Drilldown: Recipe
More actions
Use the filters below to narrow your results.
page:
DeepSeek V4 Flash 0731 on 4x 8 GB (allover326 PP4 DSpark) (1) ·
DeepSeek V4.1 Flash on 4x 8 GB (alanoo fidelity build) (1) ·
DeepSeek V4.1 Flash on 4x 8 GB plus RTX PRO 6000 (ExLlamaV3 TP5) (1) ·
GLM 5.3 Flash on 4x 8 GB (kaka86mm EXL3 + Marlin sidecar + DFlash2) (1) ·
Qwen3.8 Flash Next on 3x or 4x 8 GB (zebgop-ops FP8 PP) (1)
model:
quant:
EXL3 3.5 bpw (LibertAIDAI; routed experts 4 bpw on layers 0-9 and 30-39, 3 bpw on 10-29; attention and shared experts 6 bpw) plus the official Engram shards (1) ·
EXL3 4 bpw (Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw, 164 GB) plus an offline-generated Marlin INT4 decode sidecar (151 GB) (1) ·
Hybrid: all 40 expert layers at 3 bits (LibertAI 2.9 bpw pack with layers 18-22 grafted from the 3.5 bpw pack), official FP8 dense tensors (1) ·
Native MXFP4 experts + FP8 attention (official deepseek-ai checkpoint, about 155.4 GiB); KV cache fp8 (1) ·
Official block FP8, about 126 GiB including the about 50 GiB FP8 n-gram (PLE) table, which is served from host RAM; KV cache bf16 (the QSA layers refuse fp8) (1)
engine:
ExLlamaV3 fork (anoane/exllamav3-anemone) + TabbyAPI fork (anoane/tabbyAPI) (1) ·
GNmac-lab recipe image 1.2.0 (vLLM with sm_80 patches) plus three patches (1) ·
vLLM: haosdent/vllm branch dsv4-flash-a100 plus allover326's 7 patches (DSpark under PP, sm_80 prefill top-k fallback, logits row chunking) (1) ·
vLLM: hyd998877/vllm_170hx_glm53-flash-exl3-optimized with kaka86mm's overlay patches; exllamav3 1.4.8 (1) ·
vLLM: the vllm/vllm-openai:qwen38-flash-next image with zebgop-ops patches 0001 to 0016 (never 0008-WITHDRAWN) and three ported files (1)
engine version:
0.1.dev20073+g8e685d198; repo commit 2f551bb (patch 0016, 2026-10-02) (1) ·
base c3046d1 (recommended since 2026-08-13; first released on f8ea5bb); repo commit 3dd2d88 (1) ·
final production build v30 (numbers measured on v24); repo commit 218b47f (1) ·
GNmac-lab at 2ad5741ab1; image ds41-cmp170hx:1.2.0-fidelity (1) ·
tag recipe-09 (309ef6d1); TabbyAPI deepseek-v41 (ce250ff) (1)
cards:
card count:
topology:
PP4 (VLLM_PP_LAYER_PARTITION=12,12,12,7) with DSpark, 5 draft tokens; PCIe Gen2 x4, no P2P (1) ·
PP4 (VLLM_PP_LAYER_PARTITION=13,11,11,10), DFlash2 k=7; PCIe Gen2 x4, no P2P (1) ·
PP4, VLLM_PP_LAYER_PARTITION=12,12,12,12, MTP 3, FULL_AND_PIECEWISE CUDA graphs; PCIe Gen2 x4 per card, no P2P. Three cards: PP3 with 16,17,15 or 16,16,16 (1) ·
TP4, P2P on (1) ·
TP5 across four CMP 170HX and one RTX PRO 6000 Blackwell (native backend, reduces through host memory); MTP draft on the PRO (1)
context:
prefill tps:
vram per card:
64 GB, --gpu-memory-utilization 0.85 (1) ·
64 GB, --gpu-memory-utilization 0.93 (0.94 is the ceiling) (1) ·
64 GiB (64,911 MiB usable), GPU_MEMORY_UTILIZATION 0.98 (1) ·
64 GiB; --gpu-memory-utilization 0.94 plus per-rank KV budgets; measured 34.0-34.3 GiB per 12-layer rank, 37.9 GiB on the drafter rank (1) ·
gpu_split 58 GB per CMP, 70 GB on the PRO (1)
power cap:
driver:
unlock build:
status:
tested by:
tested on:
source:
Showing below up to 5 results in range #1 to #5.