I’ve spent several hundred thousand on Xilinx FPGAs yet they nickel and dime me for licenses. It’s not the cost that’s a problem—-it’s the hassle of making a PO for a license to set up new computers, set up CI, hiring new teammates, setting up for interns/students. Xilinx has continued to go downhill since their acquisition by AMD.. it used to feel like it was run by engineers who understood their customers, now it seems to be getting taken over by the MBA crowd who only understands pinching pennies and chiseling their own loyal customers
Yep. Strategy Letter V by Joel Spolsky (2002): "Smart companies try to commoditize their products' complements". Also, from the 2004's "How Microsoft Lost the API War":
The logical conclusion of this is that if you’re trying to sell operating systems,
the most important thing to do is make software developers want to develop software
for your operating system. That’s why Steve Ballmer was jumping around the stage
shouting “Developers, developers, developers, developers.” It’s so important for
Microsoft that the only reason they don’t outright give away development tools for
Windows is because they don’t want to inadvertently cut off the oxygen to competitive
development tools vendors (well, those that are left) because having a variety of
development tools available for their platform makes it that much more attractive to
developers. But they really want to give away the development tools. Through their
Empower ISV program you can get five complete sets of MSDN Universal (otherwise known
as “basically every Microsoft product except Flight Simulator“) for about $375.
Command line compilers for the .NET languages are included with the free .NET
runtime... also free. The C++ compiler is now free. Anything to encourage developers
to build for the .NET platform, and holding just short of wiping out companies like
Borland.
Although it also means Windows could rely on Intel so long for compiler tools, that when I was trying to build for ARM customers, I realized a lot of the expected developer tools are just barely functional or don’t exist (ifort, MKL, gdb, mingw, etc)
Similarly, AMD wonders why they can't have a bigger piece of that sweet GPU market. Maybe Nvidia is worth 5 trillion because they have a deep ecosystem (CUDA) and community and there's certainly not charging or impeding developers from using it, charging only for the hardware.
I can get parts, they're part of a BOM that gets approved, but getting POs approved for software is a pain in the ass. Been considering switching next gen stuff to microchip.
About 25 years ago I did some work on Xilinx FPGAs. The hardware was nice. The ability to get them free (as a business development strategy) as a research user was nice [0]. The quality of the software stack was awful. The FPGAs were getting big enough that the synthesis process could need more 4GB of user memory, but the tools were barely functional on 64-bit systems of any sort, and this was far from the only problem.
It’s too bad that a vendor with only slightly worse hardware hasn’t eaten Xilinx’s lunch by having a genuinely pleasant development process. These companies aren’t selling synthesis tools — they’re selling hardware. Make the developers want to use it! (Intel, for all their faults, understands this.)
[0] It takes a special sort of incompetence to take 5 figures worth of fancy chips, stick them in a poorly packaged box, write only part of the address in it, and hope that FedEx will deliver it somewhere useful.
Xilinx has often thought of the software as the special sauce that sets them apart it seems like.
At this point its a net negative. Nothing like a massive bloated Vivado that now requires a slow spywareOS to run its rotting carcass of gigabytes (100's) of java.
About the only good thing Vivado does is fail to synthesize correctly... oh wait that's a mystery bug that I and many others have run into. And impossible to understand why.
FPGA tooling desperately needs open source tooling like C needed GCC.
