georgehm

Embedded within that developer page is a good explainer of the encoder free architecture . https://newsletter.maartengrootendorst.com/p/a-visual-guide-...

I have fond memories of implementing an optimizing compiler for the CS241 compiler course offered back then by Prof Michael Franz who was a student of Niklaus Wirth, probably the most exhilarating course during my time at UC Irvine. This was in 2009 so my memory is vague but I recall he provided a virtual machine for a simple architecture called DLX and the compiler was to generate byte code for it.

Google search points me to https://github.com/cesarghali/PL241-Compiler/blob/master/DLX... for a description of the architecture and possibly https://bernsteinbear.com/assets/img/linear-scan-ra-context-... for the register allocation algorithm

>Effectively, eight CPUs run the flight software in parallel. The engineering philosophy hinges on a >“fail-silent” design. The self-checking pairs ensure that if a CPU performs an erroneous calculation >due to a radiation event, the error is detected immediately and the system responds.

>“A faulty computer will fail silent, rather than transmit the ‘wrong answer,’” Uitenbroek explained. >This approach simplifies the complex task of the triplex “voting” mechanism that compares results. > >Instead of comparing three answers to find a majority, the system uses a priority-ordered source >selection algorithm among healthy channels that haven’t failed-silent. It picks the output from the >first available FCM in the priority list; if that module has gone silent due to a fault, it moves to >the second, third, or fourth.

One part that seems omitted in the explanation is what happens if both CPUs in a pair for whatever reason performs an erroneous calculation and they both match, how will that source be silenced without comparing its results with other sources.

> Having two compilers helped ensure that the language was clearly defined.

> When the two compilers differed, we knew that we had to clarify the spec and figure out what the right behavior should be.

This idea of having two implementations for spec stood out for me.

Maybe I am being foolishly optimistic, the section in the video (~ 4:50) where it is mentioned that some giant seed corporation is staying the course betting that this thing will blow over in a quarter or so actually gives me some hope ?

promptly ordered 2 (an HD and 4K) .. its still cheaper than the new "AI" enabled experience ..

I am super curious to know what kind of software magic they did to bring the NDL system as the primary navigation system after their own system failed to function prior to landing .. From what I understood listening in to the broadcast, NDL was only a demonstration project and folks somehow managed to bring it up on demand and use 2 of the sensors on it (?) iirc .. Like how !? Was NDL set up in the same orientation as their original system ? If not, they updated their guidance system on the fly too !?? Simply amazing

No they didn’t because until the loan is repaid you can’t use whatever gold you put up for other purposes can you. What became tangible is what you used the loan for , that is what can be taxed.

Why? Here the asset used as collateral is a commitment to return the money if the loan isn’t repaid. Imagine I bought gold back when gold was 100$ , it’s current price is 1000$ and now I use it as collateral to get a loan . How is it fair that I have to pay taxes on 900$ of unrealized gains ?

I like to think that to get a sense of how this might be done, one way maybe to extrapolate from this experiment at https://til.simonwillison.net/llms/python-react-pattern .

Does the iPhone 14 still have the battery glued to the back? Couldn't figure it out from reading the article. I recently replaced the battery on my old iPhone 6s using the ifixit repair kit. Of all the steps I found removing the tape attached to the battery was the flakiest/hardest operation. After many attempts I did remove it but I also ended up cracking the battery a bit :sigh:. Any changes that make the removal of the battery easier would be a great win for repairability.

> And once we knew (p-1)(q-1), we could then use some more little tricks to recover p and q, the prime factors we wanted.

I am curious to know about these tricks to recover p, q. Does anyone know?

Adding some more related articles. this was mostly a result of me trying to find some more useful articles to better understand and it was lost in my browsing history.

https://overreacted.io/algebraic-effects-for-the-rest-of-us/

https://users.scala-lang.org/t/from-scala-monadic-effects-to...

https://dl.acm.org/doi/pdf/10.1145/3122975.3122977

well there's https://web.archive.org/web/20111012115624/http://itasoftwar...

big plus one for Kotlin as a general purpose language

Thanks @njharman , how does one consider between rewrite vs expand when it is known that some requirements will be true not for the current release but 2 or 3 releases down the line. Would you still prefer rewrite vs expansion? The reason I ask is I got severe pushback recently where I proposed building generic enough components and infra to go along with it that could have satisfied the requirements for the next 2 or 3 releases

>What if, instead, we take a position that rather than enhancing components, we replace them. Then the best way to know when something needs to be replaced, is when it doesn’t do what it says on the tin.

This is something that I have had trouble reaching a balance. Replacing components could be costly and therefore it makes sense to design components that are extensible (and seem over engineered) especially when the requirements aren’t known completely? What are your thoughts on this

Here's another one using WOOT (Without operational transform), which I read is a predecessor of CRDT - https://hal.inria.fr/inria-00071240/document . It was fun implementing it. https://github.com/hmgeorge/woot

Thank you!

His notes on programming are very well written. I recently stumbled upon them. Reading the rotate and partition sections were real eye-openers! Highly recommended http://stepanovpapers.com/notes.pdf

no mention of Yahoo Pipes?

The author answers some questions in the comments as well. Worth checking out!

Does anyone know of papers and/or example implementations of similar DNNs for accoustic modeling using tf or some other framework?

The slides for the course (cs166) themselves are a very good read.

If someone has the model running already, could you please share the captions generated for the images mentioned in page 25 of http://cims.nyu.edu/~brenden/1604.00289v2.pdf .

"After training, we can sample the network to generate synthetic utterances. At each step during sampling a value is drawn from the probability distribution computed by the network. This value is then fed back into the input and a new prediction for the next step is made. Building up samples one step at a time like this is computationally expensive, but we have found it essential for generating complex, realistic-sounding audio."

So it looks like generation is a slow process.

I personally found Andrew Ng's videos on Reinforcement Learning from cs229@stanford + inverted pole balancing programming assignment to be great intro's on the topic.

Same idea, different use? https://blogs.msdn.microsoft.com/oldnewthing/20110921-00/?p=...

Did this remind anyone of the Polis' spreading and searching the stars in Diaspora (by Greg Egan)

Just came to know that it is not possible to rearrange (organize option from G+ photos) photos in an album.. its strange Google missed it...