As someone whose 27 now, I'm pretty shocked by the longevity of some of the games that I played growing up.
Games like Minecraft, Supercell, Geometry dash just to list a few, are arguably STRONGER today than they were 10/15 years ago. Even snapchat, I thought was gonna die after 2020 but my nephews today are using it more than ever.
Whereas, for my cousins who are around 40 now, basically all of the games that they played growing up are dead.
So. I can kinda see minecraft being around for another decade to say the least. It doesn't even feel that crazy for it to be around for another century... or even the rest of humanity? (like bicycles?)
Hey HN! We're excited to share Sail 0.3, our open-source distributed computing framework that serves as a drop-in replacement for Apache Spark.
Sail is a Rust-native execution engine that speaks the Spark Connect protocol. Your existing Spark SQL and DataFrame code runs unchanged, but executes on average 4x faster while using 94% less infrastructure spend.
Here's how we are achieving that performance:
1. No JVM overhead: Rust's zero-cost abstractions and deterministic memory management eliminate GC pauses
2. Columnar processing: Apache Arrow format enables SIMD instructions to process multiple records per CPU cycle
3. Zero-copy data transfer: Python UDFs run in-process with shared memory buffers (no serialization)
4. Lightweight workers: Processes start in seconds
What's new in v0.3:
- This release is a major milestone - we now support both Spark 3.5 and 4.0, including the new lightweight pyspark-client. The framework automatically detects your Spark version and adjusts its behavior accordingly, so one Sail binary works across versions.
Why this matters:
- Spark revolutionized big data 15 years ago, but its JVM foundation struggles with modern workloads. As teams process more real-time data and AI workloads, they're hitting walls with latency, cloud costs, and operational complexity. Sail is trying to solve all of these problems while not requiring you to rewrite everything that you already did with Spark.
- We're working toward unifying batch, streaming, and AI workloads in a single framework. Imagine running your ETL, real-time analytics, and model training on the same infrastructure with predictable performance. The project is open source (Apache 2.0) and we'd love your feedback! We have a growing community on Slack where early adopters are already running production workloads.
I tried out an early version when I did some work at notion last year, just tried it again and it's honestly a lot better than I expected. The design is good as expected from a notion product, but it's actually lot snappier than I thought. Faster than the notion app lol. It's probably a consequence of not having too much legacy code, but still impressive nevertheless! Esp in a world of slow nextjs apps lol.
Vibe wise, it kinda feels kinda like superhuman but 0 instead of $30/mo lol.
Tbh, I can also kinda see the vision that notion is pitching to investors now - the entire productivity suite, but redesigned and enhanced with AI. I'm not a power user of email/calendar by any means, but find myself using their calendar and mail client. I have no doubt the next generation of Tiktok girlies will eat this up like crazy. I don't use much of their AI stuff, but can also see how business can really find it useful.
I was more talking about the optimistic updates n caching as opposed the actual rendering. From my experience, the reason why most apps are slow is more so cuz of the former than the latter(like the spinner in ur notion)
When it comes to the actual rendering performance, it’s an electron app, so … whatever you make of that
Personally, I feel like often times electron apps are actually smoother than swift apps nowadays due to the way swift renders text(ie. the ChatGPT app was so laggy when it was released, unclear if they fixed it). Start up time is probably slightly slower than native too but not a hugeee issue for me.
Probably not as fast as superhuman. But I don’t want to pay $30 for a 10ms diff(might be more idk)
I feel like it's at the point where I'm not too sure how these rankings impact the my choice of LLM. Every time a new model tops the charts, I'll try them for a bit and go back to claude-3.5-sonnet. Both for coding and day to day questions.
I don't know if I'm just getting used to the claude style of response, or the orangy UI that I kind of find cozy, but I think we need better ways to convey the difference between models.
I also think it's really cool how he parodies himself. Most of the other martial arts actors from the 80s take themselves way too seriously now, like Steven Seagal who just phones it in in B-movies. Jean Claude van Johnson was awesome.
1. just how low is the latency?
"Netris delivers a zero-latency gaming experience that won't eat up your data plan." What does this mean? Wouldn't the latency be at least from my computer, to Netris server, to the game server?
I would assume you can't play FPS/MOBA games on these?
2. "10,000+ games supported" "games that come with their own launchers are not yet supported"
I'm curious about the technical differences between steam games and games with their own launcher. I guess I was assuming that you're more or less just pixel streaming to me(?)
3. how does it compare to GeforceNOW?
Would love to see one of those long charts with checklists.
As a GFN user myself the biggest single problem is publishers getting a say in whether I can run their game on the service. With self hosted solutions like Netris you don't have those limitations because they're not baked into the software. In terms of other differences I'd imagine that you would end up paying more for this in the cloud. As someone else has mentioned, NVIDIA are likely to be subsidizing access to their GPUs.
> games that come with their own launchers are not yet supported
Please let this be the nail in the coffin for those extra clicks and annoying "Log in with our service now" for games which have no online stuff whatsoever (take2 is notorious for this)
GeForce now used to have tons of "issues" with people easily being able to run any app or program they wanted, using a few tricks. Last time I tried it, it seemed like it was all locked down a lot better, but previously it'd still be steam opening the other program.
It was nice while it lasted, since they wanted cyberpunk available day one, it was actually accessible about a week before the game released.
>"Netris delivers a zero-latency gaming experience that won't eat up your data plan." What does this mean? Wouldn't the latency be at least from my computer, to Netris server, to the game server?
I agree zero-latency sounds impossible. But why wouldn't the Netris server and the game server be the same machine? So that's 1 less latency hop.
Imo it feels more like 1-2 years away. Smaller 7B, 34B and 70B models are becoming a lot better, with more context length. Faster inference methods are coming out day by day. Better ways to quantize/distill models. All of that on top of chip advancements we saw a couple days ago with M4, Qualcomm/Googles arm chips...
I can't imagine more than 2 years for GPT4 level LLM on edge devices.
The question is, will ppl want GPT4 on the edge when GPT 6 is one request away?
"According to Precedence Research of Canada, the AI chip market is poised for explosive growth, with projections indicating a rise from $30 billion this year to over $200 billion by 2032.
Another research firm DataHorizzon Research even projected the AI Chip market to be 1,114.3 billion by 2032."
Ok, few here will deny the AI (chip) market is exploding. And CEOs will have visions of datacenters packed with expensive AI chips, extracting value from consumers. But if I were to take a guess:
AI models will be optimized the heck out of. Their architecture is only at the beginning of a long road of innovation (and in particular: simplification - aka race to the bottom). With so many heavyweights getting into the game, competition will be fierce.
Bottom line: yes, AI chips will be everywhere. But they'll also be cheap(-ish). Not unlike current day cpu's, flash, RAM etc.
With that in place, the next revolution: local-running AI models in everybody's hands. Cheap, easy to use & tinker with like mobile/PC apps today.
"Upon further investigation, we discovered that a threat actor had accessed data including Dropbox Sign customer information such as emails, usernames, phone numbers and hashed passwords, in addition to general account settings and certain authentication information such as API keys, OAuth tokens, and multi-factor authentication."
I use Dropbox Sign API, so a little fearful our private data was accessed.
API keys were leaked as part of this hack. It's unclear from press release if hackers used the API keys to access data/documents of customers.
April 24th they became aware of issue, reporting it over a week later. I'd also be curious on how long this problem went on before being detected on April 24?
This hack seems to affect the Dropbox Sign application, which is based on HelloSign which they acquired a few years ago. It’s still running on the hellosign.com domain and seems mostly separate, so it wouldn’t surprise me if they also store passwords differently.
Useful because you can support existing passwords without requiring everyone to login or reset their password. Still has flaws though, like password shucking.
I think one thing that people don't realize is that the YC application process is really one of the best tools for "sharpening" your idea/business.
The written applications forces you to articulate your ideas in a concise yet easy to understand way.
And as much as YC doesn't recommend this, the mock YC interviews we did with alums was one of those most beneficial things that happened to us. Because so rarely will you get the opportunity to ask dozens of other YC founders to grill your business, and have 80%+ of them say yes.
We did about 30 at the time, which is a lot of time to be taken off product/building, hence probably why they don't recommend it, but looking back it *really helped us understand our own business. Given how young/naive/early we were.
> I think one thing that people don't realize is that the YC application process is really one of the best tools for "sharpening" your idea/business.
I'm seeing this a lot through the comments but I have to ask, why is it? Do people just not think about how their product will make money, who the competition is or how are they going to get customers?
> Do people just not think about how their product will make money, who the competition is or how are they going to get customers?
The short answer is "yes, people don't think about those things."
The longer answer is that they think that they've thought about those things, but in reality, they put more thought into what they had for dinner last night than they did into those things.
At least for me at the time, I knew nothing about start ups. Didn't come from a prestigious school. Nor the bay or even US. Didn't work in big tech. The closest thing was that I was studying EE.
Honest it was quite inspiring to chat with these founders who were in a different world, and realizing that they're not all that different from myself.
Games like Minecraft, Supercell, Geometry dash just to list a few, are arguably STRONGER today than they were 10/15 years ago. Even snapchat, I thought was gonna die after 2020 but my nephews today are using it more than ever.
Whereas, for my cousins who are around 40 now, basically all of the games that they played growing up are dead.
So. I can kinda see minecraft being around for another decade to say the least. It doesn't even feel that crazy for it to be around for another century... or even the rest of humanity? (like bicycles?)