News for hackers. Stuff that matters.

randomgoogler1

Re:What was your "Oh Shit" moment with GenAI? on Friday June 05, 2026 @05:12PM (#48418384)
Me too, it came out of nowhere even as someone who played with GPT2 before. The moment completely changed what I have worked on since.
Re:A History of IDEs at Google on Wednesday May 13, 2026 @02:23PM (#48125547)
In my experience, antigravity IDE is much less seemless compared to Cider-V. I completely moved to using web-based antigravity for the agent and using cider-v to make manual changes and viewing code.
Re:Gemini Diffusion on Wednesday May 21, 2025 @10:32PM (#44058227)
(Disc: Googler but don't have any specific knowledge of this architecture)

My understanding of Groq is that the reason it is fast is that all the weights are kept in SRAM and since the SRAM <-> Compute bandwidth is much faster than HBM <-> Compute bandwidth, you can generate tokens faster (During generation the main bottleneck is just bringing in the weights + KV caches into compute).

If the diffusion models just do multiple unmasked forward passes through a transformer, then the activation * weights computation + (attention computation) will be the bottleneck which will make each denoising step compute bound and there won't be any advantage in storing the weights in SRAM since you can overlap the HBM -> compute transfer with compute itself.

But my knowledge of diffusion is non-existent, so take this with a truck of salt.