Courtesy of The Best Article Every day (lack of capitalization theirs, not mine): stores with punny names. I’m pretty sure I’ve seen “Curl Up and Dye” before, in New Hampshire perhaps. It’s funny how many of these are Thai restaurants. Warning: some are a bit off-color. “Lord of the Fries” is a bit too obvious, but “Luna Sea” and “Florist Gump” are pretty good.
I actually started composing this in my head before Matt Reilly’s post about Green Computing, but it kind of fits in with some of the points he makes about communication speed as a factor in overall performance. One of the properties of a Kautz graph, such as we at SiCortex use in our systems, is that
For a fixed degree M and number of vertices V = (M + 1)MN, the Kautz graph has the smallest diameter of any possible directed graph with V vertices and degree M. (from the Wikipedia article)
UPDATE 2008-11-10: I happened to read this while looking for something else, and realized that there are better torus routing methods than those I had considered. I think I’ve invented an even better one than the current general practice, but for now I’ve updated the text below to reflect the general-practice numbers (which don’t affect my argument at all).
Yeah, lovely, what does it mean? What it means is that if you want to make a system of a certain size, let’s say approximately 1000 nodes, and you’re constrained as to how many links per node you can have, you can achieve the smallest network diameter by arranging your nodes in a Kautz graph. For example, our 972-node system using three outbound links per node (i.e. degree=3) has diameter=6. That’s the maximum number of hops to get from a node X to any other node Y; the average is approximately five. By contrast, for a 10x10x10 hypertorus also using three outbound links per node, the diameter would be
27 18 and the average hop count would be over 13 almost 10. I’m not actually familiar with the proof for the statement quoted above and wouldn’t understand it even if it were shown to me, but it seems very believable based on my experience. Every time I try to think about tweaking some other topology to reduce the average hop count or bring the links per node down to reasonable levels (you can’t feasibly build a system that requires dozens of links on every node) the result starts to look more and more like a Kautz graph in terms of routing, partitioning and tiling characteristics, etc.
So far, so good, but how does that translate into anything useful? Well, if communication speed is a factor in overall performance, and communication speed falls through the floor when the system’s communication capacity is exceeded, then that capacity matters a lot. The common way to compare the communication capacity between different systems is to compare bisection bandwidth – how much capacity must be removed to divide the machine into two equal parts? It turns out that measuring the bisection bandwidth of a Kautz graph is not straightforward, and even the Kautz cognoscenti at work seem to disagree about the result. Personally, I think the whole bisection-bandwidth approach is bass-ackwards. Why measure the negative (minimum capacity you could remove) instead of the positive (maximum capacity you could use)? I think it happened because a naive attempt to measure maximum usable capacity would allow a system to be partitioned into small islands communicating amongst themselves, yielding inflated figures that don’t match what a real application running across the entire system could get. That’s fixable in a measure-the-positive approach, though. All you need to do is say that you’re going to measure the maximum bandwidth when each node in one half of the system sends equal amounts of traffic to each node in the other. That way, if you want to combine the figures for two islands (often corresponding to two switches or cabinets) then you have to pass half of your traffic between them, and what you get is a pretty fair approximation of what an actual full-scale application could get.
This is where network diameter and particularly average hop counts start to matter. The maximum usable communication capacity of a system is the aggregate link speed divided by the average hop count. Therefore, even if you can distribute your traffic perfectly across all links in the system for the test I propose above, your performance ceiling will be determined by the average hop count in your topology. For example, in our 972-node Kautz graph that’s 2916 links’ worth divided by an average hop count of five, or 583.2 links’ worth total. For that 1000-node hypertorus, it’s 3000 links’ worth divided by an average hop count of
13.5 10, or only 222.2 300 links’ worth for a similarly sized system. Looked at another way, your links would have to be about 2.6 times almost twice as fast just to keep up. If you made those links bidirectional you’d double the number of (half-duplex) links and reduce your average hop count to around 7.5, so you’d be getting 800 links’ worth of total capacity, but then you’d be talking about a system with vastly greater hardware complexity and cost than the others. If you wanted to compare apples to apples, a thousand-node Kautz graph with degree=6 would have the same 6000 links with an average hop count around 3.5 so the hypertorus would still be at a serious disadvantage. Then you have to consider that we’re talking about ceilings here, and that even in the abstract different topologies will allow for different levels of “perfection” as regards distributing traffic among all the links in the system. Maybe I’ll explore some of those issues at some time in the future. For now the key point is that, much as a more efficient algorithm will always eventually win out over a faster implementation, a more efficient topology will always eventually win out over raw bandwidth. In my opinion, bisection bandwidth as usually measured doesn’t tell that story well enough.
I first saw this on j-walk. It’s a terrible story.
Just four days after escaping a federal minimum-security work camp, “Spam King” Eddie Davidson shot his wife and child and wounded a teen-age girl before turning the gun on himself.
Sheriff’s deputies responded to a report of gunfire in the small plains town of Bennett at about 11:15 a.m. today and found Davidson, 29-year-old Amy Lee Ann Hill and their 3-year-old daughter shot to death.
Before I go on, I want to make one thing perfectly clear: the guy’s a monster. As the adoring father of a four-year-old girl, the feelings I experience when I hear of bad things happening to small children are way beyond the feelings I get from anything else. Had this guy done what he did and survived, I’d be seriously reconsidering my positions on capital punishment, torture, etc. That’s really all I need to say about that. However, I did take strong exception to the following flippant remark by John Walkenbach.
Suicide would have been a perfectly satisfactory outcome.
Since when is suicide a “perfectly satisfactory outcome” for anything? Yes, it’s preferable to the tragedy that actually occurred, but that’s not the same statement at all. To see why, consider what Walkenbach said in the context of the day before this event. Spammer escaped from prison, his whereabouts are unknown. Was suicide a “perfectly satisfactory outcome” then? Of course not. Suicide is not a perfectly satisfactory outcome for petty crime, and that’s what spamming is. As I said in a comment there,
Wild speculation about how his family might have celebrated doesn’t change that. Even when the worst people in the world commit suicide, their families rarely escape guilt and other difficult reactions, and there are quotes in the article indicating that he was hardly perceived so negatively by those around him. Suicide is a serious matter, and I continue to believe that anybody who would call it a “satisfactory outcome” is worse than any spammer. The second or two that spam from this person might have caused you to expend hitting the delete button does not outweigh even the most worthless human life.
What’s even sadder than Walkenbach being such a poor excuse for a human being is that the peanut gallery mostly seemed to approve, and attack me for dissenting. I’d like to address a couple of those cheap shots as well.
Karma’s a bitch man. — Phineas Freak
Yes, it is, and acting callously toward the innocent victims or survivors is bad karma too. My karma’s pretty good right now. Yours sucks.
” … a poor way to show we’re better than they are.” — me
LMAO. Yes – we must ALWAYS show we’re better than they. — Evil Klown
Direct your sarcasm toward Walkenbach, then. He’s the one who clearly thinks he’s better, unless he thinks that suicide is a “perfectly satisfactory outcome” for everyone including himself (and you). He thinks he’s better than a spammer. At this point I disagree. The inconsiderate nature of spamming annoys me, but the attitude that he and you represent has far more serious implications. That “the world is divided into good people and bad people with the bad ones all deserving death” attitude is responsible for many tragedies even greater than this one. Anybody making comments like Walkenbach’s, or defending them, has lost all sense of perspective and humanity. If death is the penalty for annoying people, where does that leave any of us including you? Go ahead. Cast that pebble, and watch out for the boulder that comes back. Karma’s a bitch.
Apparently a McCain aide has tried to claim that Obama’s position regarding genocide is inconsistent or insincere based on the following two statements:
Democratic presidential hopeful Barack Obama said Thursday the United States cannot use its military to solve humanitarian problems and that preventing a potential genocide in Iraq isn’t a good enough reason to keep U.S. forces there.
“Well, look, if that’s the criteria by which we are making decisions on the deployment of U.S. forces, then by that argument you would have 300,000 troops in the Congo right now — where millions have been slaughtered as a consequence of ethnic strife — which we haven’t done,” Obama said in an interview with The Associated Press.
Let our children come here and know this history so they can add their voices to proclaim ‘never again.’
I’m going to give Tucker “Out Of” Bounds the benefit of the doubt and resist the conclusion that he’s just a mean-spirited jerk who will stoop at nothing to get in a shot at Obama. Instead, I propose that the remark merely demonstrates an lack of logic and reading comprehension. “Humanitarian problems” does not equate to genocide. Even “potential genocide” does not equate to genocide. I can’t find an actual transcript of the AP interview, so I don’t actually know what criterion Obama was referring to. Neither do any of the dozens of others who’ve commented despite being unable to provide a direct quote. The exact words do matter. There’s no inconsistency involved in making two statements at two different times and places about “humanitarian problems” or “potential genocide” on the one hand and (unqualified, manifest) “genocide” on the other. Unless and until someone can definitively identify the precise context of the first remark, bashing him for any perceived inconsistency is premature, fallacious, and irresponsible.
P.S. The next pundit to talk about “walking back” a statement will be “walked back” behind my house and soundly beaten. I’ve seen that silly phrase used entirely too much this week, first applied to Maliki and now to Obama.
A couple of years ago I wrote about the Singularity project at Microsoft Research. Then I didn’t really hear anything more about it, and pretty much assumed it had run its (short) course like most research operating systems do. Today via OSNews I found a report about porting Linux’s ext2 filesystem to Singularity. The report is mildly interesting, but to me the most important piece of information is that Singularity is still alive and being used as a basis for ongoing research.
The ever-insipid Joan Vennochi was, for some unfathomable reason, given space in today’s Boston Globe where she whines about Obama’s ego.
In February, a blogger for the left-embracing Mother Jones commented on his uneasiness over the candidate’s messianic complex: “Does this post play unhelpfully into the pernicious and growing Obamaism-as-cult . . . that we’ll likely see repeated over and over by the right wing if Obama gets the nomination?” blogged Jonathan Stein.
“It does. Sorry. But Obama’s rhetoric makes an undeniable suggestion: that his election, not an eight-year administration that successfully implements his vision for America, would represent a moment in America of the grandest, most transformative kind. And that’s a bit much,” Stein wrote.
Does Vennochi seriously think it’s possible for anyone to run for president without having a sense of mission, and perhaps a bit of ego to go with it? Jonathan Stein’s criticism is almost fictional as it applies to Obama – who never made any such claims himself – but applies quite well to Reagan’s “Morning in America” campaign. At least Obama, unlike George W. Bush, doesn’t claim to be getting specific instructions directly from God. Somehow you won’t find Vennochi pointing to those examples. The simple fact is that presidential candidates are a proud lot.
Yes, Obama comes off as confident in what he says. What’s wrong with that? Should he try to sound less confident even when he’s talking about subjects he understands well? Is this another variation on the “ignorance as authenticity” meme we’ve had to put up with ever since Bush started running for president? You might think so, from the way Vennochi tries to praise McCain for a humility that would be well justified if it were sincere. Should Obama make shows of such false humility, pretending not to have knowledge or skills that in fact he does? Should he try to be “one of the boys” like the trust-fund Ivy-league false Texan in the White House today? You can’t get less authentic than that. Without defining a difference between confidence and hubris – which Vennochi doesn’t even attempt – all such criticism demonstrates is the critics’ belief in a social hierarchy where even a US senator and potential president shouldn’t presume to stand up straight and look others in the eye until the establishment has accepted him and granted him that right. I’m trying really really hard not to turn this into an accusation of covert racism, but the more I think about the mindset necessary to criticize Obama for being too proud the harder that gets so I’ll stop.
Here’s the old car – a 2000 Subaru “Outback Sport” (Impreza).
The bumper sticker, which is hard to read at this angle/resolution, says, “Buckle up. It makes it harder for the aliens to suck you out of your car.” Also, notice the lack of hubcaps. This is a common feature on 2000 Imprezas, to the extent that I always double-check any such that I find to see how many are missing. There’s almost always at least one, and as recently as this past Sunday I saw another “four-missing” parked right next to us at Smolak Farms. You can’t tell me that’s not a design flaw. That and a few other mostly-cosmetic gripes aside, though, it really has been a good car. I replaced a muffler and a couple of rear bearings, each time for reasonable prices (compared e.g. to Toyota which might need maintenance less often but costs an arm and a leg when you do). Besides commutes and errands, its 60K includes many trips to/from surrounding states and a couple to/from Michigan. Gas mileage has been decent, hovering around 25mpg for mixed driving the last few months. Overall, I still think it was a fine car for the price. It was just a little old and a little cheap and a little boring compared to . . .
For the record, that’s a 2008 MINI Cooper S, manual six-speed transmission, with stability control, automatic AC, upgraded steering wheel with audio and cruise controls, and wood dash to match – none of which you can see in the picture, of course, but I’ve been enjoying all of them except for the stability control which I don’t expect to notice until I’m in rain/snow. It’s a very nice ride, but I don’t need to say much about it because all the specs etc. are right on the MINI site. The blue is fairly new, and not the same as the older light blue you’ll see on most MINIs. I want to get a green car some day, always have, but frankly the “British racing green” that MINI has is one of the less desirable greens out there. At least this is an ever so slightly green-ish blue. Cindy has driven it too, and seems to like it. I’ll probably post an update after a couple of weeks when I really have more than first-drive impressions to share.
I’m in lovely Bangor (Maine) for work, and chose to stay at the Sheraton Four Points hotel and construction zone because it’s near the bus station. Yes, I took the bus, because flying would take just as long for ten times the price and driving wouldn’t have been good for my brand new car (more about that after I get home). One of the features here, as most everywhere nowadays, is free wi-fi via stayhome a.k.a. e-centre. I could browse OK, but I was having trouble with email and ssh tunnels. They’d just lock up. After trying many things to debug the problem, I had sort of concluded that the provider must be blocking traffic, so I googled for their name plus various relevant terms.
It turns out that they’re not blocking intentionally; they’re just screwed up. I pretty quickly found one article on an Eee forum (I’m using an Eee but wasn’t specifically searching for anything related to that) which pointed me to a solution – turn off TCP window scaling. Huh? I know a thing or two about TCP, and I’m still having trouble imagining how e-centre could have screwed things up so that mishandled window scaling affects email but not in web browsing (which is more likely to transfer large amounts of data). Clearly they’re aware of it as a problem affecting users, so why don’t they fix their routers?
Oh well, at least it works now.
Today we went to Smolak Farms in North Andover, as it turns out almost exactly a year since last time. I pretty deliberately took some pictures that were almost exact mirrors of last time, because I think it’s fun to compare how Amy looks year to year, but there are also some new ones. Before I get started on the pictures, though, I just have to say that the Treadwell’s ice cream they serve at the farm stand is awesome. Last year I got something called “purple moo” which was black raspberry with (large) light and dark chocolate chips. It was good, but the chocolate raspberry truffle this time outdid it. The raspberry base was creamy with good berry flavor, the chocolate part was somewhere between brownie and fudge, and the two flavors/textures complemented each other perfectly. Outstanding. Now, on with the pictures. I also have some video, taken with the same camera, of Amy reading. It looks good, but I’m not going to fight with video editing on Linux tonight.