MinimaxH3, using Tesla V100 GPUs, dual-card parallel generation
Tesla V100 16GB GPU, purchased for 980 yuan, miniMax H3 image-to-video generation 1280*736 resolution
［ninfer v100X2］v100 achieves new glories! qwen3.8 27b nvfp4 2000t/s prefill, 115t/s decode
V100 boosts performance by 22,100%! Is NVLink the only viable path? MoE model tests reveal the truth.
Dual 3080s running Qwen 3.8-27B at 90+ token/s, peak 200+ token/s; KV context 255k tokens
【Creator Tested】Instant output in 15s! What is the maximum resolution of Tesla V100 x MiniMax H3?
LMDeploy Local Large Model Benchmark: V100 32G Single GPU vs Dual V100 16G
［Ninfer V100X2］v100 achieves glory again! qwen3.8 27b nvfp4 peak 236t/s decoding, 2000t/s prefill
[ninfer-V100X2] V100 achieves glory yet again, real-world usage non-benchmark scenario 183t/s decoding, 2500t/s prefill
[Most Comprehensive Test] Deploying Qwen3.8 27B nvfp4 on V100 with vLLM + dFlash2: Decode speed improved by 3x, Prefill speed improved by 15x
V100 unlocked NVFP4, GLM5.3F insane output 145 toks
Quad V100 16G vs Dual V100 32G: Qwen3.5 27B / 35B Inference Performance Test
How slow is running MiniMax H3 on a V100?
V100 Purchase and Usage Guide—Extremely Detailed, Super Practical
V100 32GB PCIe using llama.cpp locally inferencing Qwen3.8-27B on Windows, averaging nearly 50 tps — sharing the optimal settings and some tips
[Chinese Dub] Qwen3.8-27B based on NInfer framework, 220 tokens per second?
70B large model, dual V100 vs. RTX4090—which is more suitable?
Meme, V100 32G dual GPUs generating minimax-h3 fp16 precision generation test
V100 running Qwen3.8 27B Qianwen nvfp4 with speed over 50 tokens, about half the speed of Pro 6000d. In the current environment of soaring graphics card prices, it is probably the most cost-effective AI operation graphics card.
Share my invention of new technology: Expert Custom Edition, qwen3.8-flush-next-W4A16-Modular, aggregated 2000/s, single stream 110/s
