Your fused attention kernel matches the reference at 512 tokens and drifts at 8,000. Find the bug.

A bug that scales with the number of tiles is invisible in a unit test that fits in one tile. The three candidates that produce exactly this signature, the measurement that separates them, and a test design that would have caught all three before the kernel shipped.

Updated Sep 2026 · Grounded in real AI infrastructure interview loops and written to a senior-engineer editorial bar, with every number worked and every diagram hand-built.

20 answers per topic instead of 10, plus saved progress and bookmarks · no cardor unlock all 283 remaining answers · ₹2,000 / $25

The concepts behind this question

Ranked by how closely each one overlaps this question's topic, so the first card is the thing to read if the answer above moved too fast.

Advanced

⚡ Kernels & Compilers🔒 Premium

FlashAttention InternalsStandard attention writes the N x N score matrix to HBM and reads it back, which makes it memory-bound and quadratic in memory. FlashAttention tiles Q, K and V through shared memory, keeps a running max and sum so the softmax never needs the full row, and recomputes scores in the backward pass. Knowing the online-softmax rescale, why FlashAttention-2 flipped the loop order, and what FlashAttention-3 overlaps on Hopper is the difference between naming the paper and being able to write the kernel.

Foundational

🖧 Hardware & Cluster Build-Out

Burn-In and Acceptance TestingNew hardware fails early or it fails late, and burn-in exists to move the early failures before the cluster is handed over rather than after. A proper acceptance test runs every layer under sustained load for days, compares every node against its siblings rather than against a specification, and produces a signed number the buyer and the vendor both agree on. The comparison is the important part: identical hardware running identical work should produce identical numbers, and the outliers are the finding.

Foundational

🧮 Open Weights & Serving Engines

Weight Formats: FP8 Blocks, MXFP4 and AWQOpen-weights models now ship pre-quantized, and the format is part of the release rather than something you choose afterwards. Block-scaled FP8 gives one byte per parameter with a scale per tile. MXFP4 gives about 0.53 bytes by pairing four-bit values with a shared exponent every 32 elements. Integer schemes like AWQ reach similar sizes with a different error profile. What decides a deployment is not which is most accurate in the abstract but which one the model was released and evaluated in, and which one your engine and hardware can execute natively.

Foundational

🧮 Open Weights & Serving Engines

Multi-Head Latent Attention and Sparse IndexersGrouped-query attention shrank the KV cache by sharing key and value heads. Latent attention goes further by caching a single compressed vector per token per layer and reconstructing the heads on the fly, which cuts the cache by tens of times rather than by a small factor. On top of that, sparse indexers pick a few thousand relevant positions per query instead of attending to all of them, turning the quadratic term linear at long context. Both are now standard in open-weights models, and both change how a serving deployment is sized.

Explore the full curriculum →

UP NEXT ON YOUR JOURNEY

Closest to this topicExplain FlashAttention. Why is it called IO-aware, and what does it actually save?Closest to this topicWhat changed between FlashAttention 1, 2 and 3, and why did each change buy what it did?Closest to this topicWhy do tensor cores accumulate in fp32 when the inputs are bf16, and why do optimizers keep fp32 master weights?

FEDITOR'S NOTE

Scored on recognising that a tile-count-dependent error points at the online-softmax state, on isolating by disabling the rescale rather than guessing, and on designing a test whose sequence length crosses the tile boundary.

DISCUSSION · 0

No comments yet — be the first to share your approach.