Closed Olegik-ah closed 2 years ago
it seems that the bug with switching is present in version 1.21 too. Here is the screen from another rig 8x1660S TBM 1.21:
Can you show us your launch config?
The 1660S only has 6gb of memory and TBM is storing both dag buffers in gpu memory. I think the increased epoc is causing this. But We will check if there is something we can do to fix it. Buffers shouldn't be stored in memory if there is not enough memory to launch them.
Thanks for your reply!
This is hiveos. All default parameters
First log (Cpu verification failed) from one rig (7x1660s) TBM 1.23. Errors when switching from ZIL to ETH. Always reproduced on my rig (every switch) on 1.22/1.23. There is no error if running 1.21 on this rig.
Second log (out of memory) from another rig (8x1660s) TBM 1.21. Errors when switching from ETH to ZIL. Not always reproduced. Approximately once every 3-5 hours.
Perhaps these errors are related, perhaps not. Previously, rigs worked on Gminer/PhoenixMiner many days (10+) without errors.
Found a problem in the code that fetch the dag buffer after zil pow. fixed in v1.24
Thanks! And when the release is planned?
v1.24 has been released now
Hello! TBM 1.21 works perfect, but if i update to 1.22 or 1.23 i have got continuous errors "Cpu verification failed. Share not sent to the pool" after switching back from ZIL to Ethash. Only miner restart helps... It seems broken DAG file. Maybe the reason is incorrect overclock settings... But i cant understand witch gpu is the reason - afrer switching back from ZIL to Ethash ALL gpus give me incorrect shares. I'am using hiveos (latest release, 5.4.0-hiveos #140 kernel) and nicehash pool. Algo Ethash. Log in attachment teamblackminer.1.log.gz