HNHacker News
TopNewBestAskShowJobs

alexhutcheson

3,780 karma · joined February 17, 2013

Nothing I post reflects the views or policies of my employers, past or present.
submissionscomments
alexhutcheson··on Linux Running on Apple M1
What instructions from an extended ISA are you referring to? The only ones I've seen written up have been:

1. Instructions to change the memory ordering mode, which allows Rosetta2 to skip some expensive memory fences when translating x86 binaries to ARM.

2. Some instructions for matrix math.

I wouldn't expect either of these to affect typical usage or battery life much, unless your workflow involves running things that do a ton of matrix math.

alexhutcheson··on BeagleV – An affordable RISC-V computer designed to run Linux
What’s the relationship between this project and BeagleBoard.org?
alexhutcheson··on Cheap and simple way to mount a smartphone directly above a laptop display
I bought a gooseneck stand to do something similar at my desk. It's very convenient because I can swivel it out of the way when I don't need it, and pull it in when I do.

I bought this one, which I've been happy with: https://www.amazon.com/gp/product/B0753GCMH5/. I'm sure there are dozens of good options, though.

alexhutcheson··on When Amazon Switched from Sun to Linux
> I recall worries that obidos might not be able to build in a few months time

What was going to cause this?

alexhutcheson··on Texas Instruments sent a DMCA takedown to a site archiving data sheets
TI might not be focused on phones and multimedia devices anymore, but their DSP chips are extremely well regarded, and some of their other chips are popular for industrial automation.

I don’t think “rentseeking on their analog dominance” is a fair characterization.

alexhutcheson··on Thonny: A hassle-free Python micro-IDE
I recommend Geany[1] to students learning Java for the same reason. Many students used to try to use Eclipse, but would get bogged down in Eclipse' notion of a "project" and other details that were unrelated to the simple tasks they were trying to do. Geany is much easier for them to get started with, and still provides a graphical way to run and debug code.

[1] https://www.geany.org/

alexhutcheson··on Simple Bank Is Closing
USAA is almost unbeatable for insurance, but I think Navy Federal has a much better banking experience.
alexhutcheson··on Simple Bank Is Closing
I managed to get my Schwab debit card eaten by ATMs twice in one trip once, and Schwab was able to quickly Fedex replacement cards to addresses I provided in Thailand and Hong Kong. It stopped my trip from turning into a big headache - I was really impressed.
alexhutcheson··on Wasmer 1.0
Also, of those options, only Qt works on mobile, and only since version 5.1 in late 2013. Qt licensing is also a bit complicated on iOS.

The web enjoyed a fairly long monopoly as the only cross-platform option if you cared about mobile.

alexhutcheson··on Wasmer 1.0
SQL UDFs might be another potential use-case in the same category.
alexhutcheson··on Wasmer 1.0
What existing sandboxing systems are you thinking of that are more efficient?

Would you be confident using them to run possibly malicious code on your system?

alexhutcheson··on WSL2 corrupting Git repositories and shell history
Windows Terminal is roughly on par with GNOME Terminal, and PowerShell is actually pretty decent once you learn it.
alexhutcheson··on WSL2 corrupting Git repositories and shell history
What version of Windows doesn’t have WSL2?
alexhutcheson··on Peer-reviewed papers are getting increasingly boring
The problem isn’t the existence of the papers, it’s the years of effort that go into producing the papers.

It’s the failure to harness the hard work and brainpower of so many brilliant people towards efforts that would meaningfully improve the lives of others.

alexhutcheson··on EU Signs €145B Declaration to Develop Next Gen Processors and 2nm Technology
The foundation just coordinates the standards process and sets rules for using the trademark. It has a tiny budget and few staff. All substantive R&D is done by its members, which are companies of various size with headquarters in various countries: https://riscv.org/members/

Changing where the foundation is incorporated doesn’t change where this R&D happens.

Also Switzerland isn’t in the EU.

alexhutcheson··on EU Signs €145B Declaration to Develop Next Gen Processors and 2nm Technology
https://nvidianews.nvidia.com/news/nvidia-to-acquire-arm-for...
alexhutcheson··on Firefox not planning on supporting PWA
The JVM world is moving away from that approach. The current recommendation is to use jlink[1] to build a custom runtime image, which doesn’t depend on a JRE being present on the system.
alexhutcheson··on Ditching Excel for Python in a legacy industry
https://readr.tidyverse.org/
alexhutcheson··on Don't Use Protobuf for Telemetry
https://research.google/pubs/pub45424/
alexhutcheson··on ACE: Apple Type-C Port Controller Secrets
Is there a specific dev board with a Cortex-M4 that you like to use, or does it vary by project?
alexhutcheson··on Bash 5.1
Rob Landley is working on implementing a full Bash clone (called toysh[1]) in his Toybox project[2], so hopefully the need to target BusyBox ash will go away in the near future.

[1] https://github.com/landley/toybox/blob/master/toys/pending/s...

[2] http://landley.net/toybox/about.html

alexhutcheson··on Essentials of Compilation
If I were teaching a compilers course, I would consider using WebAssembly as the target language. Has anyone tried this in a teaching setting?

When I took compilers, a substantial amount of student and TA time was spent learning how to use GDB to step through x86-64 machine code that had been incorrectly generated in some way. This experience was probably useful, but detracted time that could have been spent focused on the core material - optimization passes, etc. The debugging tools for WebAssembly are much easier to use.

It would also be a lot easier to set up a sandbox environment to run tests on student code - this requires a lot of work to do safely with machine code.

One major downside is that WebAssembly is a stack machine, so you wouldn't get the chance to work on register allocation, which is both interesting and important in practice.

alexhutcheson··on On the Graying of Gnome
Cinnamon is a decent alternative.
alexhutcheson··on The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster
Why would you want a 6-year-old server chip? By that point I'd expect a mid-tier consumer chip to have better performance/watt, and much better support for peripherals.
alexhutcheson··on Teardown: Identifying Apple M1's distinct circuit blocks
There are two important sets of ML workloads: training and inference. This Intel blog post does a reasonably good job summarizing the differences: https://www.intel.com/content/www/us/en/artificial-intellige....

The hardware requirements for training are much higher than for inference. Roughly:

- Training: Requires massive throughput and compute, so is done on high-end GPUs or TPUs in datacenters (or on researchers' beefy workstations)

- Inference: Compute requirement depends on the size and complexity of the model, but many models can be run on low-end smartphone hardware.

Generally speaking, any time you see "ML hardware" in a consumer device, that hardware is for inference, not training. Apple's Neural Engine is an example, of this. Other current examples include the ARM Ethos-N series and Qualcomm AI Engine, among others. Often this hardware is designed to balance inference performance with power consumption, so that inference can be done efficiently on battery-powered devices.

Training workloads on the M1 use the GPU, not the Neural Engine: https://blog.tensorflow.org/2020/11/accelerating-tensorflow-...

alexhutcheson··on Tsundoku: The art of buying books and never reading them
I now collect "sample chapters" on my Kindle, rather than buying new books. Highly recommend that approach - in my case most of the time I don't want to continue after reading the sample, so I move on to another one.
alexhutcheson··on Larry David and the Game Theory of Anonymous Donations
Another reason to give anonymously: You will avoid the flood of marketing materials soliciting more donations.

If you make a significant donation to any nonprofit that's big enough to have fundraising staff, you will continue to get mail from them for years. What's worse, some nonprofits share their mailing list with other organizations, so you may also start to get marketing material from other organizations as well.

[US-specific]: If you regularly make any non-trivial donations, it's worth considering setting up a donor-advised fund[1]. Among other benefits, it allows you to make donations anonymously, or to provide only your name, and no contact information.

[1] https://en.wikipedia.org/wiki/Donor-advised_fund

alexhutcheson··on GWSL: Run graphical Linux apps in WSL
It includes binaries for VcXsrv[1], which is an X11 server for Windows. This project itself seems to be glue code that helps you set up and manage VcXsrv and it's connection with your WSL instances.

[1] https://sourceforge.net/projects/vcxsrv/

alexhutcheson··on A new release for GNU Octave
Octave is ok for that, but not as good as Matlab. All the Matlab plotting functions work, but the plots don't look as nice, and require more tweaking to get them to "publication quality". Octave includes a Qt-based GUI, but it is not nearly as polished as the Matlab GUI.

Python is also just ok. You can use matplotlib[1] to create plots using an interface explicitly modeled on the Matlab interface, or use higher-level wrappers like Seaborn[2] which reduce the amount of work required to create nice-looking plots of certain types. "Zoom to different features" can be done via Pandas[3], or directly via NumPy, although the syntax can get a little cumbersome. The best Matlab-like GUI is probably "Scientific mode" in PyCharm[4], which requires a paid license. Jupyter notebooks[5] are a free option, but don't include all the features you'd have in a Matlab-style GUI.

R is, in my opinion, best-in-class for exploratory data analysis. The combination of ggplot2[6] and dplyr[7] makes plotting, drill-down, aggregation, etc. easier than any other environment I've used. Chapters 2 and 4 of the free "R for Data Science" book[8] provide great examples of this. The RStudio[9] GUI is comparable to Matlab in features and overall polish, and it's available with either an AGPLv3 or commercial license. Jupyter notebooks also work.

The major downside of R is that it's very optimized for interactive use, which makes deploying it to any sort of production environment tricky and error-prone. I'd mostly advise doing your ad hoc analysis in R, but rewriting things in another language if you want to deploy them. The other downside is that all the major deep learning libraries use Python, so if you want to do work in that space you should really just use Python.

[1] https://matplotlib.org/

[2] https://seaborn.pydata.org/

[3] https://pandas.pydata.org/docs/user_guide/indexing.html

[4] https://www.jetbrains.com/help/pycharm/matplotlib-support.ht...

[5] https://jupyter-notebook.readthedocs.io/en/stable/notebook.h...

[6] https://ggplot2.tidyverse.org/

[7] https://dplyr.tidyverse.org/

[8] https://r4ds.had.co.nz/

[9] https://rstudio.com/products/rstudio/

alexhutcheson··on A new release for GNU Octave
Many Chromebooks and Chromebook tablets now have a real Linux environment available: https://support.google.com/chromebook/answer/9145439
← PreviousPage 5 of 29Next →