5 ms·

Pytorch and matplotlib managed to unseat matlab, which I didn’t think would ever happen. Bitcoin, git are open source. docker, kubernetes will just add a leve

Pytorch and matplotlib managed to unseat matlab, which I didn’t think would ever happen. Bitcoin, git are open source. docker, kubernetes will just add a level of abstraction over any proprietary os. Things are not so bad!

fxtentacle 6y ago
"Pytorch and matplotlib managed to unseat matlab" From what I know, most companies use MATLAB for real-time simulation or its VHDL code generation, which means you can put your algorithms into FPGA with relative ease. I don't see Pytorch helping there anytime soon. That said, I agree with you that a lot of mathematical computing is moving to Python, especially now that there's multiple frameworks with GPU support. And yes, that's a great development :)
- im3w1l 6y ago
  Whoa, matlab has vhdl? I never knew, that's cool.
- StillBored 6y ago
  I came here to say something similar, but with a different industry. None of the people/groups I know using matlab have moved to pytorch, or even gnu octave which is a much closer fit. Because its not really about matlab so much has the huge collection of library/extensions for matlab or its close relationship to simulink. Decades of industry specific modeling/simulation/tooling built into peoples workflows are hard to change. If you look at the "applications" section here https://www.mathworks.com/products.html?s_tid=gn_ps https://www.mathworks.com/products.html?s_tid=gn_ps pytorch really only touches on a couple of them AFAIK. Deep engineering tooling is one of the places where opensource really hasn't yet made inroads. From the lack of a good solidworks/cad toolchain to the modeling/simulation space. Even EDA remains deeply dependent on proprietary tooling despite the recent success in a number of places. So while board level design with kicad may start eating into the Altium space now that its beginning to get RF/etc capabilities. I don't see it biting into the cadence/synopsis/mentor space anytime in the near future if ever due to the secrecy of the foundries. I think a large part of that is there is a lot of breath in this space, and the customer base is generally quite small. So I think a good opensource PCB layout/modeling tool, and a decent parametric 3d cad program will happen a lot sooner.

fxtentacle 6y ago

"Pytorch and matplotlib managed to unseat matlab" From what I know, most companies use MATLAB for real-time simulation or its VHDL code generation, which means you can put your algorithms into FPGA with relative ease. I don't see Pytorch helping there anytime soon. That said, I agree with you that a lot of mathematical computing is moving to Python, especially now that there's multiple frameworks with GPU support. And yes, that's a great development :)

im3w1l 6y ago
Whoa, matlab has vhdl? I never knew, that's cool.
StillBored 6y ago
I came here to say something similar, but with a different industry. None of the people/groups I know using matlab have moved to pytorch, or even gnu octave which is a much closer fit. Because its not really about matlab so much has the huge collection of library/extensions for matlab or its close relationship to simulink. Decades of industry specific modeling/simulation/tooling built into peoples workflows are hard to change. If you look at the "applications" section here https://www.mathworks.com/products.html?s_tid=gn_ps https://www.mathworks.com/products.html?s_tid=gn_ps pytorch really only touches on a couple of them AFAIK. Deep engineering tooling is one of the places where opensource really hasn't yet made inroads. From the lack of a good solidworks/cad toolchain to the modeling/simulation space. Even EDA remains deeply dependent on proprietary tooling despite the recent success in a number of places. So while board level design with kicad may start eating into the Altium space now that its beginning to get RF/etc capabilities. I don't see it biting into the cadence/synopsis/mentor space anytime in the near future if ever due to the secrecy of the foundries. I think a large part of that is there is a lot of breath in this space, and the customer base is generally quite small. So I think a good opensource PCB layout/modeling tool, and a decent parametric 3d cad program will happen a lot sooner.