For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Mojo package
builtin_kernels
Provides the graph compiler's built-in kernel registrations.
Each module binds graph op names (such as mo.matmul) to the open-source
kernels in linalg, nn, comm, and related packages using the same
@extensibility.register mechanism available to custom ops, so these
registrations double as worked examples of how kernels connect to MAX
graphs.
Modules
-
attention: Registers attention graph ops (MHA, MLA, fused QKV) and dispatches them to thenn.attentionkernels. -
conv: Registers convolution graph ops and dispatches them to thenn.convkernels. -
distributed: Registers distributed and multi-GPU collective graph ops backed by thecommandshmemkernels. -
elementwise: Registers elementwise and unary/binary math graph ops. -
ep: Expert Parallelism (EP) Communication Kernel. -
gather_scatter: Registers gather, scatter, and indexed-access graph ops backed by thenn.gather_scatterkernels. -
kernels: Registers core tensor graph ops (range, copy, reshape, and related utilities). -
kv_cache: Registers KV-cache graph ops backed by thekv_cacheandnn.kv_cachekernels. -
linalg: Registers matmul-family graph ops (matmul, grouped, batched, and quantized variants) and dispatches them to thelinalgkernels. -
logprobs: Registers log-probability computation graph ops used by sampling pipelines. -
monoids: Concrete reduction monoids conforming toalgorithm.reduce_op.ReduceOp. -
mtp: Graph-op bindings for multi-token-prediction draft layers. -
nan_check: NaN/Inf detection kernels for the max-debug.nan-check feature. -
quantization: Registers quantization and dequantization graph ops backed by thequantizationandlinalgkernels. -
reductions: Registers reduction graph ops (sum, mean, argmax, and related) over tensor axes. -
rowwise_normalization: Row-based normalization kernels (rms_norm, layer_norm and fused variants) built on thealgorithm.rowwisescaffolder and thebuiltin_kernels.monoidsmonoid library. -
rowwise_reductions: Row-based row-wise reductions (free-form body layer). -
rowwise_softmax: Row-based softmax built on thealgorithm.rowwisescaffolder and thebuiltin_kernels.monoidsmonoid library. -
short_conv_ring: Registers the ring-state short convolution graph ops; seenn.short_conv_ring.