@nabla9 have tried to tell you that for DGX Spark, you can also use optimized models; therefore, this means that Spark can also be used for inference with bigger models, such as those exceeding 200B.
Please compare the same things: carrots VS carrots, not apples VS eggs.
I don't understand what's not optimized on 5090. If we're comparing with Apple chips or AMD Strix Halo yes you will have very different hardware + software support, no FP4 etc. but here everything is CUDA, Blackwell vs Blackwell, same FP4 structured sparsity, so I don't get how it would be honest to compare a quantized FP4 model on Spark with an unoptimized FP16 model on a 5090 ?
With 128 GB of unified system memory, developers can experiment, fine-tune, or inference models of up to 200B parameters. Plus, NVIDIA ConnectX™ networking can connect two NVIDIA DGX Spark supercomputers to enable inference on models up to 405B parameters.
The datasheet isn't telling you the quantization (intentionally). Model weights at FP16 are roughly 2GB per billion params. A 200B model at FP16 would take 400GB just to load the weights; a single DGX Spark has 128GB. Even two networked together couldn't do it at FP16.
You can do it, if you quantize to FP4 — and Nvidia's special variant of FP4, NVFP4, isn't too bad (and it's optimized on Blackwell). Some models are even trained at FP4 these days, like the gpt-oss models. But gigabytes are gigabytes, and you can't squeeze 400GB of FP16 weights into only 128GB (or 256GB) of space.
The datasheet is telling you the truth: you can fit a 200B model. But it's not saying you can do that at FP16 — because you can't. You can only do it at FP4.
If the 200B model was at FP16, marketing could've turned around and claimed the DGX Spark could handle a 400B model (with an 8-bit quant) or a 800B model at some 4-bit quant.
Why would marketing leave such low-hanging fruit on the tree?
You and nabla9 are both the one comparing apples and eggs. 4x more RAM means 4x larger models when everything else is held the same to make a fair comparison.
You can look at the /DataHoarder subreddit, as I see people usually buy secondhand LTO4-LTO7 tape drives and libraries, which are affordable. Datahoarders use dozens and hundreds of cartridges for archiving/backup purposes or for making fun playing with tape drives.
The weather app doesn't give much money. The main business sells weather and climate data B2B: agro, insurance, logistics, retail, supply chains, advertisement, medical, etc.
Companies whose primary business is weather apps are small, and such areas are highly competitive.
As far as I know, AccuWeather is the main beneficiary. You can easily find reliable sources about it.
The cause is that NOAA publishes all weather data, calculated models (global coverage), meteostations data (global coverage), and weather radars to the public for free (US only, maybe also Canada, I don't remember). Therefore, many weather companies use such data to do their business and compete directly with AccuWeather. They don't like this.
On the other hand, state weather agencies that calculate global models in many countries don't provide such data for free. Therefore, startups and small companies who work in weather and climate fields use NOAA data and directly compete with AccuWeather or don't pay them for data access.
You have bizarre logic here. For example, in a topic discussed about GPUs, someone would say that it's not possible to run databases on GPUs, so GPUs don't have any chance of succeeding.
> How much do you trust your government with your money?
Do you trust crypto companies? Mt. Gox, FTX, Bybit…
Do you know that crypto companies must follow government rules, regulations, and laws? Russians were banned from using many crypto exchange platforms. China has strict rules for its citizens. You can buy and sell crypto in Brazil, but you must use only Brazilian reals.
Pix isn't global, but no one government person outside of Brazil can block this system.
MasterCard, Visa, Amex, and UnionPay work worldwide, but only a few countries regulate them, can block their usage, and can use data for tracking and statistics.
Pix is free to use, so no one needs to pay an additional "tax" to MasterCard and Visa (it's about 3%).
Google and Apple cannot say that if you want to pay, you must use only our devices.
> Now, try to use Pix outside of Brazil
Now, try to buy ice cream from street vendors using any crypto coins.
> How much do you trust your government with your money? A system like Pix don't stand a chance to get a worldwide adoption
You mischaracterized that.
> You have bizarre logic here. For example, in a topic discussed about GPUs, someone would say that it's not possible to run databases on GPUs, so GPUs don't have any chance of succeeding.
He's saying it wont get adopted worldwide. Not that it wont succeed (which is a very ambiguous metric).
Now that I've recognized this pattern I see it everywhere: someone invents part of a solution, probably including some cool technology, then hails that as the solution and insists everyone else is wrong for not getting it.
The classic one is some FOSS people inventing a protocol where servers can talk to clients, and declaring a problem solved, when in actual fact, most people don't have a server. Mastodon is this, but so is XMPP.
HTTP took off because there were servers you could fetch things from with HTTP, not because it theoretically allowed you to fetch things from servers.
Paper euros aren't cool because I can "have money". They're cool because I can go to the grocery store and trade them for something to eat. My bank card isn't cool because I can "have money". It's cool because I can go and swipe it and not have to count paper euros. If you want cryptocurrency to be cool, you're going to have to get it integrated into grocery stores, which is, of course, impossible because it can only process 7 transactions per second. You also need a way to convert my paychecks into cryptocurrency, but this is basically solved with crypto exchanges now.
> the conversation is supposed to be about cryptocurrency technology,
> but you're talking about the gross financial companies that operate in cryptocurrency as if they ARE cryptocurrency.
That's because the middlemen are inevitable and they work the way they do for a reason—governments won't let them work any other way.
Cryptocurrency is a technological solution for a human problem, and you can't analyze the impact of the technology divorced from the human reality without losing so much resolution as to make your analysis meaningless.
If we want to use technology in real world to solve real world issues, we need to consider all important non-technical things.
Nuclear Energy is great, but governments and international organizations want to control it because it is too dangerous. So, if we need to use Nuclear Energy, we must play by such rules.
Money is the same thing. Each government wants to control them, regardless of their form.
If someone wants cryptocurrencies to be widely adopted, there is no option but to give them to businesses and governments.
So, crypto would be regulated like usual money. Major blockchains have records for all transactions, which can be tracked and used by businesses and governments to implement more strict control over the whole world. Therefore, the more people use crypto, the less privacy they have.
The Internet and Web were designed to be anonymous, but cookies, IP addresses, data collection, ML/AI, IMEI, MAC, and the control of registration in ISPs and mobile operators have led us to a situation where the government and companies can easily track people. The same situation would be with crypto, which was designed to be anonymous but used in another way.
I would say your post has the logical flaws, not confronting any of the core criticism and instead misdirecting to other topics.
Creating efficient payment rails for its own currency is one of the most obvious roles for government imo. If the government provisions the currency, why would they not also provision the infrastructure (like the printing of the paper money).
That said, you’ve not offered a good rebuttal to any of OPs concerns, just repeated how good pix is within Brazil…right now with their current government.
Digital payments does present a uniquely frictionless route for tyrannical governments to assert power should they ever decide to weaponize it…unlike paper money which is harder to control.
Also, international payments is absolutely an issue with these systems. So you hate crypto due to its 2010s association with annoying Twitter bros. I get it.
But what are you offering instead as a solution to global money? Paying Wise stupid currency movement fees and waiting for them to close your account because you tried to buy a house for your family in the country you moved to?
>Digital payments does present a uniquely frictionless route for tyrannical governments to assert power should they ever decide to weaponize it…unlike paper money which is harder to control.
This is true, and there's already a proposed law to ban paper money in Brazil.
No, but abusing a payment system makes little sense. If people don't trust it they won't use it and then it serves no purpose. Incentives between government and people can have overlap
I don't offer any solution for the problems you mentioned, and I don't think it is possible.
If we want to have global money and a global payment system, they would be controlled by governments, international organizations, God, Devil, Cthulhu, Spaghetti Monster...
There is no magical solution. We, as a society, need to establish competing social institutions, and try to control them, and try to force them to compete. There is no solver bullet.
I was several times required to build simple web apps, and it wasn't hard. I used Python, Flask, Jinja2, Bootstrap, and maybe one or two functions from jQuery for AJAX. It was very easy, and I know only Python and basic HTML and CSS. I don't know any JS/TS because I'm a Computer Vision Engineer.
These web apps show tables with data, images with figures, and some controls to request calculations or data from a server. jQuery is used to update the drop-down list according to previous choices.
Some time ago I did research for myself on the topic of GUI libraries/frameworks. Below are some things from this study. The modern tendency for all frameworks to use 2D rendering engines to draw all widgets from scratch instead of using platforms' widgets and APIs. Usually, they use Skia.
I can recommend reading blog posts by Raph Levien [1], who is one of the founders of Linebender [2] and Xilem [3]. Xilem is a Rust cross-platform GUI framework intended to use GPU for rendering in the first place. Earlier, they tried to develop their own GPU meta-API, then they switched to WGPU [4] and developed their own 2D rendering engine instead of using "standard" Skia, like many others.
JetBrains develops Compose Multiplatform [5]: using the same Android API - Jetpack Compose, but for many platforms and in Kotlin. They also use Skia but are considering switching to Impeller (a new 2D rendering engine in Flutter) because of performance issues in Skia with shaders compiling.
Slint [6], was founded by 3 guys who's been working for a long time in Qt Group. It is written in Rust but uses DSL to develop GUI. As I understand, they try to solve the main fundamental issues in Qt.
MAUI, as I understand it, was Xamarin.Forms and now Microsoft tries to develop it.
Another interesting thing. There is a part of each GUI framework, a system to automatically place widgets on the screen - layouting, something like a grid, table, and so on. Several years ago, Apple had a big boom in the implementation and use of the Cassowary algorithm [7] in their GUI frameworks. It was a promising, strong math-based academia-proof algorithm, but in real life, as I understood, it can not work reliably in many edge cases [8], so Apple quietly removed it and implemented Flexbox [9].
More approaches: to use game engines as GUI frameworks: Unity, Unreal, Godot, etc.
Fun fact: Blender has its own GUI library, which is implemented on top of OpenGL.
PS If you find anything interesting, I would appreciate it if you share.
PPS As a way to find more, I can recommend using search in HN, Reddit, and GitHub. Try to find projects that people develop to solve something and look at what ideas they use.
[4] WebGPU implementation in Rust, used in Firefox and many other projects, can be used as a separate library, cross-platform, meta-API over Vulkan, DirectX, Metal etc
Sometimes, I miss paper-based magazines, and when I have an opportunity to stay and look through newsstands in supermarkets or bookstores, I do, and sometimes buy. Harvard Business Review is good one. Wired is sucks, one of the worst.
Does anyone know good magazines about tech/programming/engineering?
I found CODE Magazine [*], which looks promising, but it is primarily about C#/.NET.
Once upon a time there was Google Web Toolkit (GWT), ASP.NET WebForms. What approach do they use, the same as FastUI? Now there is Blazor for C#, how does its approach differ from FastUI? I think they all promise to write UI for Web without writing JavaScript code.
On the crowdfunding page [1] they revealed causes why they decided to run the campaign:
> Our goals with crowdfunding the development of OpenCV 5 are several: 1) To make OpenCV 5 our biggest release ever, with the most community participation of any version so far, 2) To prove to other struggling open source projects that crowdfunding is a viable option over seeking corporate donations exclusively, 3) To create a sustainable method of fundraising that OpenCV can return to year after year, making the overall organization and project more robust to global instability.
> Recently, due to global conflicts and instability, OpenCV has lost several core team members, and this has slowed progress on both new feature development and the handling of bugs and requests in the GitHub repository.
> Crowdfunding development of a big Open Source project like OpenCV is a big move. We aren’t aware of ANYONE doing it at this scale. As a non-profit organization, we are constantly fighting to find funding, but we always find support in our community. Instead of spending our time chasing big checks from billionaires, we are putting that community front and center.
There are thousands (millions?) of nonprofits who fundraise incessantly—any org that's mission driven without a profit motive needs to have that activity (and the resulting CRM management) as part of their fabric to thrive.
I think a few of the bigger software projects are realizing relying on a few generous whales is unsustainable when money faucets are turned off... and it's a good thing for the long term stability of these groups to figure out how to sustain the core project.
As long as it doesn't get annoying (multiple appeals every quarter, incessant "support us or we'll DIE!" messages...) I'm fine with fundraising initiatives for OSS.
I wonder if they were lost, as in passed away, or just lost as in they live in countries that can’t do business with the organization for legal reasons anymore. Hopefully the latter and they’ll find some way back eventually…
Please compare the same things: carrots VS carrots, not apples VS eggs.