Skip to content

Commit aca28ee

Browse files
bene2k1bene2k1
committed
docs(gpu): add L4 GPU
1 parent f7abd44 commit aca28ee

File tree

4 files changed

+26
-2
lines changed

4 files changed

+26
-2
lines changed

‎compute/gpu/reference-content/choosing-gpu-instance-type.mdx

+15
Original file line numberDiff line numberDiff line change
@@ -57,3 +57,18 @@ Remember that there is no one-size-fits-all answer, and the right GPU Instance t
5757
| Bandwidth | 1 Gbps | 10 Gbps | 20 Gbps |
5858
| Better used for | - Graphic Computer Vision<br />- General Deep Learning usage<br />- Video encoding/decoding (~4k) | - Large-size model training<br />- Fine-tune LLMs/transformer model<br />- Generative AI<br />- Optimize GPU workflows & deployments in Kubernetes thanks to MIG | - Large-size model training<br />- Fine-tune LLMs/transformers models<br />- Generative AI<br />- Optimize GPU workflows & deployments in Kubernetes thanks to MIG |
5959
| Not made for | Large models (especially LLM) | Graphic or video encoding use cases | Graphic or video encoding use cases |
60+
61+
| | L4-1-24G | L4-2-24G | L4-4-24G | L4-8-24G |
62+
|---------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------|
63+
| GPU Type | 1x [L4](https://www.nvidia.com/en-us/data-center/l4/) | 2x [L4](https://www.nvidia.com/en-us/data-center/l4/) | 4x [L4](https://www.nvidia.com/en-us/data-center/l4/) | 8x [L4](https://www.nvidia.com/en-us/data-center/l4/) |
64+
| Tensor Cores | Yes | Yes | Yes | Yes |
65+
| Performance in TFLOPS (FP16 acc 32 Tensor Cores - without sparsity) | 120 TFLOPS | 240 TFLOPS | 480 TFLOPS | 960 TFLOPS |
66+
| VRAM | 24 GB RAM | 48 GB RAM | 96 GB RAM | 192 GB RAM |
67+
| CPU Type | AMD EPYC™ 7413 (2.65GHz) | AMD EPYC™ 7413 (2.65GHz) | AMD EPYC™ 7413 (2.65GHz) | AMD EPYC™ 7413 (2.65GHz |
68+
| vCPUs | 8 | 16 | 32 | 64 |
69+
| RAM | 48 GB DDRx | 96 GB DDRx | 192 GB DDRx | 384 GB DDR3 |
70+
| Storage | Block | Block | Block | Block |
71+
| Scratch Storage | No | No | No | No |
72+
| Bandwidth | 2.5 Gbps | 5 Gbps | 10 Gbps | 20 Gbps |
73+
| Better used for | - Building Inference infrastructure<br />- Generative AI for visual communication<br />- Streaming video content | - Building Inference infrastructure<br />- Generative AI for visual communication<br />- Streaming video content | - Building Inference infrastructure<br />- Generative AI for visual communication<br />- Streaming video content | - Building Inference infrastructure<br />- Generative AI for visual communication<br />- Streaming video content |
74+
| Not made for | TBD | TBD | TBD | TBD |

‎compute/gpu/reference-content/understanding-nvidia-fp8.mdx

+1-1
Original file line numberDiff line numberDiff line change
@@ -13,7 +13,7 @@ categories:
1313
- compute
1414
---
1515

16-
With the release of the H100 GPU, NVIDIA introduced support for a new datatype called FP8 (8-bit floating point), enabling higher throughput of matrix multipliers and convolutions.
16+
Scaleway offers GPU Instances featuring L4 and H100 GPUs that support FP8 (8-bit floating point), a revolutionary datatype introduced by NVIDIA. It enables higher throughput of matrix multipliers and convolutions.
1717

1818
FP8 is an 8-bit floating point standard which was jointly developed by NVIDIA, ARM, and Intel to speed up AI development by improving memory efficiency during AI training and inference processes.
1919

‎console/account/reference-content/organization-quotas.mdx

+6
Original file line numberDiff line numberDiff line change
@@ -146,6 +146,12 @@ At Scaleway, quotas are applicable per [Organization](/identity-and-access-manag
146146
| GPU 3070 - S| To use this product, you must [validate your identity](/console/account/how-to/verify-identity/). | 1 |
147147
| H100-1-80G | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). |
148148
| H100-2-80G | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). |
149+
| | [Payment method validated](/console/account/how-to/manage-billing-information#how-to-add-a-credit-card) | Payment method and [identity validated](/console/account/how-to/verify-identity/) |
150+
| L4-1-24G | To use this product, you must [validate your identity](/console/account/how-to/verify-identity/). | 1 |
151+
| L4-2-24G | To use this product, you must [validate your identity](/console/account/how-to/verify-identity/). | 1 |
152+
| L4-4-24G | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). |
153+
| L4-8-24G | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). | To use this product, you must [contact our support team](https://console.scaleway.com/support/create). |
154+
149155

150156

151157
## Apple silicon

‎console/account/reference-content/products-availability.mdx

+4-1
Original file line numberDiff line numberDiff line change
@@ -24,7 +24,10 @@ Scaleway products are available in multiple regions and locations worldwide. Thi
2424
| Product Category | Product | Paris region | Amsterdam region | Warsaw region |
2525
|-----------------------|---------------------------------------|------------------------|-------------------------|------------------------|
2626
| **Compute** | Instances | PAR1, PAR2, PAR3 | AMS1, AMS2, AMS3 | WAW1, WAW2, WAW3 |
27-
| | GPU | PAR1, PAR2 | Not available yet | Not available yet |
27+
| | GPU RENDER-S | PAR1 | Not available yet | Not available yet |
28+
| | GPU 3070-S | PAR2 | Not available yet | Not available yet |
29+
| | GPU L4-X-24G | PAR2 | Not available yet | WAW2 |
30+
| | GPU H100-X-80G | PAR2 | Not available yet | WAW2 |
2831
| **Bare Metal** | Elastic Metal | PAR1, PAR2 | AMS1 | Not available yet |
2932
| | Apple Silicon | PAR1, PAR3 | Not available yet | Not available yet |
3033
| | Dedibox | DC2, DC3, DC5 | AMS1 | Not available yet |

0 commit comments

Comments
 (0)