This page summarizes performance measurements on AMD Instinct™ GPUs for popular AI models.

The data in the following tables is a reference point to help users evaluate observed performance. It should not be considered as the peak performance that AMD GPUs and ROCm™ software can deliver.

AI Inference

Throughput Measurements

The table below shows performance data where a local inference client is fed requests at an infinite rate and shows the throughput client-server scenario under maximum load.  

This result is based on the Docker container (rocm/vllm:rocm6.4.1_vllm_0.9.0.1_20250605), which was released on June 18, 2025.

Model

Precision

TP Size

Input

Output

Num Prompts

Max Num Seqs

Throughput (tokens/s)

Llama 3.1 70B (amd/Llama-3.1-70B-Instruct-FP8-KV)

FP8

8

128

2048

3200

3200

16581.5

     

128

4096

1500

1500

13667.3

     

500

2000

2000

2000

13367.1

     

2048

2048

1500

1500

8352.6

Llama 3.1 405B (amd/Llama-3.1-405B-Instruct-FP8-KV)

FP8

8

128

2048

1500

1500

4275

     

128

4096

1500

1500

3356.7

     

500

2000

2000

2000

3201.4

     

2048

2048

500

500

2179.7

TP stands for Tensor Parallelism.

Server: Dual AMD EPYC 9554 64-core processor-based production server with 8x AMD MI300X (192GB HBM3 750W) GPUs, 1 NUMA node per socket, System BIOS 1.8, Ubuntu® 22.04.5 LTS, Host GPU driver ROCm 6.4.1 + amdgpu driver 6.8.5 

Reproduce these results on your system by following the instructions in measuring inference performance with vLLM on the AMD GPUs user guide.

Latency Measurements

The table below shows latency measurement, which typically involves assessing the time from when the system receives an input to when the model produces a result.

This result is based on the Docker container (rocm/vllm:rocm6.4.1_vllm_0.9.0.1_20250605), which was released on June 18, 2025.

Model

Precision

TP Size

Batch Size

Input

Output

Latency (sec)

Llama 3.1 70B (amd/Llama-3.1-70B-Instruct-FP8-KV)

FP8

8

1

128

2048

15.566

     

2

128

2048

16.858

     

4

128

2048

17.518

     

8

128

2048

18.898

     

16

128

2048

21.023

     

32

128

2048

23.896

     

64

128

2048

30.753

     

128

128

2048

43.767

     

1

2048

2048

15.496

     

2

2048

2048

17.38

     

4

2048

2048

17.983

     

8

2048

2048

19.771

     

16

2048

2048

22.702

     

32

2048

2048

27.392

     

64

2048

2048

36.879

     

128

2048

2048

57.003

Llama 3.1 405B (amd/Llama-3.1-70B-Instruct-FP8-KV)

FP8

8

1

128

2048

45.828

     

2

128

2048

46.757

     

4

128

2048

48.322

     

8

128

2048

51.479

     

16

128

2048

54.861

     

32

128

2048

63.119

     

64

128

2048

82.362

     

128

128

2048

109.698

     

1

2048

2048

46.514

     

2

2048

2048

47.271

     

4

2048

2048

49.679

     

8

2048

2048

54.366

     

16

2048

2048

60.39

     

32

2048

2048

74.209

     

64

2048

2048

104.728

     

128

2048

2048

154.041

Server: Dual AMD EPYC 9554 64-core processor-based production server with 8x AMD MI300X (192GB HBM3 750W) GPUs, 1 NUMA node per socket, System BIOS 1.8, Ubuntu® 22.04.5 LTS, Host GPU driver ROCm 6.4.1 + amdgpu driver 6.8.5

Reproduce these results on your system by following the instructions in measuring inference performance with ROCm vLLM Dcoker on the AMD GPUs user guide.

Previous versions

This table lists previous versions of the ROCm vLLM inference Docker image for inference performance testing. For detailed information about available models for benchmarking, see the version-specific documentation.

Date ROCm version vLLM version PyTorch version Resources
5/27/2025 6.3.1 0.8.5 2.7.0 Documentation Docker Hub
4/10/2025 6.3.1 0.8.3 2.7.0 Documentation Docker Hub
3/25/2025 6.3.1 0.7.3 2.7.0 Documentation Docker Hub
3/11/2025 6.3.1 0.7.3 2.7.0 Documentation Docker Hub
2/5/2025 6.3.1 0.6.6 2.7.0 Documentation Docker Hub
11/7/2024 6.2.1 0.6.4 2.5.0 Documentation Docker Hub
9/4/2024 6.2.0 0.4.3 2.4.0 Documentation Docker Hub

AI Training

The table below shows training performance data, where the AMD Instinct™ platform measures text generation training throughput with a unique sequence length and batch size. It focuses on TFLOPS per second per GPU.

For FLUX, image generation training throughput from the FLUX.1-dev model with the best batch size before the runs go out of memory, and it focuses on frame per second per GPU.

PyTorch training results on the AMD Instinct™ MI300X platform

This result is based on the Docker container (rocm/pytorch-training:v25.5), which was released on April 15, 2025.

Models Precision Batch Size Sequence Length TFLOPS/s/GPU
Llama 3.1 70B with FSDP BF16 4 8192 426.79
Llama 3.1 8B with FSDP BF16 3 8192 542.94
Llama 3.1 8B with FSDP FP8 3 8192 737.40
Llama 3.1 8B with FSDP BF16 6 4096 523.79
Llama 3.1 8B with FSDP FP8 6 4096 735.44
Mistral 7B with FSDP BF16 3 8192 483.17
Mistral 7B with FSDP FP8 4 8192 723.30
FLUX BF16 10 - 4.51 (FPS/GPU)*

*Note: FLUX performance is measured in FPS/GPU rather than TFLOPS/s/GPU.

Server: Dual AMD EPYC 9654 96-core processor-based production server with 8x AMD MI300X (192GB HBM3 750W) GPUs, 2 NUMA node per socket, System BIOS 5.27, Ubuntu® 22.04.5 LTS, Host GPU driver ROCm 6.3.0 ROCm 6.3 (pre-release)

Reproduce these results on your system by following the instructions in measuring training performance with ROCm PyTorch Docker on the AMD GPUs user guide.

PyTorch training results on the AMD Instinct MI325X platform

This result is based on the Docker container (rocm/pytorch-training:v25.5), which was released on April 15, 2025.

Models Precision Batch Size Sequence Length TFLOPS/s/GPU
Llama 3.1 70B with FSDP BF16 7 8192 526.13
Llama 3.1 8B with FSDP BF16 3 8192 643.01
Llama 3.1 8B with FSDP FP8 5 8192 893.68
Llama 3.1 8B with FSDP BF16 8 4096 625.96
Llama 3.1 8B with FSDP FP8 10 4096 894.98
Mistral 7B with FSDP BF16 5 8192 590.23
Mistral 7B with FSDP FP8 6 8192 860.39

Server: Dual AMD EPYC 9654 96-core processor-based production server with 8x AMD MI325X (256GB HBM3E 1000W) GPUs, 2 NUMA node per socket, System BIOS 5.27, Ubuntu® 22.04.5 LTS, Host GPU driver ROCm 6.3.0 ROCm 6.3 (pre-release)

Reproduce these results on your system by following the instructions in measuring training performance with ROCm PyTorch Docker on the AMD GPUs user guide.

Previous versions

This table lists previous versions of the PyTorch training Docker image for training performance testing. For detailed information about available models for benchmarking, see the version-specific documentation.

Date Image version ROCm version PyTorch version Resources
3/11/2025 25.4 6.3.0 2.7.0a0+git637433 Documentation Docker Hub

Megatron-LM training results on the AMD Instinct™ MI300X platform

This result is based on the Docker container (rocm/megatron-lm:v25.5), which was released on April 25, 2025.

Sequence length 8192
Model # of nodes Sequence length MBS GBS Data Type TP PP CP TFLOPs/s/GPU
llama3.1-8B 1 8192 2 128 FP8 1 1 1 697.91
llama3.1-8B 2 8192 2 256 FP8 1 1 1 690.33
llama3.1-8B 4 8192 2 512 FP8 1 1 1 686.74
llama3.1-8B 8 8192 2 1024 FP8 1 1 1 675.50
Sequence length 4096
Model # of nodes Sequence length MBS GBS Data Type TP PP CP TFLOPs/s/GPU
llama2-7B 1 4096 4 256 FP8 1 1 1 689.90
llama2-7B 2 4096 4 512 FP8 1 1 1 682.04
llama2-7B 4 4096 4 1024 FP8 1 1 1 676.83
llama2-7B 8 4096 4 2048 FP8 1 1 1 686.25

Server: Dual AMD EPYC 9654 96-core processor-based production server with 8x AMD MI300X (192GB HBM3 750W) GPUs, 2 NUMA node per socket, System BIOS 5.27, Ubuntu® 22.04.5 LTS, Host GPU driver ROCm 6.3.0 ROCm 6.3 (pre-release)

For Deepsee-V2-Lite with 16B parameters, the table below shows training performance data, where the AMD Instinct™ MI300X platform measures text generation training throughput with GEMM tuning was on. It focuses on TFLOPS per second per GPU.  

This result is based on the Docker container(rocm/megatron-lm:v25.5), which was released on April 25, 2025.

Model # of GPUs Sequence length MBS GBS Data Type TP PP CP EP SP Recompute TFLOPs/s/GPU
Deespeek-V2-Lite 8 4096 4 256 BF16 1 1 1 8 On None 10570

Server: Dual AMD EPYC 9654 96-core processor-based production server with 8x AMD MI300X (192GB HBM3 750W) GPUs, 2 NUMA node per socket, System BIOS 5.27, Ubuntu® 22.04.5 LTS, Host GPU driver ROCm 6.3.0 ROCm 6.3 (pre-release)

Reproduce these results on your system by following the instructions in measuring training performance with ROCm Megatron-LM Docker on the AMD GPUs user guide.

Previous versions

This table lists previous versions of the Megatron-LM training Docker image for training performance testing. For detailed information about available models for benchmarking, see the version-specific documentation.

Date Image version ROCm version PyTorch version Resources
3/18/2025 25.4 6.3.0 2.7.0a0+git637433 Documentation Docker Hub