deepseek-ai/DeepEP
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepEP.git synced 2026-11-05 00:47:32 +00:00
Code Issues 308 Projects Packages Wiki Activity

[GH-ISSUE #157] test_low_latency.py Failed on 2 node H800 #138

New issue
Open
opened 2026-10-26 09:55:05 +00:00 by bsakaguchi · 5 comments
bsakaguchi commented 2026-10-26 09:55:05 +00:00
Owner
Copy link

Originally created by @liusy58 on GitHub (May 13, 2025).
Original GitHub issue: https://github.com/deepseek-ai/DeepEP/issues/157

Node1

MASTER_ADDR=WORLD_SIZE=2 RANK=0  python DeepEP/tests/test_low_latency.py

Node2

MASTER_ADDR=WORLD_SIZE=2 RANK=1  python DeepEP/tests/test_low_latency.py

Error message

/nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed 

/nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed 

/nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed 

/nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed 

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
/nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgress:59: socketProgress: Connection closed by remote peer e01-cn-e4v48e4dl07<46500>
/nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:97: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:499: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed 

/nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed 

/nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed 

/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
/nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgressOpt:39: socketProgressOpt: Call to recv from 172.18.0.177<43129> failed : Broken pipe
/nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:91: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:496: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed 

/nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed 

/nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed 

/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

/nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgressOpt:39: socketProgressOpt: Call to recv from 172.18.0.177<38567> failed : Broken pipe
/nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:91: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:496: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed 

/nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed 

/nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed 

/nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed 

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

/nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed 

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

/nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed 

/nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed

/nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgress:59: socketProgress: Connection closed by remote peer e01-cn-e4v48e4dl06<41542>
/nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:97: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:499: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed 

/nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed 

/nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed 

[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

[/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered 
/nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed 

/nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed

How can I fix this?

Originally created by @liusy58 on GitHub (May 13, 2025). Original GitHub issue: https://github.com/deepseek-ai/DeepEP/issues/157 ### Node1 ``` MASTER_ADDR=WORLD_SIZE=2 RANK=0 python DeepEP/tests/test_low_latency.py ``` ### Node2 ``` MASTER_ADDR=WORLD_SIZE=2 RANK=1 python DeepEP/tests/test_low_latency.py ``` ### Error message ``` /nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed /nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed /nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed /nvshmem/src/host/mem/mem_heap.cpp:1509: non-zero status: 2 cuMemCreate failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered /nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgress:59: socketProgress: Connection closed by remote peer e01-cn-e4v48e4dl07<46500> /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:97: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:499: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed /nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed /nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered /nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgressOpt:39: socketProgressOpt: Call to recv from 172.18.0.177<43129> failed : Broken pipe /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:91: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:496: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed /nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed /nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed /nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgressOpt:39: socketProgressOpt: Call to recv from 172.18.0.177<38567> failed : Broken pipe /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:91: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:496: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed /nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed /nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed /nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed /nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed /nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed /nvshmem/src/host/mem/mem_heap.cpp:1590: non-zero status: 7 allocate_physical_memory_to_heap failed /nvshmem/src/modules/bootstrap/uid/ncclSocket/ncclsocket_socket.cpp:socketProgress:59: socketProgress: Connection closed by remote peer e01-cn-e4v48e4dl06<41542> /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:97: non-zero status: -6 /nvshmem/src/modules/bootstrap/uid/bootstrap_uid.cpp:499: non-zero status: -6 /nvshmem/src/host/mem/mem_heap.cpp:940: non-zero status: -6 allgather of mem handles failed /nvshmem/src/host/mem/mem_heap.cpp:1098: non-zero status: 7 register heap memory failed /nvshmem/src/host/mem/mem_heap.cpp:1533: non-zero status: 7 register heap UC memory failed [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed [/nvshmem/src/host/coll/barrier/barrier.cpp:21] cuda failed with an illegal memory access was encountered /nvshmem/src/host/util/cs.cpp:21: non-zero status: 16: Device or resource busy, exiting... mutex destroy failed /nvshmem/src/host/mem/mem_heap.cpp:532: non-zero status: 1 cuMemAddressFree failed ``` How can I fix this?
bsakaguchi commented 2026-10-26 09:55:06 +00:00
Author
Owner
Copy link

@sphish commented on GitHub (May 14, 2025):

Could you try running the nvshmem perftest to see if it works properly?

<!-- gh-comment-id:2879484115 --> @sphish commented on GitHub (May 14, 2025): Could you try running the nvshmem perftest to see if it works properly?
bsakaguchi commented 2026-10-26 09:55:06 +00:00
Author
Owner
Copy link

@Iamleos commented on GitHub (May 30, 2025):

I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer

<!-- gh-comment-id:2921660466 --> @Iamleos commented on GitHub (May 30, 2025): I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer
bsakaguchi commented 2026-10-26 09:55:07 +00:00
Author
Owner
Copy link

@Iamleos commented on GitHub (May 30, 2025):

I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer

Image

<!-- gh-comment-id:2921668744 --> @Iamleos commented on GitHub (May 30, 2025): > I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer ![Image](https://github.com/user-attachments/assets/e1d76bd9-a078-416e-89cd-ed0548d8c5ba)
bsakaguchi commented 2026-10-26 09:55:07 +00:00
Author
Owner
Copy link

@thqq479 commented on GitHub (Sep 14, 2025):

I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer

how to set this？

<!-- gh-comment-id:3289564626 --> @thqq479 commented on GitHub (Sep 14, 2025): > I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer how to set this？
bsakaguchi commented 2026-10-26 09:55:07 +00:00
Author
Owner
Copy link

@alpha-baby commented on GitHub (Sep 15, 2025):

I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer我遇到了同样的问题。此问题通常是由 GPU 内存不足引起的。尝试为 DeepEP 缓冲区保留更多 GPU 内存

how to set this？  如何设置这个

Log in to the physical machine and make sure that the memory on your GPU is not used by other processes.

<!-- gh-comment-id:3290319668 --> @alpha-baby commented on GitHub (Sep 15, 2025): > > I met the same problem. This issue is usually caused by insufficient GPU memory. Try to reserve more GPU memory for DeepEP buffer我遇到了同样的问题。此问题通常是由 GPU 内存不足引起的。尝试为 DeepEP 缓冲区保留更多 GPU 内存 > > how to set this？  如何设置这个 Log in to the physical machine and make sure that the memory on your GPU is not used by other processes.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
hybrid-ep
chore/remove-clang-format-yapf
v2.5
nvDev
try-fix-cu128
mori-ep
antgroup-opt
compat-old-torch
add-linter
eager-rdma
v1.2.1
Labels
Clear labels   
pull-request
Mirrored from GitHub Pull Request

No labels
pull-request
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/DeepEP#138
No description provided.