India
4h
527
What do vegetarian Indians eat for protein?
Ask Blinders
Yesterday
872
Why is our country owned by Israel? I don't want my tax dollars fund genocide. How can we stop this nonsense?
Tech Industry
Yesterday
3232
1 vs 5 Million - no lifestyle change
Tech Industry
Yesterday
693
Database companies that pay well for Staff SWE
Tech Industry
11h
714
Offer evaluation: Amazon or Broadcom
Hello everyone 🤠 I have taken a deep learning course from Coursera a while back. I have an ok knowledge that I can use it to kick start my ML interest again. I feel my carrier and salary are reaching an end. I can't break into the 200k club. I was thinking of learning fpga on xilinx so I can build the ML interface on the FPGA directly. Is that something worth it? Would I get a better performance than Nvidia cards? Would that be cheaper? #semiconductor #machinelearningengineer #nvidia #xilinx
What's HFT?
High frequency trading
Xilinx already have Vitis-AI framework which can implement ML. If you are still interested with this field, you can go ahead with no doubt.
Learn python, opencv, gstreamer, pytorch and go to sw dev. Xilinx itself is moving out of fpga for ML.
That’s interesting. Even for the inference market?
I have used almost all of these tools. I'm thinking of building any model on FPGA directly like NLP or computer vision models.
Do you have previous experience in FPGA?
No, this is why I'm asking to start learning it.
Money or career wise, I don’t think it worths it. They are fun to play with. It may not be the case for everyone, but it took me a while to get a hang of it at the beginning. The learning curve is steeper than software. For some use cases, you pretty much have to become an electronic engineer to complete a design. I was thinking about HFT at one point, but…
Stick to software. FPGA industry has crippling H1B bloat so the pay generally sucks. Even in HFT the pay sucks relative to the amount of time they expect you to put in to make stupid micro-optimizations, so the HFT jobs are horrifically boring.
No money in hardware man. Stick to software. If at all you can try getting into compiler
Yeah, I can understand that. However, I know HFT firms will pay top dollars for FPGA programmers. I want to know if people have experience with ML on it and is it worth it or not
Very few paying top dollar amounts for FPGA and ML is not even half the battle to just get in. You'd have a better time getting into hft on the software side and you'd have way more options to boost your salary.