How to use SRBMiner for Multi-Mining? (Config Tutorial)

Understanding SRBMiner Multi-Mining Architecture

1. SRBMiner is a high-performance, open-source mining software designed specifically for AMD and NVIDIA GPUs, supporting multiple algorithms simultaneously through its multi-mining mode.

2. The core principle relies on algorithm switching based on real-time profitability calculations derived from pool APIs, exchange rates, and network difficulty metrics.

3. Unlike traditional single-algorithm miners, SRBMiner maintains persistent connections to multiple mining pools across different blockchains such as Monero (RandomX), Ethereum Classic (ETC), and Kaspa (K heavy).

4. Each mining thread operates independently with configurable GPU device affinity, enabling precise control over which GPU handles which algorithm at any given moment.

5. The miner uses a built-in profitability engine that polls every 30 seconds by default, triggering an algorithm switch only when the new target yields at least 5% higher net hash rate efficiency after accounting for pool fees and network latency.

Required Dependencies and System Preparation

1. Windows 10/11 64-bit or Linux Ubuntu 20.04+ is mandatory; legacy OS versions are unsupported due to AVX-512 instruction requirements.

2. AMD Adrenalin 23.5.1 or newer drivers must be installed for RDNA2/RDNA3 GPUs; older drivers cause kernel-level memory access violations during dual-algo execution.

3. NVIDIA driver version 525.85.12 or higher is required to prevent CUDA context corruption when running KawPow alongside RandomX in parallel threads.

4. Visual C++ Redistributable for Visual Studio 2022 must be present—missing DLLs result in immediate process termination with exit code 0xC0000005.

5. Disable Windows Fast Startup and Memory Integrity (HVCI) in Core Isolation settings, as both interfere with GPU memory mapping during multi-threaded kernel launches.

Configuring config.json for Dual-Algorithm Operation

1. Open the config.json file located in the SRBMiner-Multi root directory using a UTF-8–compliant editor like Notepad++.

2. Set 'MultiMineMode': true to activate concurrent algorithm execution instead of sequential fallback behavior.

3. Define primary and secondary pools under 'pools' array, ensuring each entry contains valid 'algorithm', 'url', and 'user' fields without trailing slashes.

4. Assign GPU indices explicitly using 'devices': [0,1] inside each pool object to avoid cross-device memory contention during simultaneous RandomX and Autolykos2 kernel invocations.

5. Enable 'enableProfitSwitching': true and adjust 'minProfitSwitchTime': 180 to enforce minimum dwell time between switches, preventing oscillation during volatile market conditions.

Runtime Monitoring and Log Interpretation

1. Launch SRBMiner-Multi.exe from command line with --log-file miner.log to capture all algorithm transitions, rejected shares, and thermal throttling events.

2. Watch for lines containing 'Switching to [ALGO] on GPU[0]' — these indicate successful profitability-driven reconfiguration without restart.

3. Rejected share counts above 3% per pool require immediate inspection of 'difficulty' mismatch or NTP sync drift exceeding ±200ms.

4. Temperature warnings showing 'GPU[1] temp=92C' mean VRAM junction sensors have exceeded safe thresholds—reduce intensity or increase fan curve before hardware degradation occurs.

5. Hashrate variance greater than 15% across identical GPUs signals memory timing instability; recalibrate timings using AMD Radeon Software’s memory tuning panel.

Frequently Asked Questions

Q: Can SRBMiner-Multi mine Ethereum (ETH) and Ravencoin (RVN) at the same time?A: No. ETH uses Ethash while RVN uses KawPow—both are memory-hard but incompatible within one instance due to divergent DAG generation logic and memory access patterns.

Q: Why does the miner show 'No suitable pool found' even with correct credentials?A: This occurs when the configured algorithm does not match the pool's accepted protocol version—verify whether the pool supports RandomX v3.0 or requires v2.4 via their API endpoint /api/status.

Q: Is it possible to limit CPU usage during RandomX mining without affecting GPU threads?A: Yes. Use 'cpu-threads': 2 in the RandomX pool section while keeping 'gpu-threads': 8 unchanged for the GPU-bound algorithm.

Q: What causes 'CUDA_ERROR_LAUNCH_TIMEOUT' during multi-mining on RTX 4090?A: This results from Windows TCC timeout violation when GPU spends >2 seconds in kernel mode—disable WDDM mode in NVIDIA Control Panel and force TCC mode via nvidia-smi -r -d 0.