary decision, so the Rambus case could still come up for further review.
The variation in headlines is somewhat interesting: "Appeals Court dismisses", "Judge dismisses", and Extreme Tech even had "FTC dismisses" which is completely wrong.

A darks spot on the landscape of litigation engineering

Fri, 13 Feb 2004 03:57:53 GMT

UPDATE 2-Rambus says Europeans revoke patent, shares tumble. Reuters Feb 13 2004 0:34AM GMT [Moreover - moreover...]

Rambus doesn't really make anything anymore except lawsuits, so I'm not too upset for them. However, I iwsh just one of the reports had bothered to write about which specific patent was struck down. Rambus has so many empty patents to chose from.

Yellow

Tue, 10 Feb 2004 14:57:12 GMT

Is Intel using sleazy tricks to slow down AMD users in its latest compiler? [Hack the Planet]

Why accuse them of being dumb for not optimizing away checks in the event they are unneeded when you can invent a conspiracy theory? Note that in the article they examined only ONE program to see if the SSE/SSE2 checks actually surrounded SSE/SSE2 calls. The accuser provides no discussion the the optimizations actually performened when these flags are used.
I'd bet good money that you could use a set non-processor specific optimization switches, corresponding to the non-processor specific optimizations that are rolled into the processor specific switch used in the article, for the particular piece of code examined and thus disprove the conspiracy. But that would be no fun.
How much effort should Intel be obligated to spend to determine if an AMD has the features that the compiler optimizations depend on? If they make a mistake at reverse engineering the AMD microarchitecture, as I think they've done here, the PR is the same as if they never tried.
Full disclosure: I work for Intel. Therefore I must be part of the cover-up, yes?

The dangers of letting an English major design your memory hierarchy

Sat, 10 Jan 2004 17:50:14 GMT

Trying to dig our from under a backlog of journals, I cam across this gem in IEEE Computer magazine: "VIM: Taming Software with Hardware" by Mark Halpern. His claim is that software would work so much better if only we'd remove the complexity of managing thememory hierarchy... by supplying infinite memory to the programmer.

Infinite. No problem.

The capacity itself isn't even the worst part. As the capacity of a storage element increases, so does its access time. (Sure, things speed up as new technologies become available. At some point, however, you have to stop waiting on new technology and actually build something.) When register files already have odd sizes to account for the buffering needed on the most distant bit lines, a multi-terabyte register file is going to be half buffers. And yes, I said multi-terabyte register file. Otherwise every singe piece of data will take multiple cycles.

Then there's the yield hit for such a large memory area. Let's say 1MB of cache in a 90mn process is roughly 1/4 cm^2, and assume a register file requires no additional devices (which is wrong, but I have no way of estimating the size of an RF even that large). A 12" wafer is only going to get you to the 10GB range, and that's if you use the whole wafer just for a single memory structure.

Of course, Halpern uses the phrase "virtual infinite memory". There is in fact a method to make a process see an infinite address space with actually supplying an infinitely large register file: it's called a memory hierarchy.

Waiting for nothing

Thu, 25 Dec 2003 19:25:31 GMT

UWB standard delay likely for a year (or more). Stalemate still stale [The Register]

The worst part is that of the two mentioned contenders neither is really UWB, just higher bandwidth versions of normal methods.

Pointing Fingers

Sat, 08 Nov 2003 03:54:05 GMT

Quite a few articles are floating around trying to pin the the blame for bad passwords/passphrases on WPA. Pay attention kids: crappy passwords are crappy universally. This has nothing to do with WPA.

Coverage of WPA Key Choice Weakness. An excellent analysis of the WPA key choice weakness problem: It's been a high-traffic week here at Wi-Fi Networking News since we posted Robert Moskowitz's paper on how short WPA passphrases comprised of words found in dictionaries could be broken. Many many thousands of people have read the paper, and a number of articles of varying levels of accuracy have been written. The IDG News Service story is about the best re-summarization in IT terms. But, IDG News Service, really: Moskowitz's paper is circulating informally on the Internet. No, it isn't. Robert provided it to me and gave me permission to post it on my site, along with a less technical summary that I wrote and had him vet before posting. This moment of wounded ego is over.... [Wi-Fi Networking News]

The first rule of Shift Key is: Don't talk about Shift Key

Fri, 10 Oct 2003 14:35:23 GMT

Dumbass litigation averted, narrowly.

SunnComm Says It Won't Sue Halderman. SunnComm, which had previously said it planned to sue Alex Halderman for publishing a critique of SunnComm's CD anti-copying technology, has now backed off. According to Josh Brodie's story in today's Daily Princetonian, SunnComm president Peter Jacobs has now said... [Freedom to Tinker]

This company claimed Halderman damaged its reputation and violated the DCMA. It's impossible to do both at once, since the damaged reputation is Halderman's fault only if his description of the "protection mechanism" is inaccurate, in which case he has not circumvented it; if he has circumvented it, then his description of how poorly it works is true and the damage to their reputation is their own fault.

This is a rare example of a vendor underestimating the intelligence of the public and losing money because of it. The difficulty with making money from dumbass copy protection scemes is that it only takes one person to identify a flaw and make a copy, at which point most people who receive a copy won't have to know how to came about to get their copy. Damned outliers! ;-)

I admire your ability to get paid for this

Tue, 26 Aug 2003 04:35:25 GMT

Good Times. Would you tolerate it if prankster script kiddies could damage your company's toilets? If not, then why tolerate allowing them to damage your computers? [Daring Fireball]

Plenty of stuf here to disagree with, but one good quote makes up for the whole thing:

The corporate IT field is in large parts comprised of men who are not smart enough to program, but yet wanted a career in computers. They are the fifth wheel of corporate America — serving no practical purpose but their own employment.

Cut off

Tue, 26 Aug 2003 01:58:47 GMT

My network port was cut off becasue IT claims I have the Nachi virus on my work-provided laptop. However, every test they publically offers says the machine does not have it, every patch they offer says it's already up to date, and every update says it's already installed.

So, I can't get any mail from them telling me if they've taken care of it, because that requires having the laptop to run Outlook to connect to the exchange server.

A completely incorrect position on quantum computing

Sun, 24 Aug 2003 16:46:35 GMT

I guess I'm not the only person skeptical about quantum computing. The head of Darpa points out that analog computing doesn't scale well, and that's what quantum computers actually are.... [Thin Film Manufacturing: Katherine's Blog]

Just because Robert Leheny says that a quantum computer is an analog computer doesn't mean that it really is. A bold statement like that will require some proof to back it up, which will be all the harder bacause it isn't true.

The article quotes Leheny as saying ""What they do is to set up an equation, and let the quantum devices solve it over time. But that is an analog computer." But that's also an digital computer, so it's pretty much a non-statement. While quantum computing isn't a magic solution to all the worlds problems, it cannot be dismissed with out even examining its solution space. I have no buy-in for quantum computing and I think any application of it is way out there, but that does not mean I can prove that it's useless.

His claims that current fabrication technology is self-assembling are equally silly. Yes they are self-aligning to the tolerances of the process. But with tolerances on the order of the features themselves, you're lucky just to get anything to show up at all.

Dismissing "capable of something approaching human thought" lines aren't interesting any more, so they are left as an exercise for the reader. Here's a hint though: how much do we know about human thought?

A completely incorrect position on security exploits and marketing

Sat, 23 Aug 2003 02:10:16 GMT

recent viruses and lost opportunities. Why, oh why, doesn’t Apple take advantage of all the current virus traffic to run ads that point out that Mac users don’t get these viruses ??? It seems so obvious. We were watching NBC news last night, and they were explaining the SOBIG virus and how it works. Not once did they mention that it only affects computers running Windows. (When I remarked about this, Gerald gently reminded me of the MS-NBC relationship. Duh.) Even on campus, none of the dire warnings about having your computer carefully checked by the tech folks before connecting it to the network mention... [mamamusings]

Without any proof that[system of choice] is not vulnerable to an equivalent exploit, trying to use viruses, worms, etc. as a marketing tactic will at best make you look foolish. At worst, it will motivate someone to put together a demonstration.

A completely incorrect position on electric power transmission grids and thier failures

Sat, 23 Aug 2003 02:05:35 GMT

Edge.org -- The Moral Sense Test: Blackout. In the latest edition of John Brockman's EDGE newsletter, conversation about the blackout of August 14th. From the contribution by Albert-Laszlo Barabasi, author of LINKED: THE NEW SCIENCE OF NETWORKS:

Once power is fully restored, it will take little time to find the culprit: most likely, it will be a malfunctioning switch or fuse, a snapped power line or some other local failure. Somebody will be fired, promotions and raises denied, and lawmakers will draw up legislation guaranteeing that this problem will not occur again.
Something will be inevitably missed, however, during all this finger-pointing: this week's blackout has little to do with faulty equipment, negligence or bad design. President Bush's call to upgrade the power grid will do little to eliminate power failures. The magnitude of the blackout is rooted in an often ignored aspect of our globalized world: vulnerability due to interconnectivity. Link, Discuss [Boing Boing Blog]

You can't use this blackout to demonstrate the danger of interconnectivity, since that was not what caused it. Such blackouts come from a) sparse connectivity, and b) poorly planned connectivity. In general, almost any additional connectivity improves robustness (although there are always pathological cases).

The engineer of my enemy is my enemy

Fri, 28 Mar 2003 03:47:59 GMT

Iraq's mobile network - Qualcomm to follow the tanks?. Congress rep denounces 'French' GSM [The Register]

...wherein the congressman blames GSM on the French, and denies companies and other nations that support the US the oportunity to compete to ensure Qualcom the oportunity for lock-in.

US Patent 6,531,910

Tue, 18 Mar 2003 03:12:58 GMT

Krawczyk; Thomas W. (Troy, NY); McDonald; John F. (Clifton Park, NY); Ernest; Matthew W. (Troy, NY) "Symmetric Multiplexer", US Patent 6,531,910, March 11, 2003

Abstract

A multiplexer is provided that is symmetric in that substantially the same delay is experienced from any input of the multiplexer to the multiplexer output. It is realized that in conventional serial transmission systems, standard Current Mode Logic (CML) multiplexers are used which are asymmetric and exhibit different delays between select and data inputs. Because of these delays, conventional transmission systems experience jitter at high frequencies. To extend the operable range of communication systems, a symmetric multiplexer may be used which has substantially the same delay from any input to the multiplexed output, thus reducing jitter. For example, the multiplexer may be part of a communication system having a serial data transmission circuit.

Tiny

Thu, 06 Mar 2003 21:12:52 GMT

USA Today: Hello, tech designers? This stuff is too small. But how will we operate such marvels? We struggle with what we own today. Our fingers are already too thick and clumsy to stab the buttons on our gadgets, and, as our eyes age, we squint even harder to see the shrinking screens on our stuff. [Tomalak's Realm]

My cell phone is quite old, and I still think it's too small. This problem even exists with wired phones. I have a trusty ol' Trimline that I like because the receiver is big and heavy.

Path of destruction

Thu, 06 Mar 2003 04:03:12 GMT

U.S. judge says Rambus destroyed evidence. Reuters Mar 5 2003 2:22PM ET [Moreover - moreover...]

A ruling against Rmabus on destruction of evidence. The FTC's motion that they forfeit trial was denied. Rambus still gets a trial on the FTC anti-trust charges, but the burden of proof will be on them to show that they were innocent... without the evidence that they destroyed.

Self-destruction

Wed, 05 Mar 2003 18:36:59 GMT

Why We Like Macintoshes. Dwight Silverman of the Houston Chronicle reminds us of why we are glad we don't do Windows: HoustonChronicle.com - Computing [Semi-Daily Journal]

Because of course a Macintosh would never eat its own entrails, right? The only operating system I've never trashed is VMS, but I know someone who has.

Branding is job 1.

Fri, 28 Feb 2003 19:15:34 GMT

Intel's Pentium-M: Too confusing?. ZDNet Feb 28 2003 8:10AM ET [Moreover - moreover...]

Analysists think Banias vs. Centrino could be confusing to consumers. That of course is the point.