60 lines
2.7 KiB
Bash
60 lines
2.7 KiB
Bash
# Tunables for frank-default-large (unsloth Qwen3.8-27B UD-Q8_K_XL, dankclaw)
|
|
# Runtime: llama.cpp llama-server (b10433+, rebuilt 2026-08-14 for bundled-MTP
|
|
# support); AMD Strix Halo (gfx1151) 96GB UMA; Ryzen AI MAX+ 395.
|
|
MODEL_DIR=/home/dank/models
|
|
MODEL_FILE=Qwen3.8-27B-UD-Q8_K_XL.gguf
|
|
ALIAS=frank-default-large
|
|
HOST=0.0.0.0
|
|
PORT=7770
|
|
# Context pool. Qwen3.8 native context = 262144 per sequence.
|
|
# 524288 total / PARALLEL 2 = 262144 per slot, matching the pi client's
|
|
# contextWindow=262144 for frank-default-large. Model is 29GB; ROCm has
|
|
# ~70GB headroom, so 2x262k quantized KV fits. Verify at startup.
|
|
CTX_SIZE=524288
|
|
PARALLEL=2
|
|
# Spec: DSpark sidecar draft (magnitudedev/Qwen3.8-27B-DSpark-GGUF Q8_0,
|
|
# 1.36B, block_size 7, Markov confidence head). Switched from bundled-MTP
|
|
# 2026-08-15 at the user's request. DSpark is the semi-autoregressive
|
|
# extension of DFlash (DeepSeek, arXiv 2607.05147): parallel diffusion
|
|
# backbone + sequential Markov head + confidence-scheduled verification.
|
|
# n_max clamps to the trained block_size 7 (server warning if set higher).
|
|
# DSpark extends DFlash (which had a shared-draft-context premature-EOS bug)
|
|
# but adds per-sequence Markov/confidence machinery; parallel safety on this
|
|
# box is still pending the two-16k-prompt test 2026-08-15. Revert to
|
|
# bundled-MTP or set PARALLEL=1 if the parallel test fails.
|
|
# DRAFT_FILE names the sidecar GGUF in MODEL_DIR (downloaded 2026-08-15,
|
|
# sha256 b007a76b2ce57c1a3ecf36046543aebf60a763861624647d653f16d336c781d2).
|
|
SPEC_TYPE=draft-dspark
|
|
DRAFT_FILE=Qwen3.8-27B-DSpark-Q8_0.gguf
|
|
SPEC_DRAFT_N_MAX=7
|
|
SPEC_DRAFT_P_MIN=
|
|
SPEC_NGRAM_MOD_N_MIN=48
|
|
SPEC_NGRAM_MOD_N_MAX=64
|
|
SPEC_NGRAM_MOD_N_MATCH=24
|
|
# Backend device from --list-devices. ROCm0 is the production backend.
|
|
DEVICE=ROCm0
|
|
# Dense model: full offload. (Not MoE.)
|
|
N_GPU_LAYERS=all
|
|
# KV on GPU, quantized (K q8_0, V q8_0). Keep K == V for Flash Attention;
|
|
# mixed types silently disable FA (long-context decode collapses).
|
|
CACHE_TYPE_K=q8_0
|
|
CACHE_TYPE_V=q8_0
|
|
# Batch sizing (verified 2026-08-14 on Qwen3.8 UD-Q8_K_XL):
|
|
# ubatch 128 with batch>=2048 -> premature-EOS under concurrency (unsafe).
|
|
# ubatch 512 -> MTP acceptance collapses (~25%) -> gen ~9 t/s (slow).
|
|
# ubatch 256 -> ~30 t/s gen, concurrent-safe, ~330 t/s prefill. WINNER.
|
|
# batch 8192 keeps single-shot prefill flexibility; ubatch drives the graph.
|
|
BATCH_SIZE=8192
|
|
UBATCH_SIZE=256
|
|
# Sampling settings — per Qwen3.8-27B model card thinking mode.
|
|
# Thinking: temp=1.0, top_p=0.95, top_k=20, min_p=0. Clients may override.
|
|
# WARNING: do not name this variable TEMP. The ROCm HIP runtime reads the
|
|
# TEMP environment variable (JIT kernel-cache dir); TEMP=1.0 made the HIP
|
|
# runtime SEGV during model load on Aug 13 2026.
|
|
TEMPERATURE=1.0
|
|
TOP_K=20
|
|
TOP_P=0.95
|
|
MIN_P=0.0
|
|
THREADS=12
|
|
THREADS_HTTP=1
|