AMD-Pro-BenchCommunity

13,812 benchmarks in 32 GPU combinations documented

Stable Diffusion XL image benchmarks on AMD GPUs

619 reported configurations · 16 sources · 626 measurements

Documented results

24 selected configurations

Selected across sources, not a performance ranking. Match model weights, resolution, steps, software and timing scope before comparing.

SourceModelGPUsOutputMeasurements
View
Report
UL Procyon
Stable Diffusion XLFP16Model reference ↗Text → imageRadeon RX 7700 XT1 GPU
Configuration
Engine
UL Procyon AI Image Generation
Backend
ONNX Runtime / DirectML
OS
Windows
Parallelization
Not reported
Memory
12 GB per GPU
Reported
2025-06-04
Full record →
1024 × 1024Batch 1100 stepsBatch 1
47.5 s/imageTime / image · reported
Whole workflow · Timing details missing
View
Report
ComfyUI
Stable Diffusion XL1.0 BaseQuant not reportedTested weights ↗Text → imageRyzen AI Max+ 395 (Radeon 8060S)1 GPU
Configuration
Engine
ComfyUI ab0d8a92
Backend
ROCm / HIP
OS
Windows11 Pro24H2 build26100
Parallelization
Not reported
Memory
Not reported
Reported
2026-07-14
Full record →
1024 × 1024Batch 125 stepsBatch 1
26.57 s/imageTime / image · reported
1.43 steps/sSampling steps · reported
Whole workflow · Cold
View
Report
ComfyUI
Stable Diffusion XLQuant not reportedModel reference ↗Text → imageRadeon RX 9070 XT1 GPU
Configuration
Engine
ComfyUI
Backend
Not reported
OS
Ubuntu Linux
Parallelization
Not reported
Memory
16 GB per GPU
Reported
2026-09-23
Full record →
1024 × 1024Batch 130 stepsBatch 1
0.137 images/sCalculated · Image rate · reported · sustained pipeline throughputExact unit normalization of reported8.22images/minute.reported images per minute / 60
View
Report
ComfyUI
Stable Diffusion XL 1.0 baseQuant not reportedModel reference ↗Text → imageRadeon AI PRO R97001 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Ubuntu 24.04.4 LTS
Parallelization
Single GPU
Memory
32 GB per GPU
Reported
2026-09-23
Full record →
1024 × 1024Batch 130 stepsBatch 1
0.151 images/sImage rate · reportedUnit conversion from reported9.03images/min; not an independently measured seconds/image latency.
View
Report
UL Procyon
Stable Diffusion XLFP16Model reference ↗Text → imageRadeon RX 9070 GRE1 GPU
Configuration
Engine
UL Procyon AI Image Generation
Backend
ONNX Runtime / DirectML
OS
Windows
Parallelization
Not reported
Memory
12 GB per GPU
Reported
Not reported
Full record →
Resolution not reportedSteps not reportedBatch not reported
46.2 s/imageTime / image · reported
Timing details missing
View
Report
Diffusion workflow
Stable Diffusion XLINT8Model reference ↗Text → imageRadeon RX 7900 XTX1 GPU
Configuration
Engine
SHARK / Shortfin / IREE
Backend
ROCm / HIP
OS
Ubuntu
Parallelization
Not reported
Memory
24 GB per GPU
Reported
Not reported
Full record →
Resolution not reportedSteps not reportedBatch not reported
6.3 s/imageTime / image · reported
Whole workflow · Timing details missing
View
Report
Diffusion workflow
Stable Diffusion XLINT8Model reference ↗Text → imageRadeon PRO W79001 GPU
Configuration
Engine
SHARK / Shortfin / IREE
Backend
ROCm / HIP
OS
Ubuntu
Parallelization
Not reported
Memory
48 GB per GPU
Reported
Not reported
Full record →
Resolution not reportedSteps not reportedBatch not reported
6.9 s/imageTime / image · reported
Whole workflow · Timing details missing
View
Report
UL Procyon
Stable Diffusion XLFP16Model reference ↗Text → imageRadeon RX 9070 GRE1 GPU
Configuration
Engine
UL Procyon AI Image Generation
Backend
Not reported
OS
Not reported
Parallelization
Not reported
Memory
12 GB per GPU
Reported
2026-06-02
Full record →
Resolution not reportedSteps not reportedBatch not reported
24.277 s/imageTime / image · reported
1,544 Procyon scoreProcyon performance index · reportedPerformance score, not image quality.
Timing details missing
View
Report
ComfyUI
SDXL (checkpoint unreported)Quant not reportedFamily reference ↗Text → imageRadeon AI PRO R97001 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Linux
Parallelization
Single GPU
Memory
32 GB per GPU
Reported
2026-03-09
Full record →
1024 × 102420 stepsBatch not reported
9.02 sGeneration time · reportedFirst run includes initialization/model loading.
Whole workflow · Cold
View
Report
ComfyUI
Stable Diffusion XLQuant not reportedModel reference ↗Text → imageRadeon RX 9070 XT1 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Kubuntu 20.04 (as reported)
Parallelization
Not reported
Memory
16 GB per GPU
Reported
2025-03-21
Full record →
1024 × 102420 stepsBatch not reported
32.25 s/imageTime / image · reported
1.03 steps/sSampling steps · reported
Timing details missing
View
Report
ComfyUI
SDXL v-pred (checkpoint unreported)Quant not reportedFamily reference ↗Text → imageRadeon AI PRO R97001 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Arch Linux
Parallelization
Single GPU
Memory
32 GB per GPU
Reported
Not reported
Full record →
1024 × 1024Batch 450 stepsBatch 4
44.5 s/batchBatch time · reportedWhole batch of4 images, not seconds per image.
1.18 steps/sSampling steps · reportedReported sampler iterations for batch 4; do not confuse normalized4.72image-iterations/s with sampler rate.
Timing details missing
View
Report
Diffusion workflow
SDXL 1.0 Base Olive optimizedQuant not reportedTested weights ↗Text → imageRadeon RX 7900 XT1 GPU
Configuration
Engine
ONNX Runtime / Microsoft Olive
Backend
DirectML
OS
Windows 11
Parallelization
Not reported
Memory
20 GB per GPU
Reported
Not reported
Full record →
1024 × 1024Steps not reportedBatch not reported
3.8 steps/sSampling steps · reportedApproximate author-reported rate.
View
Report
SD WebUI
juggernautxl_juggxibyrundiffusionQuant not reportedFamily reference ↗Text → imageRadeon RX 7900 XTX1 GPU
Configuration
Engine
SD.Next 3f360891
Backend
ZLUDA
OS
Windows
Parallelization
Not reported
Memory
24 GB per GPU
Reported
Not reported
Full record →
512 × 512Batch 1Steps not reportedBatch 1
6.34 image-steps/sPipeline throughput · reportedHarness computes steps times batch divided by process_images wall time; not sampler iterations per second or images per second.
View
Report
stable-diffusion.cpp / sd2R
Stable Diffusion XLQuant not reportedModel reference ↗Text → imageRadeon RX 7600 XT1 GPU
Configuration
Engine
stable-diffusion.cpp
Backend
ROCm / HIP
OS
Linux
Parallelization
Not reported
Memory
16 GB per GPU
Reported
2026-01-19
Full record →
Resolution not reported8 stepsBatch not reported
1.47 s/stepStep time · reported
Sampling only · Timing details missing
View
Report
UL Procyon
Stable Diffusion XLFP16Model reference ↗Text → imageRadeon RX 9060 XT1 GPU
Configuration
Engine
UL Procyon AI Image Generation
Backend
Not reported
OS
Not reported
Parallelization
Not reported
Memory
16 GB per GPU
Reported
2025-06-04
Full record →
Resolution not reportedSteps not reportedBatch not reported
33.359 s/imageTime / image · reported
1,124 Procyon scoreProcyon performance index · reportedPerformance score, not image quality.
Timing details missing
View
Report
ComfyUI
Stable Diffusion XLQuant not reportedModel reference ↗Text → imageRyzen AI Max+ 395 (Radeon 8060S)1 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Windows 11
Parallelization
Not reported
Memory
Not reported
Reported
2025-12-21
Full record →
1024 × 1024Batch 130 stepsBatch 1
24.63 s/imageTime / image · reported
Timing details missing
View
Report
UL Procyon
Stable Diffusion XLFP16Model reference ↗Text → imageRadeon RX 7800 XT1 GPU
Configuration
Engine
UL Procyon AI Image Generation
Backend
ONNX Runtime / DirectML
OS
Windows
Parallelization
Not reported
Memory
16 GB per GPU
Reported
2025-06-04
Full record →
1024 × 1024Batch 1100 stepsBatch 1
42.2 s/imageTime / image · reported
Whole workflow · Timing details missing
View
Report
ComfyUI
Stable Diffusion XL1.0 BaseQuant not reportedTested weights ↗Text → imageRyzen AI Max+ 395 (Radeon 8060S)1 GPU
Configuration
Engine
ComfyUI ab0d8a92
Backend
ROCm / HIP
OS
Windows11 Pro24H2 build26100
Parallelization
Not reported
Memory
Not reported
Reported
2026-07-14
Full record →
1024 × 1024Batch 125 stepsBatch 1
18.41 s/imageTime / image · reported
1.48 steps/sSampling steps · reported
Whole workflow · Warm
View
Report
Diffusion workflow
Stable Diffusion XLFP8Model reference ↗Text → imageRadeon RX 90701 GPU
Configuration
Engine
SHARK / Shortfin / IREE
Backend
ROCm / HIP
OS
Ubuntu
Parallelization
Not reported
Memory
16 GB per GPU
Reported
Not reported
Full record →
Resolution not reportedSteps not reportedBatch not reported
3.7 s/imageTime / image · reported
Whole workflow · Timing details missing
View
Report
Diffusion workflow
Stable Diffusion XLFP16Model reference ↗Text → imageRadeon PRO W79001 GPU
Configuration
Engine
IREE SDXL scripts
Backend
ROCm / HIP
OS
Not reported
Parallelization
Not reported
Memory
48 GB per GPU
Reported
Not reported
Full record →
Resolution not reportedSteps not reportedBatch not reported
0.318 s/forwardStep time · reportedExact ms-to-seconds conversion; isolated UNet, not image latency.
Model component · Timing details missing
View
Report
ComfyUI
SDXL (checkpoint unreported)Quant not reportedFamily reference ↗Text → imageRadeon AI PRO R97001 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Linux
Parallelization
Single GPU
Memory
32 GB per GPU
Reported
2026-03-09
Full record →
1024 × 102420 stepsBatch not reported
5.2 sGeneration time · reportedApproximately reported for warm execution.
4.6 steps/sSampling steps · reportedSampler iterations/s; a separate denoising rate, not full workflow throughput.
Whole workflow · Warm
View
Report
ComfyUI
SDXL v-pred (checkpoint unreported)Quant not reportedFamily reference ↗Text → imageRadeon AI PRO R97001 GPU
Configuration
Engine
ComfyUI
Backend
ROCm / HIP
OS
Arch Linux
Parallelization
Single GPU
Memory
32 GB per GPU
Reported
Not reported
Full record →
1024 × 1024Batch 450 stepsBatch 4
41.2 s/batchBatch time · reportedWhole batch of4 images, not seconds per image.
Timing details missing
View
Report
SD WebUI
juggernautxl_juggxibyrundiffusionQuant not reportedFamily reference ↗Text → imageRadeon RX 7900 XTX1 GPU
Configuration
Engine
SD.Next 3f360891
Backend
ZLUDA
OS
Windows
Parallelization
Not reported
Memory
24 GB per GPU
Reported
Not reported
Full record →
512 × 512Batch 2Steps not reportedBatch 2
8.15 image-steps/sPipeline throughput · reportedHarness computes steps times batch divided by process_images wall time; not sampler iterations per second or images per second.
View
Report
stable-diffusion.cpp / sd2R
Stable Diffusion XLQuant not reportedModel reference ↗Text → imageRadeon RX 7600 XT1 GPU
Configuration
Engine
stable-diffusion.cpp
Backend
Vulkan
OS
Linux
Parallelization
Not reported
Memory
16 GB per GPU
Reported
2026-01-19
Full record →
Resolution not reported8 stepsBatch not reported
1.55 s/stepStep time · reported
Sampling only · Timing details missing
Explore all 619 configurations →

All image benchmarks · Methodology and source review