The demand on memory performance degrades with lower difficulty tests.
I would not trust anything running for short amount of time. EWBF for me died when he cannot even provide a binary over Google Drive (Trojan detection.)
DTSM became the miner of my choice thanks to a benchmark posted on
http://www.zcashbenchmarks.info/Thanks for the points - suggestions are welcome ofc.
I still stand with my first assessment that the main downfall of this client is the ability to communicate and state things clearly...
- Difficulty is not a float, but hex
do you mean difficulty should be mapped to a smaller range such that it's easier to read? Float is a datatype representing real numbers it's not related to the radix/base that's used for representation.
- Log is just an extension of the command line output
- Command line output is just a status message
I'm not sure why you have them impression it's 'just a status message' - it's actually much simpler to print events on a separate line - simply at the time they happen. Currently zm collects all the performance data and events for each GPU separately during a period of time. It calculates averages etc. and presents it in a compact structured way - otherwise you'll end up with a lot of unstructured spam on your ui.
- Telemetry replaces command line output (Good thing)
- Current log makes little sense without further tools. (I would appreciate more details)
What details are you missing? And how is it related to tools you would not need otherwise?
I think the log should contain:
- TCP communication details (Hostname, IP, Port)
It does.
- Share details (More than just + or *)
What per share details are useful and what's the use case for it? This would require to print each share on a separate line. If it's useful - I'll restructure the logfile format ofc.
- no performance ticks (Could be replaced by TXT file for benchmarking.)
It's in the logs such that you're able to check if something was affecting the performance on your rig in the past which helps debugging performance issues.
Anyway I am looking forward to new versions, but I am sticking with 0.5.7. Performance on 0.5.8 is worse for me CPU wise.
Now that's is a very useful information. If there are any regressions between the versions - pls report them - I can't fix them otherwise and they might appear in later versions too. How much did the cpu load increase for you?
