Merge remote-tracking branch 'origin/v0.6.2-dev' into v0.6.2-dev

tests/kernels/test_permute_cols.py → tests/kernels/untest_permute_cols.py

@@ -3,13 +3,22 @@ import torch
 
 from tests.kernels.utils import opcheck
 from vllm._custom_ops import permute_cols
+from .utils import torch_version
 
 
 @pytest.mark.parametrize('shape', [(1, 512), (544, 4096), (67, 8192)])
 @pytest.mark.parametrize('dtype', [torch.bfloat16, torch.float16])
 def test_permute_cols(shape, dtype):
+    if torch_version.startswith("2.3"):
+        x = torch.randn(shape, dtype=dtype).cuda()
+        perm = torch.randperm(x.shape[1]).to(torch.int).cuda()
+        y = permute_cols(x, perm)
+        torch.allclose(y, x[:, perm])
+    elif torch_version.startswith("2.4"):
         x = torch.randn(shape, dtype=dtype).cuda()
         perm = torch.randperm(x.shape[1]).to(torch.int).cuda()
         opcheck(torch.ops._C.permute_cols, (x, perm))
         y = permute_cols(x, perm)
         torch.testing.assert_close(y, x[:, perm])
+    else:
+        print(f"PyTorch version {torch_version} is not specifically handled.")
\ No newline at end of file

tests/kernels/utils.py

@@ -30,6 +30,8 @@ ALL_OPCHECK_TEST_UTILS: Tuple[str, ...] = (
     "test_aot_dispatch_dynamic",
 )
 
+torch_version = torch.__version__
+
 
 class QKVInputs(NamedTuple):
     '''
@@ -974,9 +976,10 @@ def fp8_allclose(
                           equal_nan=equal_nan)).item())
 
 
-# A special version of op check that has a restricted default set of test_utils
-# and a patched version of allclose that supports fp8 types.
-def opcheck(op: Union[torch._ops.OpOverload, torch._ops.OpOverloadPacket,
+if torch_version.startswith("2.4"):
+    # A special version of op check that has a restricted default set of test_utils
+    # and a patched version of allclose that supports fp8 types.
+    def opcheck(op: Union[torch._ops.OpOverload, torch._ops.OpOverloadPacket,
                         torch._library.custom_ops.CustomOpDef],
                 args: Tuple[Any, ...],
                 kwargs: Optional[Dict[str, Any]] = None,

tests/models/decoder_only/vision_language/test_intern_vit.py

@@ -4,7 +4,7 @@ import os
 import pytest
 import torch
 import torch.nn as nn
-from huggingface_hub import snapshot_download
+# from huggingface_hub import snapshot_download
 from transformers import AutoConfig, AutoModel, CLIPImageProcessor
 
 from ....conftest import _ImageAssets, cleanup
@@ -14,10 +14,8 @@ from ....utils import models_path_prefix
 # dynamic_module and trust_remote_code for hf_runner
 DOWNLOAD_PATTERN = ["*.json", "*.py", "*.safetensors", "*.txt", "*.model"]
 models = [
-    snapshot_download(os.path.join(models_path_prefix, "OpenGVLab/InternViT-300M-448px"),
-                      allow_patterns=DOWNLOAD_PATTERN),
-    snapshot_download(os.path.join(models_path_prefix, "OpenGVLab/InternViT-6B-448px-V1-5"),
-                      allow_patterns=DOWNLOAD_PATTERN),
+    os.path.join(models_path_prefix, "OpenGVLab/InternViT-300M-448px"),
+    os.path.join(models_path_prefix, "OpenGVLab/InternViT-6B-448px-V1-5"),
 ]
 
 

tests/multimodal/test_utils.py

@@ -5,12 +5,13 @@ from typing import Dict, Tuple
 
 import numpy as np
 import pytest
+import os
 from PIL import Image
 from transformers import AutoConfig, AutoTokenizer
 
 from vllm.multimodal.utils import (async_fetch_image, fetch_image,
                                    repeat_and_pad_placeholder_tokens)
-from ..utils import urls_port
+from ..utils import models_path_prefix, urls_port
 
 # Test different image extensions (JPG/PNG) and formats (gray/RGB/RGBA)
 TEST_IMAGE_URLS = [
@@ -85,7 +86,7 @@ async def test_fetch_image_base64(url_images: Dict[str, Image.Image],
         assert _image_equals(data_image_sync, data_image_async)
 
 
-@pytest.mark.parametrize("model", ["llava-hf/llava-v1.6-mistral-7b-hf"])
+@pytest.mark.parametrize("model", [os.path.join(models_path_prefix, "llava-hf/llava-v1.6-mistral-7b-hf")])
 def test_repeat_and_pad_placeholder_tokens(model):
     config = AutoConfig.from_pretrained(model)
     image_token_id = config.image_token_index

tests/tensorizer_loader/test_tensorizer.py

@@ -88,7 +88,7 @@ def test_load_with_tensorizer(mock_agent, tensorizer_config):
 @pytest.mark.skipif(not is_curl_installed(), reason="cURL is not installed")
 def test_can_deserialize_s3(vllm_runner):
     model_ref = os.path.join(models_path_prefix, "EleutherAI/pythia-1.4b")
-    tensorized_path = f"s3://tensorized/{model_ref}/fp16/model.tensors"
+    tensorized_path = f"{model_ref}/fp16/model.tensors"
 
     with vllm_runner(model_ref,
                      load_format="tensorizer",
@@ -341,7 +341,7 @@ def test_raise_value_error_on_invalid_load_format(vllm_runner):
 def test_tensorizer_with_tp_path_without_template(vllm_runner):
     with pytest.raises(ValueError):
         model_ref = os.path.join(models_path_prefix, "EleutherAI/pythia-1.4b")
-        tensorized_path = f"s3://tensorized/{model_ref}/fp16/model.tensors"
+        tensorized_path = f"{model_ref}/fp16/model.tensors"
 
         vllm_runner(
             model_ref,

tests/tokenization/test_tokenizer.py

@@ -5,9 +5,15 @@ from transformers import PreTrainedTokenizerBase
 from vllm.transformers_utils.tokenizer import get_tokenizer
 from ..utils import models_path_prefix
 
+# TOKENIZER_NAMES = [
+#     os.path.join(models_path_prefix, "facebook/opt-125m"),
+#     os.path.join(models_path_prefix, "gpt2"),
+# ]
+
+# export HF_ENDPOINT=https://hf-mirror.com
 TOKENIZER_NAMES = [
-    os.path.join(models_path_prefix, "facebook/opt-125m"),
-    os.path.join(models_path_prefix, "gpt2"),
+    "facebook/opt-125m",
+    "gpt2",
 ]
 
 

tests/utils.py

@@ -23,7 +23,7 @@ from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.entrypoints.openai.cli_args import make_arg_parser
 from vllm.model_executor.model_loader.loader import get_model_loader
 from vllm.platforms import current_platform
-from vllm.utils import (FlexibleArgumentParser, cuda_device_count_stateless,
+from vllm.utils import (FlexibleArgumentParser, GB_bytes,cuda_device_count_stateless,
                         get_open_port, is_hip)
 import vllm.envs as envs
 import os
@@ -459,6 +459,36 @@ def fork_new_process_for_each_test(
 
     return wrapper
 
+def large_gpu_test(*, min_gb: int):
+    """
+    Decorate a test to be skipped if no GPU is available or it does not have
+    sufficient memory.
+
+    Currently, the CI machine uses L4 GPU which has 24 GB VRAM.
+    """
+    try:
+        if current_platform.is_cpu():
+            memory_gb = 0
+        else:
+            memory_gb = current_platform.get_device_total_memory() / GB_bytes
+    except Exception as e:
+        warnings.warn(
+            f"An error occurred when finding the available memory: {e}",
+            stacklevel=2,
+        )
+
+        memory_gb = 0
+
+    test_skipif = pytest.mark.skipif(
+        memory_gb < min_gb,
+        reason=f"Need at least {memory_gb}GB GPU memory to run the test.",
+    )
+
+    def wrapper(f: Callable[_P, None]) -> Callable[_P, None]:
+        return test_skipif(fork_new_process_for_each_test(f))
+
+    return wrapper
+
 
 def multi_gpu_test(*, num_gpus: int):
     """

vllm/__init__.py

@@ -11,7 +11,7 @@ from vllm.outputs import (CompletionOutput, EmbeddingOutput,
                           EmbeddingRequestOutput, RequestOutput)
 from vllm.pooling_params import PoolingParams
 from vllm.sampling_params import SamplingParams
-from vllm.version import __version__, __version_tuple__, __dcu_version__
+from vllm.version import __version__, __version_tuple__, __hcu_version__
 
 
 __all__ = [

vllm/_custom_ops.py

@@ -12,7 +12,7 @@ from vllm.platforms import current_platform
 try:
     from lmslim import quant_ops 
 except Exception:
-    print("INFO: Please install lmslim if you want to infer gptq or awq model.\n") 
+    print("INFO: Please install lmslim if you want to infer gptq or awq  or w8a8 model.\n") 
 
 logger = init_logger(__name__)
 
@@ -706,9 +706,9 @@ def cutlass_scaled_mm(a: torch.Tensor,
     # torch.ops._C.cutlass_scaled_mm(out, a, b, scale_a, scale_b, bias)
 
     # return out
+    #return quant_ops.cutlass_scaled_mm(a, b, scale_a, scale_b, out_dtype, bias)
     return quant_ops.rocblas_scaled_mm_nn(a, b, scale_a, scale_b, out_dtype, bias)
 
-
 def rocblas_scaled_mm(a: torch.Tensor,
                       b: torch.Tensor,
                       scale_a: torch.Tensor,

vllm/attention/selector.py

@@ -207,7 +207,7 @@ def which_attn_to_use(
                 # not Instinct series GPUs.
                 logger.info("flash_attn is not supported on NAVI GPUs.")
         else:
-            logger.info("%s is not supported in AMD GPUs.", selected_backend)
+            logger.info("%s is not supported in hcus.", selected_backend)
         return _Backend.ROCM_FLASH
 
     # FlashAttn in NVIDIA GPUs.

vllm/benchmarks/benchmark_throughput.py

@@ -522,7 +522,7 @@ if __name__ == "__main__":
         default="auto",
         help='Data type for kv cache storage. If "auto", will use model '
         'data type. CUDA 11.8+ supports fp8 (=fp8_e4m3) and fp8_e5m2. '
-        'ROCm (AMD GPU) supports fp8 (=fp8_e4m3)')
+        'ROCm (hcu) supports fp8 (=fp8_e4m3)')
     parser.add_argument(
         '--quantization-param-path',
         type=str,
@@ -531,7 +531,7 @@ if __name__ == "__main__":
         'This should generally be supplied, when KV cache dtype is FP8. '
         'Otherwise, KV cache scaling factors default to 1.0, which may cause '
         'accuracy issues. FP8_E5M2 (without scaling) is only supported on '
-        'cuda version greater than 11.8. On ROCm (AMD GPU), FP8_E4M3 is '
+        'cuda version greater than 11.8. On ROCm (hcu), FP8_E4M3 is '
         'instead supported for common inference criteria.')
     parser.add_argument("--device",
                         type=str,

vllm/config.py

@@ -936,7 +936,7 @@ class ParallelConfig:
             self.disable_custom_all_reduce = True
             logger.info(
                 "Disabled the custom all-reduce kernel because it is not "
-                "supported on AMD GPUs.")
+                "supported on hcus.")
         if self.ray_workers_use_nsight and not self.use_ray:
             raise ValueError("Unable to use nsight profiling unless workers "
                              "run with Ray.")

vllm/distributed/device_communicators/pynccl_wrapper.py

@@ -195,7 +195,7 @@ class NCCLLibrary:
         except Exception as e:
             logger.error(
                 "Failed to load NCCL library from %s ."
-                "It is expected if you are not running on NVIDIA/AMD GPUs."
+                "It is expected if you are not running on NVIDIA/hcus."
                 "Otherwise, the nccl library might not exist, be corrupted "
                 "or it does not support the current platform %s."
                 "If you already have the library, please set the "

vllm/engine/arg_utils.py

@@ -294,7 +294,7 @@ class EngineArgs:
             default=EngineArgs.kv_cache_dtype,
             help='Data type for kv cache storage. If "auto", will use model '
             'data type. CUDA 11.8+ supports fp8 (=fp8_e4m3) and fp8_e5m2. '
-            'ROCm (AMD GPU) supports fp8 (=fp8_e4m3)')
+            'ROCm (hcu) supports fp8 (=fp8_e4m3)')
         parser.add_argument(
             '--quantization-param-path',
             type=nullable_str,
@@ -304,7 +304,7 @@ class EngineArgs:
             'KV cache dtype is FP8. Otherwise, KV cache scaling factors '
             'default to 1.0, which may cause accuracy issues. '
             'FP8_E5M2 (without scaling) is only supported on cuda version'
-            'greater than 11.8. On ROCm (AMD GPU), FP8_E4M3 is instead '
+            'greater than 11.8. On ROCm (hcu), FP8_E4M3 is instead '
             'supported for common inference criteria.')
         parser.add_argument('--max-model-len',
                             type=int,

vllm/model_executor/layers/quantization/utils/w8a8_utils.py

@@ -4,12 +4,12 @@ import torch
 
 from vllm import _custom_ops as ops
 from vllm.platforms import current_platform
-from vllm.utils import is_hip
+from vllm.utils import is_hip,W8a8GetCacheJSON
 
 # Input scaling factors are no longer optional in _scaled_mm starting
 # from pytorch 2.5. Allocating a dummy tensor to pass as input_scale
 TORCH_DEVICE_IDENTITY = torch.ones(1).cuda() if is_hip() else None
-
+W8A8_TRITONJSON=W8a8GetCacheJSON()
 
 def cutlass_fp8_supported() -> bool:
     # cutlass is not supported on Rocm
@@ -200,12 +200,37 @@ def apply_int8_linear(
     x_q, x_scale, _ = ops.scaled_int8_quant(input, input_scale)
 
     if w8a8_strategy==1:
+        m=x_q.shape[0]
+        k=x_q.shape[1]
+        n=weight.shape[1]
+        if f"{m}_{n}_{k}" in  W8A8_TRITONJSON.triton_json_dict[0]:
+            best_config=W8A8_TRITONJSON.triton_json_dict[0][f"{m}_{n}_{k}"]
+            #print("json files:",best_config)
+        elif f"1_{n}_{k}" in  W8A8_TRITONJSON.triton_json_dict[0]:
+            if m<64:
+                m_= 32
+            elif m<128:
+                m_=64
+            elif m<256:
+                m_=128
+            elif m<512:
+                m_=256
+            elif m<1024:
+                m_=512
+            else:
+                m_=1024
+            best_config=W8A8_TRITONJSON.triton_json_dict[0][f"{m_}_{n}_{k}"]
+            
+        else: 
+            best_config=None
+            print("config not found!")
+            
         return ops.triton_scaled_mm(x_q,
                                 weight,
                                 scale_a=x_scale,
                                 scale_b=weight_scale,
                                 out_dtype=input.dtype,
-                                bias=bias)
+                                bias=bias,best_config=best_config)
     elif w8a8_strategy==2:
         return ops.cutlass_scaled_mm(x_q,
                                     weight,

vllm/model_executor/model_loader/utils.py

@@ -23,14 +23,14 @@ def get_model_architecture(
         model_config: ModelConfig) -> Tuple[Type[nn.Module], str]:
     architectures = getattr(model_config.hf_config, "architectures", [])
     visions = getattr(model_config.hf_config, "visual", []) or getattr(model_config.hf_config, "vision_config", [])
-    support_nn_architectures = ['LlamaForCausalLM', 'QWenLMHeadModel', 'Qwen2ForCausalLM', 'Qwen2MoeForCausalLM', 'Qwen2VLForConditionalGeneration', 'ChatGLMModel', 'ChatGLMForConditionalGeneration', 'BaichuanForCausalLM', 'BloomForCausalLM', 'MedusaModel', 'MixtralForCausalLM', 'MLPSpeculatorPreTrainedModel']  
+    support_nn_architectures = ['LlamaForCausalLM', 'QWenLMHeadModel', 'Qwen2ForCausalLM', 'Qwen2MoeForCausalLM', 'Qwen2VLForConditionalGeneration', 'ChatGLMModel', 'ChatGLMForConditionalGeneration', 'BaichuanForCausalLM', 'BloomForCausalLM', 'MedusaModel', 'MixtralForCausalLM', 'MLPSpeculatorPreTrainedModel', 'FalconForCausalLM']  
     if any(arch in architectures for arch in support_nn_architectures): 
         if os.getenv('LLAMA_NN') != '0': 
              if (architectures == ['QWenLMHeadModel'] or architectures == ['ChatGLMModel'] ) and visions != []:
                 os.environ['LLAMA_NN'] = '0'
              else:
                 os.environ['LLAMA_NN'] = '1'
-        if architectures == ['BloomForCausalLM'] or os.getenv('LM_NN') == '0':
+        if (architectures == ['BloomForCausalLM'] or architectures == ['FalconForCausalLM']) or os.getenv('LM_NN') == '0':
             os.environ['LM_NN'] = '0'
         else:
             os.environ['LM_NN'] = '1'

vllm/model_executor/models/falcon.py

@@ -21,6 +21,8 @@
 import math
 from typing import Iterable, List, Optional, Tuple, Union
 
+import os
+import re
 import torch
 from torch import nn
 from torch.nn import LayerNorm
@@ -47,6 +49,9 @@ from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import IntermediateTensors
 from vllm.transformers_utils.configs import RWConfig
 
+from vllm import _custom_ops as ops
+from vllm.model_executor.utils import pad_weight, gemm_bank_conf
+
 FalconConfig = Union[HF_FalconConfig, RWConfig]
 
 
@@ -176,6 +181,11 @@ class FalconAttention(nn.Module):
                                   cache_config=cache_config,
                                   quant_config=quant_config)
             
+        self.quant_method = None
+        if quant_config is not None:
+            self.quant_method=quant_config.get_name()
+            self.quant_config=quant_config
+
     def forward(
         self,
         positions: torch.Tensor,
@@ -184,6 +194,8 @@ class FalconAttention(nn.Module):
         attn_metadata: AttentionMetadata,
     ) -> torch.Tensor:
         qkv, bias = self.query_key_value(hidden_states)
+        if os.environ.get('FA_PAD') == '1' and self.quant_method is None:
+            qkv = qkv[...,:-32]
         if bias is not None:
             qkv += bias
         q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
@@ -246,6 +258,9 @@ class FalconDecoderLayer(nn.Module):
         self.mlp = FalconMLP(config, quant_config)
         self.config = config
 
+        if (not hasattr(config, "num_ln_in_parallel_attn")):
+            config.num_ln_in_parallel_attn = None
+            
         if (config.num_ln_in_parallel_attn is None
                 and config.new_decoder_architecture):
             config.num_ln_in_parallel_attn = 2
@@ -404,6 +419,17 @@ class FalconForCausalLM(nn.Module):
         self.logits_processor = LogitsProcessor(config.vocab_size)
         self.sampler = Sampler()
         
+        self.quant_method = None
+        if quant_config is not None:
+            self.quant_method=quant_config.get_name()
+            self.quant_config=quant_config
+        
+        self.use_llama_nn = os.environ.get('LLAMA_NN') == '1'
+        self.use_gemm_pad = os.environ.get('GEMM_PAD') == '1'
+        self.use_fa_pad = os.environ.get('FA_PAD') == '1'
+        self.use_awq_pad = os.environ.get('AWQ_PAD') == '1'
+        self.w8a8_strategy=int(os.getenv('W8A8_SUPPORT_METHODS', '0'))
+
     def forward(
         self,
         input_ids: torch.LongTensor,
@@ -481,3 +507,33 @@ class FalconForCausalLM(nn.Module):
             weight_loader = getattr(param, "weight_loader",
                                     default_weight_loader)
             weight_loader(param, loaded_weight)
+
+        if self.use_llama_nn and self.quant_method is None :
+            lay_key_words = [
+                "self_attention.query_key_value.weight",
+                "self_attention.dense.weight",
+                "mlp.dense_h_to_4h.weight",
+                "mlp.dense_4h_to_h.weight",
+            ]
+            combined_words = "|".join(lay_key_words)
+            
+            lay_qkv_words = ["self_attention.query_key_value.weight"]   
+            qkv_words = "|".join(lay_qkv_words)          
+            
+            for layername, weight in params_dict.items():
+                matches = re.findall(combined_words, layername)
+                if matches:         
+                    if self.use_gemm_pad and gemm_bank_conf(weight.data.shape[0]):
+                        weight.data = pad_weight(weight.data, 32)  
+                        
+                    if self.use_fa_pad and (re.findall(qkv_words, layername)):
+                        if not gemm_bank_conf(weight.data.shape[0]):
+                            weight.data = pad_weight(weight.data, 32)
+                                 
+                    _weight = torch.zeros_like(weight.data)
+                    ori_shape =_weight.shape
+                    
+                    ops.trans_w16_gemm(_weight, weight.data, _weight.shape[0], _weight.shape[1])
+                    weight.data.copy_(_weight)
+                    
+                    weight.data=weight.data.reshape(ori_shape[1], -1)
\ No newline at end of file