Skip to content
Updated minutes ago
amd

Radeon PRO W7900

amd · rdna3 · 48 GB GDDR6 · 295W TDP

VRAM

48 GB

BF16 TFLOPS

122

Bandwidth

864 GB/s

From

$0.85/hr

Calculate ROI with this GPU →

Spec Sheet

VRAM48 GB GDDR6
Memory Bandwidth864 GB/s
BF16 TFLOPS122
FP16 TFLOPS122
FP8 TFLOPS122
INT8 TOPS122
TDP295W
InterconnectPCIE
Max per Node1
PCIe Gen4
Tensor CoresNo

Pricing by Provider

ProviderOn-DemandReservedSpotBadge
retail$0.85/hr--Cheapest

Compatible Models (226)

Training Capabilities

Estimated GPU count for full fine-tuning (AdamW, BF16) and QLoRA

Model SizeFull Fine-TuneQLoRA
7B model3 GPUs1 GPU
13B model6 GPUs1 GPU
70B model28 GPUs1 GPU

Energy Efficiency

Estimated tokens/second per Watt for popular models

Mistral 7B
0.40 t/s/WFP8
Qwen 2.5 7B
0.39 t/s/WFP8
Llama 3.1 8B
0.36 t/s/WFP8

Similar GPUs

GPUVRAMBF16 TFLOPSBW (GB/s)From
RX 7900 XTX24 GB123960$0.20/hr
RTX A600048 GB38.7768$0.49/hr
A4048 GB37.4696$0.42/hr
L40S48 GB362864$0.85/hr
L4048 GB362864$0.75/hr

Methodology Note

Performance estimates for the Radeon PRO W7900are based on InferenceBench's roofline performance model with CUDA kernel-level optimization including FlashAttention v2 and PagedAttention. Memory calculations account for model weights (48 GB GDDR6 available), KV-cache allocation, and activation memory. Throughput predictions use the Radeon PRO W7900's rated 864 GB/s memory bandwidth and 122 BF16 TFLOPS compute capacity as roofline ceilings, with empirical correction factors per GPU architecture (rdna3). See our full methodology.

Frequently Asked Questions

How many AI models can run on Radeon PRO W7900?

The Radeon PRO W7900 can run 226 AI models from our database within a single node. Compatible models range across various parameter sizes depending on the quantization precision (BF16, FP8, INT4). Smaller models fit on a single GPU while larger models may require multi-GPU setups up to 1x Radeon PRO W7900.

What is the Radeon PRO W7900 inference throughput?

The Radeon PRO W7900 delivers 122 BF16 TFLOPS and 122 FP8 TFLOPS with 864 GB/s memory bandwidth. Actual inference throughput (tokens/sec) depends on the model size, precision, and batch size. Use our calculator for model-specific throughput estimates.

How much does Radeon PRO W7900 cost per hour?

The Radeon PRO W7900 is available starting from $0.85/hour via retail. Prices vary by provider and pricing tier (on-demand, reserved, spot). Compare pricing across all providers in the table above.