mirror of
https://github.com/luxfi/zap.git
synced 2026-10-30 15:04:16 +00:00
zeekay
d14b104790
Three orthogonal additions, one wire format. v2 package — generic ad-hoc schema dispatch (~/work/lux/zap/v2): - README + IMPOSSIBILITY + BENCH_RESULTS document the v1↔v2 split. - builder/field/iter/list/pool/registry/schema/view + codegen package emit v1-equivalent fast paths from declarative .zap sources. - _compile_fail_test/ + examples/ cover negative and positive cases. - v2_test.go exercises the generic dispatch path. - doc.go in v1 marks v1 deprecated for *new* schema authoring; existing v1 buffers stay parseable by both v1 and v2. bufpool — pooled read slabs (bufpool.go + bufpool_test.go): - Message gains optional *bufRef; Release() returns the slab to its pool when sourced from the pooled read path, GC-managed otherwise. - ParseHeader returns (data, rootOffset) without allocating *Message, used by zapv2 to avoid two allocations per parse. - bench_read_test.go + node_quic_stream_test.go cover both paths. Transport allocator + UMA/RDMA cgo (transport/*): - Transport.Caps()/Buffer interface: capability + slab-typed buffer contract. Bytes() vs DevicePtr() decoupled so UMA and GPUDirect can share the same call surface. - DPDK PMD + GPUDirect RDMA + CUDA UMA real cgo backends gated behind build tags; stub paths remain default on darwin / non-Linux. - transport_test.go covers Caps reporting + Buffer lifecycle on the default transport. LLM.md + go.mod follow the additions.
30 lines
1.3 KiB
Go
30 lines
1.3 KiB
Go
// Copyright (C) 2025-2026, Lux Industries Inc. All rights reserved.
|
|
// See the file LICENSE for licensing terms.
|
|
|
|
package transport
|
|
|
|
// uma.go is the cross-platform doc comment for the UMA transport. The
|
|
// real factory lives in:
|
|
//
|
|
// uma_linux_cuda.go — //go:build cgo && linux && cuda
|
|
// Real cudaMallocManaged-backed slab pool.
|
|
// uma_linux.go — //go:build cgo && linux && !cuda
|
|
// Clean "not available" so Pick() falls through.
|
|
// uma_darwin.go — //go:build cgo && darwin
|
|
// Metal-backed UMA shim (hook for the Apple Silicon path).
|
|
// uma_other.go — //go:build !cgo || (!darwin && !linux)
|
|
// Clean "not available" for every other combo.
|
|
//
|
|
// What UMA means here: CPU and GPU share physical memory. Packet bytes
|
|
// land in shared RAM (cudaMallocManaged on Grace-Hopper / GB10, Metal-
|
|
// shared MTLBuffer on Apple Silicon). The GPU kernel parses the ZAP
|
|
// message in place. No memcpy from kernel space to GPU. This is the
|
|
// fast path on Lux validator nodes that ship with the GB10 or M3 Ultra
|
|
// hosts (per LP-203).
|
|
//
|
|
// Build matrix:
|
|
// linux+cuda → real UMA (lcudart, slab pool, GPU-resident Recv)
|
|
// linux+!cuda → not available (factory returns error)
|
|
// darwin (cgo) → metal hook (current state: stub, will land with M3 wire-up)
|
|
// anything else → not available
|