一、问题现象（附报错日志上下文）：
单机8卡分布式训练报错，

二、软件版本:
torch=2.4.0
torch-npu=2.4.0.post2
CANN version=8.0.RC3
arch=aarch64
python=3.10.16
npu-smi=23.0.6
ubuntu=22.04
910B2 64G 8卡

三、测试步骤：
使用模型并行+数据并行，环境变量export HCCL_CONNECT_TIMEOUT=6000
报错代码如下：

def evaluate_ppo(self):  # noqa: C901
        # self.model.eval()
        """Samples model on `eval_prompts`, logs stats with `reward_fn` or `metric_fn` if provided"""
        stats = {}
        all_full_ids = []
        all_rev_kl = []
        all_lens = []
        
        table = []

        with torch.no_grad():
            for batch in tqdm(self.eval_dataloader, "Generation Evaluation", disable=(not get_rank() == 0)):
                batch, no_model_batch = batch
                batch, _ = self.eval_pipeline.move_to_device(batch, no_model_batch, self.device)
                
                 # 2. 插入NPU流同步（新增）
                stream = torch.npu.current_stream()
                stream.synchronize()
                # 获取学生模型的输出
                gen_out = self.generate(
                    **batch,
                    return_dict_in_generate=True,
                    output_scores=True
                )
                full_ids = gen_out.sequences
                gen_logits = gen_out.scores # NOTE: [b, s, h_p]
                inf_mask = torch.isinf(gen_logits)
                
                 # 5. 同步NPU流后再收集数据（新增）
                stream = torch.npu.current_stream()
                stream.synchronize()

                all_full_ids.append(full_ids)
                
                input_ids = batch["input_ids"]
                gen_ids = full_ids[:, input_ids.size(1):]
                mask = self.get_mask(full_ids)
                mask = mask[:, input_ids.size(1)-1:input_ids.size(1)+gen_ids.size(1)-1]
                lens = torch.sum(mask, dim=-1)
                
                teacher_rewards = self.reward_fn(input_ids, gen_ids)["rewards"] # \log p(y_t | y_{<t}, x)
                _, logprobs = self.compute_logits_and_log_probs(input_ids, gen_ids, inf_mask=inf_mask, base="base") # \log q_{\theta}(y_t | y_{<t}, x)
                
                kl = get_rev_kl(teacher_rewards, logprobs, mask) # 获取反向KL分数
                kl = kl.sum(-1)
                
                if self.args.length_norm:
                    kl = kl / lens

                all_rev_kl.append(kl)
                all_lens.append(lens)
            
            # 7. 添加分布式同步点（新增）
            torch.distributed.barrier()
            
            all_full_ids = torch.cat(all_full_ids, dim=0)
            all_rev_kl = torch.cat(all_rev_kl, dim=0)
            all_lens = torch.cat(all_lens, dim=0)

            full_ids = all_gather(all_full_ids, dim=1, world_size=self.dp_world_size, group=self.dp_group, op="stack")
            full_ids = full_ids.view(-1, full_ids.size(-1))

            prompt_ids = full_ids[:, :self.eval_pipeline.max_prompt_length]
            all_rev_kl = all_gather(all_rev_kl, dim=0, world_size=self.dp_world_size, group=self.dp_group)
            stats["rev_kl"] = all_rev_kl.mean()
            all_lens = all_gather(all_lens, dim=0, world_size=self.dp_world_size, group=self.dp_group)
            stats["lens"] = all_lens.float().mean()
            
            # 7. 添加分布式同步点（新增）
            torch.distributed.barrier()
            
            response_texts = []
            if get_rank() == 0:
                # 解码前同步流（新增）
                stream = torch.npu.current_stream()
                stream.synchronize()
                
                prompt_texts = self.tokenizer.batch_decode(prompt_ids, skip_special_tokens=True)
                response_texts = self.tokenizer.batch_decode(full_ids[:, self.eval_pipeline.max_prompt_length:], skip_special_tokens=True)
                gen_texts = [p + g for p, g in zip(prompt_texts, response_texts)]

                columns = ["prompts"]
                columns_data = [prompt_texts]
                # in online setting, compute the reward for validation
                columns.append("samples")
                if isinstance(gen_texts[0], str):
                    columns_data.append(gen_texts)
                else:
                    columns_data.append(gen_texts.tolist())
                table.append(list(zip(*columns_data)))
        # 9. 添加全局同步点（关键修复点）
        torch.distributed.barrier()

        # Log and display evaluation metrics
        if get_rank() == 0:
            rows = sum(list(map(list, zip(*table))), [])

            # Add metrics/rewards to the table's title
            table_title = f"Evaluation #{self.nth_evaluation}"
            for k, x in stats.items():
                if k.startswith("reward") or k.startswith("metrics"):
                    table_title += f" {k}: {significant(x)}"

            rich_table = Table(*columns, title=table_title, show_lines=True)

            for ix in range(min(3, len(rows))):
                rich_table.add_row(*[str(significant(x)) for x in rows[ix]])

            try:
                Console().print(rich_table)
            except:
                pass
        # 10. 返回前同步所有进程（新增）
        torch.distributed.barrier()
        
        self.nth_evaluation += 1
        return stats, table, response_texts

四、报错信息：
在以下这几行都报过一模一样的错误：all_lens.append(lens)、torch.distributed.barrier()、full_ids = all_gather(all_full_ids, dim=1, world_size=self.dp_world_size, group=self.dp_group, op="stack")、prompt_texts = self.tokenizer.batch_decode(prompt_ids, skip_special_tokens=True)

【03-11补充】：当eval dataloader样本数量少（如64）的时候不会报错，样本数量增多至200就报错

Generation Evaluation: 0%| | 0/106 [00:00<?, ?it/s]len(self.train_dataloader), 4
len(self.train_dataloader), 4
len(self.train_dataloader), 4
len(self.train_dataloader), 4
Generation Evaluation: 37%|███████████████████████████████████████████████████████▏ | 39/106 [39:39<1:11:12, 63.76s/it]
Generation Evaluation: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 106/106 [1:26:00<00:00, 48.68s/it]
Traceback (most recent call last):
Traceback (most recent call last):
Traceback (most recent call last):
File "/home/jovyan/gqy/minillm/./train_minillm.py", line 113, in
File "/home/jovyan/gqy/minillm/./train_minillm.py", line 113, in
File "/home/jovyan/gqy/minillm/./train_minillm.py", line 113, in
Traceback (most recent call last):
File "/home/jovyan/gqy/minillm/./train_minillm.py", line 113, in
main()main()

  File "/home/jovyan/gqy/minillm/./train_minillm.py", line 99, in main

File "/home/jovyan/gqy/minillm/./train_minillm.py", line 99, in main
main()
File "/home/jovyan/gqy/minillm/./train_minillm.py", line 99, in main
main()
File "/home/jovyan/gqy/minillm/./train_minillm.py", line 99, in main
train(train(

File "/home/jovyan/gqy/minillm/minillm/init.py", line 52, in train
File "/home/jovyan/gqy/minillm/minillm/init.py", line 52, in train
train(
File "/home/jovyan/gqy/minillm/minillm/init.py", line 52, in train
train(
File "/home/jovyan/gqy/minillm/minillm/init.py", line 52, in train
trainer.train()trainer.train()

File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 244, in train
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 244, in train
trainer.train()
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 244, in train
trainer.train()
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 244, in train
self.global_iter_count = 1self.global_iter_count = 1

File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 432, in evaluate
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 432, in evaluate
self.global_iter_count = 1
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 432, in evaluate
self.global_iter_count = 1
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 432, in evaluate

File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 505, in evaluate_ppo
File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 505, in evaluate_ppo

File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 505, in evaluate_ppo

File "/home/jovyan/gqy/minillm/minillm/trainer.py", line 505, in evaluate_ppo
all_lens.append(lens)all_lens.append(lens)
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
all_lens.append(lens)all_lens.append(lens)

File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
return func(*args, **kwargs)return func(*args, **kwargs)return func(*args, **kwargs)

File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3703, in barrier
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3703, in barrier
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3703, in barrier
return func(*args, **kwargs)
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3703, in barrier
work.wait()work.wait()

work.wait()RuntimeError
RuntimeErrorRuntimeError: : npuSynchronizeDevice:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:467 NPU function error: AclrtSynchronizeDeviceWithTimeout, error code is 507048
[ERROR] 2025-03-10-04:19:37 (PID:1185975, Device:2, RankID:2) ERR00100 PTA call acl api failed
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EI0002: [PID: 1185975] 2025-03-10-04:19:37.576.321 The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[4265454028], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
Possible Cause: 1. An exception occurs during the execution on some NPUs in the cluster. As a result, collective communication operation failed.2. The execution speed on some NPU in the cluster is too slow to complete a communication operation within the timeout interval. (default 1800s, You can set the interval by using HCCL_EXEC_TIMEOUT.)3. The number of training samples of each NPU is inconsistent.4. Packet loss or other connectivity problems occur on the communication link.
Solution: 1. If this error is reported on part of these ranks, check other ranks to see whether other errors have been reported earlier.2. If this error is reported for all ranks, check whether the error reporting time is consistent (the maximum difference must not exceed 1800s). If not, locate the cause or adjust the locate the cause or set the HCCL_EXEC_TIMEOUT environment variable to a larger value.3. Check whether the completion queue element (CQE) of the error exists in the plog(grep -rn 'error cqe'). If so, check the network connection status. (For details, see the TLS command and HCCN connectivity check examples.)4. Ensure that the number of training samples of each NPU is consistent. For details:https://www.hiascend.com/document
TraceBack (most recent call last):
The error from device(chipId:2, dieId:0), serial number is 3, hccl fftsplus task timeout occurred during task execution, stream_id:4, sq_id:4, task_id:13, stuck notify num:1, timeout:1836.[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1645]
The 0 stuck notify wait context info:(context_id=4, notify_id=3).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[4265454028], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
npuSynchronizeDevice:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:467 NPU function error: AclrtSynchronizeDeviceWithTimeout, error code is 507048
[ERROR] 2025-03-10-04:19:37 (PID:1185974, Device:1, RankID:1) ERR00100 PTA call acl api failed
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EI0002: [PID: 1185974] 2025-03-10-04:19:37.566.086 The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[3917014364], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
Possible Cause: 1. An exception occurs during the execution on some NPUs in the cluster. As a result, collective communication operation failed.2. The execution speed on some NPU in the cluster is too slow to complete a communication operation within the timeout interval. (default 1800s, You can set the interval by using HCCL_EXEC_TIMEOUT.)3. The number of training samples of each NPU is inconsistent.4. Packet loss or other connectivity problems occur on the communication link.
Solution: 1. If this error is reported on part of these ranks, check other ranks to see whether other errors have been reported earlier.2. If this error is reported for all ranks, check whether the error reporting time is consistent (the maximum difference must not exceed 1800s). If not, locate the cause or adjust the locate the cause or set the HCCL_EXEC_TIMEOUT environment variable to a larger value.3. Check whether the completion queue element (CQE) of the error exists in the plog(grep -rn 'error cqe'). If so, check the network connection status. (For details, see the TLS command and HCCN connectivity check examples.)4. Ensure that the number of training samples of each NPU is consistent. For details:https://www.hiascend.com/document
TraceBack (most recent call last):
The error from device(chipId:1, dieId:0), serial number is 3, hccl fftsplus task timeout occurred during task execution, stream_id:4, sq_id:4, task_id:13, stuck notify num:1, timeout:1836.[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1645]
The 0 stuck notify wait context info:(context_id=4, notify_id=11).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[3917014364], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
:
work.wait()npuSynchronizeDevice:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:467 NPU function error: AclrtSynchronizeDeviceWithTimeout, error code is 507048
[ERROR] 2025-03-10-04:19:37 (PID:1185976, Device:3, RankID:3) ERR00100 PTA call acl api failed
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EI0002: [PID: 1185976] 2025-03-10-04:19:37.588.583 The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[3517536076], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
Possible Cause: 1. An exception occurs during the execution on some NPUs in the cluster. As a result, collective communication operation failed.2. The execution speed on some NPU in the cluster is too slow to complete a communication operation within the timeout interval. (default 1800s, You can set the interval by using HCCL_EXEC_TIMEOUT.)3. The number of training samples of each NPU is inconsistent.4. Packet loss or other connectivity problems occur on the communication link.
Solution: 1. If this error is reported on part of these ranks, check other ranks to see whether other errors have been reported earlier.2. If this error is reported for all ranks, check whether the error reporting time is consistent (the maximum difference must not exceed 1800s). If not, locate the cause or adjust the locate the cause or set the HCCL_EXEC_TIMEOUT environment variable to a larger value.3. Check whether the completion queue element (CQE) of the error exists in the plog(grep -rn 'error cqe'). If so, check the network connection status. (For details, see the TLS command and HCCN connectivity check examples.)4. Ensure that the number of training samples of each NPU is consistent. For details:https://www.hiascend.com/document
TraceBack (most recent call last):
The error from device(chipId:3, dieId:0), serial number is 3, hccl fftsplus task timeout occurred during task execution, stream_id:4, sq_id:4, task_id:13, stuck notify num:1, timeout:1836.[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1645]
The 0 stuck notify wait context info:(context_id=4, notify_id=17).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[3517536076], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]

RuntimeError: npuSynchronizeDevice:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:467 NPU function error: AclrtSynchronizeDeviceWithTimeout, error code is 507048
[ERROR] 2025-03-10-04:19:37 (PID:1185973, Device:0, RankID:0) ERR00100 PTA call acl api failed
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EI0002: [PID: 1185973] 2025-03-10-04:19:37.591.361 The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[3366440588], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
Possible Cause: 1. An exception occurs during the execution on some NPUs in the cluster. As a result, collective communication operation failed.2. The execution speed on some NPU in the cluster is too slow to complete a communication operation within the timeout interval. (default 1800s, You can set the interval by using HCCL_EXEC_TIMEOUT.)3. The number of training samples of each NPU is inconsistent.4. Packet loss or other connectivity problems occur on the communication link.
Solution: 1. If this error is reported on part of these ranks, check other ranks to see whether other errors have been reported earlier.2. If this error is reported for all ranks, check whether the error reporting time is consistent (the maximum difference must not exceed 1800s). If not, locate the cause or adjust the locate the cause or set the HCCL_EXEC_TIMEOUT environment variable to a larger value.3. Check whether the completion queue element (CQE) of the error exists in the plog(grep -rn 'error cqe'). If so, check the network connection status. (For details, see the TLS command and HCCN connectivity check examples.)4. Ensure that the number of training samples of each NPU is consistent. For details:https://www.hiascend.com/document
TraceBack (most recent call last):
The error from device(chipId:0, dieId:0), serial number is 3, hccl fftsplus task timeout occurred during task execution, stream_id:4, sq_id:4, task_id:13, stuck notify num:4, timeout:1836.[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1645]
The 0 stuck notify wait context info:(context_id=9, notify_id=14).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The 1 stuck notify wait context info:(context_id=11, notify_id=24).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The 2 stuck notify wait context info:(context_id=13, notify_id=20).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The 3 stuck notify wait context info:(context_id=15, notify_id=18).[FUNC:ProcessStarsHcclFftsPlusTimeoutErrorInfo][FILE:device_error_proc.cc][LINE:1652]
The wait execution of the Notify register times out. Reason: The Notify register has not received the Notify record from remote rank [unknown].base information: [streamID:[3366440588], taskID[13], tag[AllReduce_100.124.104.201%eth0_60000_0_1741569404003403], AlgType(level 0-1-2):[fullmesh-ring-ring].] task information: []
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]

[W compiler_depend.ts:487] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:19:38.591.420 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeUsedDevices)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:19:58.399.507 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:19:58.943.347 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:19:59.775.426 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:00.287.260 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:00.803.311 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:01.311.298 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:01.823.326 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:02.335.348 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:02.851.347 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:03.359.328 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:03.875.310 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:04.383.291 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:04.895.384 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:05.407.455 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[W compiler_depend.ts:122] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:05.923.406 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function empty_cache)
[W compiler_depend.ts:469] Warning: NPU warning, error code is 507048[Error]:
[Error]: The execution of the internal task times out.
Rectify the fault based on the error information in the ascend log.
EH9999: Inner Error!
rtDeviceSynchronizeWithTimeout execute failed, reason=[fftsplus timeout][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
EH9999: [PID: 1185973] 2025-03-10-04:20:06.467.417 wait for compute device to finish failed, runtime result = 507048.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
TraceBack (most recent call last):
(function npuSynchronizeDevice)
[2025-03-10 04:20:20,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1185977 closing signal SIGTERM
[2025-03-10 04:20:20,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1185978 closing signal SIGTERM
[2025-03-10 04:20:20,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1185979 closing signal SIGTERM
[2025-03-10 04:20:20,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1185980 closing signal SIGTERM
[2025-03-10 04:20:27,429] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1185973) of binary: /home/jovyan/.conda/envs/model_dis/bin/python
Traceback (most recent call last):
File "/home/jovyan/.conda/envs/model_dis/bin/torchrun", line 8, in
sys.exit(main())
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 346, in wrapper
return f(*args, **kwargs)
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
run(args)
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
elastic_launch(
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in call
return launch_agent(self._config, self._entrypoint, list(args))
File "/home/jovyan/.conda/envs/model_dis/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:

./train_minillm.py FAILED
Failures:
[1]:
time : 2025-03-10_04:20:20
host : nb-546241047869523123-0
rank : 1 (local_rank: 1)
exitcode : 1 (pid: 1185974)
error_file: <N/A>
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
[2]:
time : 2025-03-10_04:20:20
host : nb-546241047869523123-0
rank : 2 (local_rank: 2)
exitcode : 1 (pid: 1185975)
error_file: <N/A>
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
[3]:
time : 2025-03-10_04:20:20
host : nb-546241047869523123-0
rank : 3 (local_rank: 3)
exitcode : 1 (pid: 1185976)
error_file: <N/A>
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
Root Cause (first observed failure):
[0]:
time : 2025-03-10_04:20:20
host : nb-546241047869523123-0
rank : 0 (local_rank: 0)
exitcode : 1 (pid: 1185973)
error_file: <N/A>
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html

五、日志信息:
日志太大，点击链接查看
日志文件链接