So far we've been focused on edge devices like the Zynq, Artix and Zynq Ultrascale+ families. Tensil certainly works on larger devices but it's not as optimized there as we'd like it. If that's interesting to you, I'd love to talk and understand your use case in more depth.
The Intel FPGA side is interesting, as you say there are fewer projects targeting their technologies for ML use cases. We haven't tested support for their boards yet, but there is nothing in our generated RTL that is exclusive to Xilinx. The only thing we'd need to add is new drivers for their platforms.
Sorry if this was unclear - in a datacenter use case you are right, but for an edge deployment, you will usually need to quantize, prune or compress your ML model to get it working as fast as you'd like on a sufficiently small CPU/GPU. Compared with running your ML model unchanged on those platforms, Tensil can run with the performance ranges listed above. You can also quantize and use Tensil too!
FINN is a very cool project, but usually requires big changes to your model in order to work, e.g. quantizing down to 1 or 2 bit weights. It also works best on large FPGAs which are unsuitable for edge deployments. Tensil works out of the box with any model (no need to quantize / compress) and on small edge FPGAs.
Cheers, and great question! FPGAs are pretty amazing devices, but one thing that's been holding them back is how difficult they have been to work with. Typically to actually make use of an FPGA you'd need to have an FPGA expert and an embedded software engineer on your team, along with all the requisite tools and materials.
That has started to change dramatically in the last decade, with open source FPGA toolchains like yosys, runtimes like the PYNQ framework and RTL generator tools like Tensil being developed. When you put these things together, working with FPGAs starts to become as easy as using any other compute platform. For that reason, I think there are lots of applications involving FPGAs that will soon be invented to take advantage of this trend. One could speculate that the reason Intel and AMD are buying up FPGA vendors is because they see the potential there.
As far as head-to-head comparisons go, as long as you're running the workload it was designed for in the environment it was designed for, an ASIC will always be the best possible perf per watt. The question is what happens when you go outside those bounds. Can you take your model, swap out a layer, and have it run just as fast on your Coral or NPU? Probably not, at least right now. But with Tensil, you can re-run your architecture search to find the best accelerator, and take advantage of it right away.
I don't have hard numbers at hand, but I'd estimate something like an order of magnitude improvement for using DSP for multiplication vs not. If they're available on the fabric, you'll definitely want to use them! If this is an experiment you want to run, I'd be very happy to help you figure out how to do it.
Just saw your edit re: model compression. One thing that Tensil can do is help you avoid the need to quantize or compress your model entirely! For example, we've found that using a 16-bit fixed point numeric data type preserves almost all the model accuracy while not sacrificing performance thanks to the huge amount of parallelism available on FPGA.
The broader point is that Tensil is extremely flexible, so you can try out lots of different accelerator configurations to find the one that works best for your ML model. Think of it as optimizing the hardware first, then the software if needed.
We're actually working on a tool to manage and automate this hardware architecture search - watch this space!
That's excellent - feel free to join our Discord if you'd like to brainstorm ideas or get help choosing models and boards https://discord.gg/TSw34H3PXr
Even if you agree with Taleb's arguments in that article, he still admits that low IQ is correlated with low wealth, low income and other such social measures of low success. So the hypothesis that IQ may be declining should still be concerning, since more and more people will be in that low IQ range predictive of low success.
I'm not GP but as I understand it most religions contain an idea that selflessness is the highest good that a person can attain. Further they often maintain that only by becoming truly selfless can you make your spirit one with God (I don't specifically mean the Christian God here, just the concept of God). Becoming truly self-less literally means eliminating the self and its wants and impulses (really what is meant by this is typically something like the ego). Such an elimination has been called a "dying to self" or a mortification.
I will say that in my experience acting in selfless ways is indeed the most spiritually fulfilling thing you can do, regardless of your specific theology. The paradox is that you have to really be selfless: if you have any expectation of a reward (even a spiritual one) then you are in fact still acting for your self. Consequently the only true source of selflessness is love for others that exceeds love for yourself. Cultivate such love and you can't go wrong.
How does this square with intelligence studies of deaf people? Despite being half as likely to complete tertiary education, the deaf exhibit the same IQ distribution as the general population. Are they doing something else that is just as effective as education at increasing IQ?
> In sum, the implication of the research of the last fifty years which compared the IQ of the deaf with the hearing and of subgroups of deaf children indicates that when there are no complicating multiple handicaps, the deaf and hard-of-hearing function at approximately the same IQ level on performance intelligence tests as do the hearing. [1]
I agree with the prescription to stay off weaponized social propaganda platforms and focus on your local community, but I don't see how you can ask people to blindly trust the CIA. Too many misdeeds to count.
- Fabricating justification for war (Vietnam, Iraq)
- Fabricating evidence of treason by a sitting President (Trump-Russia)
- Overthrowing democratically elected governments (e.g. Iran)
- Starting civil wars (e.g. Nicaragua)
- Assassinating thousands of people, often with collateral killing of innocents
- And drugs drugs drugs: https://en.wikipedia.org/wiki/Allegations_of_CIA_drug_traffi...
Ultimately, do we live in a better world thanks to the CIA? From where I sit, I'd need a lot of convincing to believe that.
> but in those cases i dont think its entirely fair to expect schools to fully accommodate such outliers
This sentiment would equally support removing special education programs for students with learning disabilities, yet those are not being called into question.
This already exists and it doesn't quite work. School funding and outcomes are less well correlated than you might expect. Some of the worst performing schools and districts are also the best funded. The top causal factor appears to be parental involvement: highly involved parents in even poorly funded schools is better than uninvolved parents in well funded schools.
Prizes are incentives. AIs can respond to incentives just as we can, but for now a Nobel prize is a type of incentive that only humans care about. Unless the AI shows that it cares about receiving a Nobel, the prize should go to the humans who made the AI.
No one cares what outsiders think when they are being assaulted on their walk home from work. If you want justice you must first create order.
In addition, it's not like these people don't have sympathy for the plight of the poor: they voted to spend vast sums on tax money trying to fix the root cause of the problem. It's not their fault that these efforts are failing.
LA city budget, not California state. The amount spent is so high that you could just give every homeless person in LA $2000 every month and it would be less expensive.
The Intel FPGA side is interesting, as you say there are fewer projects targeting their technologies for ML use cases. We haven't tested support for their boards yet, but there is nothing in our generated RTL that is exclusive to Xilinx. The only thing we'd need to add is new drivers for their platforms.