System Updates
Scheduled Update for December 2024
- The home area of ABCI 3.0 will be mounted on the ABCI 2.0 interactive nodes. Details, including the schedule and the mount point, will be provided at a later date.
Scheduled Update for November 2024
With the introduction of the next ABCI system, ABCI 2.0 operations will be terminated on October 31, 2024. In preparation for this, several services will be phased out.
During November, the following services will remain available:
- Users can continue to log in to the ABCI 2.0 interactive nodes.
- Storage associated with ABCI 2.0 including the group area(
/groups
), Fast data area(/bb
), and ABCI Cloud Storage will remain accessible from the ABCI 2.0 interactive nodes. However, storage services will be phased out gradually. For specific end dates, please refer to "Notification of upcoming storage services".
The following services will be discontinued:
- ABCI 2.0 Compute Nodes (V and A)
- ABCI 2.0 Local Storage and Global scratch area(
/scratch
)
Data Migration from ABCI 2.0 to ABCI 3.0
- The system administrators will handle the migration of data from the ABCI 2.0 home area(
/home
).- During the data migration from the ABCI 2.0 home area, logging in to the ABCI 2.0 interactive nodes will not be possible. Details regarding the migration schedule will be provided at a later date.
- The ABCI 3.0 group area will be mounted on the ABCI 2.0 interactive nodes at the
/groups-new
mount point.- Users are requested to transfer any necessary data from ABCI 2.0 storage(such as
/groups
and/bb
) to the ABCI 3.0 group area (/groups-new
) via the interactive nodes.
- Users are requested to transfer any necessary data from ABCI 2.0 storage(such as
Restrictions on the ABCI User Portal Starting in November
Starting in November, certain application services will no longer be available in the ABCI User Portal.
The following applications will still be accepted. However, while these changes will be reflected on the ABCI 2.0 system, they will not be reflected on ABCI 3.0.
- Cloud Storage Usage
- User Password Reissue
- Usage History for ABCI Point
- Point usage history for ABCI 2.0 up to October 31, 2024 is available until March 30, 2025 at 17:00.
- ABCI 3.0 point usage history not visible.
The following applications will be accepted. These changes will be reflected on the ABCI 2.0 system, but a separate application will be required to reflect them on ABCI 3.0. Application procedures for ABCI 3.0 will be provided at a later date.
- Edit User
- Only email address changes will be accepted.
- Delete User from ABCI Group
- Register Public Key and Delete Public Key
- Edit User
- Set Administrator Authority
- Change Responsible Person
- Change a bill address or budget code
- Disk Addition or Reduction Request
The following applications will not be accepted. The application buttons for these services will be hidden in the User Portal, and any applications submitted will not be approved.
- New Application
- Node Reservation / Cancellation
- Additional History for ABCI Point
- ABCI Point Addition Request
- Add User to ABCI Group
- Change theme and overview
2024-10-02
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.6.1 | |
Add | cudnn | 9.3.0 9.4.0 |
2024-08-30
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | openjdk | 1.8.0.422 | 1.8.0.402 |
Update | openjdk | 11.0.24.0.8 | 11.0.22.0.7 |
Update | openjdk | 17.0.12.0.7 | 17.0.10.0.7 |
2024-08-08
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | nccl | 2.22.3-1 |
2024-07-31
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.5.1 | |
Add | cudnn | 9.2.1 |
2024-06-28
- The specific group area (/projects) is no longer available.
- The service for the global scratch area (/scratch) was discontinued on June 28, 2024, and will resume with new storage space on July 12. However, the disk quota for each customer will be reduced from 10TiB to 5TiB. In addition, the automatic deletion feature for usage over 80% will still apply. Previously, files that were 40 days past their last access or creation date would be considered for deletion, but this has been reduced to 7 days.
2024-06-18
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.4.1 12.5.0 |
|
Add | cudnn | 9.1.1 | |
Add | nccl | 2.21.5-1 |
2024-05-28
- In Spot and On-demand Service on Compute Node (V), the upper limits for the number of running jobs at the same time per system have been changed as follows.
Resource type name | Limitations | Changed upper limit |
---|---|---|
rt_F | Maximum number of running jobs at the same time per system | 918 |
rt_G.large, rt_C.large | Maximum number of running jobs at the same time per system | 170 |
rt_G.small, rt_C.small | Maximum number of running jobs at the same time per system | 680 |
2024-04-12
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Delete | cuda | 10.2.89 11.0.3 11.1.1 11.3.1 11.4.4 11.5.2 12.0.0 12.1.0 12.3.0 |
|
Delete | cudnn | 7.6.5 8.0.5 8.2.4 8.5.0 8.9.1 8.9.2 8.9.5 |
|
Delete | nccl | 2.5.6-1 2.6.4-1 2.7.8-1 2.9.9-1 2.10.3-1 2.18.1-1 2.18.3-1 |
|
Update | gdrcopy | 2.4.1 | 2.3 |
Update | intel | 2024.0.2 | 2023.0.0 |
Update | intel-advisor | 2024.0.0 | 2023.0 |
Update | intel-inspector | 2024.0 | 2023.0 |
Update | intel-itac | 2022.0 | 2021.8.0 |
Update | intel-mkl | 2024.0.0 | 2023.0.0 |
Update | intel-vtune | 2024.0.0 | 2023.0.0 |
Update | intel-mpi | 2021.11 | 2021.8 |
Add | gcc | 13.2.0 | |
Delete | gcc | 12.2.0 | |
Update | cmake | 3.29.0 | 3.26.1 |
Add | python | 3.12.2 | |
Update | python | 3.10.14 | 3.10.10 |
Update | python | 3.11.9 | 3.11.2 |
Update | R | 4.3.3 | 4.2.3 |
Update | openjdk | 1.8.0.402 | 1.8.0.362 |
Update | openjdk | 11.0.22.0.7 | 11.0.18.0.10 |
Update | openjdk | 17.0.10.0.7 | 17.0.6.0.10 |
Update | Go | 1.22.2 | 1.20 |
Update | julia | 1.10 | 1.8 |
Update | aws-cli | 2.15.35 | 2.11 |
Update | s3fs-fuse | 1.94 | 1.91 |
Update | DDN Lustre | 2.14.0_ddn134 | 2.12.8_ddn23 |
Update | SingularityPRO | 4.1.2-2 | 3.11-5 |
Update | ABCI Singularity Endpoint | 2.3.0 | 2.1.5 |
2024-03-27
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.4.0 | |
Add | cudnn | 9.0.0 | |
Add | nccl | 2.20.5-1 |
- We have confirmed that when cuDNN 9.0.0 is used with CUDA 11.0 to CUDA 11.3, an error occurs when calling the
cudnnRNNBackwardWeights_v8
function.
2024-01-23
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.3.2 | |
Add | cudnn | 8.9.7 | |
Add | nccl | 2.19.3-1 |
2024-01-09
- Due to the termination of the "Compute Node (V) Usage Promotion Program", the Resource Type Charging Factor for each resource type in the Computation Node (V) that has been changed was restored from 12:00 on January 9, 2024.
Resource type | Resource type name | Previous resource type charge coefficient | Changed resource type charge coefficient |
---|---|---|---|
Full | rt_F | 0.50 | 1.00 |
G.large | rt_G.large | 0.50 | 0.90 |
G.small | rt_G.small | 0.20 | 0.30 |
C.large | rt_C.large | 0.30 | 0.60 |
C.small | rt_C.small | 0.10 | 0.20 |
2023-12-12
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | rclone | 1.61.1 | |
Update | SingularityPRO | 3.11-5 | 3.9-10 |
- Add
/lib64/libibverbs/libmlx5-rdmav34.so
to libraries that are bind-mount from the runtime host side of the Singularity container.
2023-11-14
- Added new option
-v ALLOW_GROUP_SSH=1
to allow SSH logins to the compute nodes for other ABCI accounts that belong to the ABCI group.
For more information about the new option-v ALLOW_GROUP_SSH=1
, please refer to the SSH Access to Compute Nodes. - Added new option
-v ALLOW_GROUP_QDEL=1
to qsub that allows job deletion for other ABCI accounts that belong to the ABCI group.
For more information about the new qsub option-v ALLOW_GROUP_SSH=1
and the qdel option-g group
, please refer to the Job Execution Options and Delete a batch job.
2023-11-07
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.3.0 | |
Add | cudnn | 8.9.5 | |
Add | nccl | 2.18.5-1 |
2023-10-27
- The memory-intensive node service ended at 15:00 on October 27, 2023.
The target resource type names are rt_M.large and rt_M.small.
2023-10-02
- Due to the "Compute Node (V) Usage Promotion Program", the following changes were made to the resource type charge coefficients for resource type in the Computation Node (V). The change period is scheduled from 12:00 on October 2, 2023 to January 9, 2024.
Resource type | Resource type name | Previous resource type charge coefficient | Changed resource type charge coefficient |
---|---|---|---|
Full | rt_F | 1.00 | 0.50 |
G.large | rt_G.large | 0.90 | 0.50 |
G.small | rt_G.small | 0.30 | 0.20 |
C.large | rt_C.large | 0.60 | 0.30 |
C.small | rt_C.small | 0.20 | 0.10 |
- Resource type charge coefficients other than Compute Node (V) are the same as before.
- The charge coefficients for reservation of Compute Node (V) by Reserved Service is also 0.50.
For the calculation formula, see "Accounting"
2023-09-21
- The upper limits of Spot Service on Compute Node (A) and Memory-intensive Node have been changed as follows.
Service | Resource type name | Limitations | Previous upper limit | Changed upper limit |
---|---|---|---|---|
Spot | rt_AF | Number of nodes available at the same time | 90 nodes | 64 nodes |
Spot | rt_AF | Limit of elapsed time | 168 hours | 72 hours |
Spot | rt_AF | Limit of node-time product | 15120 nodes · hours | 2304 nodes · hours |
Spot | rt_M.large, rt_M.small | Limit of node-time product | 2304 nodes · hours | 72 nodes · hours |
- The new limits will be effective for jobs submitted after the limit change on Thursday, September 21.
If you want to use the previous limit, please submit your job before the limit change.
2023-08-01
- Due to the "1st ABCI Large-scale Language Model Building Support Program", Change the following limits for Reserved services on compute node (A) until the end of September 2023.
Service | Resource type name | Limitations | Previous upper limit | Changed upper limit |
---|---|---|---|---|
Reserved | rt_AF | Maximum number of nodes can be reserved at once per system | 50 nodes | 30 nodes |
2023-07-10
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.2.0 | |
Add | nccl | 2.18.3-1 |
2023-06-12
- ABCI User Portal Updates
- The "Position" field is required in the "User input" form.
- Change process binding for HPC-X module
- Changed the default value of the parameter
hwloc_base_binding_policy
that controls process binding in the HPC-X module fromcore
that binds processes to CPU cores tonone
that does not bind processes.- Default value before change:
core
- Default value after change:
none
- Default value before change:
- Changed the default value of the parameter
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cudnn | 8.9.2 |
2023-06-02
- In the Spot service, the limits of elapsed time and node-time product have been changed as follows.
Resource type name | Limitations | Previous upper limit | Changed upper limit |
---|---|---|---|
rt_F, rt_G.large | Limit of elapsed time | 72 hours | 168 hours |
rt_G.small, rt_AG.small, rt_C.small | Limit of elapsed time | 168 hours | 72 hours |
rt_F | Limit of node-time product | 2304 nodes · hours | 43008 nodes · hours |
- In the Reserved service, the limits of the number of reserved nodes and reserved node-time have been changed as follows.
Resource type name | Limitations | Previous upper limit | Changed upper limit |
---|---|---|---|
rt_F | Maximum reserved nodes per reservation | 34 nodes | 272 nodes |
rt_F | Maximum reserved node time per reservation | 13056 nodes · hours | 45696 nodes · hours |
rt_F | Maximum number of nodes can be reserved at once per system | 442 nodes | 476 nodes |
2023-05-18
- Change the following limits for Spot and Reserved services on compute node (A) until the end of August 2023.
However, depending on power and congestion conditions, it is possible to restore the settings before the end of August 2023.
Service | Resource type name | Limitations | Previous upper limit | Changed upper limit |
---|---|---|---|---|
Spot | rt_AF | Number of nodes available at the same time | 64 nodes | 90 nodes |
Spot | rt_AF | Limit of elapsed time | 72 hours | 168 hours |
Spot | rt_AF | Limit of node-time product | 288 nodes · hours | 15120 nodes · hours |
Reserved | rt_AF | Maximum reserved nodes per reservation | 18 nodes | 30 nodes |
2023-05-16
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.1.1 | |
Add | cudnn | 8.9.1 | |
Add | nccl | 2.18.1-1 |
2023-04-07
-
Change the OS for compute nodes (V) and interactive nodes (V) from CentOS 7 to Rocky Linux 8.
- This change requires you to recompile your programs or rebuild the Python virtual environments.
-
The following tools are no longer supported on 2023/03/31. For modules that are no longer supported, please use container images or previous ABCI Environment Modules. For more information, please refer to the Modules removed and alternatives.
- Compilers:PGI
- Development Tools:Lua
- Deep Learning Frameworks:Caffe, Caffe2, Theano, Chainer
- MPI:OpenMPI
- Utilities:fuse-sshfs
- Container Engine:Docker
-
The maximum number of nodes that can be reserved at the same time for each ABCI Group was set.
- The maximum number of the Compute Node (V) that can be reserved at the same time for each ABCI Group: 272 nodes
- The maximum number of the Compute Node (A) that can be reserved at the same time for each ABCI Group: 30 nodes
-
The inode quota limit for groups area was set.
- The inode quota limit for groups area was set to 200 millions on April 2023.
- For more information about checking the number of inodes, please refer to the Checking Disk Quota.
-
Updates the ABCI Singularity Endpoint.
- With this update, you will need to recreate the access token.
- With this update, the SingularityPRO Enterprise Plugin is available. As a result, the following overlapping functions have been removed.
- list_singularity_images
- revoke_singularity_token
-
ABCI User Portal Updates
- The following functions have been added to the Declaration regarding the applicability of specific categories.
- The "Declaration Concerning Applicability to Specified Categories" for "Japanese Students, etc." can be applied for from the ABCI User Portal.
- All users other than "Japanese Students, etc." and "Non-residents" can apply for the "Declaration Concerning Applicability to Specified Categories" from the ABCI User Portal. (Note: Users who have not applied for the "Declaration Concerning Applicability to Specified Categories" cannot use the ABCI.)
- The following functions have been added for public key operations.
- The ABCI group's responsible person/administrator can refer to the public key operation history of the ABCI group's users.
- When a user in the ABCI group registers or deletes a public key, a notification e-mail will be sent to the responsible person/administrator of the ABCI group. By default, no notification is sent.
- The following functions have been added to the Declaration regarding the applicability of specific categories.
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | gcc | 12.2.0 | |
Delete | gcc | 9.3.0 11.2.0 |
|
Add | python | 3.11.2 | |
Update | python | 3.10.10 | 3.10.4 |
Delete | python | 2.7.18 3.7.13 3.8.13 |
|
Delete | cuda | 9.0.176.1 9.1.85.3 9.2.148.1 10.0.130.1 10.1.243 11.7.0 |
|
Delete | cudnn | 7.0.5 7.1.4 7.2.1 7.3.1 7.4.2 7.5.1 |
|
Delete | nccl | 2.4.8-1 | |
Add | hpcx | 2.12 | |
Add | hpcx-debug | 2.12 | |
Add | hpcx-mt | 2.12 | |
Add | hpcx-prof | 2.12 | |
Update | intel | 2023.0.0 | 2022.2.1 |
Update | intel-advisor | 2023.0 | 2022.3.1 |
Update | intel-inspector | 2023.0 | 2022.3.1 |
Update | intel-itac | 2021.8.0 | 2021.7.1 |
Update | intel-mkl | 2023.0.0 | 2022.0.2 |
Update | intel-vtune | 2023.0.0 | 2022.4.1 |
Update | intel-mpi | 2021.8 | 2021.7 |
Delete | pgi | 20.4 | |
Update | cmake | 3.26.1 | 3.22.3 |
Update | go | 1.20 | 1.18 |
Update | julia | 1.8 | 1.6 |
Update | openjdk | 1.8.0.362 | 1.8.0.332 |
Update | openjdk | 11.0.18.0.10 | 11.0.15.0.9 11.0.15.0.10 |
Update | openjdk | 17.0.6.0.10 | 17.0.3.0.7 |
Update | R | 4.2.3 | 4.1.3 |
Delete | openmpi | 4.0.5 | |
Delete | openmpi | 4.1.3 | |
Update | aws-cli | 2.11 | 2.4 |
Delete | fuse-sshfs | 3.7.2 | |
Update | SingularityPRO | 3.9-10 | 3.9-9 |
Update | ABCI Singularity Endpoint | 2.1.5 | 1.7.2 |
Update | DDN Lustre | 2.12.8_ddn23 | 2.12.8_ddn10 |
Update | Scality S3 Connector | 7.10.6.7 | 7.10.2.2 |
Update | BeeOND | 7.3.3 | 7.2.3 |
2023-03-08
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.1.0 | |
Add | cudnn | 8.8.1 | |
Add | nccl | 2.17.1-1 |
2023-02-03
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | intel | 2022.2.1 | 2022.0.2 |
Update | intel-advisor | 2022.3.1 | 2022.0 |
Update | intel-inspector | 2022.3.1 | 2022.0 |
Update | intel-itac | 2021.7.1 | 2021.5.0 |
Update | intel-mkl | 2022.0.2 | 2022.0.0 |
Update | intel-vtune | 2022.4.1 | 2022.0.0 |
Update | intel-mpi | 2021.7 | 2021.5 |
- Programs compiled with previous version of the Intel oneAPI may contain vulnerabilities, so please recompile with the newer version.
intel/2022.0.2
and earlier Intel oneAPI modules containing vulnerabilities have been deprecated. Programs compiled with previous version of the Intel oneAPI modules, which was deprecated on Feb 6, may no longer run, so please recompile with the newer version.
2023-01-05
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | SingularityPRO | 3.9-9 | 3.9-8 |
2022-12-23
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 12.0.0 | |
Add | cudnn | 8.7.0 | |
Add | nccl | 2.16.2-1 |
2022-12-13
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | SingularityPRO | 3.9-8 | 3.9-4 |
Update | ABCI Singularity Endpoint | 1.7.2 | 1.2.5 |
2022-10-25
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 11.8.0 | |
Add | cudnn | 8.6.0 | |
Add | nccl | 2.15.5-1 |
2022-09-02
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 11.7.1 | |
Add | cudnn | 8.5.0 | |
Add | nccl | 2.13.4-1 2.14.3-1 |
2022-07-29
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | cudnn | 8.4.1 | 8.4.0 |
2022-06-24
- Changed the job execution option for change GPU Compute Mode to EXCLUSIVE_PROCESS mode from
-v GPU_COMPUTE_MODE=1
to-v GPU_COMPUTE_MODE=3
. For more information, please refer to the Changing GPU Compute Mode.
2022-06-21
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuda | 11.7.0 | |
Update | nccl | 2.12.12-1 | 2.12.10-1 |
Update | Altair Grid Engine | 8.6.19_C121_1 | 8.6.17 |
Update | openjdk | 1.8.0.332 | 1.8.0.322 |
Update | openjdk | 11.0.15.0.9(Compute Node (V)) 11.0.15.0.10(Compute Node (A)) |
11.0.14.1.1 |
Update | openjdk | 17.0.3.0.7 | 17.0.2.0.8 |
Update | DDN Lustre | 2.12.8_ddn10 | 2.12.6_ddn58-1 |
- Altair Grid Engine has been updated. The job queue and job reservations are not preserved. Please resubmit your batch job(s). Please recreate your reservation(s).
- Some of Known Issues have been resolved in this update.
- Reinstalled R (4.1.3) with --enable-R-shlib enabled.
- The update of ABCI Singularity Endpoint has been postponed.
2022-05-26
- Product names documented in this User Guide have been renamed to reflect the acquisition of Univa by Altair.
Current | Previous |
---|---|
Altair Grid Engine | Univa Grid Engine |
AGE | UGE |
2022-05-10
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | gcc | 9.3.0 | |
Add | cudnn | 8.4.0 | |
Update | nccl | 2.12.10-1 | 2.12.7-1 |
- Deleted
gcc/9.3.0
module has been restored to the current environment modules.
2022-04-06
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | Scality S3 Connector | 8.5.2.2 | 7.4.9.3 |
Update | SingularityPRO | 3.9-4 | 3.7-4 |
Update | DDN Lustre (Compute node (V)) | 2.12.6_ddn58-1 | 2.12.5_ddn13-1 |
Update | OFED (Compute node (V)) | 5.2-1.0.4.0 | 5.0-2.1.8.0 |
Update | gcc | 11.2.0 | 9.3.0 |
Delete | gcc | 7.4.0 | |
Update | intel | 2022.0.2 | 2020.4.304 |
Delete | nvhpc | 20.11 21.2 |
|
Delete | openjdk | 1.7.0.171 | |
Update | openjdk | 1.8.0.322 | 1.8.0.242 |
Update | openjdk | 11.0.14.1.1 | 11.0.6.10 |
Update | openjdk | 17.0.2.0.8 | 15.0.2.0.7 |
Delete | lua | 5.3.6 5.4.2 |
|
Delete | julia | 1.0 | |
Update | julia | 1.6.6 | 1.5 |
Update | intel-advisor | 2022.0 | 2020.3 |
Update | intel-inspector | 2022.0 | 2020.3 |
Update | intel-itac | 2021.5.0 | 2020.0.3 |
Update | intel-mkl | 2022.0.0 | 2020.0.4 |
Update | intel-vtune | 2022.0.0 | 2020.3 |
Add | python | 3.10.4 | |
Update | python | 3.7.13 | 3.7.10 |
Update | python | 3.8.13 | 3.8.7 |
Delete | python | 3.6.12 | |
Update | R | 4.1.3 | 4.0.4 |
Delete | cuda | 8.0.61.2 9.2.88.1 11.4.1 11.6.0 |
|
Update | cuda | 11.4.4 | 11.4.1 |
Update | cuda | 11.5.2 | 11.5.1 |
Update | cuda | 11.6.2 | 11.6.0 |
Delete | cudnn | 5.1.10 6.0.21 8.2.0 8.2.1 8.2.2 |
|
Update | cudnn | 8.3.3 | 8.3.2 |
Delete | nccl | 1.3.5-1 2.1.15-1 2.2.13-1 2.3.7-1 2.9.6-1 |
|
Add | nccl | 2.12.7-1 | |
Update | gdrcopy | 2.3 | 2.0 |
Update | intel-mpi | 2021.5 | 2019.9 |
Add | openmpi | 4.1.3 | |
Delete | openmpi | 2.1.6 | |
Delete | openmpi | 3.1.6 | |
Update | aws-cli | 2.4 | 2.1 |
Update | fuse-sshfs | 3.7.2 | 3.7.1 |
Update | f3fs-fuse | 1.91 | 1.87 |
Delete | sregistory-cli | 0.2.36 | |
Update | NVIDIA Tesla Driver | 510.47.03 | 470.57.02 |
- Maximum reserved node time per reservation of compute node (V) is changed in the Reserved Service from 12,288 to 13,056.
- Maximum reserved nodes per reservation of compute node (A) is changed in the Reserved Service from 16 to 18.
- Maximum reserved node time per reservation of compute node (A) is changed in the Reserved Service from 6,144 to 6,912.
- The installation of Singularity Enterprise CLI has been postponed.
- One of known issues has been resolved in this update.
- We have reconfigured the Environment Modules. If you would like to use modules prior to FY2021, please refer to the FAQ (How to use previous ABCI Environment Modules).
- Due to the reconfiguration of the Environment Modules, some modules have been removed. For more information, please refer to the Modules removed and alternatives.
2022-03-03
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Delete | hadoop | 3.3 | |
Delete | spark | 3.0 | |
Update | DDN Lustre (Compute Node (A)) | 2.12.6_ddn58-1 | 2.12.5_ddn13-1 |
Update | OFED (Compute Node (A)) | 5.2-1.0.4.0 | 5.1-0.6.6.0 |
- One of known issues has been resolved in this update.
2022-01-27
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | CUDA | 11.3.1 11.4.1 11.4.2 11.5.1 11.6.0 |
|
Add | cuDNN | 8.2.2 8.2.4 8.3.2 |
|
Add | NCCL | 2.10.3-1 2.11.4-1 |
2021-12-15
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | OFED | 5.1-0.6.6.0 | 5.0-2.1.8.0 |
Update | Scality S3 Connector | 7.4.9.3 | 7.4.8.4 |
Update | NVIDIA Tesla Driver | 470.57.02 | 460.32.03 |
Add | ffmpeg | 3.4.9 4.2.5 |
- Maximum reserved nodes per reservation of compute node (V) has been changed in the Reserved Service from 32 to 34.
- With the addition of the Global Scratch Area, we added Global scratch area section.
2021-08-12
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | BeeOND | 7.2.3 | 7.2.1 |
Update | DDN Lustre | 2.12.5_ddn13-1 | 2.12.6_ddn13-1 |
2021-07-06
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | SingularityPRO | 3.7-4 | 3.7-1 |
2021-06-30
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuDNN | 8.2.1 | |
Add | NCCL | 2.9.9-1 |
2021-05-10
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuDNN | 8.2.0 | |
Add | NCCL | 2.9.6-1 |
- The documentation has been revised with the addition of a compute node (A) with NVIDIA A100.
2021-04-07
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | NVIDIA Tesla Driver | 460.32.03 | 440.33.01 |
Update | OFED | 5.0-2.1.8.0 | 4.4-1.0.0.0 |
Update | Univa Grid Engine | 8.6.17 | 8.6.6 |
Update | SingularityPRO | 3.7-1 | 3.5-6 |
Update | BeeOND | 7.2.1 | 7.2 |
Update | Docker | 19.03.15 | 17.12.0 |
Update | Scality S3 Connector | 7.4.8.1 | 7.4.8 |
Add | gcc | 9.3.0 | |
Add | pgi | 20.4 | |
Add | nvhpc | 20.11 21.2 |
|
Add | cmake | 3.19 | |
Add | go | 1.15 | |
Add | julia | 1.5 | |
Add | lua | 5.3.6 5.4.2 |
|
Add | python | 2.7.18 3.6.12 3.7.10 3.8.7 |
|
Add | R | 4.0.4 | |
Add | CUDA | 11.0.3 11.1.1 11.2.2 |
|
Add | cuDNN | 8.1.1 | |
Add | NCCL | 2.8.4-1 | |
Add | openmpi | 4.0.5 | |
Add | mvapich2-gdr | 2.3.5 | |
Add | mvapich2 | 2.3.5 | |
Add | hadoop | 3.3 | |
Add | spark | 3.0 | |
Add | aws-cli | 2.1 | |
Add | fuse-sshfs | 3.7.1 | |
Add | s3fs-fuse | 1.87 | |
Add | sregistry-cli | 0.2.36 | |
Delete | intel | 2017.8.262 2018.5.274 2019.5.281 |
|
Delete | pgi | 17.10 18.10 19.1 19.10 20.1 |
|
Delete | nvhpc | 20.9 | |
Delete | cmake | 3.16 3.17 |
|
Delete | go | 1.12 1.13 |
|
Delete | intel-advisor | 2017.5 2018.4 2019.5 |
|
Delete | intel-inspector | 2017.4 2018.4 2019.5 |
|
Delete | intel-itac | 2017.0.4 2018.0.4 2019.0.5 |
|
Delete | intel-mkl | 2017.0.4 2018.0.4 2019.0.5 |
|
Delete | intel-vtune | 2017.6 2018.4 2019.6 |
|
Delete | julia | 1.3 1.4 |
|
Delete | python | 2.7.15 3.4.8 3.5.5 3.7.6 |
|
Delete | R | 3.5.0 3.6.3 |
|
Delete | cuda | 10.0.130 | |
Delete | cudnn | 7.1.3 7.5.0 7.6.0 7.6.1 7.6.2 7.6.3 7.6.4 8.0.2 |
|
Delete | nccl | 2.3.4-1 2.3.5-2 2.4.2-1 2.4.7-1 2.8.3-1 |
|
Delete | intel-mpi | 2017.4 2018.4 2019.5 |
|
Delete | openmpi | 4.0.3 | |
Delete | mvapich2-gdr | 2.3.3 2.3.4 |
|
Delete | mvapich2 | 2.3.3 2.3.4 |
|
Delete | hadoop | 2.9 2.10 3.1 |
|
Delete | singularity | 2.6.1 | |
Delete | spark | 2.3 2.4 |
|
Delete | aws-cli | 1.16.194 1.18 2.0 |
|
Delete | fuse-sshfs | 2.10 | |
Delete | s3fs-fuse | 1.85 | |
Delete | sregistry-cli | 0.2.31 |
2021-03-13
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | SingularityPRO | 3.5-6 | 3.5-4 |
Update | DDN Lustre | 2.12.6_ddn13-1 | 2.10.7_ddn14-1 |
2020-12-15
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | go | 1.14 | |
Add | intel | 2020.4.304 | |
Add | intel-advisor | 2020.3 | |
Add | intel-inspector | 2020.3 | |
Add | intel-itac | 2020.0.3 | |
Add | intel-mkl | 2020.0.4 | |
Add | intel-mpi | 2019.9 | |
Add | intel-vtune | 2020.3 | |
Add | nvhpc | 20.9 | |
Add | cuDNN | 8.0.5 | |
Add | NCCL | 2.8.3-1 | |
Update | BeeOND | 7.2 | 7.1.5 |
Update | Scality S3 Connector | 7.4.8 | 7.4.6.3 |
Additional Feature: SSH Access to Compute Nodes
We have added the feature to enable SSH login to the compute nodes. See SSH Access to Compute Nodes for details.
2020-10-09
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | SingularityPRO | 3.5-4 | 3.5-2 |
2020-08-31
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | Scality S3 Connector | 7.4.6.3 | 7.4.5.4 |
2020-07-31
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | SingularityPRO | 3.5-2 | |
Add | cuDNN | 8.0.2 | |
Add | NCCL | 2.7.8-1 | |
Add | mvapich2-gdr | 2.3.4 | |
Add | mvapich2 | 2.3.4 |
2020-06-01
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | BeeOND | 7.1.5 | 7.1.4 |
2020-04-21
Update MVAPICH2-GDR 2.3.3
MVAPICH2-GDR 2.3.3 for gcc 4.8.5 was updated to the fixed version about the following issue.
- MPI_Allreduce provided by MVAPICH2-GDR may raise floating point exceptions
On the other hand, MVAPICH2-GDR 2.3.3 for PGI was uninstalled. If you need MVAPICH2-GDR for PGI, please contact Customer Support.
2020-04-03
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | DDN GRIDScaler | 4.2.3.20 | 4.2.3.17 |
Update | Scality S3 Connector | 7.4.5.4 | 7.4.5.0 |
Update | libfabric | 1.7.0-1 | 1.5.3-1 |
Add | intel | 2018.5.274 2019.5.281 |
|
Add | pgi | 19.1 19.10 20.1 |
|
Add | R | 3.6.3 | |
Add | cmake | 3.16 3.17 |
|
Add | go | 1.12 1.13 |
|
Add | intel-advisor | 2017.5 2018.4 2019.5 |
|
Add | intel-inspector | 2017.4 2018.4 2019.5 |
|
Add | intel-itac | 2017.0.4 2018.0.4 2019.0.5 |
|
Add | intel-mkl | 2017.0.4 2018.0.4 2019.0.5 |
|
Add | intel-vtune | 2017.6 2018.4 2019.6 |
|
Add | julia | 1.0 1.3 1.4 |
|
Add | openjdk | 1.8.0.242 11.0.6.10 |
|
Add | python | 3.7.6 3.8.2 |
|
Add | gdrcopy | 2.0 | |
Add | nccl | 2.6.4-1 | |
Add | intel-mpi | 2017.4 2018.4 2019.5 |
|
Add | mvapich2-gdr | 2.3.3 | |
Add | mvapich2 | 2.3.3 | |
Add | openmpi | 3.1.6 4.0.3 |
|
Add | hadoop | 2.9 2.10 3.1 |
|
Add | spark | 2.3 2.4 |
|
Add | aws-cli | 1.18 2.0 |
|
Delete | gcc | 7.3.0 | |
Delete | intel | 2018.2.199 2018.3.222 2019.3.199 |
|
Delete | pgi | 18.5 19.3 |
|
Delete | go | 1.11.2 | |
Delete | intel-mkl | 2017.8.262 2018.2.199 2018.3.222 2019.3.199 |
|
Delete | openjdk | 1.6.0.41 1.8.0.161 |
|
Delete | cuda | 9.0/9.0.176.2 9.0/9.0.176.3 |
|
Delete | gdrcopy | 1.2 | |
Delete | intel-mpi | 2018.2.199 | |
Delete | mvapich2-gdr | 2.3rc1 2.3 2.3a 2.3.1 2.3.2 |
|
Delete | mvapich2 | 2.3rc2 2.3 2.3.2 |
|
Delete | openmpi | 1.10.7 2.1.3 2.1.5 3.0.3 3.1.0 3.1.2 3.1.3 |
|
Delete | hadoop | 2.9.1 2.9.2 |
|
Delete | spark | 2.3.1 2.3.2 2.4.0 |
2019-12-17
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | DDN Lustre | 2.10.7_ddn14-1 | 2.10.5_ddn7-1 |
Update | BeeOND | 7.1.4 | 7.1.3 |
Update | Scality S3 Connector | 7.4.5.0 | 7.4.4.4 |
Update | NVIDIA Tesla Driver | 440.33.01 | 410.104 |
Add | CUDA | 10.2.89 | |
Add | cuDNN | 7.6.5 | |
Add | NCCL | 2.5.6-1 |
Other fixes are as follows:
2019-11-06
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | GCC | 7.3.0, 7.4.0 | |
Add | sregistry-cli | 0.2.31 |
Other fixes are as follows:
- Fixed cuda/* modules to set the paths to
extras/CUPTI
. - Fixed python/3.4, python/3.5, and python/3.6 to solve the problem that error occurred when executing
shutil.copytree
on Home area.
2019-10-04
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | Univa Grid Engine | 8.6.6 | 8.6.3 |
Update | DDN GRIDScaler | 4.2.3.17 | 4.2.3.15 |
Update | BeeOND | 7.1.3 | 7.1.2 |
Add | CUDA | 10.1.243 | |
Add | cuDNN | 7.6.3 7.6.4 |
|
Add | NCCL | 2.4.8-1 | |
Add | MVAPICH2-GDR | 2.3.2 | |
Add | MVAPICH2 | 2.3.2 | |
Add | fuse-sshfs | 2.10 |
Other fixes are as follows:
- Add CUDA 10.1 support to cuDNN 7.5.0, 7.5.1, 7.6.0, 7.6.1, 7.6.2
- Add CUDA 10.1 support to NCCL 2.4.2-1, 2.4.7-1
- Add CUDA 10.0 and 10.1 support to GDRCopy 1.2
- Add CUDA 10.1 support to Open MPI 2.1.6
- Increase /tmp capacity of interactive nodes from 26GB to 12TB
- Add process monitoring and process cancellation mechanism on the interactive node
Start process monitoring on the interactive nodes
Process monitoring started on the interactive nodes.
High load or lengthy tasks on the interactive nodes will be killed by the
process monitoring system, so use the compute nodes with the qrsh/qsub
command.
Change the job submission and execution limits
We changed the job submission and execution limits as follows.
Limitations | Current limits | Previous limits |
---|---|---|
The maximum number of tasks within an array job | 75000 | 1000 |
The maximum number of any user's running jobs at the same time | 200 | 0(unlimited) |
About known issues
The status of following known issues were changed to close.
- A comupte node can execute only up to 2 jobs each resource type "rt_G.small" and "rt_C.small" (normally up to 4 jobs ).
2019-08-01
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuDNN | 7.6.2 | |
Add | NCCL | 2.4.7-1 | |
Add | s3fs-fuse | 1.85 |
Other fixes are as follows:
- Add CUDA 10.0 support to Open MPI 1.10.7, 2.1.5, 2.1.6
2019-07-10
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | CUDA | 10.0.130.1 | |
Add | cuDNN | 7.5.1 7.6.0 7.6.1 |
|
Add | aws-cli | 1.16.194 |
2019-04-05
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | CentOS | 7.5 | 7.4 |
Update | Univa Grid Engine | 8.6.3 | 8.5.4 |
Update | Java | 1.7.0_171 | 1.7.0_141 |
Update | Java | 1.8.0_161 | 1.8.0_131 |
Add | DDN Lustre | 2.10.5_ddn7-1 | |
Update | NVIDIA Tesla Driver | 410.104 | 396.44 |
Add | CUDA | 10.0.130 | |
Add | Intel Compiler | 2019.3 | |
Add | PGI | 18.10 19.3 |
Other fixes are as follows:
- Migrate Home area from GPFS to DDN Lustre
2019-03-14
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | Intel Compiler | 2017.8 2018.3 |
|
Add | PGI | 17.10 | |
Add | Open MPI | 2.1.6 | |
Add | cuDNN | 7.5.0 | |
Add | NCCL | 2.4.2-1 | |
Add | Intel MKL | 2017.8 2018.3 |
Other fixes are as follows:
- Add PGI 17.10 support to MVAPICH2-GDR 2.3
- Add PGI support to Open MPI 2.1.5, 2.1.6, 3.1.3
- Change the default version of Open MPI to 2.1.6
- Fix typo in MVAPICH2 modules, wrong top directory
2019-01-31
User/Group/Job names are now masked when displaying the result of 'qstat'
We changed the job scheduler configuration, so that User/Group/Job names are masked from the result of qstat
command. These columns are shown only for your own jobs, otherwise these columns are masked by '*'. An example follows:
[username@es1 ~]$ qstat -u '*' | head
job-ID prior name user state submit/start at queue jclass slots ja-task-ID
------------------------------------------------------------------------------------------------------------------------------------------------
123456 0.28027 run.sh username r 01/31/2019 12:34:56 gpu@g0001 80
123457 0.28027 ********** ********** r 01/31/2019 12:34:56 gpu@g0002 80
123458 0.28027 ********** ********** r 01/31/2019 12:34:56 gpu@g0003 80
123450 0.28027 ********** ********** r 01/31/2019 12:34:56 gpu@g0004 80
2018-12-18
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Add | cuDNN | 7.4.2 | |
Add | NCCL | 2.3.7-1 | |
Add | Open MPI | 3.0.3 3.1.3 |
|
Add | MVAPICH2-GDR | 2.3 | |
Add | Hadoop | 2.9.2 | |
Add | Spark | 2.3.2 2.4.0 |
|
Add | Go | 1.11.2 | |
Add | Intel MKL | 2018.2.199 |
cuDNN 7.4.2
The NVIDIA CUDA Deep Neural Network library (cuDNN) 7.4.2 was installed.
To set up user environment:
$ module load cuda/9.2/9.2.148.1
$ module load cudnn/7.4/7.4.2
NCCL 2.3.7-1
The NVIDIA Collective Communications Library (NCCL) 2.3.7-1 was installed.
To set up user environment:
$ module load cuda/9.2/9.2.148.1
$ module load nccl/2.3/2.3.7-1
Open MPI 3.0.3, 3.1.3
Open MPI (without --cuda option) 3.0.3, 3.1.3 were installed.
To set up user environment:
$ module load openmpi/3.1.3
MVAPICH2-GDR 2.3
MVAPICH2-GDR 2.3 was installed.
To set up user environment:
$ module load cuda/9.2/9.2.148.1
$ module load mvapich/mvapich2-gdr/2.3
Hadoop 2.9.2
Apache Hadoop 2.9.2 was installed.
To set up user environment:
$ module load openjdk/1.8.0.131
$ module load hadoop/2.9.1
Spark 2.3.2, 2.4.0
Apache Spark 2.3.2, 2.4.0 were installed.
To set up user environment:
$ module load spark/2.4.0
Go 1.11.2
Go Programming Language 1.11.2 was installed.
To set up user environment:
$ module load go/1.11.2
Intel MKL 2018.2.199
Intel Math Kernel Library (MKL) 2018.2.199 was installed.
To set up user environment:
$ module load intel-mkl/2018.2.199
2018-12-14
Add / Update / Delete | Software | Version | Previous version |
---|---|---|---|
Update | Singularity | 2.6.1 | 2.6.0 |
Delete | Singularity | 2.5.2 |
Singularity 2.6.1 was installed. The usage is as follows:
$ module load singularity/2.6.1
$ singularity run image_path
The release note will be found:
And, we uninstalled version 2.5.2 and 2.6.0 because severe security issues (CVE-2018-19295) were reported. If you are using Singularity with specifying version number, such as singularity/2.5.0
or singularity/2.6.0
, please modify your job scripts to specify singularity/2.6.1
.
ex) module load singularity/2.5.2 -> module load singularity/2.6.1