{"tag": "cold-shim-on", "model": "/data/zhoutaichang/embedding_infer/models/spark2_5-gptq-g128", "init_s": 93.77, "rss_mb": 8502.3, "rss_peak_mb": 8502.6, "maxrss_mb": 8502.6, "kvcache_space_gib": "4", "cache_root": "/tmp/claude-2005/-data-zhoutaichang-embedding-infer/e695f57b-0163-4287-b1be-1d0ec9330768/scratchpad/vllm_cache_fresh_1870348", "aot_shim": "1", "max_model_len": 2048, "tokens_generated": 8} {"tag": "warm-shim-on", "model": "/data/zhoutaichang/embedding_infer/models/spark2_5-gptq-g128", "init_s": 14.87, "rss_mb": 8335.3, "rss_peak_mb": 8346.9, "maxrss_mb": 8346.9, "kvcache_space_gib": "4", "cache_root": "/tmp/claude-2005/-data-zhoutaichang-embedding-infer/e695f57b-0163-4287-b1be-1d0ec9330768/scratchpad/vllm_cache_fresh_1870348", "aot_shim": "1", "max_model_len": 2048, "tokens_generated": 8} {"tag": "warm-shim-off", "model": "/data/zhoutaichang/embedding_infer/models/spark2_5-gptq-g128", "init_s": 49.83, "rss_mb": 8473.4, "rss_peak_mb": 8473.4, "maxrss_mb": 8473.4, "kvcache_space_gib": "4", "cache_root": "/tmp/claude-2005/-data-zhoutaichang-embedding-infer/e695f57b-0163-4287-b1be-1d0ec9330768/scratchpad/vllm_cache_fresh_1870348", "aot_shim": "0", "max_model_len": 2048, "tokens_generated": 8} {"tag": "mem-kv4gib", "model": "/data/zhoutaichang/embedding_infer/models/spark2_5-gptq-g128", "init_s": 15.11, "rss_mb": 8329.1, "rss_peak_mb": 8356.6, "maxrss_mb": 8356.6, "kvcache_space_gib": "4", "kv_cache_memory_bytes": null, "cache_root": "(default)", "aot_shim": "1", "max_model_len": 2048, "tokens_generated": 8} {"tag": "mem-kv128mib", "model": "/data/zhoutaichang/embedding_infer/models/spark2_5-gptq-g128", "init_s": 14.24, "rss_mb": 5254.5, "rss_peak_mb": 5359.9, "maxrss_mb": 5359.9, "kvcache_space_gib": "1", "kv_cache_memory_bytes": 134217728, "cache_root": "(default)", "aot_shim": "1", "max_model_len": 2048, "tokens_generated": 8} {"tag": "mem-kv64mib", "model": "/data/zhoutaichang/embedding_infer/models/spark2_5-gptq-g128", "init_s": 76.57, "rss_mb": 5394.2, "rss_peak_mb": 5394.4, "maxrss_mb": 5394.4, "kvcache_space_gib": "1", "kv_cache_memory_bytes": 67108864, "cache_root": "(default)", "aot_shim": "1", "max_model_len": 1024, "tokens_generated": 8}