HNHacker News
TopNewBestAskShowJobs

glangdale

5,457 karma · joined October 19, 2015

I work at Intel on a range of things, including the Hyperscan project:

https://github.com/01org/hyperscan

Worked on JSON parsing with Daniel Lemire at:

https://github.com/lemire/simdjson

Blog: branchfree.org

https://twitter.com/geofflangdale

submissionscomments
glangdale··on So you want to go to grad school (in the academic humanities)?
It depends on field and more importantly, country. I was the last person in my year (that I know of) at CMU to get a PhD, a whopping 9 years after we started. A friend had to re-enroll to get his 10.5 or 11 years in.

Meanwhile, I knew tons of people who got PhD's in CS in 4-5 years back in Australia. They typically did zero coursework or breadth stuff, and they were able to start dissertation work within a few months of the clock starting on their degree, without the undignified business of having to propose their thesis formally in public or even defend it in public.

The process of getting a PhD isn't order-of-magnitude different from field-to-field or place-to-place, but there's definitely a good factor of 2-3x between the most arduous process and the least (which does not mean that one degree is 2-3x better, of course).

glangdale··on When users never use the features they asked for
We found (in the world of building Hyperscan, a high-performance regular expression matching library) that asking users what features they wanted was a disaster.

You have to ask them what they want to achieve. Their suggestions as to how to do it were almost always terrible: partly, they don't have the domain knowledge to know what's going to be simple and practical vs complex and nasty.

Also, they usually have a comparatively clear picture of what they want to achieve, but their notions of how best to break that into "your problem (library)" vs "our problem (app)" are often very naive - you wind up owning way too much of the problem (usually), or occasionally going in the other direction and not really getting enough context information into your library.

glangdale··on Free Parking Is Killing Cities
Agreed, and even more drastically, taxes on cars are an incredibly poor ROI on the massive infrastructure given to roads. Obviously we all benefit from this public good - even non-drivers are getting the benefit from things being moved around on the road network, so it's not cut-and-dried.

But in Australia, we tend to raise money on road taxes that supposedly balances spending on roads - ignoring the humongous investment that giving so much of our space to roads represents (if you give me several billion dollars worth of real estate, I can probably break even too!).

It frustrates me, because spending on public transport is usually treated in a cost/benefit fashion in a very short-term way, but the massive capital investment into roads is just regarded as a given.

glangdale··on Dear Mom, I'm Dropping Out
The kid, unless he is very stupid, won't falsely claim to have graduated. They don't background check you on things that you don't claim.

That said, I wonder if a university will confirm a partially completed degree. If not, it becomes something that rounds down to zero at anywhere that takes background checking seriously.

glangdale··on Dear Mom, I'm Dropping Out
A common thread in these endless discussions is what a shallow and reductive view the pro-dropout people have of university. This guy thinks the biggest dig against McGill is that they teach web development (!) with an old stack, and he "knows" that he can "replicate everything that college can give and hopefully more".

I hope Facebook can pay this kid in Dunning-Krugerrands.

What a lot of early dropouts don't seem to realize is that the normal university experience in is very tail-loaded towards the tough, interesting courses. You can blitz through a couple years of basics and fume about how slow stuff is, or that the Intro Programming is in the "Wrong Language (TM)", but most of what I remember being good were 300- and 400-level courses. As you specialize, you also pick up a lot of these courses, relatively speaking.

There's also a lot to be said for, I dunno, all the other courses that you could be doing at a university for intellectual breadth. It's not a vocational school.

Oh well, he has a starter job at a tech company, which will probably be for life. That's how that works, right?

glangdale··on Women are nearly half of new gun buyers, study finds
I should note, in the interest of fairness, that about half of the top 10 cities in the USA by homicide rate are in states that could be described as "gun control states". Thus "no pattern". Although, I will reiterate the idea that any city in the continental USA is really not that long a drive away from a state with lax gun laws.
glangdale··on Women are nearly half of new gun buyers, study finds
It's remarkable how little you engaged with what I actually wrote, in your haste to deploy the usual LARPing tropes about freedom fighters and "armed civilians" and such nonsense.

Once again, for the slow kids with reading comprehension problems:

Armed civilians are not a reliable source of resistance against certain types of tyranny - notably right-wing authoritarian tyranny. They are, in fact, more likely to crop up in pro-government paramilitaries who are willing to do the dirty work that a tyranny doesn't want to get around to.

In short, for every gun in the hands of a genuine libertarian, there are 2-3 guns in the hands of hard right authoritarians whose embrace of 'libertarianism' only extends as far as "government letting people exactly like me do whatever they want".

glangdale··on Women are nearly half of new gun buyers, study finds
It's actually advantageous to have a separate "mob" as well as controlling the military/police; the last thing you want is to formally deputize or enlist them.

This way you can go to international events and join the hand-wringing over lynchings, torture and murder ("we can't control our overenthusiastic supporters and of course mistakes were made on both sides, etc"). You can also preserve your control of rule-followers and humanitarians in the police and military.

As a rule, weapons in private hands seem to be handy for overriding the policies of democratically elected governments while complaining about "tyranny" (c.f. Ammon Bundy).

glangdale··on Women are nearly half of new gun buyers, study finds
I always enjoy watching Americans accuse other Americans of living in places "with strict gun laws", as if nearly all of the US isn't a moderate drive away from a place with lax gun laws.

It's also fairly absurd to refer to places with few gun laws as "much safer". The only way of making that trick work is to compare big cities in areas with stronger gun control with rural areas and small towns elsewhere. The worst US cities for murder rates show no real pattern of being in anti-gun jurisdictions (or the opposite).

St Louis has the highest homicide rate in the US. You figure Missouri is anti-gun? New Orleans, Kansas City, Memphis and Las Vegas also make the top by murder rates.

glangdale··on Women are nearly half of new gun buyers, study finds
The thing I always find astonishing about the LARPing element of "taking up arms against the state" is how one-sidedly political it is. An authoritarian right-wing state - say, an alternate world where Trump had somehow seized power and suspended democracy - would find large numbers of American gun owners who would be very happy about this and serve as a pro-government paramilitary (coordinating the police/military and doing all the Bad Things that the government doesn't want to be seen doing, like human rights abuses and ethnic cleansing).

It shows an astonishing lack of historical awareness to imagine that gun ownership in private hands will be default mobilized "against tyranny" - like tyranny is always going to be some sort of sneering baddies that the public is 95% against. There are obviously some genuine libertarians among gun owners, but there's also a pretty large contingent of authoritarian right wingers (or people who are libertarian only in respect to their own freedoms).

glangdale··on Why doesn’t natural immunity count in the US?
Aside from DUI risk and passive smoking, most of these risks are specific to the person. I can't catch obesity from being in the same room as a obese person, and while public health costs go up from drinkers/smokers/heavy people, I haven't heard of ambulances being gridlocked outside a hospital due to a spike of lifestyle diseases.
glangdale··on Ask HN: Graduating CS soon, how do I find a job?
I shouldn't have made it sound like I was sneering at "web stuff" - just that people I work with who are meant to be doing other things have been regularly dragged into fairly menial "web stuff" (i.e. they're not specialists at that, and they're meant to be spending their time on other things). If you are doing a bunch of random firefighting then you don't get good at what your chosen specialty is (it would be similar to having a bunch of web specialists and making them do sysadmin chores half the time).
glangdale··on Ask HN: Graduating CS soon, how do I find a job?
Agreed. It's hard to hire senior people with experience in anything much. I strongly suspect that our industry has a self-inflicted injury where we make developers - even people who are meant to be doing deep algorithmic stuff - do web stuff and devops and 80 other ancillary things - then we jump up and down because we can't find people with experience at higher level things.

Even at a very senior level, I've been asked to do some ridiculously sysadmin-ish stuff that I'm not good at, and I've seen more junior people loaded up with web/testing/devops/admin work to the point that the majority of their time is no longer being spent on what is ostensibly their job.

glangdale··on Phony diagnoses hide high rates of drugging at nursing homes
Two points.

First of all, a lot of the grandiose claims here about how "if I was demented I would just get myself put down" are coming from tech people who could afford this end of the spectrum.

Second, 24 hour home care - or 'family care plus a certain amount of professional relief' is, while expensive, not much more expensive than medicalized fulltime care in a lot of facilities. You're not paying the overheads of a bunch of salaries and potentially the profit margin.

At the lowest end of the spectrum, things are indeed terrible. Aside from fairly extreme small government people, I think most people would want the state to step in. A lot of the places that are over-medicating their patients are collecting vast amounts of subsidies and pocketing the improved profits that come from getting to run staffing levels as low as they can get away with.

glangdale··on Phony diagnoses hide high rates of drugging at nursing homes
Agreed. I think people change their minds about how bad life would be if you can't function at a high level. A good deal of this is pride.

On Monday I will be an Intel Principal engineer again - but there's some Monday where I may need someone else's help to get the shit off my bum. From the perspective of that next Monday the indignity of the latter one is astonishing and humiliating, but there will be many intermediate stages where I get used to the idea that I am not what I was and to see that life is still quite pleasant.

glangdale··on Phony diagnoses hide high rates of drugging at nursing homes
The drugging in those places is an obscenity. If you're going to do that, you should be intentionally shortening people's lives with opiates, which at least is a lot more pleasant than getting loaded up with anti-psychotics to make people compliant. Unfortunately, the jig is generally up with that - everyone knows what opiates do.

I'm seeing a lot of bullshit on all these threads from people who are doing the whole "fake bravery about how they would just be euthanized" in this situation, and how life isn't worth living if you're demented, etc.

Here's how life goes for a demented person close to me:

A "fulfilling life" with dementia isn't that different from one without. You wake up, someone makes you a nice meal, you read the papers, shuffle around a bit, get helped to take a shower and dress, watch a movie (which you won't really take in, but it's some nice pictures), get a call from your daughter (who either lives down the road or thousands of miles away, depending on what you're remembering today), etc. On some days your carers will take you for a drive to look at the ocean, or out to a dinner.

You might dream that you went on an exciting adventure down the California coast and tell everyone about it; surprising that an 80-year-old man who can't walk unaided can take a sailboat out by himself. Everyone you talk to will enjoy your story and no-one will try to tell you that you are deluded.

Some days will be better than others, which is the same for anyone.

That's 24 hour home care - and it's not cheap. Neither is living at a facility at this stage. Often the money spent on medical heroics would be better spent on quality of life. Of course, packing folks off to a facility is necessary for some (in this case, we're talking about outright home ownership plus a modest level of wealth). And then a lot of other people frankly want to put hands on grandpa/grandma's money and/or house which is tricky if they are still living in it.

glangdale··on Phony diagnoses hide high rates of drugging at nursing homes
Speaking from experience, people are not nearly as enthusiastic to make the decision to kill themselves (as dementia sets in) as every likes to imagine that they will be themselves. I have seen multiple cases of people who - in healthy middle age - made bold statements ("just shoot me when I get like that"), who later went on to write documents insisting that "all measures be taken to preserve my life". So not even "just let me die peacefully when it's my time", much less "load me up on opiates and finish me off in a month when I don't know who/where I am anymore".

I don't know why this is, but dementia is often a continuum, and the decision making that people make seems to change in early dementia.

Unless you are proposing that others get to make this executive decision that people with dementias lives aren't worth living, this solution is not practical.

More practical would be to accept that we are far better off over-medicating with pain meds to "keep people comfortable" with the implicit idea that it will shorten their lives. Everyone can pretend we're doing it for pain. Effectively it's like what you said but with some face-saving bullshit where no-one has to admit we're killing grandpa/grandma over the course of a few months. I'd much rather a few months on opiates at the end of my life than a few years on Haldol (JFC). Ugh.

glangdale··on “Nihilism is OK” is not OK
Seriously. I couldn't stop hearing The Dude from The Big Lebowski ("Yeah? Well, you know, that's just like uh, your opinion, man") while I was reading this. An astonishing amount of the heavy lifting in this article comes from assertions that we're just mean to accept, even when they aren't even explained, much less proved:

"Unfortunately, existentialism is difficult to maintain, because mostly we can’t choose meanings. Some meanings are nearly impossible to either alter or overlook, and so are some meaninglessnesses."

WUT

glangdale··on The Problems with Deliberate Practice (2020)
Thanks, good thoughts there. I'm off BJJ for a while (lockdowns in Australia), and won't be getting to decide how classes go for a long time if ever, but it's good to hear people thinking about it rather than blindly replicating "what I saw in Brazil once".
glangdale··on The Problems with Deliberate Practice (2020)
I like this approach. I always wanted more positional sparring and less free sparring. I've noticed that high pressure "expected to win" situation (which, frankly, I love in mount/side control as I've got a better game, relative to expectation of a submission, while my back control/submissions game underperforms).

I've noticed the defender from 'bad positions' does better in specific training than they do in sparring in the same position, probably because they can put everything into getting the "W" - while the attacker in sparring would probably be cooking them or switching to a different position.

There are still weird ego-driven things going on here. My least favorite: being on bottom open guard and having people disengage and run around to win - it's like, bro, if we weren't positional sparring I can just stand up and we can see who has better takedowns. Personally I think this one should begin with at least one grip to avoid time-wasting.

Do you do the "5 minutes on top / 5 minutes on bottom" approach? Or alternating? Or "3-5 people down with a line of people coming in". I think each of them has good/bad elements (without "3 wins and you're out", the latter tends to reduce to the top 3 students annihilating the newer people and not facing each other :-) ).

glangdale··on The Problems with Deliberate Practice (2020)
Not a black belt myself, but agree with GP. Sparring frequently tends to be "people trying to murder each other on the mats in pursuit of their egos". There's a level of dynamism in good sparring that can immediately get shut down if one or both practioners are overly motivated by "winning". You need to take a lot of risks to learn in practice even if it means getting clowned by someone who you could beat if you had to, but you do see people sticking to conservative games in order to dominate. Which often doesn't teach them (as they aren't learning new aspects) but it also doesn't teach their opponent (as it's very hard to learn new stuff when trapped in someone's well-rehearsed "A Game").

The advantage of positional sparring (i.e. go from one position and stop when someone gets out of it to some outcome - not necessarily a tap) is that everyone gets to sharpen a single position. The ego motivation isn't as strong there as it is with full sparring.

glangdale··on RISC-V Scalar Cryptography Extension reaches public review
Maybe you're right, although 'Stockholm syndrome' feels a bit strong.

Agreed that stuff like CLZ/popcount/rotate etc should not be tucked away in a weird extension. I'm a bit alarmed at RISC-V's tendency to fragment into a gazillion subsets (the argument is that it's better to at least define the subsets rather than have a nightmare of incompatible and overlapping extensions, but still).

glangdale··on It's Time for Operating Systems to Rediscover Hardware
I feel you are focusing on the literal aspect of what I wrote and ignoring the broader implications. The tradeoff here between "student OS kernel principles" and "grind through a USB driver implementation" will recur as you try to add drivers and hardware to run meaningful-to-users workloads.

So you wind up with a research operating system that has finger-painting graphics, can't talk to modern SSDs or networks cards... or you somehow conjure up the resources to build an enormous number of drivers.

If you "hack Linux or a bsd" - or even more timidly - use ebpf - the constraints of what you build are going to result in you basically building "more UNIX stuff". This isn't bad research, but the sheer mass of constraints you're taking on if you accept all the design tradeoffs of Linux/bsd/eBPF/whatever means you're certainly not doing the sort of OS research that the original author talked about.

glangdale··on RISC-V Scalar Cryptography Extension reaches public review
Pretty much every instruction that doesn't start with the name of some known crypto algorithm (and maybe some that do) are useful for general-purpose stuff. I've had a good deal of success making Intel's GFNI do "weird off-label things" (bit-matrix transpose and a lot of the missing byte shift/rotate operations just scratches the surface). CLMUL is a good one for all sorts of things, as it can be used for XOR-parallel-prefix (we used it to detect quote pairs in simdjson).

I don't know whether I resent crypto because it gets the cool instructions at low latency because it's so important, or whether I love it due to the fact that even the "leavings at the crypto table" are computationally useful.

glangdale··on It's Time for Operating Systems to Rediscover Hardware
No. Driver work is painful but it mostly teaches you about the h/w in question; you don't learn deep principles from a lot of it. You just grind away, spec in hand, until stuff works. A lot of this winds up being "one damn thing after another". There's a good deal of craft to it, and I don't mean to deprecate it, but it's not broadly applicable.

Conversely, the work they did in kernels has a huge amount of transfer to thinking about concurrency, which is truly valuable - and deep. It also meant that they acquired a far less magical idea of where things like processes and threads come from, having actually built the mechanisms to make them happen (you build a thread library in a warm-up project).

As for microkernels - they often punt a lot of the hard concurrency stuff up to user space servers, which will wind up needing exactly the kind of concurrency that the students learned to build.

glangdale··on It's Time for Operating Systems to Rediscover Hardware
It was a fantastic experience, although TA'ing it was daunting. Like the apocryphal Ledru-Rollin quote: "There go the people. I must follow them, for I am their leader", it was frankly hard to keep up with all the different approaches students conjured up.

I believe that the course number changed at some point - or some shallow aspect of the course was set to change - and a bunch of industry people called the university to express concern. It is/was (I can't speak for the current iteration of the course, although I have no indication that quality has slipper or anything) truly transformative.

I offered to run a 'franchise' of the course at the University of Sydney a while back and was informed that anything quite that transformative wasn't really an option; our job was at least in part to pass engineering students who didn't care that much about computing.

glangdale··on It's Time for Operating Systems to Rediscover Hardware
A perennial complaint; Rob Pike had a trollier but similar version years back. I'm not nearly as optimistic that we have the resources to cope with the monstrous complexity of hardware now.

An anecdote I like to share is that while I was TA'ing CMU's 15-410 operating systems course (a great course - build a preemptive multitasking operating system from scratch) was that the student projects could run on real - if old - hardware. There was a PC that could boot your stuff if you put it on a disk.

This PC had to have PS/2 interface to the keyboard, though. A newer PC would be all USB. Apparently, the complexity of the codebase to talk to the USB device was around the similar level of complexity to the entire preemptive multitasking operating system the students were building (and, of course, considerably less educational). I mean, this wasn't a full-functioned OS, but it allowed multitasking and preemption of kernel calls, so seriously non-trivial.

Multiply this story by all the devices a reasonable OS would be expected to talk to and it's a scary prospect for the OS researcher... and if you don't have those devices up and running, good luck supporting most workloads anyone cares about.

glangdale··on I believe California is the dumping ground for America's homeless problem
This seems naive. I'm sure the people looking for SROs or other low income choices are not the people buying $1M houses, but they probably have a good deal of overlap with people who are buying houses that are at the lowest rung on the ladder, who themselves have overlap with the next cohort, etc.

It's a market. Changing it in a major way at the bottom will have knock-on effects through the whole market.

Of course, more directly, people with $1M houses might also be highly motivated to avoid "the wrong kind of person" being able to afford housing any where near them, their schools, etc.

glangdale··on Home Lab Hardware Guide
Agreed. The startup I used to work for (acquired by Intel) both before and after acquisition had a good deal of lab hardware. Between evaluation boards and rack mount hardware our server room was insanely loud - like "it's 25M away and behind a closed door, but you can still tell when the lunchtime QA and benchmarking runs kick off".

There's no way I could live with that in my house, or even a fraction of it.

glangdale··on AstraZeneca Scientist Says Delta Variant Makes Herd Immunity Impossible
So, all you have to do is experience a 15x worse outcome in terms of deaths and illness early, in order to have a delta-bump that's half the size (per capita) now. I guess that passes for "enlightenment", at least among the people who are disposed to find that anything Sweden has done is wonderful.
← PreviousPage 2 of 34Next →