Algorithms + Data Structures = Programs - Episode 243: Learning Languages

Episode Date: July 18, 2025

In this episode, Conor and Bryce chat about language learning apps, recent C++/CUDA/Python meetups and more!Link to Episode 243 on WebsiteDiscuss this episode, leave a comment, or ask a question (on G...itHub)SocialsADSP: The Podcast: TwitterConor Hoekstra: Twitter | BlueSky | MastodonBryce Adelstein Lelbach: TwitterShow NotesDate Generated: 2025-07-01Date Released: 2025-07-18MondlyduolingoBabbelADSP Episode 213: NumPy & Summed-Area TablesADSP Episode 227: Re: The CUDA C++ Developer’s ToolboxIntro Song InfoMiss You by Sarah Jansen https://soundcloud.com/sarahjansenmusicCreative Commons — Attribution 3.0 Unported — CC BY 3.0Free Download / Stream: http://bit.ly/l-miss-youMusic promoted by Audio Library https://youtu.be/iYYxnasvfx8

Transcript
Discussion (0)
Starting point is 00:00:00 London for the London Cuda meetup and Pi Data London, then Budapest for a day for the Budapest Python meetup, then Hamburg for ISC and the Hamburg C++ meetup, and then Ramona and her 14-year-old niece joined me there. Welcome to ADSP the podcast episode 243 recorded on July 1st, 2025. My name is Connor and today with my co-host Bryce, we chat about learning languages. We recap some recent C++, CUDA and Python meetups and more. Say something. Hello. Hello. All right. Shush. Am I saying that right?
Starting point is 00:00:47 I've been learning Polish. I've been learning Polish with the with the apps and that is not a word. Yeah, it is. Yeah. Oh, it's chest. There we go. Shush. I love this.
Starting point is 00:00:55 So so really you're recording. Yeah. I'm recording. I'm recording. I'm recording. I'm recording. I'm recording. I'm recording.
Starting point is 00:01:03 I'm recording. I'm recording. I'm recording. I'm recording. I'm chest. There we go. I love this. So, so reading from the recording. I am. Yeah. Okay. Okay. Good. Reading from the transcript, the AI transcript, it's S H I S H as what I thought I said. And then it says Connor. What is chic? Chic.
Starting point is 00:01:27 And then Connor says, oh, it's chest. And then Bryce, chest. There we go, chest. Well I'm probably, I haven't, I mean I did like a ton of Polish at one point. Yeah, I remember. Chest, it's one of the harder things to say actually because it's like CZESC or something like that. Yeah, it's got two, it's got like three of the letters that we don't have. Yeah. Anyways, it's definitely not sheish though, I can tell you that. I even practiced before we got on the call and you can see how that
Starting point is 00:02:05 went. An easier one and more impressive in my opinion is Shepresham which I believe is I'm sorry or excuse me so if you're like walking around you just say oh Shepresham. I'll have to remember that one because you know I'm a very clumsy person, so I'm constantly bumping into my girlfriend around the apartment. Yeah, to take a small little tangent, even though we have lots of more exciting things to talk about, I have such a bone to pick with all these language learning apps. Yeah, I was just about to ask, the reason I tried to speak Polish to you was I was going to ask for your advice on what's the best way to learn because I tried Duolingo and
Starting point is 00:02:52 it immediately bothered me because I like I, when I tried Duolingo I just put in that like I have no knowledge because like I know a couple of Polish words but like I just wanted to start from like the most basics. And so it started like teaching me, but it doesn't explain anything. It doesn't explain conjugation. It doesn't like, it just does everything through examples. And you're supposed to learn through inference, which is, maybe that's the way, but I wanted something
Starting point is 00:03:20 where it's going to explain to me, these are the different forms and that's how you do like conjugation and something like that. And also I'm terrible at pronunciation. I want an app where it's going to have where like for me to make progress, I have to speak the words correctly and it will grade how well I've spoken them. So Duolingo does do that at some point. It just gets introduced a
Starting point is 00:03:45 little bit later. And so yeah, for I mean for a background I mean I used to speak Chinese very well like back in 2012 and I spoke a little bit of Polish at one point. I've also so I've done Chinese on Duolingo. I've done Polish. I've and sometimes I've messed around like learning a little bit of Spanish or Dutch or Japanese. But my and so they they do do speaking they do do the You know spelling if you want kind of but you can kind of toggle that in the settings my main issue Though is the words that they start you with like I'm trying to think what the most recent one I think it was I
Starting point is 00:04:20 Redownloaded it and started learning a bit of Japanese and like, they don't, in the first like seven or ten lessons you don't learn yes or no or thank you or like, and like I, usually you want it for conversational purposes or like, you know, it's like, and in the first seven or ten words I'm learning doctor and lawyer. And I'm just like, I don't like, I don't need like the first things you should learn are like, yes, no, excuse me, I'm sorry. And then it's like, like, even they even asked, like, what are you using this for? And it's like, oh, usually what what people choose is like travel or like basic conversation. And I understand that they need to introduce like a couple nouns like Apple or something like that. But I'm always shocked like how far you get into it before they start like introducing you like in numbers is also very important. Like a lot of times when you're ordering something you want to be able to say like one two or anyway. So I know Dua Dua Lingo is very successful and I should mention to I also am actively learning Persian, but Duolingo doesn't support Persian. Interesting.
Starting point is 00:05:26 So I have to use a different app called Mondly. And I've tried a couple others. All of them are terrible. Duolingo is the best. But I think this space is ripe for disruption with like some LLM where basically you can just say like, this is, I want to have this kind of conversation and teach me these kinds of words. And you're definitely correct. They never, like in in Mondly,
Starting point is 00:05:50 with all the Persian that I do daily, it never teaches you the conjugations. Basically like luckily my partner is Persian. And so I can just ask. And a lot of times when I'm like hanging out with your family, I'll like take the opportunity to, and the pronunciations that the app gives you are very very formal and so
Starting point is 00:06:08 they always give me a heart and they're like ah that's not how you pronounce house it's who then that honey and I'm like well I mean the lady on the app says honey so like what am I supposed to know and anyways so it's just like there is a massive opportunity here for some app to come in like Duolingoolingo is amazing at the gamification, but in my opinion, they're actually quite terrible at getting you in a place where you understand, like you said, conjugations. And also, you have your core set of vocabulary. The path, I'm sure they've done studies, but I guarantee you they're optimizing for retention, not actually...
Starting point is 00:06:44 Yeah, yeah. Well, I have a theory about so it's interesting. I felt the same way because like for the Polish, the first couple words it taught is like man, woman, boy, girl, and then. Oh, yeah. And those are very hard. Jivchenka like for girl is. It's not a simple word. And then they also teach you the word for eating, because the like for girl is it's not a simple word and then they also teach you the word
Starting point is 00:07:06 for eating because the word for eating is very similar to the word for my theory is that the reason that they teach you the word for eating is that it's very similar to how you say am and so you get if you didn't get taught what the word for eating meant early on, you would get confused. Because it's like, yeah, right? For like, is eating. And then that's very similar to how you would say, like, yes, them, which is like, I am. Oh, yes, them, yeah. Yeah, yeah.
Starting point is 00:07:39 And then they, but then they also teach you like, you know, like milk, bread, apple. I think apple man, boy, woman, girl were the first five words. I did find it a little weird. One thing that occurred to me is that perhaps they start off with concrete nouns because they have this pictorial element where they show you the sentence like you know in the sentence like you're gonna translate is like I am a boy and it's got a little cute like pictograph of a little boy and then for like I am a woman you know they've got a cute little pictograph of a woman and maybe they did some research and they determined that you if if like you make a visual connection you're gonna learn better.
Starting point is 00:08:26 But I tried another app called Babble. And I- Yeah, try that one as well. The first thing that Babble taught you was how to say yes and no. And I liked Babble better because I felt like it did a better job of explaining things. It explained that like, nay,
Starting point is 00:08:44 or I'm sure I'm pronouncing that wrong. It's how you say no, but then like it explains that like, it can mean no and also not, and that like it's used for negation, and it did a good job of explaining the differences. And then it also started explaining some of the different sounds in Polish like, zh. And it like made analogies to like this sound
Starting point is 00:09:04 for like the Polish letter Polish letter RZ. It made the analogy like this sounds like how you'd say the S in pleasure. And I was doing this particular one with my girlfriend and she was like, oh, that's very clever. I've never thought of explaining it to people like that. But that's exactly how I would think of it. That's what really good language teachers, like when I was learning Chinese,
Starting point is 00:09:26 that was one of the things like the R, they said exactly the same thing, like the shir, which is like a really soft R. That's how words that start with R in Chinese, a lot of the times, like I think urban is Japan, and it's like, it's one of the harder words to pronounce. And so they do tricks like that. So like, that's actually what really good
Starting point is 00:09:46 like language tutors and language teachers do. They give you some analogy that ties back to what you know how to say in your own language, right? Yeah. But it's funny, because I've always thought I've been like really bad at languages. And the thing that finally inspired me to do this was we were at the C++ committee meeting in bulk area.
Starting point is 00:10:04 And before that, we were at ISC in Hamburg. I've been to eight countries in the past month for this CUDA tour of a year. Surprise, surprise. Yeah, so the whole itinerary was London for the London CUDA meetup and Pi Data London, then Budapest for a day for the Budapest Python Meetup, then Hamburg for ISC and the Hamburg C++ Meetup.
Starting point is 00:10:30 And then Ramona and her 14-year-old niece joined me there. And her niece, she speaks some English, but she's not fluent yet. And so if her and I are talking one-on-one, we're using the phone, we're using Google Translate on the phone. And then we went from Hamburg to the committee meeting in Bulgaria. And then there was another meetup there in Sofia. It was sponsored by NVIDIA, although I didn't actually speak it. My colleague, Gunzel, ended up speaking at it instead. And then I was in Turkey, in Madrid. But I was just like, every time I run into Ramona's family, most of them just speak Polish.
Starting point is 00:11:18 And it's like you're sitting and having lunch, and everyone else talking in Polish, and you're not going to be able to keep up. So I was like, all right, finally it's time. After three years, I need to start learning. But I always had this perception that it was just going to be impossible for me to learn other languages because I'm really bad at pronunciation. And I took eight or 10 years of Spanish from elementary through middle school through high school.
Starting point is 00:11:48 And I've retained like no Spanish. And then I took Latin in college because you had to have a language requirement. And my theory behind Latin was it's a dead language. They can't make you speak a dead language and I'm bad at pronunciation. But it turns out I was wrong. They didn't make us speak Latin in class. But I also retained like none of that. It's probably I mean it sounds like you do have a history of struggling but it is it is the apps are not helping you in my opinion. I I think that there's an opportunity for
Starting point is 00:12:19 an app out there that is really like kind of immerses you in what it's like to just go to a country and having to start speaking. Like it is like I studied Chinese for like seven months before I moved there. I think I learned more in the first week having moved to China than I did in the seven months leading up to it because it's like you know one two three hours a week like and and that's the only time and then sure you're studying a tiny bit on the side versus you go there and every single interaction, you basically like, sure, like almost everyone
Starting point is 00:12:52 in China speaks English, but they don't expect to switch to English. Like you definitely run into people that wanna practice their English, but for the most part, when you're going on, you know, subways or out to eat, like that's the language you have to operate in. So very quickly, like you are forced and you'd be surprised how fast once you
Starting point is 00:13:09 start to do that it starts to reinforce all the things that you kind of studied in your head but it like it's just like it's skyrockets and like Duolingo is like a terrible version of basically like doing that one to two three hours a week like my joke I always used to say when I was learning Polish was, ja lubię kaczka, which means I like ducks. Because I knew ducks so quickly and I was like, why do I know the word? And they teach you all these animals.
Starting point is 00:13:34 I can't remember elephant, but I used to say I like pink elephants or whatever it was because the set of words they would teach you and I wouldn't know how to order something, I wouldn't know how to... And then at some point, it taught because like the set of words they would teach you and like I wouldn't know how to order something. I wouldn't know how to and then at some point, you know, it taught you like, oh, I'm sorry. And I was like, oh, like an actually very useful thing that you will say on like the daily basis. And like at one point, they also teach you credit card. And I was like, okay, like another very useful thing. Like when you're in a store there,
Starting point is 00:14:00 they always ask, like, do you want cash or credit card? And usually, like as a foreigner, you just You just like point to your card and it's very it's very like embarrassing. You're just like yes. I don't live here I have no idea what you're saying I'll just point at this thing, but if you know how to say credit card now It's like you say credit card you say thank you and then poof you're done. It's like you're local So I know about I know a collection of random Polish words That for amusing reasons. First of all, most importantly, lodi, which is ice cream. I'm very troubled by this.
Starting point is 00:14:34 According to ChatGPT, Google Translate, and my girlfriend, there is no distinct word in Polish for gelato, which I think is like a food crime. Your language needs to have distinct words for gelato and ice cream. But I know Lodi. And then I know a variety of names or nicknames that Ramona has had throughout the years. So her family used to call her Brudronka, which is like the Polish word for ladybug, which kind of sounds like Ramonka, which is the like casual like family version of Ramona's name. So they would call her like Ramona Bajronka. And then I call
Starting point is 00:15:14 her sometimes Ribka, which is like the diminutive of fish, because she just drinks like a ton of water. So I'm like, you're my fish. And she says that like in Polish, like people, like couples nicknames for each other will be to like call each other something like ribca, like which in English would be like, you know, like, I don't know, it is to fish what like doggy is to dog. But yeah, and it was interesting what you're saying about the immersion part of it because Ramona's niece, she came to visit us two years ago in New York and I noticed that her English improved when she was here for the trip. And when she met up with us in Hamburg, the first like three to four days, she was, like her English was rustier.
Starting point is 00:16:09 Maybe it was just a confidence thing. I always feel like she understands more than she's willing to speak because she can understand a bit, but she doesn't feel comfortable speaking. But by the end of the trip, I could tell that she had, like her English apprehension was much better. And I don't think it was that like, yeah, she learned a lot of things just like chatting, like talking with us day to day for seven or eight days. But also it was just like being immersed.
Starting point is 00:16:37 I'm sure there are a lot of things that she already knew, you know, sort of came, came to the front of her head. Yeah, it's it, it's very common. Like I don't typically break out into a Persian when I'm with Shima's family. But a lot of the times, like I can pick up on not full sentences, but I'll hear like one or two words, and then you can infer the rest. And so like every once in a while, like, you know, I'll say something and then they'll be like, like, her mom's always
Starting point is 00:17:03 like, Oh, it's not safe to talk around him anymore. Like, he's learning too quickly. Yeah, yeah. I mean, I definitely feel like even with the little bit that I know, I can suss out a little bit. And it's interesting. It was interesting being in Spain for a couple of days, because it reminded me that I do actually
Starting point is 00:17:21 know a little bit more Spanish than I think. So what do you recommend for learning, if not the apps? I was thinking to see whether Ramona's niece would want to hop on WhatsApp with me every now and then, and she could practice her English and I could practice my Polish, because I know that's something that I've had colleagues in the past, I had a colleague back at LSU who would wake up at like 5 a.m. and go into the office to Skype his friend in China to for them to mutually practice their Chinese and English. Yeah, I would 100% recommend, if you can do that, recommend that.
Starting point is 00:18:00 I mean, when I was, I mean, my best recommendation is move to the country where they speak the language It just skyrockets and I actually when I was in China I built a program in Excel that basically was a compilation of all the vocab that we would get in Class and then I had two different modes one where it would feed me the English and then I would have to say out loud the the Chinese pronunciation, and then the other one was where it would give me the Chinese character, and then I'd have to do the same thing,
Starting point is 00:18:31 and the way it would work is, after I would answer it out loud, I then click a button to see if I'm right, and then I could click, like, did I get the pronunciation right or everything right? And if I got everything right, it would put it on some list where it wouldn't show me anymore. And if I got everything right, it would put it on some list where it wouldn't show me that anymore.
Starting point is 00:18:46 And if I got it wrong, it would go back in the list. And so over time, you know, there's a couple thousand words in there. It's only showing me the ones that I have trouble remembering. And that, that is like a, that's specifically like, it's a type of memorization technique that like they put some fancy name on, but it's just like a, it's like a, what do you call it? Very intuitive way, like if you have flashcards and every single time you get a flashcard and you always know it, you put it in a pile
Starting point is 00:19:10 and you don't look at it anymore. Yeah, that's what I used to do when I was studying in school. The second part of school, the two years where I tried in college is I would do either flashcards on the computer with some new flashcard app. This was back when I used Linux. Or I would actually write out physical flashcards. And I'd do the same thing.
Starting point is 00:19:30 But what I would do is I would go through the entire set three times. And when I was going through, each time I would go through the set of flashcards, I would, once I got the flashcard right, I would set it aside. And then I would continue going through the ones that I missed. And I would just do that three times. And one technique that I used, which I still use when I'm giving a talk, is I try to go
Starting point is 00:19:54 over the talk the night before I'm going to give it. Because I find, or I would do the flashcards right before I go to sleep the night before the test because I just find that doing the like practicing the material the night before it helps me retain it better like doing it right. I sleep. Yeah. I think there's some science behind that. I don't know. Yeah. Yeah. There's there's definitely like research. So I mean that so that's like that's not really built into any of the apps that I know of.
Starting point is 00:20:25 I would say just use whatever app that you like. Even though they're terrible, there's still a good way to pick up vocab or whatever. Are you taking a class for Persian? I had a tutor at one point, but it lasted one session, even though I paid for five, because I'm so socially awkward and I couldn't ask for what I wanted. So it was virtual. I mean it was virtual
Starting point is 00:20:47 but I didn't want to learn how to write and she was like forcing me to write and I kind of was I tried to like Articulate like can we just skip this part? I just want to learn how to speak and She was like, yes. Yes, and then just ignored what I said and and so that was that but I don't I don't really practice with Shima because she speaks Persian, but not like super fluently. And so honestly, my recommendation is if you have access and you do via Ramona
Starting point is 00:21:15 and Ramona's family is is when you hang out like, you know, you silently listen and try and pick up things. But like slowly over time, there's small things that you can do to just like you can say one off words. When you say goodbye, say goodbye in Polish. In Persian, it's chora hafez. Everyone's saying it. And so it's like, oh, right, I know that word. And so then you say that and then everyone, oh, it's very cute that the person's learning. But then slowly over time, you just like, so then those words, I don't have to think about like goodbye and purges not Persian like I just it's automatically like it goes
Starting point is 00:21:48 and when you're when you're learning a language there's certain phrases that they like it's what they talk about like when you're really fluent you can dream in the language which is like when I was learning Chinese like I got to the point where I was doing that and like it even now like there's certain things like when you say like I don't know that's probably like one of the most common things you're gonna say when learning a language. And in Chinese, like I'll never forget that. Will brujie dao?
Starting point is 00:22:09 Yeah, I'm probably butchering the pronunciation of that now. But like you want to start like getting those phrases and and saying them as like much as possible. And yeah, if you're going to do something with like your niece, I would say learn the phrase like you want to get to the point where the snowball really starts rolling. And that's where you can basically you have enough like fluency although fluency is the wrong word like you know enough the language that even if you don't understand what she's saying you can say like in Chinese I think it was like a well booming by Nikoi jedzima which means I don't understand but could you explain and then as soon as you have that that like enough Chinese or enough?
Starting point is 00:22:47 Persian or enough Polish that you can say that they'll switch into simpler language of The vocab that you probably know and then you're off to the races and that's when it like really starts snowballing because you have enough to get By you just like you just don't know all the words and and you have trouble listening sometimes Which is all to say that like you need have access to these people, whether you go move to the country or have family or have a partner or someone out there builds an app that basically can have this conversation. And literally I have a list of 50 things now and in the top five is build a specifically a Persian learning app that does exactly
Starting point is 00:23:26 this kind of thing. And that basically like, I just want to sit there and converse and be like, oh, sorry, I didn't understand that. Because there's some apps like Monly does have this kind of back and forth conversation where you have a conversation, but you have to read the words that it's giving you. You don't get to make up what you want to say. And so it's a scripted conversation, which still is. You do kind of like, it's like, oh, hello, hello. How's your day? Bah, bah, bah.
Starting point is 00:23:51 But it's not exactly what you would be doing, right? It's going to give you some set of words. And it's like, well, I don't actually know what that means. So you're just kind of guessing. Anyways, so. So Persian was not Shema's first language, right? Her native tongue is English? Uh...
Starting point is 00:24:10 I should know this. I mean, she was born in Persia, so... Oh, so maybe, but didn't she move when she was real young? Yeah, yeah, she moved when she was very young, though. So I think after six months, they moved to Germany for a little bit, and then they ended up moving to Canada, I think she was two so you never tell when did she start learning English that's a good question I don't actually know I mean her English is perfect right so I mean people have heard her speak from time to time on the podcast so I would immediately yeah because it's interesting
Starting point is 00:24:41 with Ramona because Ramona, you know, her native language is Polish. She lived in Poland and she was 17. She didn't speak any English until she came to the US. And she came to the US when she was 17, went into high school, learned English. And then for most of her life in the US, she's not really had that many, or at least the last couple of years, she's not had a ton of Polish speakers to interact with. And so sometimes when she's talking to her family, even though Polish is her native tongue,
Starting point is 00:25:13 she's got like, she can't find the words that she needs. She's like, oh, I have to go look on my phone to find this word that I know in English, but I don't know in Polish. And I just think it's a fascinating phenomenon because Polish was her native language. And it's not like she's entirely lost it, but I think that at the very least her Polish and English are equally
Starting point is 00:25:37 strong, but I think her English is actually stronger than her Polish. I think that English is more of her her, you know, it's not her native tongue, but it's her first language now. Yeah, well I mean if you think about it, like in the Scrabble Dictionary, just as a reference of a dictionary, I think there's like a hundred thirty-eight thousand words. I mean most people don't know like half of those, but the ones that you use commonly,
Starting point is 00:26:00 I think there's like two to four thousand of them or something. And so there's so many words that are on the edge of like, you know it, but like you'll use it once a year, once every a couple years when it's like it's, you know, if you're talking about some political thing or, you know, some domain specific thing, like, you know the words it's in the back of your brain, but like, it's so easy to forget that stuff and also not know that stuff if you're learning the language., like that's the thing is people used to say, oh, like your Chinese is amazing. You're fluent.
Starting point is 00:26:30 But I was like, if I can't have a conversation about the current like political climate and like the politicians that are running for what party and whatever party like you're not actually fluent, right? You can't have a conversation about like anything. It's like no, I can go out and have a good time. Go to some dance class, go to some board game event and like totally converse with people. But if they start dropping down into like the intricacies of like the, I don't know, the collegiate badminton scene
Starting point is 00:26:52 in China and the up and coming people, I know the word for badminton or did at the time, but like, I don't know how to follow that conversation. Anyways, so my point being is it's very easy. Like, I don't think we realize like how expansive like our vocab set is, but like, the percentage that we actually use on a weekly basis is exactly it's a super long tail. So like, it's so easy to forget that stuff.
Starting point is 00:27:15 Is that true of all languages where they have the folks that the common vocab size is like two to 4k? I mean, it's probably order of magnitude, it's true. But I wonder whether there's languages where the common vocab size is a lot smaller or a lot larger. I think that's, I actually stole that from learning Chinese, is that they say you need to know like 2 to 4,000 characters because a single character maps to a word, which a lot of people struggle with, which is why I specifically built that program to practice
Starting point is 00:27:44 my character recognition. But I think in general it's true, right? And they say it's 2,000 to 4,000 characters or words to read a newspaper. And the idea is about newspapers, even though they're talking about a bunch of different things, in general actually, and I know this because my dad's a reporter, is you're supposed to lean towards simpler language. My dad used to always joke that he would say when he was younger, there's a veritable plethora
Starting point is 00:28:08 of options, or maybe it was my aunt that would say that to bug my dad, because those are veritable and plethora are two nice sounding words. But I remember my dad saying once, he was like, you'd never put that in a newspaper article because you're writing to inform the masses. And so you're not going to reach for words that are four or five syllables that might not be in everyone's lexicon or vocabulary
Starting point is 00:28:31 because people aren't going to understand it. I wonder if there is some lesson here for interactions with large language models. Should you use simpler language when talking to an LLM because the simpler language is going to activate more in its weights than more obscure words that it just has less information about. I don't know. I do not know.
Starting point is 00:28:57 Although I do know that lately with like esoteric stuff, if you're super specific about it, it is actually amazing. Like I asked it a question about BQN the other day, which notoriously these models aren't fantastic with array languages, let alone the bleeding edge ones that have been invented in the last like five years. And it absolutely nailed it. Like I was asking if you could do some pattern matching, destructuring thing in the header of a function, which is specific to BQN, like APL and friends don't have it. And it was like, unfortunately, like you can't do what you want.
Starting point is 00:29:27 Like you can do it this way, you can do it this way. And it was it was 100% correct. And so the point being is, I think it is actually if like if you go to some corner of the model, it is actually still quite good, even if it's got a smaller corpus of stuff. Speaking, actually, well, should we got this? Well, it is a perfect transition. All right, maybe we'll transition. We might come back to the to hear more about the Kuta trip.
Starting point is 00:29:49 You know, I mentioned that it's on my list. Ask about the Kuta trip. What were you going to say? Okay, yeah, spend five minutes. Just I'm curious. How did that go? You know, because you were giving talks at most of these places, correct? True or false?
Starting point is 00:29:59 Yeah, it was great. I was just putting my trip report together yesterday. Yeah, great audiences. It was interesting in Hamburg. There were some number of people from ISC, the International Supercomputing Conference, which is like the European version of the supercomputing conference that we have in the US. And so a bunch of people from that came to the C++ user
Starting point is 00:30:24 group, and then there were some locals too. And so a bunch of people from that came to the C++ user group and then there were some locals too. And just great questions, ton of great questions there. Like people were grilling me. We spent like an extra 30 minutes there on Q&A. But all around, yeah, pretty much all the meetups, very engaged audiences. You know, it was a mix of some of the meetups for people who weren't that familiar with CUDA, weren't that familiar with HPC. Some of them were people who were very familiar with CUDA and HPC.
Starting point is 00:30:57 But it was sort of applicable for everybody because we were essentially showing, here's how we want you to think about teaching and learning and using CUDA today. So it's like we're rethinking how we educate. And I think the dog is here because it's around time for her to get food and she wants to remind me that she's hungry. But she still has at least 22 minutes before she gets to eat and hopefully you're not interfering with the mic front too much.
Starting point is 00:31:27 But yeah, and a lot of students, in particular at the London meetup, it was at University College of London. And we did a raffle at the end of the meetup, where one of the things we raffled away was lunch with Andy Terrell and me and two undergraduate students one and so we had lunch with them on Sunday and that was really nice. One of them was a law student who was into tech and so she came with all the CS students to our very, very deeply technical meetup. And yeah, I mean, it was great.
Starting point is 00:32:14 I think I had a great time. I'm really happy with the talks that I gave there. I'm going to be doing a few more meetups later this year. Was this the toolbox talk that we did a whole episode on earlier, or was this a very, Yeah, so the C++ talk I was giving was the toolbox talk. And I want to, and basically the, that toolbox talk is presenting a new view of how you should learn CUDA and think
Starting point is 00:32:42 about CUDA. So it's like, if you like if you, it's an overview of like, here's how you should write CUDA C++ code. And like, if you don't know anything about CUDA C++, here's a good place to start. But even if you do know stuff about CUDA C++, it's still good content to watch, to understand what levels of abstraction should you be working at. And one of the key takeaways is you should start at the
Starting point is 00:33:08 high level of abstractions. You should start with using libraries, et cetera. And I want to make a Python version of that talk, which is sort of like an intro to CUDA Python. But the Python version of the talk that I was giving was about how you can write kernels in CUDA Python and exploring a couple different cool kernels. I think we talk about softmax and a row-wise sum.
Starting point is 00:33:40 I think I cut out the summed area table problem from that talk, which is one that I think you and I have talked about in the past. But that's more of like a price algorithms talk. I mean, okay, they're all kind of price algorithms talks. Yeah, I'll leave links to both the sum area table episode. And we also did an episode on the toolbox. Be sure to check these show notes either in your podcast app or at adspthepodcast.com for links to anything we mentioned in today's episode, as well as a link to a GitHub discussion Cool Bucks.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.