Xliinx/AMD: no more free Linux licenses...

Darn. No more lifelong licenses, it's all annualized now (so they can change the terms more easily I suppose), and the new free version is only on Windows not Linux.

https://www.amd.com/en/products/sof...ions.html#tabs-cbefba2790-item-c03136817a-tab

Well, I guess it's the end of the line for my Artix-7 base projects :-( Version 2025.2 will be enough to work with the current hardware, but not a lot of point in creating new hardware when the software support is gone.
 
It was bad enough as was - the amount of effort with signups and agreements and the pure size of the installs meaning I had to keep deleting them to get space back were part of what held me back learning FPGA stuff for both main manufacturers.

Guess that they have no interest in home users, which is sort of fair enough, except that once in a while, playing at home is where a big product comes from. I have no doubt that Arduinos drive decades of AVR sales.
 
Guess that they have no interest in home users, which is sort of fair enough, except that once in a while, playing at home is where a big product comes from. I have no doubt that Arduinos drive decades of AVR sales.
Yes, it's very short-sighted. Having really great support for your FPGA freely accessible to those who won't/can't pay is a great way to lock-in mindshare and expertise.

Hobbyists, students, small companies were using those licenses. Without them, people are going to look at alternatives, including boosting support for open-source tools. Which will help their competitors using those tools as part of their toolchains, and so lowering the barrier of entry for new players - including many of those based in China.

It's probably good for open-source tools eventually (and therefore everyone long-term), but I can't understand what idiotic money-saving rational Xilinx/AMD has to do this stupid of a move... You'd think no-one at AMD noticed that CUDA is why people buy truckloads (literally these days) of NVidia GPU, and that CUDA is free from day one because NVidia was smart enough to realize it's just necessary NRE to help the primary business, selling hardware.
 
Oh, what a bummer. I was just looking at your IIsiA7 project with an eye towards a potential quadra pds version (using an edge connector like the Q610 PDS and nubus adapter boards to sidestep the unobtanium KEL/Robinson-Nugent/3M connector problem). Sad that this might be the end of the line.
 
Oh, what a bummer. I was just looking at your IIsiA7 project
The linux 2025.2 might remain usable (though it calls home, so you never now...), and the 2026.1 on Windows still has a free license though for one year only. So it's not entirely dead.

with an eye towards a potential quadra pds version (using an edge connector like the Q610 PDS and nubus adapter boards to sidestep the unobtanium KEL/Robinson-Nugent/3M connector problem). Sad that this might be the end of the line.
I'm not sure what you mean; you want to plug a card-edge in the PDS slot ?
 
I'm not sure what you mean; you want to plug a card-edge in the PDS slot ?

The Quadra/Centris 610 (aka WLCD) has an MCA-style edge connector socket on its logic board that exposes the PDS signals. As far as I know it's only used by 3 cards: the passive adapter/riser for the Q610 DOS Compatible (just routes the signals to a standard quadra PDS connector), the active Centris/Quadra 610 nubus adapter, and the Asante MacCon 610 ethernet card (e.g. https://www.ebay.com/itm/226177508165 - this is a weird one as I don't think very many 610s were sold without onboard ethernet populated)

I'm working (gradually, as time permits) on a couple of Wombat-derived board designs, and was considering adding an edge-connector style PDS using something widely available (so, not the Q610 board connector, sadly, as it has an uncommon pin count/layout and is just as hard to source as the standard quadra PDS connectors) to accommodate fun retro projects. But a Q610-A7 would be a nice starting point since those slots already exist.
 
and was considering adding an edge-connector style PDS using something widely available
I'd recommend a real connector, preferably some sort of modern high-speed high-density connector. It helps with signal integrity issue and with mechanical support. Look at what is available in large quantity at JLCPCB, and that should be popular enough to be somewhat future-proof.

(edit: i will admit to a bias, I deeply loathe card-edge, it's just a cost-cutting measure from the 70's that should not have survived the 80s IMHO...)
 
Yes, it's very short-sighted. Having really great support for your FPGA freely accessible to those who won't/can't pay is a great way to lock-in mindshare and expertise.

<snip> look at alternatives, including boosting support for open-source tools <snip> based in China.
The lack of OSHW FPGA tools is mostly why I've shied away from FPGA development. I just don't want to do it on a PC/Windows. I understand Tang Nano has OSHW tools?

@Melkhior does this mean your FPGA Sparc is on hold?
 
Quartus lite is still free at least, for Alteras. Might be the only way to go now. There’s a whole lot of used chips on the grey market these days (see: all the mister fpga clone boards).

I have wondered about the idea of a nubus riser for the de10 nano since they’re so easy to get clones of. But I’m still no hardware designer.
 
The lack of OSHW FPGA tools is mostly why I've shied away from FPGA development. I just don't want to do it on a PC/Windows. I understand Tang Nano has OSHW tools?
I think the Gowin stuff is somewhat supported by open-source tools, yes. The current issue is that the FOSS tools are not at the level of the incumbent's proprietary solutions that have been in the work for decades. So for really small FPGA it's OK, but the bigger the FPGA, the bigger the gap between proprietary and FOSS.

Which IMHO is a big reason for incumbent to make their tools available. Again - the more users, the better FOSS tools become, the easier for new players to to come in, and the bigger the FPGA they can be competitive with...

AMD was loosing to intel for decades in the HPC market because of Intel ICC, IFORT ant the MKL; they have been loosing for decades to NVidia because of CUDA; and in the one area were they had the best software, they decide to make it less accessible... they have no clue about software, do they!

@Melkhior does this mean your FPGA Sparc is on hold?
Not working on SPARC at the moment (temlib works fine AFAICT and has a version for MiSTer), I'm working on Sun-3 (and Sun-2) based on the MC68020 (resp. MC68010). The network is the troublesome parts at the moment...

It's not yet affected - the software still works and for now I don't think they're planning to brick older versions. Doing new projects is sketchier, though the 7-series are supposed to be available up to at least 2040.
 
Back
Top