Hacker News new | past | comments | ask | show | jobs | submit | beyondzero's comments login

Spot on. This could be a game changer for us ($xB 100+ year old manufacturing company), where we are on a long, slow journey to digitize every piece of manufacturing (long and slow because manufacturing runs almost 24x7x365). Wifi, even the top enterprise systems, is not as resilient, cheap and quickly installed as we'd like.

So now I can use 5G instead, and template and deploy it via the cloud? Yes, yes! We'll put this through some cost models, but it will likely jump NPV of IoT and automation projects by pushing down the initial capital costs (fiber runs pulled by union electricians to wifi gear installed by a vendor vs. 5G base stations and servers installed and configured by plant electricians and corporate IT).


If you have problems with wifi resiliency I've got bad news for you about the wavelengths 5G uses.


I think GC might be talking about how the reliability of a large-scale mesh network using wifi isn't great (mostly because that's not what wifi was designed for), while this is literally what 5G was designed for.


I think 5G will replace wifi within next 10 years. We don't need WiFi anymore


As somebody not very well-versed in this, a sincere question. Would something like 5g be able to replace a home intranet / network? Eg I have a personal media server that I stream to devices on my wifi--would this be able to work with 5g? From my perspective 5g can only replace internet access, yes? Devices that want to communicate with each other securely would still prefer to be connected over wifi or other personal network like Ethernet?


The part of 5G that gets thrown around a lot with high speed and capacity is mostly based on mmWave bands(think 10's of Ghz) which attenuate incredibly quick. Most of them are nearly direct line of sight. You already see some issue with 5Ghz wifi not going well through walls, there's a high chance you'd be falling back to one of the lower speed bands inside.

The other thing is there's a non-zero amount overlap in the way spectrum is used in wifi and 3/4/5G, MIMO, spread spectrum and other approaches are all trying to get as close to the theoretical limit of the channel bandwidth. When you start looking at large scale wifi deployments it starts getting split into cells of channels not to dissimilar to cellular deployments. There are some differences and certain bands benefit from discrete allocation/cell management but at the end of the day it's all radio waves over the air.


This would seem like an argument for installing wired / wire->mesh 5G base stations around our houses instead of the wifi repeaters we currently use..

Amazon in the past flirted with the idea of building a mesh network of Amazon IOT devices so that your neighbours Echo could connect yours to the internet even if you didn't have a connection. BT in the UK offered reduced price broadband if other BT customers could use a WiFi Guest network from your connection - so BT customers had like 90% free wifi coverage in cities.

I can see a amazon deploying Amazon Echos and automation poducts as 5G enabled IOT devices that backhaul to your home internet but also mesh with neighbours, then eventually deploying its own outdoor 5G coverage and suddenly becoming a mobile network.


Yes it's possible but AWS focus is to be the Layer 1. They're doing it for data and can easily replicate it for voice if there is a use case but at that time they're directly competing with Telco, a fight that they may not pickup at this time. Ideally they would utilize all the unused bandwidth & reuse it. So one neighbour can provide internet + mobile to entire street


Yes it should. Wifi isn't great with security + authentication. So ideally data would transfer over 5G.


5G is typically deployed in CBRS bands - i.e. 3.5 GHz. This isn't mmWave. Will have meaningfully higher coverage areas than WiFi (but mostly because WiFi is power-limited, not to do with frequency).


Nokia has been the pioneer in Private wireless networks. They have already deployed production ready private LTE / 5G systems in ports and factories.

For more info: https://dac.nokia.com/private-wireless/

Disclaimer: I work for Nokia.


"deploy it via the cloud" doesn't actually mean anything and confuses people. You still need to set up hardware at your own enterprise! It's NOT the same as renting a VM in a datacenter and "deploying" it in the cloud.

Sure, it might have a good UX and not require as much expertise to manage, but it's not "deploy via cloud". That's just marketing.


Telecom networks have a seriously complex back-end, that's what makes them better for larger areas than Wifi. A commercial mobile network typically has a few racks of machines below the antenna mast to run the radio systems, and then uses a fiber connection back to a mode central place to control those radios. In this case you're deploying (much smaller) local radios and AWS runs everything else in the cloud.


Challenges are features like HDR and having four HDMI inputs.

Planar makes incredible and also expensive displays for the commercial market. They offer a "luxury living" solution but the smallest is 100".

https://www.planar.com/markets/luxury-living/


> I think this digital child managing system sounds moderately dystopian to be honest.

A lot of the reactions and rebuttals to this comment are from HN childless people, whose perspective is their memory of being a child age 12-17, talking past HN people with children, whose perspective is about their kids age 5-12. At one end of the range you are educating about drugs and sex and good decisions, on the other end of the range you are worried about clean butts and walking across busy streets.

The method of CREATING an older child who can be an independent and functional adult is by "MICROMANAGING" early-on so they develop good habits (especially good habits of independence!). And I am a Montessori parent which is fairly radical compared to the normal US system.


Thank you for making this comment, it very much rings true as a parent whose child is far to young to ever be thinking about planning her own day.


This is silly but sort of relevant.

During college junior year (1993) as a physics major I took a class in digital electronics (which included 68000 assembler programming). We had a lab contest to create the fastest sorting routine for a set of random unique numbers between x and y.

I won the contest by setting to 0 a y-x register range of memory and then inserting each number into the range based on the number itself ("27" into register 27, "18" into register 18, etc.). Then I printed out all non-zero numbers in the range.

The other 20 or so students did versions of bubble-sorting and called my solution a cheat. The professor defended my victory as I had not broken any of the rules of the contest...


Haha. Reminds me of a lab where we had to program a 6502 to play some sort of game, rendered on an oscilloscope. My genius friend stayed up all night writing code that rendered the snake by dynamically writing code, rather than the obvious approach of reading the “snake” coordinates and rendering those to the screen.

Our snake code ran at something obscene like 1,000 Hz. So much faster than anyone else’s. The professor couldn’t understand how we did it, so gave a low mark.


>The professor couldn’t understand how we did it, so gave a low mark.

That was genuinely upsetting to read.


In High School 2004 I made a license plate detection system for our IT class project. It was not perfect, granted, but was able to locate license plates in a photograph pretty reliably and do OCR on them.

Teacher was like, nah, that's not possible, and clearly hadn't even read it. Gave me an A just to stop me complaining, still not understanding the project.


My Fortran professor gave me a low mark because I used a lookup table for octal to binary conversion, instead of using division and modulo.


In a teaching lab in 1995, I had to make a digital thermometer using a mcu. I wrote code that dynamically read the voltage drop on the sensor and calculated the temperature. It was fast enough and easy to calibrate. The code was small and easy to understand.

I didn’t pass because the point of the assignment was to use lookup tables, even though that was a more complex approach in this case :(


"It was fast enough and easy to calibrate."

Are you sure about easy to calibrate 8) Did your solution have the same level of accuracy across the range as the MCU would provide?

Bizarrely, you sometimes have to follow the standards because that is what everyone does. I can't think of an example now but there will and have been cases in engineering where the wrong answer is the right one because that is what is done. I know of mensuration devices that "model" some curves with a linear approximation and paper over the errors with cough error estimates and/or keeping within the nearly linear range. That too is fine if everyone understands what is going on.

Accuracy is a funny old thing.

However, I like your approach and it shows you probably understand the principles involved. Ideally, if you are going to be a smart arse like that, then submit two solutions - the proscribed one and your clever one.

You mention a MCU but not what sort of sensor ...


Curious as to what you mean by "dynamically writing code"


Like meta programming, creating a string to define a query and then executing the string. In the case above, you dynamically create the program you need to solve the problem and then execute it. This is different than "solving the problem". You are writing a program to "write the program to solve the problem".


Doesn't sound like this kind of complicated string eval would be able to do 1000Hz but then again I also have no clue where you could possibly need that for a snake game. On an oscilloscope you'd need to produce vector lines to move from/to right?


More likely, I would wager that they were just overwriting the instructions to draw with new instructions that just inline everything so it didn't need to fetch out to anything.


Yes. This.


This is a slightly simpler (because of unicity) version of bucket sort or counting sort.


I remember seeing a (gag) variation of counting sort that just provided each element as an argument to `sleep`.



That’s more like heapsort, because the OS will use a heap to implement a priority queue that contains times for when the next process needs to be scheduled.


Brilliant!

Although if it took a long time to enumerate the list of elements…


Pass them in as milliseconds or nanoseconds?


The "set of random unique numbers between x and y" part stands out so strongly that I would be shocked if the whole point wasn't to encourage a non-comparative sort.


> but sort of relevant

Hahahahaha :D

> I won the contest by setting to 0 a y-x register range of memory and then inserting each number into the range based on the number itself ("27" into register 27, "18" into register 18, etc.). Then I printed out all non-zero numbers in the range.

That's really smart. The others were just angry because they couldn't think outside of the box. Nice work!


That's not silly, that's the right way to solve that problem, unless y - x is large enough that you'd benefit from using one bit per number instead of one word. Well, usually calling qsort() would be fast enough, and less code.


That brings up memories! I did the same thing. We had to sort a million integers in the range 0-1000 (exclusive range). I did a counting sort, and beat every other solution handily.

In my case, though, the teacher disqualified my solution since the resulting lists didn't contain the same fixnums as the one i sorted, which I argued was stupid.

In the end I implemented a merge sort with some tricks and won anyway.


The same fixnums!? Fixnums are distinct from bignums in that they're small enough to fit in a machine word and therefore don't need to be objects allocated in the heap... Was this an odd implementation wherein fixnums were put into the heap anyway?


Nope. This was in PLT scheme. Fixnums were eq? to other fixnums, meaning they satisfied the most basic equality.

Arguing was futile. When we started benchmarking very large lists, the counting sort was the only one that did it in reasonable time. Despite being disqualified.


It's not silly. That's basically the "counting sort" and it definitely has uses.


The observation that "random numbers can be sorted in linear time" is actually often useful in algorithm design. A typical application is storing a sorted list of hash values.


Basically pigeon hole or bucket sorting https://en.wikipedia.org/wiki/Pigeonhole_sort which, as another commenter mentioned, are types of non-comparison sorting so you can get near linear times

I think the mainstream sorting algorithms tend to cater towards "mainstream" ordering which tends to be somewhat sorted versus purely random. An example is https://en.wikipedia.org/wiki/Timsort


I have also heard about it as radix sort : http://www.codercorner.com/RadixSortRevisited.htm


This is not the same as radix sort but you're right that radix is a non-comparative sorting algorithm just like the one the parent commenter described.


So, you implemented bucket sort?


Isn’t this usually taught right before sort algorithms? It’s a simple lead-in to the topic and is great for reasoning about memory overhead.


Sounds like an index sort. Your number space must have been fairly dense for this to work.


I like this! Could this be a digital implementation of the analog Spaghetti Sort? [https://en.wikipedia.org/wiki/Spaghetti_sort]


That's just the right solution if you know x and y (or even just x?) ahead of time isn't it?

And don't you mean "27" into register 27-x, do. 18 etc.?


Or even neither!

Much faster to find upper and lower bounds and essentially construct a hash table than to manually sort a whole list of non-trivial length.


Is it really? Building the hash can be made in linear time but iterating the values in order, especially when there might be gaps in the interval? Could be messy


The overhead of hashing in a hash table makes it worse than sorting algorithms without hashing.


ha, I did exactly the same thing!


Construction materials supplier here. That is not necessarily true. There are two major drivers for open office plans:

1) The notion in architecture that open plans, by literally removing barriers, figuratively remove them and create a more open and inclusive office environment. This is a popular belief in the biggest architectural firms in the world, even though the principals at those firms all have luxurious private offices.

2) Recently some offices have flipped to open plans to accommodate hybrid working. E.g. allowing employees to WFH 2-3 days / week but then expecting them to hot-desk when they come in. This reduces the total footprint required, which can reduce costs...except many firms are using the saved funds to enhance the office in other ways (more natural light, better furniture, plants and fountains, catering, etc.) in order to attract talent.

Gensler has published a lot of this research and positioning publicly in reports and in their podcasts.


Lancaster, PA resident here. Many, possibly most, Amish are not farmers anymore. To your point, a lot own businesses like lawn care, grocery and hardware stores, tourist traps, and construction companies.

Since childcare is mentioned, I would be remiss if I did not point out that the Amish have a disproportionate rate of child sex abuse. Offenders often get off because the Amish community pressures victims not to testify. Victims are sent to re-education centers. It is pretty gross what goes on in plain (pardon the pun) sight. But they bring in tourist dollars. Most Amish are lovely of course, but their lifestyle has a very dark side.


> Curious on your age range.

This is going to be the dominant driver for female infertility. My wife and I waited until she was 32 and I was 40 and we struggled for two years. Eventually we talked to a doctor and both got tested. I was in the 98th percentile for sperm health, but unfortunately her egg production was closer to a woman 10 years older. We did IVF and got very lucky on the first try, with one viable embryo, who is now a curious and amazing four-year-old.


Maybe google had to quickly remove solarwinds from their network. :)


Re: which CAD to use...

I work in a +1$B company in AEC (Architecture, Engineering and Construction) and our designers and engineers use AutoCAD, Revit, Inventor, 3DSMax, KeyShot, Unity, Solidworks, Rhino, SketchUp, plus the entire Adobe Suite. The Autodesk tools represent probably 75% of the install base.

I agree that learning the concepts is more important on where and what to click in any particular software set. But in AEC, Autodesk is king.


This school district in the USA almost exclusively uses Free Software:

TED Talk: https://www.youtube.com/watch?v=f8Co37GO2Fc&app=desktop

RedHat Video: https://youtu.be/Nj3dGK3c4nY

The IT Director there has also written a book: https://www.amazon.com/Open-Schoolhouse-Building-Technology-...

[Full Disclosure: He and I are good friends.]


Join us for AI Startup School this June 16-17 in San Francisco!

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

Search: