User Guide (Latest Version)

Training Accuracy: NVIDIA DGX SuperPOD (8 x 8 x A100 80GB for CLIP B/32 Model)

We followed the training recipe from Open CLIP blog to verify our training pipeline. Our results are displayed in the table below:



Model Name

Batch Size

Samples Seen

ImageNet Top-1

OpenCLIP LAION 400M B/32 32k 12B 62.90%
NeMo Our Multimodal Blend* B/32 32k 12B 60.13%

Our multimodal dataset is originated from Common Crawl with custom filtering and contains 670M image-caption pairs.

We believe the final accuracy difference is due to the dataset, as LAION 400M is filtered with CLIP scores. To ensure our implementation is consistent with OpenCLIP, we trained OpenCLIP with our dataset and found out that the loss curve and validation accuracy were nearly identical to NeMo’s CLIP.

We measured the throughput of training CLIP models on different numbers of DGX A100 nodes and DGX H100 nodes, and we achieved near-linear scaling on both platforms.

We are comparing the out-of-box performance on DGX H100 machines with the same configuration from DGX A100 machines. This comparison is an apple-to-apple assessment, ensuring that we evaluate the relative performance of the two machine types under equivalent conditions and configurations.

The tables and charts below show the performance results.

NVIDIA DGX SuperPODs (16 x 8 x A100 80GB for CLIP g/14 model)






Samples per Second 559 1115 2190 4407 8633
CLIP g/14 Perfect Linear Scaling (Samples) 559 1119 2237 4475 8950
Speedup 1x 1.99x 3.92x 7.88x 15.43x
CLIP g_14 NeMo Throughput (A100) (2308).svg

NVIDIA DGX SuperPODs (16 x 8 x H100 80GB for CLIP g/14 model)






Samples per Second 935 1795 3502 6771 13829
CLIP g/14 Perfect Linear Scaling (Samples) 935 1869 3739 7478 14955
Speedup 1x 1.92x 3.75x 7.24x 14.8x
CLIP g_14 NeMo Throughput (H100) (2308).svg

DGX A100 vs. DGX H100: A Comparative Analysis of CLIP Training



Global Batch Size

Micro Batch Size


Global Batch / Sec (A100)

Global Batch / Sec (H100)

Speedup (x)

CLIP B/32 4 16000 500 bf16 (O2) 2.12 5.26 2.5
CLIP H/14 4 3584 112 bf16 (O2) 0.88 1.92 2.2
CLIP g/14 4 2560 80 bf16 (O2) 0.86 2.25 2.6
CLIP Training Throughput Comparison (2308).svg

Latency times are taken as starting with an image on CPU and text input (of length 64) and stopped on output. For framework we use the Torch Automated Mixed Precision (AMP) for FP16 computation. For TRT, we export the various models with the FP16 acceleration. We use the optimized TRT engine setup present in the deployment directory to get the numbers in the same environment as the framework.

GPU: NVIDIA DGX A100 (1x A100 80 GB) Batch Size: Number of Images in a Batch


Batch Size

TRT FP16 Latency (s)

FW FP 16 (AMP) Latency (s)

TRT vs FW Speedup (x)

1 0.014 0.032 2.3
2 0.014 0.033 2.4
CLIP B/32 4 0.014 0.028 2.0
8 0.015 0.028 1.9
Previous Framework Inference
Next Vision Transformer
© | | | | | | |. Last updated on Jun 14, 2024.