←
Pablo T.2,017 posts
Pablo T. avatar

Pablo T.

@controlpaths

Full time FPGA designer and DSP enthusiast. Also I write a blog about FPGA and DSP where I share projects and treaks.

◷ Joined September 2021▧ 475 Photos and videos
Pablo T. avatar
Pablo T.@controlpaths·13h

DSP algorithms are one of the reasons because #FPGA are so important, and that's because a piece of hardware that is inside of the FPGA, the #DSP Blocks. Here you can find the FPGA with most DSP blocks in the market! thegatearray.com

Post attachment
Pablo T. avatar
Pablo T.@controlpaths·Nov 13

La #IA ha cambiado nuestra forma de desarrollar hardware, pero no siempre podemos delegarlo todo a modelos en la nube. En este curso te muestro como montar tu propio sistema de inferencia de IA para desarrollo de FPGA en local con Ollama y Beellama. controlpaths.com/academy/flu…

Pablo T. avatar
Pablo T.@controlpaths·Nov 13

There was a time when companies had their own email servers and Word installed on their computers, so a downfall of Microsoft did not affect them in any way.

Pablo T. avatar
Pablo T.@controlpaths·Nov 10

No vendor tools needed: the 10 biggest FPGAs you can build for with an open-source flow (Yosys + nextpnr, F4PGA, openXC7). #1: the Cologne Chip CCGM1A25, ≈1.02M LUT4 (not released yet). #OpenSource

Post attachment
Pablo T. avatar
Pablo T.@controlpaths·Nov 8

Vendors use to talk about the biggest, the one with more DSP, the faster, but there is another interesting raking. Here you can find the smallest #FPGA in the market. The Lattice LCMXO2-256 has just 256 LUT4 equivalents. thegatearray.com/small-fpga/

Post attachment
Joel Serna Moreno avatar
⟳ Pablo T. retweeted
Joel Serna Moreno@JoelSernaMoreno·Nov 6

Introducing the new Evil Crow Keylogger V2: - Keylogger and BadUSB with Wi-Fi - P2P VPN for remote access - AI-Powered Payload Generation - OS detection & AutoExec - USB Network Implant (NCM) - USB Keyboard Autoclone Shipping: China | EU | UK | Mexico 
github.com/joelsernamoreno/E…

Post attachment
Pablo T. avatar
Pablo T.@controlpaths·Nov 5

Today I realized that I have some unused Raspberry Pi 4s, and they can be used as nodes to integrate into a CI/CD workflow for FPGAs, and it seems to be working well! I can even deactivate the USB ports to power up the board remotely.

Post attachment
Pablo T. avatar
Pablo T.@controlpaths·Oct 31

I guess IBM thought the same when someone talked about personal computers in the 40s; as I see it, future models will be as efficient as we cannot imagine today, so even a very cheap GPU will run models with even more capabilities than current frontier models. They will be used for some kind of centralized intelligence, but for 90% of users, AI will be local.

sam: local inference is a terrible allocation of scarce resources “a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment somehow many people's intuition is the opposite of that”

0:45