vzidex

I'm very happy with my Leopold with Cherry MX Clears, having come from a Model M as my first board. Alternatively, at this price point one could put together a very decent custom from a kit + switches + keycaps.

I have an original production Model M and don't think I would ever go back to it. It's extremely loud (friends would always complain if we were gaming together), keypresses are very heavy, and it takes up a ton of deskspace.

If the Model M or F are appealing to you, at this price point you should also consider some more modern production keyboards or even building your own from a kit + switches + keycaps. If you're after the heavy & tactile keypresses of buckling springs a board with Cherry MX Clear (common) or Green (less common) will probably satisfy you - I can highly recommend any of Leopold's boards with Clears in them.

Fascinating, thank you! Admittedly I don't keep the closest tabs on what Xilinx is doing.

Disclaimer: I work in this space (not at Xilinx), comments are strictly my own opinions and do not reflect any positions of my employer, etc.

Broadly speaking, FPGA-based ML model accelerators are in an interesting space right now, where they aren't particularly compelling from a performance (or perf / Watt, perf / $, etc.) perspective. If you just need performance, then a GPU or ASIC-based accelerator will serve you better - the GPU will be easier to program, and ASIC-based accelerators from the various startups are performing pretty well. Where an FPGA accelerator makes a lot of sense is if you otherwise need an FPGA anyways, or the other benefits of FPGAs (e.g. lots of easily-controlled IO) - but then you're just back to square 1 of "there's some cases where an FPGA makes sense and many where it doesn't". Besides that, a few niche cases where a mid-range FPGA might beat a mid-range GPU on perf / Watt or whatever metric is important for you.

Again, opinions are my own and all that. As someone in the space, I am very much hoping that someone - whether an ASIC startup or Xilinx / Intel come up with a "better" (performant, cheaper, easier to use, etc.) solution than GPUs for ML applications. If the winner ends up being FPGAs, that would be really really cool! Just at the moment it's not too compelling, and I'm trying to be realistic.

All that said, FPGAs and their related supports (software, boards, etc.) are an $Xb / Y market - nothing to shake a stick at, and there are many cases where an FPGA makes sense. Just doesn't currently make sense for every dev to buy an FPGA card to drop in their desktop to play with.

You've struck on the fundamental problem that the FPGA industry has been trying to solve for 30+ years - how to get an FPGA into the hands of every developer, like how GPUs have propagated to be essential tools.

Nobody has come up with a good answer yet. Developing for an FPGA still requires domain-specific knowledge, and because place & route (the "compile" for an FPGA) is a couple of intertwined NP-hard problems development cycles are necessarily long. Small designs might take an hour to compile, the largest designs deployed these days ~24H.

All this to say is that while they are neat, nobody has found the magic bullet use case that will make everyone want one enough to put up with the pain of developing for them (a la machine learning for GPUs). Simultaneously, nobody has found the magic bullet to make developing for them any easier, whether by reducing the knowledge required or improving the tooling.

Effort has been made in places like High-Level Synthesis (HLS, compiling C/C++ code down to an FPGA), open-source tooling, and (everyone's favorite) simulation, but they all still kinda suck compared to developing software, or even the ecosystem that exists around GPUs these days. You'll often hear FPGA people saying stuff like "just simulate your design during development, compiling to hardware is just a last step to check everything works" - but simulation still takes a long time (large designs can take hours) and tracking down a bug in waveforms is akin to Neo learning to see the Matrix.

I've done it as well with my parents, who love wine. At the shops I've been to, you buy the grape concentrate up front and they handle to process of making it up until bottling. Once the juice has fermented, you go in and they walk you through cleaning and sanitizing the bottles, filling them, corking them, and usually shrink-wrapping the tops. They provide all the supplies except the bottles themselves - customers bring their own, saved from buying wine the "normal" way.

Never seen a shop do ageing, so the wine will be noticeably "young". My parents like dryer and sharper white wines anyways (Pinot Grigio, Riesling, etc.) so it doesn't bother them. Also note that due to taxes and such, the cheapest wine you'll find commercially is C$11 a bottle, so even at C$20 / gallon you're getting a great deal if you like the resulting wines.

Personally, I quite like the wines my folks get through these shops - properly chilled they make a wonderfully refreshing beverage in the summer, and we'll often drink a few bottles on the back deck together when I go to see them.

It might very well be the DAC doing the work then, or maybe the placebo effect is just making me pay more attention to music I first heard years ago and I'm picking up more :)

Thanks for the information though, I'll do some A/B testing to see if it makes any difference to me.

Anecdotally, recently my dad got me a Tidal subscription (which has masters at supposedly higher quality than CDs) and an entry-level DAC. The two put together have blown me away and I've been having a blast the last couple of weeks re-listening to all my favourite records from as far back as the 70s.

I think the 70s are also around the same time that a lot of familiar genres started to emerge, while music from before then is often dismissed as "oldies" or saved for special occasions - e.g. old crunchy recordings of Christmas songs.

Adding to the comment about difficulty, the number and ratio of likes : dislikes on a problem is a good measure of difficulty. A Medium with 5K likes and 300 dislikes is probably quite a good (well-written, constrained, fair) problem. A Medium with 300 likes and 800 dislikes I'd steer clear of.

Yes, Intel will still sell you a license for NIOS if you want to harden it into your product.