It Looks Like The Open-Source GPU Will Fail Again

[Mad Hobo]

LINK

Stop linking to yourself phaggot

[[wrd]]

who would sponsor that?

I think this was a nice idea with the wrong timing, the wrong code and the wrong and the wrong estimation.

There is a series of things I really don't about this project:

- there isn't an open source verilog compilers that will synthesize this into an FPGA (gate count to big for the hardware)
- hardware isn't being sold with the backing ( I mean it's not reasonable, there is not support except for VGA iirc and you need to use PCI instead of PCIe). there isn't any eval board or similar where you can run that thing out of the box. This is completely not "buy and use" it requires skills in board design fitting the code to your board and lots of other things.
- the code base solves problems really old! it's not as if an fpga will produce anything similar to even an intel graphics chip.

We seen in the past that code-dump projects (like open64?) are not likely to drive momentum to open source community.
Furthermore FPGA based products are considerably more expensive than ASIC based solutions which are unlikely to be open sourced.
On top of that FPGA designs are currently developing at quite a pace and are tightly coupled to their development environments.


In my opinion focus should be driven to open source tooling first. It's useless if single CPU licenses to compile the code cost >1000eur and the board to run the code just as much. Unless everybody who can afford a desktop can put 200eur aside and try it himself projects like this are doomed to fail.

[Ibidem]

In my opinion focus should be driven to open source tooling first. It's useless if single CPU licenses to compile the code cost >1000eur and the board to run the code just as much. Unless everybody who can afford a desktop can put 200eur aside and try it himself projects like this are doomed to fail.

Have you perhaps heard of ISE Webpack?

[[wrd]]

Have you perhaps heard of ISE Webpack?

Yes I know of that. I don't know how Xilinx handles that, but it might be less bright on the Altera side.

The gate count of the strechgoals will most likely exceed the free quartus versions. From what I got they're not targetting the cyclone FPGAs.

Furthermore who guarantees that the free synthesis tools stay available?

[Kivada]

Putting a link to the kickstarter page in the article might help.

(Could someone make a firefox plugin that replaces internal links on phoronix pages with useful links to the projects)

Stop linking to yourself phaggot

Oh hi, you guys must be new here. Larabel only posts click bait and only links back to himself because you see it's far too difficult for him to make money running a legitimate site.

[Kivada]

Also, I'd like to gloat this time and say I told you so to everybody that tried to shoot down my questions about how this project could ever make money seeing as far as I could tell there was absolutely no market for it and as such would have no funding backers.