Nexus/GPUs - Revision history

Mbaney at 19:21, 17 April 2026

2026-04-17T19:21:30Z

← Older revision		Revision as of 19:21, 17 April 2026
Line 8:		Line 8:
	! [https://www.nvidia.com/en-us/technologies Architecture]		! [https://www.nvidia.com/en-us/technologies Architecture]
	! [https://developer.nvidia.com/cuda-toolkit CUDA] Cores		! [https://developer.nvidia.com/cuda-toolkit CUDA] Cores
	! ~~GPU~~ Memory		! Memory Amount and Type
	! Memory Bandwidth		! Memory Bandwidth
	! FP32 Performance (TFLOPS)		! FP32 Performance (TFLOPS)

Mbaney at 16:52, 17 October 2025

2025-10-17T16:52:09Z

← Older revision		Revision as of 16:52, 17 October 2025
Line 186:		Line 186:
	\|}		\|}

	[0] - This GPU is actually a pair of two physical cards connected over [https://www.nvidia.com/en-us/data-center/nvlink NVLink] bridges. NVIDIA's provided specifications for this GPU type are for one physical card; to get these specs, we have hence doubled NVIDIA's provided values.		[0] - This GPU type is actually a pair of two physical cards connected over [https://www.nvidia.com/en-us/data-center/nvlink NVLink] bridges. NVIDIA's provided specifications for this GPU type are for one physical card; to get these specs, we have hence doubled NVIDIA's provided values.

Mbaney at 19:58, 9 October 2025

2025-10-09T19:58:36Z

← Older revision		Revision as of 19:58, 9 October 2025
Line 1:		Line 1:
	There are several different types of [https://www.nvidia.com/en-us/ NVIDIA] GPUs in the [[Nexus]] cluster that are available to be scheduled. They are listed below in order of newest to oldest architecture, and then ~~alphabetically~~ by name.		There are several different types of [https://www.nvidia.com/en-us/ NVIDIA] GPUs in the [[Nexus]] cluster that are available to be scheduled. They are listed below in order of newest to oldest architecture, and then alphanumerically by name.

	The exact quantities of GPUs per type are not listed here since these numbers may change frequently due to additions to or removals from the cluster or during compute node troubleshooting. To see which compute nodes have which GPUs and in what quantities, use the <code>show_nodes</code> command on a submission or compute node. The quantities are listed under the <tt>GRES</tt> column.		The exact quantities of GPUs per type are not listed here since these numbers may change frequently due to additions to or removals from the cluster or during compute node troubleshooting. To see which compute nodes have which GPUs and in what quantities, use the <code>show_nodes</code> command on a submission or compute node. The quantities are listed under the <tt>GRES</tt> column.
Line 46:		Line 46:
	\| 80GB HBM3		\| 80GB HBM3
	\| 3.35TB/s		\| 3.35TB/s
			\| 66.9
			\| not officially published/989
			\|-
			\| H200 SXM
			\| <code>h200-sxm</code>
			\| Hopper
			\| 16896
			\| 141GB HBM3e
			\| 4.89TB/s
	\| 66.9		\| 66.9
	\| not officially published/989		\| not officially published/989

Mbaney at 19:43, 8 October 2025

2025-10-08T19:43:16Z

@@ Line 12: / Line 12: @@
 ! FP32 Performance (TFLOPS)
 ! [https://developer.nvidia.com/blog/accelerating-ai-training-with-tf32-tensor-cores/ TF32] Performance ([https://developer.nvidia.com/blog/accelerating-inference-with-sparsity-using-ampere-and-tensorrt Dense / Sparse TOPS])
 |-
 | H100 NVLink [0]
@@ Line 30: / Line 48: @@
 | 66.9
 | not officially published/989
 |-
 | A100 PCIe 80GB

Mbaney at 19:41, 8 October 2025

2025-10-08T19:41:54Z

← Older revision		Revision as of 19:41, 8 October 2025
Line 19:		Line 19:
	\| 188GB HBM3		\| 188GB HBM3
	\| 7.87TB/s		\| 7.87TB/s
	\| ~~134~~		\| 133.8
	\| not officially published/1671		\| not officially published/1671
	\|-		\|-
Line 28:		Line 28:
	\| 80GB HBM3		\| 80GB HBM3
	\| 3.35TB/s		\| 3.35TB/s
	\| 67		\| 66.9
	\| not officially published/989		\| not officially published/989
	\|-		\|-

Mbaney at 14:34, 29 August 2025

2025-08-29T14:34:23Z

← Older revision		Revision as of 14:34, 29 August 2025
Line 177:		Line 177:
	\|}		\|}

	[0] - This GPU is actually a pair of two physical cards connected over [https://www.nvidia.com/en-us/data-center/nvlink NVLink] bridges. NVIDIA's provided specifications for this GPU type are for one physical card; to get these specs, we have hence doubled ~~their advertised~~ values.		[0] - This GPU is actually a pair of two physical cards connected over [https://www.nvidia.com/en-us/data-center/nvlink NVLink] bridges. NVIDIA's provided specifications for this GPU type are for one physical card; to get these specs, we have hence doubled NVIDIA's provided values.

Mbaney at 18:20, 26 November 2024

2024-11-26T18:20:38Z

← Older revision		Revision as of 18:20, 26 November 2024
Line 1:		Line 1:
	There are several different types of [https://www.nvidia.com/en-us/ NVIDIA] GPUs in the [[Nexus]] cluster that are available to be scheduled. They are listed below in order of newest to oldest architecture, and then alphabetically by name.		There are several different types of [https://www.nvidia.com/en-us/ NVIDIA] GPUs in the [[Nexus]] cluster that are available to be scheduled. They are listed below in order of newest to oldest architecture, and then alphabetically by name.

	~~We do not list the~~ exact quantities of ~~GPU~~ here since ~~they~~ change frequently due to additions to or removals from the cluster or during compute node troubleshooting. To see which compute nodes have which GPUs and in what quantities, use the <code>show_nodes</code> command on a submission or compute node. The quantities are listed under the <tt>GRES</tt> column.		The exact quantities of GPUs per type are not listed here since these numbers may change frequently due to additions to or removals from the cluster or during compute node troubleshooting. To see which compute nodes have which GPUs and in what quantities, use the <code>show_nodes</code> command on a submission or compute node. The quantities are listed under the <tt>GRES</tt> column.

	{\| class="wikitable sortable"		{\| class="wikitable sortable"

Mbaney at 18:16, 26 November 2024

2024-11-26T18:16:16Z

← Older revision		Revision as of 18:16, 26 November 2024
Line 1:		Line 1:
	There are several different types of [https://www.nvidia.com/en-us/ NVIDIA] GPUs in the [[Nexus]] cluster that are available to be scheduled. They are listed below in order of newest to oldest architecture, and then alphabetically.		There are several different types of [https://www.nvidia.com/en-us/ NVIDIA] GPUs in the [[Nexus]] cluster that are available to be scheduled. They are listed below in order of newest to oldest architecture, and then alphabetically by name.

	We do not list the exact quantities of GPU here since they change frequently due to additions to or removals from the cluster or during compute node troubleshooting. To see which compute nodes have which GPUs and in what quantities, use the <code>show_nodes</code> command on a submission or compute node. The quantities are listed under the <tt>GRES</tt> column.		We do not list the exact quantities of GPU here since they change frequently due to additions to or removals from the cluster or during compute node troubleshooting. To see which compute nodes have which GPUs and in what quantities, use the <code>show_nodes</code> command on a submission or compute node. The quantities are listed under the <tt>GRES</tt> column.

Mbaney at 21:04, 6 November 2024

2024-11-06T21:04:53Z

← Older revision		Revision as of 21:04, 6 November 2024
Line 30:		Line 30:
	\| 67		\| 67
	\| not officially published/989		\| not officially published/989
			\|-
			\| L40S
			\| <code>l40s</code>
			\| Ada Lovelace
			\| 18176
			\| 48GB GDDR6
			\| 864GB/s
			\| 91.6
			\| 183/366
	\|-		\|-
	\| RTX 6000 Ada Generation		\| RTX 6000 Ada Generation

Mbaney at 17:28, 17 September 2024

2024-09-17T17:28:05Z

← Older revision		Revision as of 17:28, 17 September 2024
Line 5:		Line 5:
	{\| class="wikitable sortable"		{\| class="wikitable sortable"
	! Name		! Name
	! GRES string (~~Slurm~~)		! GRES string ([[SLURM]])
	! [https://www.nvidia.com/en-us/technologies Architecture]		! [https://www.nvidia.com/en-us/technologies Architecture]
	! [https://developer.nvidia.com/cuda-toolkit CUDA] Cores		! [https://developer.nvidia.com/cuda-toolkit CUDA] Cores