Is this fgpa / vhdl shit still used ?

Well MS definitely uses them on some work, but I don't know if that includes any of their windows 10 work.

Yes, FPGAs are still used. I've been looking into buying a DE10-Nano to use with the MiSTer project and to fuck around with.

this is not an industrial board, it's an educational board.
It's still used to educate people about fpgas.

I never understood these things.
Are they just coprocessors?
Can you program a basic 32 bit added with them and have it be faster than your main processor?
No one ever seems to talk about what they do with them and hos they do it. I almost never see discussion beyond "yea I have one" and "yea I use one" but never anything specific about them outside of the fact that they are expensive as fuck online

>DE10-Nano
myirtech.com/list.asp?id=502

What makes the Z-Turn so special?

build an entire Ardunio processor on the FPGA and run Arduino code on your FPGA running a Arduino CPU.

next step:
build the FPGA running Arduino in Minecraft

Yes. It's not at the "forefront" of public tech, but that's because it's above the level of Linux Text Tats

They're used for prototyping, DSP, SDR, high speed custom interfaces. More hardware / electrical engineering than software

I learned on this exact one 3-4 years ago

Can this thing run a single synth VST without major latency with midi?

FPGA is actually taking over the crypto mining scene too for "non-ASIC" minable coins

Digibyte is implementing an ASIC safe algorithm which re-adjust itself every 10 days slightly.
So mainly FPGAs can be used for mining since ASICs are fucked every 10 days.
or GPUs if re-working FPGAs taking too long.

you can execute multiple instruction sets simultaneously rather then having to follow the cpu pipeline since they are just logic gates

Why do motherboards tend to have only one CPU slot to begin with?

An example of this?
So I can basically make my own simd coprocessors?

Seems like it may just cause more crashes having too many processors at once, but I could see how this could effectively compute one set of data one way in the most efficient processor for that data, then one set of data a different way on another processor that suits it.
But then again, we do seem to have one cpu for "good reason" so I think the benefit of doing this is limited.

You can make pretty much any logic circuit that doesn't exceed the available number of LUTs and FFs, it's literally a programmable circuit.

It's still a thing, just look at this beast, 8 FPGAs from Xilinx and dual 32-core EPYC CPUs.

Attached: Xilinx-Alveo-EPYC-BOXX-Top.jpg (800x635, 166K)

Yep. Faggots get attracted to the field cause "muh blank-slate do-anything silicon" but quickly bounce off it when they discover the complexity of implementation.

>No one ever seems to talk about what they do with them and hos they do it. I almost never see discussion beyond "yea I have one" and "yea I use one" but never anything specific about them outside of the fact that they are expensive as fuck online
As my master's thesis I designed some parts of the firmware for an FPGA that will be used in the Large Hadron Collider at CERN. They are used there because they needed extremely low latency and accurate timings of triggers, and very high bandwidth when recording the data from the particle collisions. That is something FPGAs excel at because you can configure them to do exactly what you want on the exact clock pulses you want to do it, and because you can parallelize data paths as much as you want. It would be impossible to achieve their requirements a normal CPU-based system.

Because unless you're a server fielding hundreds or thousands of requests and operations concurrently, one cpu is more than enough and more cost and performance effective than having say two 4 cores vs one 8 core. Plus the fact that the os and programs run by the layman are usually unable to multithread well or at all, so you have 47 cores idling while #48 does everything on a specific program

Okay but do people post this shit on git or sonething
How do you get it to read and write to ram? Or do floating point shit? Or even write your stuff to it?
Why do fpgas seem like such a secretive culture where no one talks about it outside of "I have one. I made a thing" but there is zero social aspect of people sharing fpga programs on git or talking about algorithms or anything

It's not a "secretive culture", just most of the this shit by it's nature tends to be for work/under NDA. There's also plenty of VHDL and Verilog to be found on GitHub if you're not a drooling retard.

What if your computer was working on something in non-binary?

There isn't that big of a hobbyist community because its real uses are in specialized systems that you mostly see in business. And obviously most businesses don't discuss or share their developments, it's trade secrets.

>How do you get it to read and write to ram? Or do floating point shit? Or even write your stuff to it?
Just watch some basic tutorials or read a textbook, there are plenty of them online. Basically most such stuff involve implementing it on a protocol level, e.g. writing the address to a register connected to a RAM module, then reading the data it sends out through a different register. But for common cases like that, most design tools have premade modules you can use,

Because not enough people understand the potential of it enough to put the effort into it. Alternatively, the use of it might be kept secret to those that seek to benefit from keeping it secret (e.g. if it is useful for gaining leverage over others).

The latter seems far more likely.

Frivolous example: You want to add 10 numbers together. A typical general purpose CPU doesn't have a special instruction for that, so it has to do multiple operations (and possibly wait for IO in between) to get it done i.e. you have to implement the functionality in software. With an FPGA you can create custom hardware that can take 10 numbers at a time, add them together, and spit out the result. This is way faster

And also has potential for the way we construct computers and their programs entirely if you can optimise different set ups efficiently and if have the flexibility to change from set up to set up.

what?

Attached: questioning.jpg (329x389, 44K)

Programming for parallel/multiple CPUs is pretty difficult.

>because its real uses are in specialized systems that you mostly see in business
At the moment the real uses might be for businesses, but that's only because not enough people understand the potential for these systems. A time may come where more and more people get on board with it.

Just look how quickly people flocked to bitcoin when it mooned, I imagine the same would happen when people understand the individual profits that could be made with these systems.
(and it might actually have to do with bitcoin, such as with mining ...)

en.wikipedia.org/wiki/Three-valued_logic

I'm just thinking that this could revolutionise how we make abstracted programs.

Yes, they are only getting more popular. Especially as an alternative computational accelerator after the death of Moore's Law.

They're expensive for one, and it's not exactly braindead easy to use one, since it requires extensive EE knowledge
It's not like bitcoin because of the reasons I outlined. I can see electronics hobbyists getting into these though

>It's not like bitcoin because of the reasons I outlined
No I meant it could be profitable and if it does become so it will explode.

good amount of stuff on opencores.org

Yes user, it is standard practice that all board level components deployed by companies are first tested on FPGAs

Pretty much every ASIC system started as an FPGA implementation to get it tested.