Profiling model using genai-perf

[nv-braf]

Successfully profiled and checkpointed a a vLLM run through MA using genai-perf. Here is the resulting table output:

Models (Inference):
Model       Batch   Concurrency   Model Config Path          Instance Group   Max Batch Size   Satisfies Constraints   Throughput (infer/sec)   p99 Latency (ms)   p99 Inter Token Latency (ms)   p99 Time To First Token (ms)   Output Token Throughput (infer/sec)  
gpt2_vllm   1       1             gpt2_vllm_config_default   1:MODEL          0                Yes                     8.9                      117.0              11.3                           47704.6                        76192.3                              

Models (GPU Metrics):
Model       GPU UUID                                   Batch   Concurrency   Model Config Path          Instance Group   Satisfies Constraints   GPU Memory Usage (MB)   GPU Utilization (%)   GPU Power Usage (W)  
gpt2_vllm   GPU-8557549f-9c89-4384-8bd6-1fd823c342e0   1       1             gpt2_vllm_config_default   1:MODEL          Yes                     12768.5                 20.4                  65.2                 

Server Only:
Model           GPU UUID                                   GPU Memory Usage (MB)   GPU Utilization (%)   GPU Power Usage (W)  
triton-server   GPU-8557549f-9c89-4384-8bd6-1fd823c342e0   12768.0                 0.5                   42.1                 

[tgerdesnv]

Great job! Awesome to see it able to run and print out results.