Merge pull request #122 from LucianoThunderCompute/main

Add Thunder Compute to GPU comparison table

Author: sergeyk

docs/cloud-gpus/cloud-gpus.csv

@@ -256,3 +256,8 @@ RunPod,A40 (48 GB),Ampere,8,384,40,502,6.32,0.79,,
 RunPod,A4000 (16 GB),Ampere,8,128,64,234,2.72,0.34,,
 RunPod,A5000 (24 GB),Ampere,8,192,64,313,3.52,0.44,,
 RunPod,A6000 (48 GB),Ampere,8,384,40,502,6.32,0.79,,
+Thunder Compute,A6000 (48 GB),Ampere,1,48,4,32,0.27,0.27,,
+Thunder Compute,A100 (80 GB),Ampere,1,80,4,32,0.78,0.78,,
+Thunder Compute,H100 (80 GB),Hopper,1,80,4,32,1.38,1.38,,
+Thunder Compute,A100 (80 GB),Ampere,8,640,144,720,14.32,1.79,,
+Thunder Compute,H100 (80 GB),Hopper,8,640,144,720,19.92,2.49,,

docs/cloud-gpus/index.md

@@ -56,7 +56,7 @@ not just inference in particular models as a service.
 
 - Direct price comparisons are trickier for serverless offerings: cold boot time and autoscaling logic can substantially impact cost-of-traffic.
 - Some of the providers allow configuration of CPU and RAM resources. We have selected reasonable defaults, generally comparable to the fixed offerings of other providers.
-- You can find pricing pages for the providers here: [Banana](https://banana.dev#pricing), [Baseten](https://docs.baseten.co/settings/pricing), [Beam](https://beam.cloud/pricing), [Covalent](https://www.covalent.xyz/), [Modal](https://modal.com/pricing), [OVHcloud](https://www.ovhcloud.com/en/public-cloud/prices/#11264), [Replicate](https://replicate.com/pricing), [RunPod](https://www.runpod.io/serverless-gpu)
+- You can find pricing pages for the providers here: [Banana](https://banana.dev#pricing), [Baseten](https://docs.baseten.co/settings/pricing), [Beam](https://beam.cloud/pricing), [Covalent](https://www.covalent.xyz/), [Modal](https://modal.com/pricing), [OVHcloud](https://www.ovhcloud.com/en/public-cloud/prices/#11264), [Replicate](https://replicate.com/pricing), [RunPod](https://www.runpod.io/serverless-gpu), [Thunder Compute](https://www.thundercompute.com/pricing)
 - Serverless GPUs are a newer technology, so the details change quickly and you can expect bugs/growing pains. Stay frosty!
 - Raw data can be found in a [csv on GitHub](https://github.com/full-stack-deep-learning/website/blob/main/docs/cloud-gpus/serverless-gpus.csv).
 
@@ -89,6 +89,7 @@ Below are the raw TFLOPs of the different GPUs available from cloud providers.
 
 | Model | Arch   | FP32 | Mixed-precision | FP16 | Source             |
 | ----- | ------ | ---- | --------------- | ---- | ------------------ |
+| H100  | Hopper | 51   | 756             | 1513 | [Datasheet][h100]  |
 | A100  | Ampere | 19.5 | 156             | 312  | [Datasheet][a100]  |
 | A10G  | Ampere | 35   | 35              | 70   | [Datasheet][a10g]  |
 | A6000 | Ampere | 38   | ?               | ?    | [Datasheet][a6000] |
@@ -99,6 +100,7 @@ Below are the raw TFLOPs of the different GPUs available from cloud providers.
 | K80   | Kepler | 8.73 | N/A             | N/A  | [Datasheet][k80]   |
 | A40   | Ampere | 37   | 150             | 150  | [Datasheet][a40]   |
 
+[h100]: https://resources.nvidia.com/en-us-hopper-architecture/nvidia-tensor-core-gpu-datasheet?ncid=no-ncid
 [a100]: https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Center/a100/pdf/nvidia-a100-datasheet-us-nvidia-1758950-r4-web.pdf
 [a10g]: https://d1.awsstatic.com/product-marketing/ec2/NVIDIA_AWS_A10G_DataSheet_FINAL_02_17_2022.pdf
 [a6000]: https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/proviz-print-nvidia-rtx-a6000-datasheet-us-nvidia-1454980-r9-web%20(1).pdf