[refact] unified soc_version code (#4359)

### What this PR does / why we need it?

Currently, there are two paths to judge the chip type in code,
`get_ascend_soc_version` use `get_soc_version` api in torch_npu, and
`is_310p` `use _build_info.__soc_version__`, which generate when
install. We need to unify the two paths.

We need to unify these codes based on the following points:

1. We need to ensure consistency in chip type judgment between compiling
and running states;
2. In compiling state, we need chip type to complete op's compilation,
but in running state, we only need device
type(910B/910_93/310P/910_95/etc) to make code branch judgement;
3. In compiling state, torch_npu may not have been installed yet, so we
can't use torch_npu's api.

Based on the above points, we have made the following changes:

1. When user set env `SOC_VERSION`, use it; when not set, query
soc_version by `npu-smi`;
2. generate device_type based on soc_version when compiling, and write
`__device_type__` instead of `__soc_version__` in `_build_info.py`;
3. In running state, use `__device_type__` to judge code branch.

### Does this PR introduce _any_ user-facing change?

When not set env `SOC_VERSION`, it will not be `ASCEND910B1` by default,
we will query soc_version by `npu-smi`. And env `SOC_VERSION` must be in
the list `soc_to_device` in `setup.py`.

- vLLM version: v0.11.0
- vLLM main:
2918c1b49c

Signed-off-by: zzzzwwjj <1183291235@qq.com>

vllm_ascend/worker/model_runner_v1.py

@@ -138,9 +138,9 @@ from vllm_ascend.spec_decode.interface import SpecDcodeType
 from vllm_ascend.spec_decode.mtp_proposer import MtpProposer
 from vllm_ascend.torchair.torchair_mtp_proposer import TorchairMtpProposer
 from vllm_ascend.utils import (ACL_FORMAT_FRACTAL_ND, ACL_FORMAT_FRACTAL_NZ,
-                               AscendSocVersion, ProfileExecuteDuration,
-                               enable_sp, get_ascend_soc_version, is_310p,
-                               is_enable_nz, is_moe_model, lmhead_tp_enable,
+                               AscendDeviceType, ProfileExecuteDuration,
+                               enable_sp, get_ascend_device_type, is_enable_nz,
+                               is_moe_model, lmhead_tp_enable,
                                prefill_context_parallel_enable)
 from vllm_ascend.worker.npu_input_batch import CachedRequestState, InputBatch
 
@@ -161,7 +161,7 @@ import torch_npu
 # if true, allow tensor initialization and casting with internal format (e.g., NZ)
 torch.npu.config.allow_internal_format = True
 
-if is_310p():
+if get_ascend_device_type() == AscendDeviceType._310P:
     torch_npu.npu.set_compile_mode(jit_compile=False)
     ACL_FORMAT = ACL_FORMAT_FRACTAL_NZ
 else:
@@ -2226,14 +2226,14 @@ class NPUModelRunner(LoRAModelRunnerMixin):
         if not is_moe_model(self.vllm_config):
             return None
 
-        soc_version = get_ascend_soc_version()
+        soc_version = get_ascend_device_type()
         quant_type = getattr(self.vllm_config.model_config.hf_config,
                              'moe_quantize', None)
         model_type = self.vllm_config.model_config.hf_config.model_type
 
         if not self.parallel_config.enable_expert_parallel:
             moe_comm_type = MoECommType.ALLGATHER
-        elif soc_version in {AscendSocVersion.A2}:
+        elif soc_version in {AscendDeviceType._910B}:
             if (num_tokens <= self.mc2_tokens_capacity
                     and self.parallel_config.world_size_across_dp >= 16):
                 moe_comm_type = MoECommType.MC2
@@ -2244,7 +2244,7 @@ class NPUModelRunner(LoRAModelRunnerMixin):
                 else:
                     moe_comm_type = MoECommType.ALLGATHER
 
-        elif soc_version in {AscendSocVersion.A3}:
+        elif soc_version in {AscendDeviceType._910_93}:
             moe_comm_type = (MoECommType.MC2
                              if num_tokens <= self.mc2_tokens_capacity else
                              MoECommType.ALLTOALL)
@@ -3183,7 +3183,7 @@ class NPUModelRunner(LoRAModelRunnerMixin):
             self.model = get_model(vllm_config=self.vllm_config)
             if self.dynamic_eplb:
                 model_register(self.model, self.model_config)
-            if is_310p():
+            if get_ascend_device_type() == AscendDeviceType._310P:
                 from vllm.model_executor.layers.linear import (
                     MergedColumnParallelLinear, QKVParallelLinear,
                     RowParallelLinear)

vllm_ascend/worker/worker_v1.py

@@ -50,7 +50,7 @@ from vllm_ascend.cpu_binding import bind_cpus
 from vllm_ascend.device_allocator.camem import CaMemAllocator
 from vllm_ascend.distributed.parallel_state import init_ascend_model_parallel
 from vllm_ascend.platform import NPUPlatform
-from vllm_ascend.utils import (init_ascend_soc_version, is_enable_nz,
+from vllm_ascend.utils import (check_ascend_device_type, is_enable_nz,
                                prefill_context_parallel_enable,
                                register_ascend_customop, sleep_mode_enabled,
                                try_register_lib)
@@ -91,7 +91,7 @@ class NPUWorker(WorkerBase):
         register_ascend_customop(vllm_config)
         # init ascend config and soc version
         init_ascend_config(vllm_config)
-        init_ascend_soc_version()
+        check_ascend_device_type()
         use_sparse = False
         if vllm_config.model_config is not None:
             use_sparse = hasattr(vllm_config.model_config.hf_config,