Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Async/Await on the GPU

vectorware.com

Post reply on HN
Print view View original

41–50 of 61 posts

Re: Async/Await on the GPU

#41

Post by andrepd »

magicalist wrote: post #39
Arifcodes wrote: post #33

[flagged]

I had a longer, snarkier response to this the (as I'm writing) top comment on this thread. I spent longer than I'd like to have trying to decode what insight you were sharing here (what exactly is inverted in the GPU/CPU summaries you give?) until I browsed your comment history and saw what looks like a bunch of AI-generated comments (sometimes less than a minute apart from each other) and realized I was trying to de…

This is what I fucking hate about this AI craze. It's all [1], fundamentally, about deception. Trying to pass off word salad as a blogpost, fake video as real, a randomly generated page as a genuine recipe, an LLM summary as insight.

[1] Nearly all.

Re: Async/Await on the GPU

#42

Post by NewsaHackO »

magicalist wrote: post #39
Arifcodes wrote: post #33

[flagged]

I had a longer, snarkier response to this the (as I'm writing) top comment on this thread. I spent longer than I'd like to have trying to decode what insight you were sharing here (what exactly is inverted in the GPU/CPU summaries you give?) until I browsed your comment history and saw what looks like a bunch of AI-generated comments (sometimes less than a minute apart from each other) and realized I was trying to de…

I see this accusation a lot, and admittedly, I defended someone who later on was shown to use AI to generate comments, but I am still missing a motivation for this. Is your argument that he is using AI to copyedit his posts, or that he is asking AI to write a response to a random thread that looks insightful? Because I cannot fathom why someone would ever do that.

Re: Async/Await on the GPU

#43

Post by magicalist »

NewsaHackO wrote: post #42
magicalist wrote: post #39

Earlier quoted context omitted.

I had a longer, snarkier response to this the (as I'm writing) top comment on this thread. I spent longer than I'd like to have trying to decode what insight you were sharing here (what exactly is inverted in the GPU/CPU summaries you give?) until I browsed your comment history and saw what looks like a bunch of AI-generated comments (sometimes less than a minute apart from each other) and realized I was trying to de…

I see this accusation a lot, and admittedly, I defended someone who later on was shown to use AI to generate comments, but I am still missing a motivation for this. Is your argument that he is using AI to copyedit his posts, or that he is asking AI to write a response to a random thread that looks insightful? Because I cannot fathom why someone would ever do that.

I have no idea what their motivation is and no idea if they're using an LLM to tune their prose or write comments whole cloth (considering the four recent comments, each two paragraphs, within 2.5 minutes, though, I'm guessing fully generated).

I was just annoyed enough by spending a couple of minutes trying to decode what had the semblance of something interesting that I felt compelled to write my response :)

There are a ton of interesting top-level comments and questions posted in this thread. It's such a waste this one is at the top.

Re: Async/Await on the GPU

#44

Post by lmeyerov »

We have been curious about this for rapids cudf. We built the graphistry stack 10 years ago (!) for GPU native end-to-end acceleration - data loading, wrangling, analytics, enrichment, viz, etc, all the way server GPUs to client GPUs - but this has been a huge sticking point. Major constant overheads for smaller workloads that seem avoidable.

Essentially, we solved the problem of writing our stack in a bulk-oriented way that Nvidia kernels can optimize. Think apache arrow, pure vectorized dataframe pipelines, etc. However, cudf is 'eager' with per-step CPU/GPU control plane coordination, even if the data plane lives on the GPU. Polars in theory moves to lazy scheduling that can allow deforesting optimizations for more bulk GPU-side control macro steps, but not really. Nvidia efforts to cut python asyncio costs for multitenant etc flows didn't pan out either. So enabling moving more to the GPU here is super interesting.

Will be watching!

Re: Async/Await on the GPU

#45

Post by Arifcodes »

magicalist wrote: post #39
Arifcodes wrote: post #33

[flagged]

I had a longer, snarkier response to this the (as I'm writing) top comment on this thread. I spent longer than I'd like to have trying to decode what insight you were sharing here (what exactly is inverted in the GPU/CPU summaries you give?) until I browsed your comment history and saw what looks like a bunch of AI-generated comments (sometimes less than a minute apart from each other) and realized I was trying to de…

[flagged]

Re: Async/Await on the GPU

#46

Post by thousand_nights »

Arifcodes wrote: post #45
magicalist wrote: post #39

Earlier quoted context omitted.

I had a longer, snarkier response to this the (as I'm writing) top comment on this thread. I spent longer than I'd like to have trying to decode what insight you were sharing here (what exactly is inverted in the GPU/CPU summaries you give?) until I browsed your comment history and saw what looks like a bunch of AI-generated comments (sometimes less than a minute apart from each other) and realized I was trying to de…

[flagged]

> I have been experimenting with AI-assisted drafting for HN comments

forgive the hyperbole but this seems completely insane to me. like is the purpose of a forum not to share our collective human experiences? or do you get off on some internetpointmaxxing side game instead

i just don't get it, what are you optimizing for here exactly. are you trying to remove every ounce of autonomy from your life or what

Re: Async/Await on the GPU

#47

Post by lmeyerov »

the__alchemist wrote: post #13

Et tu, GPU? I am, bluntly, sick of Async taking over rust ecosystems. Embedded and web/HTTP have already fallen. I'm optimistic this won't take hold in GPU; well see. Async splits the ecosystem. I see it as the biggest threat to Rust staying a useful tool. I use rust on the GPU for the following: 3d graphics via WGPU, cuFFT via FFI, custom kernels via Cudarc, and ML via Burn and Candle. Thankfully these are all Async…

For graphistry at least, I care less about the surface syntax (async/await) and more about getting gpu-side work stealing, dynamic task scheduling, etc. Our code is written at such a much higher level that these are primitives needed by our runtime, not most of our developers & users. Imagine SQL, cypher, etc on GPUs and our implementations of those being able to use the gpu-side libraries when coordinating 1M+ threads.

Re: Async/Await on the GPU

#48

Post by latchkey »

LegNeato wrote: post #36
monster_truck wrote: post #9

Earlier quoted context omitted.

I don't have anything to offer but my encouragement, but there are _dozens_ of ROCm enjoyers out there. In years prior I wouldn't have even bothered, but it's 2026 and AMD's drivers actually come with a recent version of torch that 'just works' on windows. Anything is possible :)

Thank you! We're small so have to focus. If anyone from AMD wants to reach out, happy to chat.

Anush is who you want to ping, he's motivated and will take care of you.

https://x.com/AnushElangovan

Re: Async/Await on the GPU

#49

Post by magic_at_nodai »

latchkey wrote: post #48
LegNeato wrote: post #36

Earlier quoted context omitted.

Thank you! We're small so have to focus. If anyone from AMD wants to reach out, happy to chat.

Anush is who you want to ping, he's motivated and will take care of you. https://x.com/AnushElangovan

yes lmk how i can help. at the minimum i can get you hw and help with PRs etc. firstname at amd.com to reach me.

Re: Async/Await on the GPU

#50

Post by bschwindHN »

thousand_nights wrote: post #46
Arifcodes wrote: post #45

Earlier quoted context omitted.

[flagged]

> I have been experimenting with AI-assisted drafting for HN comments forgive the hyperbole but this seems completely insane to me. like is the purpose of a forum not to share our collective human experiences? or do you get off on some internetpointmaxxing side game instead i just don't get it, what are you optimizing for here exactly. are you trying to remove every ounce of autonomy from your life or what

Seriously people, type out your own thoughts. If I want to chat with a clanker I can do that on my own time.
Post reply on HN