Hacker News new | past | comments | ask | show | jobs | submit | dahart's comments login

> the total fraud of academics with their names on 20+ papers a year

What fraud? It’s normal for academic advisors to at least be last author, and everyone knows that. And why shouldn’t they, if they helped fund the research, guided the topic, pointed at references, contributed to the research, edited the paper and presentation, etc., etc.? I was more than happy to put my advisor on my first paper after only the first couple of hours of his work, as he did more to make it acceptable for publication than I did in a month. And he did a lot more than that.

Also, some people are legitimately prolific enough to write a paper every 2-3 weeks. Not me, but I’ve seen it.

Publication rate alone doesn’t reflect on quality nor suggest fraud.


I wrote a series of long replies, but then read this blog that says things that I think quite well.

https://blogs.lse.ac.uk/impactofsocialsciences/2025/02/17/do...

I am afraid that there is, likely, a correlation between publication rate and fraud. I agree that a very high publication rate doesn't necessarily mean fraud, but I am afraid that it does cast suspicion, in my mind, on the totality of the output of the author.

In my field I do know some legitimately high output authors. I know a lot of authors who think that they are legitimately high output, when in fact they are simply gaming the system. The sad thing is that they don't know better. I know a lot of people who believe that they have no option but to go with the flow as well - but know perfectly well that they are acting badly.

This is not just a question of academic morals. There are children who today will receive medicines that have no value, and may harm them, because of this practice. There are lines of research that will lead no where and produce no value that are being funded because of this practice. There are lines of research that would provide significant societal benefit that are not being funded because of this practice.


You’re jumping to some pretty big and possibly unsupported conclusions about citation gaming, which is what that blog post is about. People gaming citations are likely trying to fly under the radar and don’t want to draw public attention. Maybe also pay attention to what fields these are in too; gaming in social science might be less impactful than gaming in medicine. One of the examples in the blog post you cited was ghost authors in medical journals which means people who contributed but were not listed; this is almost the opposite problem of what you’re worrying about and it does not amount to bad science.

Medicines aren’t created from the results of a single paper, especially an obscure one with unexplained obscure citations. There are checks and balances. Medicines go through trials which don’t depend on citations. We’ve had ineffective medicines in the past, and it’s happened for other reasons. Notably, consider that the portion of ineffective and actively harmful medications were dramatically higher 50 and 100 years ago than today. If you’re worried about the effectiveness of medicines, then spend your limited time worrying about the anti-vax crowd. They are doing far more damage than people gaming academic citations.

There will always be lines of research that lead nowhere, that’s an inherent feature of the system. Experimental research into unknown topics carries risk, and it should, otherwise it’s not research. If we knew the answer, then we wouldn’t need research.

For the same reason, there will also always be lines of research that don’t get funded. Citation gaming might have a small effect, but there are dozens of other ways human behavior affects what gets funded. And things that work tend to attract people that feel strongly and tend to attract research, so citation gaming doesn’t necessarily lead to strong research getting pushed out.

Gaming of papers is definitely a problem for academics and their careers, and it’s a problem that does need to be fixed, but it’s premature to think the sky is falling. Good science isn’t ending just because some people do bad or mediocre science.


> Edit: children have helped point out the precise notion of ethics - no exploit is unethical on its face, so long as it is disclosed

(Edit: I know I goofed here) Calling people children for correcting your assumptions breaks HN guidelines, please review: https://news.ycombinator.com/newsguidelines.html

Whether video game exploits are ethical was never a question.


Children comments to theirs, not childish commenters. For what it’s worth, I thought the same thing on first read before I realized another interpretation.

Sorry! You’re right. My bad assumption, I’m embarrassed. Apologies to parent @szvsw.

FWIW, I completely misinterpreted their comment too. I thought they were talking about actual young people and the intersection of innocence and ethics or something. Never would have connected it to "child comments."

> Calling people children for correcting your assumptions

I'm pretty sure that's an abbreviation for “children comments”.


Yep, my bad.

What is interesting or unique about honesty as a shared norm?

You are, again here, despite trying to deny it elsewhere in this thread, claiming the “anything goes” part within the bounds of the game and within the rules of the competition is somehow reflecting on the community norms and making this “interesting”.

It’s like trying to say “Oh wow it’s so interesting that marathon runners have their own ethics; they can just run as fast as they want as long as they don’t take shortcuts”.


> they can just run as fast as they want as long as they don’t take shortcuts

In fact that is not how marathon running works. If you take certain drugs that is frowned uppon, even if you are honest about it. Similarly you can’t break the legs of the competitors. That is heavily frowned uppon even if you are honest about it.

These are not obvious. There could be a full-doping version of marathon running but there isn’t. There could be a combined martial art&marathon running competition where you try to hurt your opponent without getting hurt while running a marathon distance but there isn’t.


Frowned upon? Not obvious? What are you talking about? This is straw man. Doping is against marathon rules and assault is both against the rules and also straight up illegal. Maybe you should read some marathon rules https://www.baa.org/sites/default/files/2021-08/Boston_Marat...

There could be a doping-allowed marathon, indeed, if anyone actually wanted that. And that still wouldn’t demonstrate any unique community norms or change my point at all.


> Doping is against marathon rules

Yes. That is what I'm saying. Thank you for repeating it.

> assault is both against the rules and also straight up illegal.

This is coupled. What is or isn't assault depends on the rules of the sport. What is absolutely a-okay in boxing would be considered assault in other situations.

If there would be thousands and thousands of people with long history who think it is just and right to do a kind of marathon where they stop occasionally and have a kickboxing match on the side of the road then it would not be considered assault to participate in it.

> Maybe you should read some marathon rules

Doesn't appear that we have a disagreement on marathon rules. So why would I?

My point is that in marathons the people running and organising them have very specific ideas about what is or isn't a valid kind of marathon. "Doping-allowed marathon" is not a thing, even if all participants were honest about it. "boxing while distance running" events are not a thing.

The norm "anything goes as long as you are honest about it" has two parts. Part 1 "anything goes". Part 2 "as long as you are honest about it". You think part 2 is the curious thing. What I'm pointing out is that Part 1 is the curious thing. In most kind of competitions people are a lot lot more picky about the circumstances which they feel "okay" to compete under.

> There could be a doping-allowed marathon, indeed, if anyone actually wanted that.

And that is my point. People don't want that. People are repulsed by the idea and feel a strong disdain towards others who would participate or organise a "doping-allowed marathon". If such an event were ever contemplated very likely there would be protest and pressure on the organisers to abandon the plan. Do you agree with that?

In other words in long distance running they do not have a norm of "anything goes as long as you are honest about it". Not because they are not honest, or don't require honesty (part 2). But because they all agree to a sufficient degree that "anything does not go" (part 1). Therefore the curious thing is part 1 in speed running. That they celebrate, and consider to be appropriate, a much wider variation of circumstances than in other competitions.


This discussion is sloppy and failing to define or be careful with ethics and how ethics is distinct from competition rules. Let’s drop the word norms since it’s ambiguous, and be clear about ethics. My point from the top and throughout is that video game exploits don’t raise any ethical questions outside of competition. Your examples do raise some interesting ethical questions on their own, but that’s a different subject than speedrunning, and they don’t actually challenge what I was trying to say.

Outside of fighting sports, assault is default illegal and considered unethical, and there are very few exceptions for it. Importantly, a sport’s rules must make an explicit exception for assault because it’s illegal, otherwise assault is not allowed. Marathon isn’t a fighting sport, and like any sport it does not need to state that assault is not allowed. Marathon’s ethics on assault match society’s ethics on assault, and therefore the marathon does not have unique ethics when it comes to assault.

Boxing is an interesting example, one where the sport’s rules do bend social ethics. Speedrunning is not an interesting example, as glitch exploits and tool assisted speedruns aren’t considered unethical outside of competition - they don’t harm anybody. The only ethical question anyone’s raised so far about speedrunning is whether the speedrunners have followed the competition rules, or whether they’ve cheated.

There is chess boxing, which is perhaps similar to your idea of a boxing while distance running event. People have suggested doping-allowed sports https://www.cnn.com/2023/10/30/sport/enhanced-games-olympics...

Looks like the people in the article objecting are people who make a living doing sports, so perhaps unsurprising. The primary objection to doping is that it’s unfair to competitors. I’m not sure the general public would be repulsed. Whether doping is generally unethical is debatable since outside of competition, the main harm a doper does is to themselves. Some ethicists consider self-harm a type of social harm, but it’s different than directly harming others, and challenging the right to self-harm may raise ethical questions about freedom of choice. But, if you’re right about people considering doping to be generally unethical, then that means it’s in the same bucket as assault, and so does not help support the idea that speedrunning represents unique or flexible ethics.

Anyway, there are no competitions where anything goes, they all have rules. Most of the rules aren’t based on ethics, they’re just rules. When rules do bend society’s ethics, I agree it can be interesting. When rules don’t bend normal ethics, then we’re only talking about a different competition and not a case of flexible ethics. I think video game speed running falls in that latter group, and I haven’t yet seen any examples or reasoning here that demonstrate otherwise.


> This discussion is sloppy and failing to define or be careful with ethics and how ethics is distinct from competition rules

Because I think what we are talking about here is not ethics. Do note that I haven't used that term at all. What I'm talking about is the participants shared sense of "what goes" and "what doesn't". What I'm talking about is norms. Norms of the activity. vanderZwan uses the word "ethics" in quotes, but from the context, and the quotes, I think it is clear that it's not about literal ethics.

> Anyway, there are no competitions where anything goes, they all have rules.

Of course. The observation is about what kind of things are okay according to the shared sense of the participants. In some activities this space is very narrow, in some it ends up much more expansive. In my observation it seems speedrunning has a more expansive notion of "a-okay rules" than what I have experienced in other activities. I find this interesting. You don't find this interesting.

It sounds like you want to litigate this and convince those who find this interesting that it is not interesting. That's a fool's errand.


> Because I think what we are talking about here is not ethics.

This is why I said your argument is a straw man in relation to mine. I am and always was talking about ethics, because that’s what the top comment said. The top comment used the word “ethics”. You jumped into a conversation about ethics and without being clear until now and tried changing topics to the ill-defined “norms”, which is why we have some miscommunication. I assumed incorrectly that your use of norms meant ethics, because we were talking about ethics from the start.

> It sounds like you want to litigate this and convince those who find this interesting that it is not interesting. That's a fool's errand

SMH You’re here debating too, I didn’t ask for you to misinterpret and try to contradict me. I’m not trying to convince you about what you find interesting, I’m pointing out that speedrunning is not interesting from an ethical point of view. I don’t think it’s particularly interesting from a norms point of view either, but you do you. I don’t feel like you’ve even demonstrated that speedrunning competition rules represent norms at all, your argument is making more unstated assumptions.


As a disinterested third party I’m not sure what your point is. Honesty isn’t unique? I don’t think anyone is saying otherwise.

You seem to be proving the other poster’s point that people interested in marathons landed on a class of rules that is specific and that nobody wants the alternatives.

It’s the difference between that and the speed-running community’s “anything goes, we’ll just classify it” rules that people find are finding interesting.


I don’t see how you’ve arrived at marathons having any specific norms. No doping and no assault is pretty universal. Can you explain it better?

The point was that speed runners are not operating with different ethics than any other competition, or society in general. “Anything goes” isn’t accurate as a description, and the top poster maybe didn’t realize that suggesting that exploiting game glitches somehow represented flexible ethics might be a bit presumptuous and patronizing, and also incorrect. It’s interesting to me that people would assume otherwise.


> I don’t see how you’ve arrived at marathons having any specific norms. No doping and no assault is pretty universal. Can you explain it better?

Sure. First, I'm going to leave 'assault' because I think it sort of changes the mapping between marathon running and speedrunning in an analogy-breaking way and adds ethical concerns that aren't really relevant to the point and practically don't exist for speedrunning anyway. I understood what the poster was saying, but I think it's not a great example.

Second, I'm going to suggest that like speedrunning, a marathon run is really against the clock. This isn't strictly true, but it's true enough for the analogy to work.

With all that in place, the apparent universality of 'no doping' is half the point. There are no marathons where you can choose to dope or not dope and they simply categorize you that way. Doping has a fair chance of ruining a career and disqualifying the participant from future marathons.

The other half of the point is that it doesn't have to be this way. We could have separate marathons for dopers, allow them to race together with non-dopers but categorize them differently, or even have specified paths to run but no organized marathon and allow people to get creative in speedrunning them. We could allow people to find shortcuts on trails to improve their time and then categorize them differently from those who follow the proscribed path. Yet we don't do any of it.

In contrast, speedrunning does allow this sort of activity. Tools are okay, but now you're categorized as TAS. You don't have to 100% the game, but now you're doing an Any% run. This does not exist in marathons. You can't take a short cut and get an "Any%" time, for instance.

Simply put, if a speedrunner finds a shortcut it is celebrated, but if a marathon runner happens to find a path down a vine that saves him 10 minutes of running, he's a cheater. These are two different ethical systems.

> The point was that speed runners are not operating with different ethics than any other competition, or society in general.

As you can see above, I disagree with this statement. I haven't read all the sibling and cousin posts, but neither the poster you responded to nor the top poster said the ethical system was bad; in fact both found it interesting. Both the marathon system of ethics and the speedrunning system of ethics seem to have the same moral or ethical basis in fairness, but the approach is different and therefore is the ethical system.

> “Anything goes” isn’t accurate as a description, and the top poster maybe didn’t realize that suggesting that exploiting game glitches somehow represented flexible ethics might be a bit presumptuous and patronizing, and also incorrect.

This is not my interpretation of the top post and I think it's extremely ungenerous. He expressed surprised that a system designed to work around the rules still develops a system of ethical rules. His edit then points out that honesty is a core tenet of the ethics of the group.


The edit was a response to multiple people including me calling out the framing of game exploits as a reflection of ethics at all, and it was an explicit admission of having done that, and perhaps indicated a change of perspective. You are contradicting that post.

I can see why you’re dropping assault now, since it undermined @krisoft’s argument. It was a mistake to defend it. And as for doping, you agree with me that anti-doping rules are universal? Okay so marathoners don’t have different norms than other sports, or at least you’ve run out of examples.

I disagree with your equating of TAS and doping. But that’s mostly irrelevant anyway. There is no ethics of TAS outside of a competition that disallows TAS. The only ethics we’re talking about is the ethics of cheating, which is universal; whether or not the agreed upon rules were broken. It’s the same for speed runners as it is for marathoners.


That’s a wholly inadequate and inappropriate response to my post.

I have only read the edited post and don’t know if in addition to the addendum he also edited the original post. If that is the case then there may be context here I don’t have, so feel free to educate me there, but I stand by my interpretation of what I’ve read here. I think you might be confused and trying to make this personal between us doesn’t convince me otherwise.


I’ve edited my post above, before I saw your reply. I’m not making anything personal here, I’m debating you. If you don’t like that, then don’t state your disagreement with me and invite argument.

BTW, you lobbed an ad-hominem with “extremely ungenerous” which is why you got my retort that was edited out. Maybe you didn’t realize you were already making it personal yourself?

Personally, I think you’re confused about the difference between rules and ethics. Yes, we’re talking about different sets of rules. But we have not been talking about different sets of ethics. The difference between a speedrunner’s shortcut being good and a marathoner’s shortcut being bad is rules, not ethics.


You ignored the entire content of my post to make essentially a personal attack because you think someone recognizing that your interpretation of a post is ungenerous is ad hominem. I wasn’t making an argument. I was telling you what you’re arguing against isn’t what was said or at least wasn’t what was meant.

Then you edited your post completely to essentially attack a point that wasn’t really mine, that you wouldn’t see the way it was actually meant, and that I was already conceding for expediency since you yourself claimed you didn’t understand. I posted to you because regardless of who posts you seem to not understand their point of view and want to “debate” about things they didn’t say or if they did say them they didn’t mean them the way you took them.

You’re doing this in several response not just to me. But I’m sure it’s just because I don’t understand the intersection of rules, ethics, and morals. I’m just confused.

Have a nice life.


Whether someone was honest or lied in competition isn’t a unique set of ethics anywhere. Yes they use ethics, and those ethics are shared with society’s notion of ethics. You’re failing to make a case that speed running has their own norms.

You’re framing this in a very funny way.

This is not at all the same thing as the Astros scandal. Astros didn’t break the agreed upon rules (edit: I’m wrong - human sign stealing was allowed, but electronic sign stealing was not). Groobo did break the agreed upon rules.

There is no “own notion” of ethics here, and it’s presumptuous and incorrect to suggest that exploiting games during a game exploit competition somehow reflects on ethics and makes the cheating “interesting”. Cheating and ethics in a competition is defined by what the agreed upon rules are. Perhaps ironically, your comment is developing it’s own notion of ethics.


> Astros didn’t break the agreed upon rules.

They did. Using electronic devices to communicate during a game (I guess besides the dugout phone) is banned. They were watching the video feed in the clubhouse. https://en.m.wikipedia.org/wiki/Houston_Astros_sign_stealing...


Ah, you’re right, thanks. I read that sign stealing was allowed, and missed the parts about devices being banned.

Runners on 2B are allowed to try to steal signs if they can crack whatever set the pitcher and catcher have switched to (well they use pitch com now); the astros used live video feeds to pick up the signs even with no one on base as well as computer assisted methods to crack the codes, and then related the pitches to the batter via the dugout.

> This is not at all the same thing as the Astros scandal. Astros didn’t break the agreed upon rules.

Sheesh, didn’t expect to re-litigate WS‘17 on HN today… I concede that there was at least some gray area in the codified official rules and the 2001 directive around electronic transmission, but it very clearly was a violation of the norms, IMO, and the Giants-Dodgers binocular incident in the 50s and its reception is good precedent.

> There is no “own notion” of ethics here, and it’s presumptuous and incorrect to suggest that exploiting games during a game exploit competition somehow reflects on ethics and makes the cheating “interesting”.

That’s not what I was suggesting was interesting to me. What interests me is the social phenomenon that is what the actual members of the community deem to be acceptable or in this case unacceptable, how the the community works to identify violations of norms, and then handle with the fallout.


> That’s not what I was suggesting was interesting to me. What interests me is the social phenomenon that is what the actual members of the community deem to be acceptable or in this case unacceptable, how the the community works to identify violations of norms, and then handle with the fallout.

You cited the exploiting as the one and only reason that cheating is interesting here, and explicitly implied that it’s somehow different from other kinds of competition. After this new comment, I don’t see how to interpret your top comment any differently.

This community competition didn’t do anything differently than any other competition. Someone entered the competition claiming to be adhering to the rules (https://kb.speeddemosarchive.com/Rules), they flagrantly broke the competition rules, tried to hide their cheating, and people got upset when it was uncovered. No different than any competition cheating, I don’t see what you’re implying about what people deem to be acceptable, or norms. The norms here are no different than anywhere else.


> You cited the exploiting as the one and only reason that cheating is interesting here

As far as I can tell, I did not… this is what I said:

> it’s very interesting to see how a practice that is fundamentally about exploiting generally unintended mechanisms within a system design still develops its own notion of ethics, legitimate and illegitimate exploits, etc

Removing extra clauses:

> it’s very interesting to see how a practice […] develops its own notion of ethics, legitimate and illegitimate exploits, etc


> Removing extra clauses

Oh, “fundamentally about exploiting” is just a “extra clause”? Hahaha.

Your sentence intentionally juxtaposed game exploits with ethics. If you disagree with that, you’re contradicting yourself and not me.

What’s left over is still the part I’m objecting to. Nobody developed their own notion of ethics, the community has the same competition ethics that practically any competition anywhere uses.


I’m surprised nobody has yet (as of this writing) pointed out that Moore’s Law never claimed anything about single threaded execution or clock rates. Moore’s Law is that the number of transistors doubles every two years, and that trend has continued since 2012.

It looks like maybe the slope changed slightly starting around 2006, but it’s funny because this comment ends complaining that Moore’s Law is too good after claiming it’s dead. Yes, software needs to deal with the transistor count. Yes, parallel architectures fit Moore’s law. The need to go to more parallel and more parallel because of Moore’s Law was predicted, even before 2006. It was a talking point in my undergrad classes in the 90s.

https://upload.wikimedia.org/wikipedia/commons/0/00/Moore%27...


Sadly, Moore’s law is often cited, but never read. It’s only 3 pages.

http://cva.stanford.edu/classes/cs99s/papers/moore-crammingm...

But to further needle, the law is


computer go faster over time

> “The big question is always the mechanism,” Goforth said. “What’s the mechanism and what is the sensor? We know that for the visual sense, you have eyes; for the sense of smell, you have a nose; and for hearing, you have ears. But no receptor like that has been identified for the magnetic sense, and the mechanism remains unknown.”

Interesting, I just barely listened to a Radiolab episode about magnetoreception in birds today [1]. The mechanism is thought to be cryptochrome, a type of protein that exhibits quantum effects when exposed to blue light [2]. The RadioLab story mentioned this mechanism was first proposed in the 70s (I think?) and it was heresy at the time a didn’t get published. Now there’s a SciAm article about it [3], and lots of blog posts. I wonder why this isn’t mentioned? Isn’t there a pretty good chance the mechanism is the same?

* edit: BTW while listening to the podcast I was reminded of Haidinger’s Brush [4] - the fact that many humans can see polarization, once they know what to look for. I wonder if it’s potentially related to navigation since the polarization effect is strongest at 90 degrees from the sun… i.e., generally north and south, in the morning and evening, for much of the earth.

[1] https://radiolab.org/podcast/quantum-birds

[2] https://en.wikipedia.org/wiki/Cryptochrome#Magnetoreception

[3] https://www.scientificamerican.com/article/how-migrating-bir...

[4] https://en.wikipedia.org/wiki/Haidinger's_brush


I’m unconvinced. Working hours in industrial countries have gone down over the last century or two, not up. Things were even worse for the vast majority of people before the 1800s, most people didn’t even get “careers” or have the luxury of thinking in terms of “accomplishments”.

> Working hours in industrial countries have gone down over the last century or two, not up.

Wages stagnated while productivity doubled. What you could afford on one salary 50 years ago is unaffordable for a 2 salaries household, &c.

We peaked sometime between WWII and 1990, since then it's not obvious that things improved in the west


I think waking time away from work, and not work pay, is the most relevant metric to answer the question raised in this sub-thread. (Though there is obviously overlap, what you can do with your free time and how much you have is obviously driven by your income.)

While you have a good point - there is real inequality and it’s still a struggle to be poor, no question - this doesn’t actually contradict my point, which is that people have on average more free time than they used to, and that free time is what I think @v3xro is referring to.

“Productivity” is a loaded word, it means dollars collected, and a productivity increase does not necessarily mean workers worked harder (though it does happen sometimes). While it sucks for workers if companies can collect more dollars without paying more for production, it’s not directly relevant to worker’s lifestyle.

This chart shows adjusted wages increasing at a very slow moderate rate for the last 50 years: https://fred.stlouisfed.org/series/LES1252881600Q. It’s not that bad if wages “stagnate” if we’re talking inflation adjusted purchasing parity, which I think you are. While it might be unfair for capitalists to reap relatively more, that doesn’t actually reflect on the question of what human life is about and whether “accomplishments” at work are the primary meaning of our lives.


What GPUs aren’t IEEE compliant? I don’t think that’s been true for more than a decade.

After a bunch of testing and looking around I think I should actually change my statement. GPUs do offer IEEE floating point compliance by default, but don't strictly adhere to it. Multiple optimizations that can be applied by the driver developers can massively effect the floating point accuracy.

This is all kind of on the assumption that the accuracy of floating point multiplication and division is in the IEEE spec, I was told before that it was but searching now I can't seem to find it one way or the other.

I believe one of the optimizations done by nvidia is to drop f32 variables down to f16 in a shader. Which would technically break the accuracy requirement (as before if it exists). I don't have anything I can offer as proof of that due to NDA sadly though. I will note that most of my testing and work is done in PIX for Windows, and most don't have anti-cheat so they're easy to capture.


What shaders (presumably GLSL & HLSL) do precision wise isn’t an IEEE compliance issue, it’s either a DX/Vulkan spec issue, OR a user compiler settings issue. Dropping compliance is and should be allowed when the code asks for it. This is why GLSL has lowp, mediump, and highp settings. I think all GPUs are IEEE compliant and have been for a long time.

I agree on the dropping compliance when asked for aspect, the problem I'm referring to more is the driver dropping compliance without the game asking for it. If the underlying system can randomly drop compliance when ever it thinks it's fine without telling the user and without the user asking, I would not consider that compliant.

That is fair, if true. But I’m very skeptical any modern drivers are dropping compliance without being asked. The one possibility I could buy is that you ran into a case of someone having dropped precision intentionally in a specific game in order to “optimize” a shader. Otherwise, precision and IEEE compliance is the prerogative of the compiler. The compiler is sometimes in the driver, but it never decides on it’s own what precision to use, it uses either default or explicit precision settings. The only reason it would not produce IEEE compliant code is if it was being asked to.

If someone says branching without qualification, I have to assume it’s incoherent. The branching mechanics might have lower overhead today, but the basic physics of the situation is that throughput on each side of the branch is reduced to the percentage of active threads. If both sides of a branch are taken, and both sides are the same instruction length, the average perf over both sides is at least cut in half. This is why the belief that branches are slow on GPUs is both persistent and true. And this is why it’s worth trying harder to reformulate the problem without branching, if possible.

Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: