B200
B200 at Beyond Market Intelligence is a file of 3 stories. The newest of them: “Looped transformers share weights but memory limits keep them from staying on chip”, “Explore how a Mac Mini and iPad setup can complement your university AI access”, and “Porting Doom's renderer into a transformer with zero training”. Shared weights sound like a free lunch, but memory bandwidth has other plans. Balancing a Mac Mini and iPad against a single MacBook Pro is a smart dilemma for local inference, especially when you already have B200s and A100s at university for training. Acme AI is the next-generation, AI-powered spreadsheet platform built to replace Excel and redefine how analysts, data scientists, and enterprise teams work… The list below is every B200 story on Beyond Market Intelligence, newest first.
Looped transformers share weights but memory limits keep them from staying on chip
Shared weights sound like a free lunch, but memory bandwidth has other plans. A chain of eight matmuls with one 72 MiB weight ran 24% faster than eight identical copies, reading almost nothing from HBM. Scale that weight to 288 MiB, larger than L2 cache, and the advantage vanishes. Eight iterations of a full Huginn Core took eight times as long and read eight times the data. Sharing saves storage, but the GPU still fetches those weights from HBM every single time.
Explore how a Mac Mini and iPad setup can complement your university AI access
Balancing a Mac Mini and iPad against a single MacBook Pro is a smart dilemma for local inference, especially when you already have B200s and A100s at university for training. The Mac Mini with 32 GB RAM offers a stable, always-on workstation for running models, while the iPad adds portable flexibility for testing outputs on the go. This setup complements your existing access rather than competing with it.
Porting Doom's renderer into a transformer with zero training
A 21-billion-parameter transformer just rendered Doom's iconic first frame, and it never saw a single training example. The trick: a custom compiler that bakes a computation graph directly into weights, then runs the classic renderer inside that fixed structure. The result is a standard Hugging Face checkpoint, no special code needed. One frame takes 3,614 prompt tokens plus 53,747 generated ones, about 40 minutes on a B200. That is 35 frames per day, a fitting pace for a project this gloriously impractical.