Common cause
Unstable overclocks
If reject rates increase after tuning memory or core settings, reduce the overclock and retest. A miner can look stable for a short period while still returning bad work under load.
Common cause
Network latency and timeouts
High latency, intermittent packet loss, or endpoint resolution problems can lead to stale or timed-out shares. Verify that the miner resolves the correct pool host and that the selected region is appropriate.
Checklist
What to test first
Check the endpoint, lower the overclock, test another network path if possible, and confirm the miner version is appropriate for the algorithm you are using. If the issue remains, compare the miner-side error timestamps with the public pool page and then contact support with the exact error output.