[Usage]: How to use AsyncLLMEngine in a multithreaded environment

[pvardanis]

Your current environment

The output of `python collect_env.py`

Collecting environment information...
PyTorch version: 2.4.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Debian GNU/Linux 11 (bullseye) (x86_64)
GCC version: (Debian 10.2.1-6) 10.2.1 20210110
Clang version: Could not collect
CMake version: version 3.18.4
Libc version: glibc-2.31

Python version: 3.9.20 (main, Oct 25 2024, 11:23:40)  [GCC 10.2.1 20210110] (64-bit runtime)
Python platform: Linux-5.10.0-32-cloud-amd64-x86_64-with-glibc2.31
Is CUDA available: True
CUDA runtime version: 11.8.89
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: GPU 0: NVIDIA A100-SXM4-40GB
Nvidia driver version: 550.90.07
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture:                         x86_64
CPU op-mode(s):                       32-bit, 64-bit
Byte Order:                           Little Endian
Address sizes:                        46 bits physical, 48 bits virtual
CPU(s):                               12
On-line CPU(s) list:                  0-11
Thread(s) per core:                   2
Core(s) per socket:                   6
Socket(s):                            1
NUMA node(s):                         1
Vendor ID:                            GenuineIntel
CPU family:                           6
Model:                                85
Model name:                           Intel(R) Xeon(R) CPU @ 2.20GHz
Stepping:                             7
CPU MHz:                              2200.160
BogoMIPS:                             4400.32
Hypervisor vendor:                    KVM
Virtualization type:                  full
L1d cache:                            192 KiB
L1i cache:                            192 KiB
L2 cache:                             6 MiB
L3 cache:                             38.5 MiB
NUMA node0 CPU(s):                    0-11
Vulnerability Gather data sampling:   Not affected
Vulnerability Itlb multihit:          Not affected
Vulnerability L1tf:                   Not affected
Vulnerability Mds:                    Not affected
Vulnerability Meltdown:               Not affected
Vulnerability Mmio stale data:        Vulnerable: Clear CPU buffers attempted, no microcode; SMT Host state unknown
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed:               Mitigation; Enhanced IBRS
Vulnerability Spec rstack overflow:   Not affected
Vulnerability Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:             Mitigation; Enhanced / Automatic IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Vulnerable: Clear CPU buffers attempted, no microcode; SMT Host state unknown
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm mpx avx512f avx512dq rdseed adx smap clflushopt clwb avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves arat avx512_vnni md_clear arch_capabilities

Versions of relevant libraries:
[pip3] mypy==1.7.1
[pip3] mypy-extensions==1.0.0
[pip3] numpy==1.25.0
[pip3] nvidia-cublas-cu12==********
[pip3] nvidia-cuda-cupti-cu12==12.1.105
[pip3] nvidia-cuda-nvrtc-cu12==12.1.105
[pip3] nvidia-cuda-runtime-cu12==12.1.105
[pip3] nvidia-cudnn-cu12==********
[pip3] nvidia-cufft-cu12==*********
[pip3] nvidia-curand-cu12==**********
[pip3] nvidia-cusolver-cu12==**********
[pip3] nvidia-cusparse-cu12==**********
[pip3] nvidia-ml-py==12.560.30
[pip3] nvidia-nccl-cu12==2.20.5
[pip3] nvidia-nvjitlink-cu12==12.6.77
[pip3] nvidia-nvtx-cu12==12.1.105
[pip3] pyzmq==25.1.2
[pip3] torch==2.4.0
[pip3] torchvision==0.19.0
[pip3] transformers==4.46.0
[pip3] triton==3.0.0
[conda] numpy                     1.23.5                   pypi_0    pypi
[conda] nvidia-ml-py              11.495.46                pypi_0    pypi
[conda] pyzmq                     26.0.3                   pypi_0    pypi
ROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.6.3.post1
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	0-11	0		N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

How would you like to use vllm

I use grpc for serving AsyncLLMEngine and the RPC that uses the engine on a separate thread as follows for every request that arrives:

request_id = str(uuid.uuid4())

        
        logger.info(f"Generating text for request_id: {request_id}")
        results_generator = await self.model.add_request(request_id, prompt, SamplingParams(temperature=0.8, top_p=0.95, max_tokens=128))
        logger.info("Generator initialized.")

        final_output = None

        cursor = 0
        async for request_output in results_generator:
            logger.info("Received output.")
            text = request_output.outputs[0].text
            # logger.info(f"Received for request_id: {request_id} -> `{text[cursor:]}`")
            final_output = request_output
            cursor = len(text)

        prompt = final_output.prompt
        text_output = [output.text for output in final_output.outputs]

It seems, this messes up with self.background_loop in AsyncLLMEngine. When the first request arrives it runs fine, but the second one just hangs. Is there a way I can run this correctly on a multithreaded environment? Seems the same loop cannot be shared across multiple threads.

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[njhill]

@pvardanis I would recommend using the async flavour of gRPC, then it should fit naturally with the async vLLM interface. You can see an example of this here.

We are however reworking these interfaces and you should soon also be able to use the synchronous LLM.generate for this.