Hacker Newsnew | past | comments | ask | show | jobs | submit | eulgro's commentslogin

Do databases do matrix multiplication? Why would they even use floats?

That's a great question. I never worked on any cool NASA stuff which would involve large scale number crunching. In the corpo space, that's not been my experience at all. We were trying to solve big data problems of like, how to report on medical claims that are in flight (which are hardly ever static until much later after the claim is long completed and no longer interesting to anyone) and do it at scale of tens of thousands per hour. It never went that well, tbh, because it is so hard to validate what a "claim" even is since it is changing in real time. I don't think excess GPUs would help with that.

lot's of columns are float valued, GPU tensor cores can be programmed to do many operations between different float/int valued vectors. Strings can also be processed in this manner as they are simply vectors of integers. NVidia publishes official TPC benchmarks for each GPU release.

The idea of a GPU database has been reasonably well explored, they are extremely fast - but have been cost ineffective due to GPU costs. When the dataset is larger than GPU memory, you also incur slowdowns due to cycling between CPU and GPU memory.


what do you think vector databases are? absolutely. i think the idea of a database and a "model" could start to really be merged this way..

To be honest, given the efficiency of the drug and the huge benefit it could be to society, I feel like if I had been the employee in charge of filing patents I would've more than ready to lose my job in exchange for low cost general availability in the US via (illegal maybe, whatever) cross-border market. It's a nice loop hole and a great thing that once the delay expired they can't file ever again.

One's got to find ways to feel like the good guy when working for Big Pharma . That's probably not what happened but it's nice imagining it.


Maybe they'd even do it in exchange for just low-cost general availability in Canada!

I still use Acrobat for two things:

- Viewing Altium generated schematics, which have some macros that only work in Acrobat.

- Printing stuff. Acrobat print dialog is pretty good.



Avoiding ultra-processed food is completely realistic.

It's a question of priorities. If she really wanted to avoid them she would take the time to do so, even if that means working less.

What's unrealistic is expecting everyone to have the drive to do what's necessary to avoid ultra-processed food.

Tangentially, kids will always want to eat what everyone else is eating. If you start feeding them home cooked Golden fish or whatever, people at school are going to judge them, your kids will feel excluded because they won't be eating the same things as everyone else, and they'll end up resenting you somewhat.

For me the challenge here is really how to give your kids healthy food while also not excluding them from the normal kid experience, which, for better or worse, is going to include consuming ultra-processed until the day we have proper regulations.


In Vim: * to highlight a word, n/C-n to move between them. You need to have the hlsearch option set.

* searches for the word under the cursor and jumps to the next instance (as if you had pressed n). So it's basically what GP asked about (as long as we're talking basic string matching and not symbol matching).

And hlsearch isn't needed at all; that just controls whether or not the word + its search results get highlighted.


Can you give at least a few examples? I have no understanding of ergonomics it seems


There literally is a plugin [1] containing sensible defaults that everyone in the community agrees would be good as default but “backspace” and “incsearch” are the most obvious. “Backspace” allows you to delete with the backspace key beyond the point where you pressed “insert”. I don’t think I’ve ever met someone who thinks the vim default (not allowing this) is ergonomic.

[1] https://github.com/tpope/vim-sensible


Have these already been adopted in neovim by default? It seems like I already have this behavior without needing any plugins.


There are quite a lot of default changes so go :he nvim-defaults to get them all but yes, the default setup in neovim is much more sane.


I wouldn't argue that the vim default (of not allowing backspace beyond where you entered insert mode) is ergonomic, but I do prefer and exclusively use the default vim behavior. It's just what I'm used to at this point (24 years using vim/nvim) and if I set it the "ergonomic" way, it's disorienting.

So, ergonomic or not, some people do prefer the default -- at least one person :).


You can start with the original example 4 comments above and try to understand how unergonomic the default way of learning/remembering hundreds of key combinations in vim is.

Then think about the basics: why some of the most frequently used commands w/b are located so inconveniently and far away from each other (and if you reconfigure them, how unergonomic the config language is where instead of reading a sensible name like 'move_prev_word_start' you can only reference 'b' that you'd never use since, well, you've changed it to something else!)


You don't need hundred of commands to use vim, have you at least finished vimtutor? Finish it in the way it proposes, then start it again and think on your own, what commands for more effective passing the problems from vimtutor might have been there as well.

Letters w and b may be located wherever they are located, if you can not touchtype vim is not for you for the same reason we do not learn how to run before we master how to walk.


A more efficient way might be to train them in simulation. If you simulate a warehouse environment and use that to pre-train a million robots in parallel at 100x real time learning would go much faster. Then you can fine tune on reality for details missed by the simulation environment.


What exactly was the process to choose the 50 machines?

I wonder how "Hay rake" got on the same list as "CNC precision multimachine" for example? If you asked me the former is probably more useful than the latter.


I don't know if you've been on a farm, but things like gasoline engines, welders, and tractors are pretty important there. To make them, you need machining, and the idea of GVCS/OSE/FeF is to be able to do everything.

The "hay rake" in question is supposed to be something comparable to https://www.greenmarkequipment.com/new-equipment/agriculture..., which is a large and complex machine that requires a tractor to tow it. It involves parts such as tires, bearings, nuts, bolts, chains, and wiring, and its main structural frame is rectangular structural framing welded together. You aren't going to be able to make anything comparable by lashing together bamboo. You need a machine shop.


Oh right. I was imagining this: https://www.fruithillfarm.com/media/catalog/product/cache/a8...

Which is the hay rake I used in the past to collect hay...


Yeah, I was puzzled about why modo_mario thought you might need a machine shop to make a hay rake — sounds like the kind of thing Primitive Life might make with stone tools. So I checked out the John Deere model number mentioned in the spreadsheet, and it turns out that the hay rake in question can rake hay at utterly inhuman speeds over a wide swath of your field. Handy if your cows are depending on you for winter fodder.


I'd say the former won't last if you can't fabricate or fix a simple part for which you might need the later.


When C++23 is fully supported functional style will be somewhat bearable, but even then a lot is missing. Lambdas are just too verbose to make this fun.


Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

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

Search: