Hacker News new | past | comments | ask | show | jobs | submit | mg794613's comments login

It doesn't harm diversity. The world is a diverse place.

Also, America's idea about "diversity" is extremely non-diverse. And quite frankly rather racist.


Its also a anti meritocratic caste system. That desperatly tries to ignore the different academic outcomes various cultures create . So it may be racist, but at least its culture blind.


True, the "culture" of oppressed poverty really does produce poor outcomes...

It's unfortunate that the Affirmative Action initiative wasn't strictly economically based.

I wonder what the academic outcommes will be for the children of Gaza over the next few years? An extreme example to be sure...


Ok wow, I did not expect to be actually entertained!


Yeah me, ndd couldn't be happier about it. Let others slave away for such horrible employers.


On their example page you can sometimes see what data it got to train on. https://imgur.com/a/CuXNATp for example is without a doubt "Ragnaros" from the game "World of Warcraft".

Probably due to not having much different sources for "A fiery humanoid monster with flames burning".


What is a good fpga starter board that's big enough to load this? It's not obvious from the repo as a complete novice in fpga land.


The author is targeting the DE-10 Nano which is also used in MiSTer setups so that's probably the best choice


$300? Christ… I bought by MegaDrive+32X+MegaCD for less than that combined recently.

Tho I’m glad this project exists that’s… a huge chunk of change for a 30 yr old console


It's tens-to-hundreds of 30 year old consoles though, https://mister-devel.github.io/MkDocs_MiSTer/cores/console/

but yes, it is fairly excessive if you just want to play games, emulation is absolutely "good enough" and even indistinguishable for most. We do it for the love of the hobby ;)


Replicating the physical interfaces does a lot for the quality of the emulation experience. Take, for instance, the C64 maxi: it’s physically a C64 (dimensions closer to a VIC-20, but that’s negligible) and, while the emulation is good, it feels much closer to using a cycle-accurate C64 over a PS/2 keyboard.

I don’t think the potential market for Amiga and ST keyboards for use with MiSTer justifies the investment, but I’d LOVE to see LK-411, Atari ST, Amiga, Sun, Symbolics, and other pre-PC-101 layouts as reasonably priced USB keyboards.


Also, those original components have more than 25 years-old... They can fail anytime. DE10-nano is new hardware with new components which last other 25 years at least.


Realistically the only components that are likely to fail though are the capacitors, which are trivially replaced.

While other parts CAN fail it’s physical damage that’s much more likely a problem


I didn’t think MiSTer was particularly concerned with accuracy like this project is? Just implementation on FPGA.

Am I wrong there?


It depends on the core. For example Furrtek decapped the Neo Geo chips to write that core, and thus is extremely accurate.


Gotcha. So core dependent basically, rather than a specific project goal?

That tracks with what I thought. Great to know about the NeoGeo!


Yeah, the MiSTer project is really just a loose collection of people with a loose common goal. So the true accuracy of cores does vary a lot. Most users don't really care, as long as they provide a pretty accurate experience. I think most MiSTer owners (myself included) prefer it over emulation due to having no lag. I also really like that the MiSTer does not buffer any writes, so there is no shut down. Just turn it off when you are done, allowing my setup to be very console-like.


My feeling is that it’s more reverse engineering than this, which aims to be an exact replica of the original.


Unfortunately the original consoles will stop working and existing outside of museums over time; I see projects like this as archival work, so that in tens, hundreds, maybe thousands of years people can still rebuild and play these games.

Someone elsewhere pointed out that PS1 disks (or any disk based system for that matter) suffers from disk rot, so likewise, what is now considered piracy should also be considered archival work since the original disks will stop working.

something something entropy, something something the inexorable passage of time.


Oh absolutely: I fully support the project existing & data archival as well.

Just… ouch on the cost of running it. Limits the effective impact somewhat.


That price, isn't that just because you're looking at a development system?


Which you need to run this project.

I get why it’s expensive, but that doesn’t mean it’s not expensive.


Can the same FPGA be placed on a board more targeted to emulation that could, maybe, be a lot cheaper?


very same fpga chip alone is more expensive than DE-10 nano https://octopart.com/5cseba6u23i7n-intel+%2F+altera-88365716 Be sure to send your thanks to EC, FTC and the Antitrust Division of the DOJ. Altera and Xilinx mergers clearly were super beneficial to customers and the market, competition has never been better with both competing by rising prices and cutting volume in unison.

https://ec.europa.eu/commission/presscorner/detail/en/IP_15_...

https://www.sec.gov/Archives/edgar/data/768251/0001193125153...


Yikes.

But I wouldn't place too much blame on the companies - COVID wrecked a lot of supply chains as well - as anyone who runs emulators on RPis will be able to confirm.


Covid showed companies new way forward. Cut supply and bump prices, this is the source of all semi shortages. There was no earthquake, no one ran out of rare ingredients, corporations simply cut fab orders.

rpi shortage is a direct result of Broadcom (Avago) deciding to stop manufacturing chips that dont generate hundreds of millions in revenue a year. FPGAs were same deal, post merger Xilinx/Altera cut manufacturing.

Here is Nvidia not making more GPUs instead of lowering prices to something market would bear https://www.extremetech.com/gaming/report-nvidia-has-practic...


I think maybe that board is so expensive precisely because it's used for MiSTer.


FPGA vendors being taken over by cpu companies and generally not giving two Fs about manufacturing enough to satisfy demand is the reason https://octopart.com/5cseba6u23i7n-intel+%2F+altera-88365716


The de-10's price has more than doubled since covid hit.


I don't know the precise answer, but my guess is that it would be ported to the MiSTer or something like it.

https://mister-devel.github.io/MkDocs_MiSTer/#what-is-mister...

The core of the MiSTer looks pretty beefy.

https://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=...


But what would it offer over this? https://github.com/MiSTer-devel/Genesis_MiSTer


AFAIK, an FPGA board is not enough, you also need the "compiled netlist" from the Verilog files, clock configuration and routing info, tailored to the specific FPGA/board you choose.


Kinda, sorta? This is like saying "a computer is not enough, you need a binary." What's here is the source code in Verilog. Once this is compiled by a Verilog toolchain, this produces a netlist (set of gates) which is synthesized into an actual bitstream (proprietary gate configuration file) for a specific FPGA hosted on a specific board.

In this case, the project isn't very documented but it looks like fairly generic Verilog without a lot of vendor specific extensions. So, what you need is a Verilog toolchain which can synthesize the source code into a netlist, and then into a bitstream, and the right set of extra code to target an actual physical piece of hardware.

Right now, it looks like the only board support that's checked into the repository is for the Icarus Verilog simulation environment: https://github.com/nukeykt/Nuked-MD-FPGA/tree/main/icarus .

But, the overall setup looks pretty simple and generic, so it should (hopefully) be possible to synthesize to your board of choice by reimplementing run.v and memstubs.v towards an actual hardware configuration.

Basically what you'd want to do to start trying to run this on real hardware is to build a hosting environment which wired the inputs and outputs from `md_board.v` into the real hardware provided in your environment, possibly by integrating other soft-IP (ie - you could attach a cartridge emulator of some sort to the cartridge lines, your choice of video encoder to the video output, and so on).


> proprietary gate configuration file

A fully open-source FPGA/CPLD and toolchain would be very disruptive. How hard would it be?

It could be manufactured by any number of smaller foundries with older processes. No need for a 3nm node for this.


Aka the software that comes with the FPGA vendor. I think Intel, AMD, the Chinese vendors and some parts from Microchip have the software available for free.


Before intel acquired Altera they had some free (as in beer) software for it, helped me lots in grad as I could design, verify and simulate at home.


This is such a parallel universe of computing


Parallel, and much worse. Everyone out in the software land with their open languages, free compilers, and package repositories doesn't know how bad it is in FPGA land.


It feels like embedded development 20 years ago.


making notes


TBF he said commit, not attempt.


I don't however, it's not my cup of tea, but since when are payment processors, which are into money, into applying cultural norms?


Since forever. There are a few reasons why - one is cultural the other is risk. On the cultural side industries that are "controversial" tend to be avoided. Payment processors don't like to be associated with porn, gambling, drugs etc. because of potential negative PR. If you're a payment processor and trying to land big conservative clients, having press about how you support the porn industry doesn't do you any favors.

The other side is risk, meaning industries that have an above average chargeback (fraud) rate. Many times these industries overlap - porn for instance. Lots of stolen credit card numbers are used to by porn, but also legitimate purchases can easily end up as a chargeback. The classic example is that someone buys porn on their credit card and then their partner finds out. Instead of admitting they bought porn, they claim someone must have stolen their credit card to save face.


I’d upvote this if I could. Chargebacks aren’t impossible after paying via bank transfer - it might be called another name, but it has been done before to us by a scammer (but we can dispute and fight it). And one time after a customer added an extra 0 to their transfer and immediately tried to get their bank to claw it back without contacting us.

But we basically never had issues with chargebacks. Even when we used PayPal and Stripe for two years, we only had two chargebacks and we disputed one of them and won through PayPal as we proved it was fraudulent.


Nor sure what this article is trying to convey? Is it about grepping wrong defaults for bash? Is it about mentioning the protagonist as much as often? I've read it twice now, and I feel ignorant.


The environment variables count towards the argument limit. To a casual shell user, this is non-obvious and typical error messages are bad and don't even mention it. They say "Argument list too long" even when the argument list (argv) isn't too long, the environment list (envp) is.

So the author went through an investigation to figure out what was happening, and decided to write down her findings.

----

On a personal note, reading this caused me to look at fish's error message in this case. It mentioned "environment list", but tbh that's still too jargony and non-obvious.

So the next version of fish will now explicitly mention "environment variables" in the error message here, and it will print a hint if the environment variables account for over half the limit.


Nice to watch the gears of open source and open ideas turn.


Ah yes, the no-code dream of shareholders. ”Why pay those pesky developers if you can just click it together, how hard can it really be, I became CEO/sales/random-other-thing because I'm really smart, so this will be easy too.”

Every year I see this post rear its head, can we agree we only post it again when it's actually possible and not some CEO 'has a great idea that nobody thought about'?


Nice how this article pus together people who want more flexibility in work location with people who hate work and try to work as less as possible. Makes you almost guess where theguardian.com owners stand on this issue.


I think Gavin’s probably an ass but don’t see anything wrong with being anti-work in general. I would be pro work if most work was meaningful, but am generally anti-work because most of it is exploitative and we do far too much of it in the US.

I’m not out here stealing from my job like Gavin but my attitude towards work is probably the same as him. Some of us just grin and bear it instead.


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

Search: