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

Can you help me understand, how can the V100 have 10x the TFLOPs of the P100, but only get a 2.5x speed increase in training a neural net according to nvidia's docs? https://devblogs.nvidia.com/parallelforall/inside-volta/

Do we need significant software changes to take advantage of the new power? Are the TFLOPs somehow not directly comparable?


Most published numbers aren’t actually using the tensor cores. We’re using dlib (which is in c++) and gives us more direct control, but surely Tensorflow will eventually do this too.



Probably memory bound.


EOL for Chrome Apps outside of Chrome OS was announced in August 2016 https://blog.chromium.org/2016/08/from-chrome-apps-to-web.ht...


But OP is specifically calling out how this hurts Chrome OS, but it doesn't seem to affect that platform, from what I can tell.


One way it affects it is to reduce the overall market for Chrome apps, reducing the incentive to make an app in the first place.


Exactly. I doubt most of the ChromeApps that exist today were made for ChromeOS. They were made because it was an easy and straight forward way of making a webapp on the desktop. Now, as OP mentions, everyone is moving to Electron and NWJS, neither of which works directly on ChromeOS.

The only upside I can see is that CrOS is soon going to support running Android Apps which may save it, but even then... Maybe they'll figure out a way to run Electron/NWJS apps on ChromeOS?


This might also motivate better support for running ChromeOS apps under standard Linux.


Maybe Google is looking to run Android apps in chrome?


The massive ChromeOS market as compared to the Chrome extension app store?


OP isn't calling out how it hurts CrOS, they're describing how it's the place that Chrome Apps are still available, and thus could have that functionality copied out of CrOS and rejiggered to (continue to) work in Linux.



I wish more web services provided opensource versions (Service-as-a-Software). Large orgs would still pay for service contracts even if they self host, and just having the option makes everyone more likely to invest on top of your service.


This is a very good point. Even if you don't want to self-host and you'd rather use a service, having the software available as open source so you can self-host if you so choose, makes it more rational to use the service, because it makes it harder for you to be screwed the way OP's startup was.


But you can bet what then most of your customers do, right? This sounds good in theory but is really tough in practice ...


Seems to be correct, went from $120 to $200 by the time I made it to checkout... which I then abandoned, but I was buying it for personal stuff.


Chat at "indie scale" is fairly easy. 10-20k concurrent users fit into one or two machines with off the shelf software (ejabberd) with minimal expertise. 1k-2k concurrent users could easily be done in almost any stack.

If you're doing more than 20k concurrent users, odds are you can afford to hire someone to scale up or build out a team to do it, or find a 3rd party service and pay them to host your chat.


Indie don't have 20k CCU.

Openfire is a known server for XMPP:

http://www.igniterealtime.org/projects/openfire/


> Openfire is a known server for XMPP

Yes it is, and so is ejabberd, in GP's comment. (And the most popular iirc)

They were talking about indie game studies, not indie chat protocols.


The author works for mozilla, and firefox is a common theme on his blog. I think it's for regular readers of his blog to understand the practical impact (readers he assumes care more than average about firefox).


He didn't post it by mistake. He asked Visual Studio to create a private repo and it created a public one. A bug which he reproduced.


It wasn't even visual studio but an extension created by github that caused the issue. Visual studio does not come with github integration by default


Sure, but he shouldn't have checked keys in to any repo, public or private!


The sources I could find say 4% is used for plastic and another 4% for the energy to make the plastic. Nowhere near 50%.

http://www.plasticoceans.net/the-facts/energy-consumption/


Found this after Joël tweeted the quote "The dilemma every artist confronts, again and again, is when to stick with familair tools and materials, and when to reach out and embrace those that offer new possibilities." So true.

If there's one thing I've learn from reading about Systems Theory / reading Russell Ackoff, it's that every field, every practice, every art is far more similar than they are different.


"This method takes a foo object, applies HTML encoding, and returns an array of the original and the clean code." could be a 3-line unit test that's just as clear to read, but has the side effect of actually being true instead of maybe being true!

I would consider that comment just as bad as the ones in my examples, but maybe not quite as obviously so.


The main benefit of comments that you seem to be missing is that they are in the right context. Unit tests, while useful and required, make for boring reading and are not right where you need them. If I need to open up a separate file and scroll through tons of other tests to find the unit test "docs" on a function, you've lost me already.


Of course a better comment could be written. I was using a short and concise comment to illustrate how you can gain a huge amount of understanding in a single line versus parsing a long function definition. Other comments here make suggestions for writing good comments.


> could be a 3-line unit test that's just as clear to read, but has the side effect of actually being true instead of maybe being true!

It also has the side effect of being in a different file to the one you're trying to read.


Hey, actually, I think I found exactly what you want, maybe!

http://docs.python.org/2/library/doctest.html


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

Search: