ROCm vs. RustiCl
CPU: AMD Ryzen 9 745HX 64GB
GPU: AMD RX 9060XT 8GB
Active Modules
Select Runs to Compare:
Click to select runs...
Drop your log files here
Supports multiple dt_bench.txt files at once.
Module Processing Time Comparison
Timeline Comparison (Compute ↔ Transfers ↔ CPU)
Performance Analysis: ROCm vs. RustiCL (RX 9060XT)
In this scenario, the performance gap between the two frameworks is even more drastic than in the mobile system. ROCm dominates the field with massive superiority.
1. ROCm Performance (Extremely Fast)
- Best Result: 2.776 seconds (Version 5.5.0).
- Consistency: ROCm delivers very stable values across all tested versions (5.4.0 to 5.5.0), ranging between 2.77s and 2.89s.
- Efficiency: The GPU memory is almost fully utilized (7544 MB out of 8144 MB). The architecture is addressed here as
gfx1200.
2. RustiCL Performance (Significantly Lagging)
- Best Result: 7.247 seconds (Version 5.4.1).
- Comparison: ROCm is approximately 2.6 times faster (approx. 160% more performance) than RustiCL on this hardware.
- Latency: While ROCm stays well under 3 seconds, RustiCL requires over 7 seconds for the exact same image processing task.
3. Version Differences
- ROCm: Versions 5.4.0 and 5.5.0 are nearly neck-and-neck (2.77s), while 5.4.1 is minimally slower (2.89s).
- RustiCL: A similar pattern to this test emerges—version 5.4.1 (7.24s) is slightly faster than the newer 5.5.0 (7.46s).
February 12, 2026