And you are basically a single consumer with a personal car relative to those data centers and cloud computing providers.

YOUR workload works well with an FPGA. Good for you, take advantage of that to the best degree you can.

People;/companies who want to run newer models that haven't been optimized for/don't support FPGAs? You get back to the case of "Well... I can run a 25% cheaper node for twice as long?". That isn't to say that people shouldn't be running these numbers (most companies WOULD benefit from the cheaper nodes for 24/7 jobs and the like). But your use case is not everyone's use case.

And, it once again, boils down to: If people are going to require the latest and greatest nvidia, what incentive is there in spending significant amounts of money getting it to work on a five year old AMD? Which is where smaller businesses and researchers looking for a buyout come into play.

At the end of the day, every company will choose to do it faster and cheaper, and nothing about Nvidia hardware fits into either of those categories unless you’re talking about milliseconds of timing, which THEN only fits into a mold of OpenAI’s definition.

Faster is almost always cheaper. There have been decades of research into this and it almost always boils down to it being cheaper to just run at full speed (if you have the ability to) and then turn it off rather than run it longer but at a lower clock speed or with fewer transistors.

And nVidia wouldn't even let the word "cheaper" see the glory that is Jensen's latest jacket that costs more than my car does. But if you are somehow claiming that "faster" doesn't apply to that company then... you know nothing (... Jon Snow).

unless you’re talking about milliseconds of timing

So... its not faster unless you are talking about time?

Also, milliseconds really DO matter when you are trying to make something responsive and already dealing with round trip times with a client. And they add up quite a bit when you are trying to lower your overall footprint so that you only need 4 notes instead of 5.

They don't ALWAYS add up, depending on your use case. But for the data centers that are selling computers by time? Yeah,. time matters.

So I will just repeat this: Your use case is not everyone's use case.

FPGAs have been a thing for ages.

If I remember it correctly (I learned this stuff 3 decades ago) they were basically an improvement on logic circuits without clocks (think stuff like NAND and XOR gates - digital signals just go in and the result comes out on the other side with no delay beyond that caused by analog elements such as parasitical inductances and capacitances, so without waiting for a clock transition).

The thing is, back then clocking of digital circuits really took off (because it's WAY simpler to have things done one stage at a time with a clock synchronizing when results are read from one stage and sent to the next stage, since different gates have different delays and so making sure results are only read after the slowest path is done is complicated) so all CPU and GPU architecture nowadays are based on having a clock, with clock transitions dictating things like when is each step of processing a CPU/GPU instruction started.

Circuits without clocks have the capability of being way faster than circuits with clocks if you can manage the problem of different digital elements having different delays in producing results I think what we're seeing here is a revival of using circuits without clocks (or at least with blocks of logic done between clock transitions which are much longer and more complex than the processing of a single GPU instruction).

Huh. Everything I'm reading seems to imply it's more like a DSP ASIC than an FPGA (even down to the fact that it's a VLIW processor) but maybe that's wrong.

I'm curious what kind of work you do that's led you to this conclusion about FPGAs. I'm guessing you specifically use FPGAs for this task in your work? I'd love to hear about what kinds of ops you specifically find speedups in. I can imagine many exist, as otherwise there wouldn't be a need for features like tensor cores and transformer acceleration on the latest NVIDIA GPUs (since obviously these features must exploit some inefficiency in GPGPU architectures, up to limits in memory bandwidth of course), but also I wonder how much benefit you can get since in practice a lot of features end up limited by memory bandwidth, and unless you have a gigantic FPGA I imagine this is going to be an issue there as well.

I haven't seriously touched FPGAs in a while, but I work in ML research (namely CV) and I don't know anyone on the research side bothering with FPGAs. Even dedicated accelerators are still mostly niche products because in practice, the software suite needed to run them takes a lot more time to configure. For us on the academic side, you're usually looking at experiments that take a day or a few to run at most. If you're now spending an extra day or two writing RTL instead of just slapping together a few lines of python that implicitly calls CUDA kernels, you're not really benefiting from the potential speedup of FPGAs. On the other hand, I know accelerators are handy for production environments (and in general they're more popular for inference than training).

I suspect it's much easier to find someone who can write quality CUDA or PTX than someone who can write quality RTL, especially with CS being much more popular than ECE nowadays. At a minimum, the whole FPGA skillset seems much less common among my peers. Maybe it'll be more crucial in the future (which will definitely be interesting!) but it's not something I've seen yet.

Looking forward to hearing your perspective!