T-Rex 0.19.4

Bug fixes:

• (Windows) Not showing temperature / power consumption details when using v460.79 video drivers
• (Linux) Miner not starting with various TREX: xxxx (base64 encoded) errors
• --keep-gpu-busy not working under some conditions

T-Rex 0.19.3

• (octopus) Performance improvements 1-2% on some configurations
• (octopus, ethash) Reduce the amount of invalid shares
• Add --no-hashrate-report parameter to disable hashrate reporting to the mining pool
• Add --gpu-report-interval parameter removed in an earlier version
• Add --keep-gpu-busy parameter to continue mining even in case of connection loss (useful if pausing GPUs causes instability)
• Add --api-read-only parameter to forbid applying config changes via API and web-monitoring page
• Add CUDA Compute Capability 8.0 video cards support (NVIDIA A100)
• (UI) Various cosmetic changes

Bug fixes:

• (Windows) Miner crashes when using v460.79 video drivers

T-Rex 0.19.1

• (octopus) Minor performance improvements
  NOTE: use CUDA 10.0 build for 16 and 20 series cards for better performance

T-Rex 0.19.0

• Add octopus algorithm (dev fee 2%).
• Load NVRTC library only when required reducing RAM footprint for non-ProgPoW based algorithms
• Remove --gpu-report-interval parameter, now GPU stats summary table is printed every 30s. Other cosmetic changes

NOTE: according to user reports CUDA 10.0 build is faster than CUDA 11.1 on octopus for 16 and 20 series cards.

T-Rex 0.18.11

Bug fixes:

• (Windows) Miner is not displaying GPU power consumption for some video driver versions
• can't pair device with KEY error (see #89)
• Display GPU names when NVML library can't be loaded or disabled

T-Rex 0.18.10

• Add optional --no-strict-ssl parameter to disable certificate validation for SSL connections
• Change --fork-at parameter syntax to the following:
  Epoch number: <algo_name>=epoch:<epoch_number> (eg: --fork-at etchash=epoch:390 <-- use this one for upcoming ETC hardfork).
  Block number: <algo_name>=block:<block_number> (eg: --fork-at x16rv2=block:6526421).
  Time: <algo_name>=time:<YYYY-MM-DDTHH:MM:SS>. Time must be set in UTC+0.

T-Rex 0.18.9

• (ethash) Increase pool side hashrate by 1-2% by reducing the amount of stale shares the miner drops before sending to the pool
• (ethash, kawpow, progpow) Set worker name to %hostname% if not specified

Bug fixes:

• (ethash) Low pool side hashrate when mining ETH+ZIL due to incorrect difficulty parsing
• (ethash, kawpow, progpow) Miner crashes on low RAM multi-GPU rigs during DAG rebuild

T-Rex 0.18.8

• Add etchash algorithm for upcoming ETC fork

Bug fixes:

• (ethash) Compatibility with Mining Rig Rentals pool

T-Rex 0.18.7

Bug fixes:

• (API, Web UI) Miner crashes when changing intensity on the fly
• (ethash, kawpow, progpow) Worker name is not being passed for some mining pools
  From now on the username (-u) for these algorithms is no longer parsed as <wallet_address>.<worker_name>. If the pool requires the worker name to be sent, it has to be set with -w parameter. If -w is not set, the miner will set it to <algorithm>_worker0 by default.

T-Rex 0.18.6

• (ethash, kawpow, progpow) Reduce GPU memory consumption (by ~6MB)
• Cache DAG for epoch 0 when mining ETH+ZIL (run with --extra-dag-epoch 0 parameter)

Bug fixes:

• (ethash) kernels 4 and 5 generate incorrect shares if DAG size > 4GB (e.g. when mining ETC)
• (ethash) out of memory error in "zombie" mode
• worker parameter is ignored if placed under pools section in a config file
• Various web UI fixes
• The miner now reports hashrate every 2 seconds in benchmark mode