Skip to main content

Module cuda_lane_stub

Module cuda_lane_stub 

Source
Expand description

Stub when CUDA feature is off — CUDA lane APIs return false / no-op.

Structs§

MegaPassLayerDims
MegaPassLayerWeights

Constants§

MAX_DENSE_ELEMS

Functions§

cache_dense_weight
No-op — no device weight cache exists without CUDA.
clear_weight_cache
No-op — no cache to clear without CUDA.
dense_weight_cached
Returns false — nothing is ever cached without CUDA.
device_kv_ready
Returns false — device KV cache never exists without CUDA.
ensure_device_kv_cache
Returns false — no device KV cache can be allocated without CUDA.
preload_q4k_soa_weights
Returns 0 — no device weight preloading without CUDA.
q4k_device_weight_count
Returns 0 — no Q4K weights are resident on-device without CUDA.
try_cuda_batch_gemv
Returns false — CUDA GEMV unavailable; caller falls back to CPU.
try_cuda_batch_gemv_cached
Returns false — CUDA GEMV unavailable; caller falls back to CPU.
try_cuda_batch_gemv_cached_only
Returns false — no cached device weights exist without CUDA.
try_cuda_mega_pass
Returns None — the fused mega-pass is unavailable without CUDA.
try_q4k_soa_attention_device
Returns false — device-side attention unavailable without CUDA.
try_q4k_soa_ffn_block
Returns false — fused FFN block unavailable without CUDA.
try_q4k_soa_ffn_block_residual
Returns false — fused residual FFN block unavailable without CUDA.
try_q4k_soa_fused_swiglu
Returns false — fused SwiGLU unavailable without CUDA.
try_q4k_soa_gemv
Returns false — Q4K device GEMV unavailable without CUDA.
try_q4k_soa_qkv
Returns false — fused QKV projection unavailable without CUDA.
warm_cuda_context
Returns false — no CUDA context to warm.
weight_cache_len
Returns 0 — the weight cache is empty by definition without CUDA.
weight_fingerprint
Returns 0 — no weight fingerprinting without CUDA (feature cuda off or wasm).