Best mining intensity settings for T-Rex miner? (Tuning)

Understanding Mining Intensity Fundamentals

1. Intensity in T-Rex Miner directly controls how many parallel workgroups the GPU processes per kernel launch.

2. Higher intensity values increase GPU utilization but may trigger instability if memory bandwidth or VRAM capacity is exceeded.

3. Lower intensity reduces hash rate but often improves stability and lowers power consumption per MH/s.

4. The optimal setting varies significantly across GPU models — NVIDIA RTX 3090 behaves differently than AMD RX 6800 XT or older GTX 1070 units.

5. Intensity does not scale linearly with hash rate; diminishing returns appear rapidly beyond certain thresholds.

GPU-Specific Intensity Ranges

1. For NVIDIA RTX 3060 Ti: recommended intensity spans from 20 to 26, with 22–24 delivering best balance for ETHash and KawPoW.

2. For NVIDIA RTX 4090: intensity values between 28 and 34 are common, though memory clock tuning often matters more than intensity alone.

3. For AMD RX 6700 XT: intensity 18–22 yields stable operation on Ergo and BeamV3 without excessive core voltage spikes.

4. For legacy GTX 1060 6GB: intensity above 16 frequently causes DAG load failures or stale shares due to VRAM bandwidth constraints.

5. Dual-GPU rigs require individual per-card intensity calibration — identical cards may need different settings due to PCIe lane allocation or thermal throttling variance.

Interaction With Other T-Rex Parameters

1. Increasing intensity while leaving --lhr-tune unchanged can push LHR GPUs into permanent lock mode on Ethereum Classic or Ravencoin.

2. When using --mclock, higher intensity demands tighter memory clock alignment to prevent memory errors and rejected shares.

3. Enabling --cclock offset without adjusting intensity may result in underutilized compute units despite elevated core clocks.

4. The --maxdualimpact flag modifies how intensity distributes load between dual-algo mining threads — misalignment here degrades combined efficiency.

5. Using --temp-stop with aggressive intensity forces abrupt shutdowns instead of graceful throttling unless --temp-limit is precisely calibrated.

Stability Testing Protocol

1. Begin testing at intensity = 16 and increment by 2 until rejected share rate exceeds 0.8% over a 30-minute window.

2. Monitor GPU memory temperature separately — memory junction temps above 95°C during sustained mining indicate intensity is exceeding safe thermal envelope.

3. Log --log-path output to detect intermittent “GPU timeout” or “PCIe error” entries that correlate with intensity jumps.

4. Run --benchmark only after confirming stability — synthetic benchmarks ignore real-world DAG fragmentation effects.

5. Validate settings across multiple pool connections — some pools reject shares generated under high-intensity configurations due to timestamp skew or nonce range violations.

Frequently Asked Questions

Q: Does changing intensity affect power draw linearly?A: No. Power draw increases disproportionately beyond intensity 24 on most modern GPUs due to memory controller saturation and repeated retry cycles.

Q: Can I use the same intensity value across all algorithms in dual mining mode?A: Not reliably. KawPoW intensity settings often cause instability when applied to Autolykos2 without adjustment due to differing memory access patterns.

Q: Why does intensity behave differently on Windows versus Linux with identical hardware?A: Windows WDDM driver introduces additional GPU scheduling latency and memory management overhead, reducing effective intensity ceiling by 2–4 points compared to Linux’s bare-metal access.

Q: Is intensity related to overclocking safety margins?A: Indirectly. Higher intensity amplifies voltage fluctuations caused by unstable OC profiles — even if core clock appears stable alone, intensity stress exposes hidden VRM or memory timing weaknesses.