frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Ask HN: Who is using FPGA for ML inference?

7•softwarewright•6h ago
With RAM price inflation, I wonder if FPGAs can be used to offload inference processing without keeping weights in RAM? The available RAM would be for activations, KV Cache, context but not static weights. Weights could be streamed from disk. This approach is not for tokens/second but throughput at a lower cost. Possibly better answers/kHh? I've started researching this, but wonder if others have considered/tried this?

Comments

wmf•6h ago
FPGAs are far more expensive than GPU+DRAM, even at today's inflated prices.
softwarewright•4h ago
Tang nano FPGAs are around $10, the idea being to couple with a $25 ARM or RISC-V dev board and offload the math from a streaming store. Coupling this with an older 8GB VRAM board can theoretically run a model whose weights do not fit. The weights get repeatedly run through the MCU/FPGA and the VRAM is used for KV Cache and context. If you connected several via USB and streamed a MoE model expert per FPGA, you could achieve parallelism. Won't be fast, but, many requests could be processed in parallel, as each request shares the large static weights.

The reason for this is, model weights do not need to be randomly accessed. So why store them in expensive RAM.

Cerebras and Qrok seem to be using a very different approach than NVIDIA to get orders of magnitudes speed ups. I'm trying to explore other alternative approaches.

wmf•3h ago
I predict that the FPGA adds no value in this scenario. Just process inference on the CPU.
softwarewright•2h ago
I am doing the initial work on many core dual CPUs. I already have systems with large system RAM and older GPUs with less VRAM, and large SAS HDs, fast SSDs running large open weight LLMs that do not really fit any GPU I own (they offload and/or run on CPU and system RAM, but slowly).

I have the MCUs and FPGAs (in a drawer) and I am retired, and this is my idea of fun.

I am trying to generalize an approach to use large MoE models (with possibly small quants) to run many agents in parallel without spending more on more or bigger GPUs.

I am also doing some edge ML (bird species recognition near the camera) using NPUs (in design phase, yet untested). I have an electronics lab, and I've emulated soft CPUs and built software that runs on FPGAs and in my emulators.

Instead of assuming my approach won't work or is too expensive, I choose to be optimistic. Also, failures are educational. I'm trying to gain more FPGA experience.

pugfugly•4h ago
Train circuits, not weights; then use an FPGA
softwarewright•2h ago
Can you suggest what circuits? examples?
mathisfun123•4h ago
No one because FPGAs are 10-20x less dense than ASICs actually designed for purpose. The tooling is also complete shit. Signed someone that did (part of) their PhD on this topic.
softwarewright•2h ago
But I can afford my already purchased drawer full of cheap FPGAs for experiments.

I guess I am too risk adverse to bet $10,000 on an ASIC run (having no experience doing that and no desire to go that route).

If I can demonstrate a proof-of-concept in a reproducible research way, others can then advance to the ASIC level.

This is just a hobbyist experiment looking for other hobbyists who can afford a cheap FPGA and have some free time and interest.

b89kim•2h ago
You could build small ML model like talos-v2. But larger model/LLM requires much more engineering cost than you expected. Optimizing HDL has too many control knob/param to solve by RL.
addag•1h ago
I thought the same when I saw that the financial industry was hiring FPGA people for low-latency algorithms.

My understanding (as a non-FPGA expert) is that currently FPGA beats generic hardware (CPU,GPU) for "small size algorithm" (i.e that do not need GB of weights), while enabling a certain flexibility vs ASIC.

My guess is that you cannot bake all the weights into the circuit topology, so you are still bound by the memory transfer speed (to be double checked).

Ask HN: Who is using MCP in production?

34•sukit•17h ago•41 comments

Ask HN: Why were OpenAI, Claude, and Grok simultaneously down?

355•halcdev•14h ago•532 comments

Ask HN: Are others seeing Google's reCAPTCHA rejecting Firefox users?

48•Animats•9h ago•13 comments

Tell HN: NVIDIA's Acquisition of HuggingFace was for $HuggingFace

24•MontagFTB•7h ago•4 comments

Ask HN: 30, burnt out, with 6-10 months of runway. What would you do?

17•throwaway_30yo•1d ago•11 comments

Ask HN: Who is using FPGA for ML inference?

7•softwarewright•6h ago•10 comments

Tell HN: Both recent GCP outages caused by fiber optic maintenance

8•fastest963•7h ago•1 comments

Ask HN: Why do many websites use a sign-in code instead of a password?

5•bariumbitmap•1h ago•2 comments

Ask HN: What is the ideal email domain setup?

2•bix6•5h ago•1 comments

Ask HN: How do you market Open Source Software?

5•Daniel-Pan•2h ago•4 comments

Ask HN: Who is hiring? (September 2026)

242•whoishiring•2d ago•318 comments

Ask HN: Founders that raised funding and shutdown, did you raise again?

17•asim•15h ago•1 comments

Ask HN: Who wants to be hired? (September 2026)

107•whoishiring•2d ago•423 comments

Ask HN: Laptop or Desktop?

6•Gecko4072•12h ago•4 comments

Ask HN: How do you guys keep up with the most cost-effective AI models?

4•solomonmwalker•14h ago•4 comments

Tell HN: The Thoughts and Woes of a Full-Time Founder Part 1

3•lcampbellsoup•5h ago•0 comments

Ask HN: How do we build a team? (bootstrapped B2B startup)

3•theonlinx•8h ago•0 comments

Ask HN: iPhone Camera on indicator

2•zahirbmirza•7h ago•1 comments

Ask HN: Is there a test for measuring cognitive affects from LLM usage?

5•niwrad•7h ago•0 comments

Ask HN: Could AI eventually build god?

5•olirex99•7h ago•8 comments

Ask HN: Are technical blogs dead?

11•quaintdev•20h ago•19 comments

One-person startup: a Cornell postdoc wants to put AI in everyone's hands

2•modelslive•8h ago•0 comments

Ask HN: Accountants working with ecommerce

3•vasyl_kyryliuk•11h ago•0 comments

Ask HN: How do you gate an autonomous coding agent's shell access?

2•alanfuNZ•8h ago•1 comments

Ask HN: Why does free time become another thing you have to use correctly?

2•nullifypreset•10h ago•0 comments

Ask HN: Are guardrails being baked into models, or enforced outside them?

2•arkj•10h ago•0 comments

Tell HN: Xbox Game Pass now limits the amount of hours each month

7•nodesocket•4h ago•3 comments

Beware: Malicious Profiles on YC Co-Founder Matching

5•pain_perdu•13h ago•0 comments

Ask HN: Can we classify AI as: subhuman, quasihuman, human, superhumam?

3•aunekiaru•13h ago•5 comments

Resources to Get Good at Soldering?

9•tosmatos•1d ago•9 comments