Page 1 failed to load.
1usmus - Bringing back the good old OC days — Radeon Advanced Control
More from 1usmus
1usmus - SQ AGENT: how one core can hold boost back
About This Title
Remember when GPU tuning meant more than moving a few sliders? When exploring voltage, power limits and memory timings could reveal something your card’s default settings left behind?
That’s the spirit we’re bringing back with HYDRA.
After bringing advanced NVIDIA unlocks back to enthusiasts, we’re taking the next step with Radeon. This is our first generation of Radeon Advanced Control — and, for us, one of the most exciting steps toward meaningful GPU unlocks in years.
Modern GPUs are incredibly capable, but their standard tuning controls leave little room to explore. Here’s what we’re opening up:
Six voltage-offset points instead of one global slider
Different parts of the frequency range can need different adjustments. A six-point curve lets you pursue a stronger undervolt where your GPU tolerates it while preserving voltage headroom where stability needs it.
Instead of accepting one compromise across the entire range, you get more freedom to tune around your particular chip — whether your priority is efficiency, lower temperatures or sustained performance.
Expanded PPT control
More freedom to adjust your GPU’s power budget — beyond the usual percentage slider. Combined with Overdrive, the expanded base PPT offers some interesting possibilities:
RX 7900 XTX: 350 W base PPT + 15% OD = 402.5 W calculated PPT.
RX 9070 XT: 340 W base PPT + 10% OD = 374 W, demonstrated on our test card.
Whether you’re chasing additional performance or building a more efficient everyday profile, you get more room to tune. This first generation will retain temporary limits, with availability and adjustment ranges depending on the card.
Memory timing adjustments
Veteran Radeon owners will remember how useful timing optimization could be. Memory frequency is only part of the equation: supported timing adjustments can improve how efficiently memory handles work, potentially helping performance without simply pushing MHz higher.
It brings another worthwhile area of experimentation back into GPU tuning.
FCLK, UCLK and SOCCLK control
We’re also opening up clock controls beyond the GPU core: FCLK for the internal fabric, UCLK for the memory subsystem and SOCCLK for the SoC domain.
These controls let you explore how different parts of the GPU work together. Depending on the card and workload, tuning their limits can help you find a better balance between performance and power consumption. Higher is not automatically better — the goal is to find what works for your setup.
HYDRA brings these capabilities together in one interface, going beyond the options exposed by standard Radeon tuning software.
This is the first generation, with more work ahead. Features and adjustment ranges will vary by GPU, firmware and driver. We’re continuing compatibility and stability testing, and individual results will depend on your hardware and workload.
HYDRA build 2320, bringing expanded capabilities for RDNA 3 and RDNA 4, is scheduled to begin rolling out this Friday, barring any unforeseen issues.
Thank you for supporting the research that makes this possible. We’re bringing back more of the freedom that made GPU tuning fun.
Stay tuned — the update is on its way!
Remember when GPU tuning meant more than moving a few sliders? When exploring voltage, power limits and memory timings could reveal something your card’s default settings left behind?
That’s the spirit we’re bringing back with HYDRA.
After bringing advanced NVIDIA unlocks back to enthusiasts, we’re taking the next step with Radeon. This is our first generation of Radeon Advanced Control — and, for us, one of the most exciting steps toward meaningful GPU unlocks in years.
Modern GPUs are incredibly capable, but their standard tuning controls leave little room to explore. Here’s what we’re opening up:
Six voltage-offset points instead of one global slider
Different parts of the frequency range can need different adjustments. A six-point curve lets you pursue a stronger undervolt where your GPU tolerates it while preserving voltage headroom where stability needs it.
Instead of accepting one compromise across the entire range, you get more freedom to tune around your particular chip — whether your priority is efficiency, lower temperatures or sustained performance.
Expanded PPT control
More freedom to adjust your GPU’s power budget — beyond the usual percentage slider. Combined with Overdrive, the expanded base PPT offers some interesting possibilities:
RX 7900 XTX: 350 W base PPT + 15% OD = 402.5 W calculated PPT.
RX 9070 XT: 340 W base PPT + 10% OD = 374 W, demonstrated on our test card.
Whether you’re chasing additional performance or building a more efficient everyday profile, you get more room to tune. This first generation will retain temporary limits, with availability and adjustment ranges depending on the card.
Memory timing adjustments
Veteran Radeon owners will remember how useful timing optimization could be. Memory frequency is only part of the equation: supported timing adjustments can improve how efficiently memory handles work, potentially helping performance without simply pushing MHz higher.
It brings another worthwhile area of experimentation back into GPU tuning.
FCLK, UCLK and SOCCLK control
We’re also opening up clock controls beyond the GPU core: FCLK for the internal fabric, UCLK for the memory subsystem and SOCCLK for the SoC domain.
These controls let you explore how different parts of the GPU work together. Depending on the card and workload, tuning their limits can help you find a better balance between performance and power consumption. Higher is not automatically better — the goal is to find what works for your setup.
HYDRA brings these capabilities together in one interface, going beyond the options exposed by standard Radeon tuning software.
This is the first generation, with more work ahead. Features and adjustment ranges will vary by GPU, firmware and driver. We’re continuing compatibility and stability testing, and individual results will depend on your hardware and workload.
HYDRA build 2320, bringing expanded capabilities for RDNA 3 and RDNA 4, is scheduled to begin rolling out this Friday, barring any unforeseen issues.
Thank you for supporting the research that makes this possible. We’re bringing back more of the freedom that made GPU tuning fun.
Stay tuned — the update is on its way!
Comments (0)