T-Rex – universal program for cryptocurrency mining. It supports many algorithms and we, as developers, do our best to make it as fast and easy to use as possible.
Developer fee is 1% (3% for Tensority). When the miner starts and stops mining, the developer board reports this to the console.
What’s new in T-Rex v0.18.2
- (ethash, kawpow, progpow) Up to 3% performance improvement in Windows for some configurations
- New parameter –watchdog-exit-mode to perform actions when the miner is restarted multiple times
- (UI) Print the current mining pool in the GPU hashrate table.
- (Windows) Running multiple instances of the miner is not possible since version 0.18.1.
- progpow-veil algorithm (for the upcoming VEIL fork)
- GLT algos: astralhash, jeonghash, padihash, pawelhash
t-rex -a ethash -o stratum+tcp://ru-eth.hiveon.net:4444 -u <wallet> -p x
t-rex -a kawpow -o stratum+tcp://us.rvn.minermore.com:4501 -u <wallet> -p x
t-rex -a mtp -o stratum+tcp://xzc.2miners.com:8080 -u <wallet> -p x
- For a complete list of supported options, run:
* (ethash, kawpow, progpow) Performance improvements on Windows for some configurations up to 3%
* New parameter “–watchdog-exit-mode” to perform actions when miner gets restarted multiple times
* (UI) Print current mining pool in the GPU hashrate table
* (Windows) Running multiple miner instances is not possible since 0.18.1
* “progpow-veil” algorithm (for VEIL’s upcoming fork)
* Add “progpow-veriblock” algorithm (for Veriblock)
* Add “progpow-veil” algorithm (for VEIL’s upcoming fork)
* Add “megabtx” algorithm (for Bitcore)
* Remove CUDA dependency for “kawpow”/”progpow” algorithms
* From now on “stratum+tcp” automatically selects the most commonly used pool protocol for a given algorithm (use “stratum1+tcp” and “stratum2+tcp” to override it)
* ethash optimisations (1-2% on some cards, see –kernel in the help file for more details)
* Add x33 algorithm (for Qureno)
* Add –worker parameter to mitigate the issue with usernames containing dots
* Improved compatibility with various mining pools / protocols
* Add “ethash” algorithm (for Ethereum). For Nicehash mining use “nicehash+tcp://” pool url prefix.
* Add “progpowz” algorithm (for Zano)
* Add “–pci-indexing” parameter: orders GPUs by PCI Bus Id (same as “–ab-indexing” but starts with 0)
* Add “–gpu-init-mode” parameter: enables sequential DAG generation to reduce load on power supplies
* Print hash rate if no shares have been found for more than 1 minute to indicate miner’s activity
* CUDA 11 build not working on some algorithms (x25x and similar), and with RTX cards
* “GPU is idle” error when generating DAG on low-end cards
* “nan(ind)” in share rejection GPU stats
* Log files not rotated correctly
* (UI, API) Show which GPU generated nonce/solution for share submission
* (UI, API) Show per-GPU rejected/total share statistics
* (Web monitoring page) Auto-saving web monitoring page preferences (dark theme, chart on/off etc.)
* Add “–low-load” cmd line parameter to optionally opt in for using CUDA kernels that put less stress on the GPUs where running regular kernels cause instability. See “t-rex -h” or “t-rex-help.txt” for more details. Currently this is only applicable to “kawpow” and DAG generation in particular.
* “can’t init enumerate, code -6” error when applying memory tweaks to P102/104/106 cards
* “can’t init 999, 0x800b01a” error when applying memory tweaks on Windows 7
* memory tweaks stuck after the miner termination
* memory tweaks becoming inactive in case of network issues or overclock adjustments
* setting intensity through the web monitoring page is broken
Add memory tweaks for GDDR5 and GDDR5X cards (10xx series)
–mt Memory tweak mode (default: 0 – disabled). Range from 0 to 6.
(kawpow, progpow) Faster DAG generation
Extranonce support for Nicehash
(Windows) Miner won’t start when using v450+ video drivers
* (KAWPOW) “kawpow validation failed” rejects (especially on high intensity and low difficulty)
* Use %HOSTNAME%, %USER% environment variables in parameters or config file
* Miner name/version not shown at some pools
* Add algorithms: kawpow (upcoming RVN fork), progpow, mtp-tcr
* Add SNI support for SSL
* Improve stat table (now showing percentage of rejected shares)
* (Windows) High CPU usage on NVIDIA drivers 431+
* (Linux) SSL connections not working on HiveOS
* (Linux) “T-Rex instance wasn’t validated” error despite firewall allowing all connections
* CUDA 9.1/9.2 versions of the miner will no longer work with RTX (Compute Capability 7.5+) cards
This is done because RTX cards either hash slower or don’t work correctly (depending on the algo) on these versions, producing “result for … does not validate on CPU” errors. Use the version built with CUDA 10.0 instead.
* Add x16rv2 algorithm (upcoming RVN fork)
* Add ability to auto-switch mining algorithm using new –fork-at parameter:
RVN fork: –fork-at x16rv2=2019-10-01T16:00:00
See the help file for more details
Tensority nonce fix for antpool (BTM)
(URGENT) Tensority max nonce limit (BTM)
“Unregistered stratum method” error on some pools
Add tensority algorithm (RTX cards, 3% dev fee) – used by Bytom coin
Add dedal algorithm back (removed in previous versions)
Add SSL support (url scheme should be ssl, e.g. “stratum+ssl://veil-eu.coinblockers.com:4244”)
Cosmetic changes and bug fixes:
Watchdog displays GPU restart stats
Miner restart automatically performed by the watchdog after an update no longer increases the WD restart counter
Fix miner shutdown API command when invoked with watchdog enabled
Improve MTP by 1-3% on some cards
Add ability to view T-Rex log file in the web browser (by default at
Fix benchmark (regression)
Add honeycomb algorithm (Beenode coin)
Fix “GPU … is idle” error when there is a pool connection problem
Significant x25x speed increase
Remove dedal algo
Use environment variables in config files. Use %YOUR_ENV_VAR% in any config file field and it will substitute it with the value of YOUR_ENV_VAR environment variable set in your system
Temperature limit bug (GPU got disabled if there was problems with NVML)
Show NVML errors and unsupported features
Truncate MTP share log message when using –protocol-dump
Fix start-up failure in some cases for CUDA 9.1 builds
Add x25x algo (will be used by SUQA/SIN after the fork)
Bug fixes (built-in watchdog):
Miner fails to start when using Windows task scheduler or Awesome Miner
Miner restarts on Linux/Hive OS
Preserve a single log file across miner restarts
NOTE: Using built-in watchdog is encouraged. If you experience any issues with it, please let us know
Add MTP algo (Zcoin)
Add Tesla P100 support
Add built-in watchdog (can be disabled using –no-watchdog parameter)
Default network timeout increased from 3 to 5 minutes
New parameter –hide-date to hide date part of the timestamp in the console
Prevent Windows from going to sleep mode while T-Rex is running
“does not validate on CPU” error on RTX cards (CUDA 10 builds)
Long dev fee sessions if system clock “jumps”
GPUs keep hashing even if the pool connection is lost
Add support for GIN (x16rt)
1) New parameter –ab-indexing to enable Afterburner-like GPU indexing
2) Web-monitoring page enhancements
CentOS 7 compatibility issue
Add algos: x16rt (Veil), sha256q (Pyrite)
Minor performance improvements for x16r, bcd, x22i and other x16-like algorithms
Multiple web monitoring page improvements
* Switch between multiple config files dynamically (put your *.config files into the miner directory)
* Dark theme
Linux watchdog is now available. Download at
cURL/SSL compatibility issue on newer Linux distros (Ubuntu 18, Debian 9)
Miner wouldn’t start if NVML wasn’t installed
Performance improvements: +2-3% for x16 family of algorithms (x16r, bcd, x17, x22i etc.)
Add config parameter: –no-nvml – Disable NVML GPU stats.
Display miner version next to “Uptime” info.
Fix “result for … does not validate on CPU!” errors on x21s and dedal
Add algos: astralhash, jeonghash, padihash, pawelhash (used by GlobalToken, can be mined at
Performance improvements: a few percent to x16r family of algorithms, skunk, x22i
Highlight GPU temperatures in green/yellow/red. Temperature ranges are configured with –temperature-color parameter
Invalid shares on Dedal for some hashorders
Not identifying a found block when mining x22i
Performance improvements: x22i +5%
Performance improvements: x22i +1-4%
New algo: x21s (Ritocoin).
New config parameter: –sharerate-avr – Sliding window length in seconds used to compute sharerate (default: 600).
New algo: dedal
Improve x22i: +1-3%
Fix skunk performance issue introduced in 0.8.1
Performance improvements: x22i +10-20%
Miner hangs on start up when it doesn’t receive authorize message response
New algo: geek (GeekCash coin)
Performance improvements (1-3%) of x16-like algorithms for GTX/RTX cards.
Added watchdog. Replace “t-rex.exe” with “watchdog.exe” in your bat file in order to use it.
Watchdog lets you update T-Rex using the monitoring web page or perform auto-updates if launched with “–autoupdate” parameter, and also restarts the miner if it hangs for any reason.
Default value for “-N” paramater increased from 30 to 60 seconds.
Added “shares per minute” stats read-out.
“stratum+tcp://” pool URL prefix is no longer mandatory
Reduce CPU load when using large “-N” parameter values
“Clean jobs” feature removed in the previous release is back
Hung on Ctrl+C in Windows
Minor performance improvements for skunk and x16 family of algorithms for Pascal and Turing cards
Crashes 20 sec after dev fee session when using -r 0
Temporary backed out “clean job” feature that was causing rejects on some pools e.g. beepool and bitpoolmining
Fix average hashrate reporting when using large -N values (over -N 4294)
Improve skunk, phi, polytimos; small improvements for x16r/s etc on some configurations.
Performance improvements for Turing GPUs (20xx series)
Add functionality to return to the main pool after mining on failover ones (see –back-to-main-pool-sec launch parameter)
Handle client.reconnect stratum messages (used by MiningRigRentals)
Handle clean jobs stratum flag correctly (resulting in fewer stale jobs) – this feature is supported by coinblockers.com
Add file handler to API. Allows to create custom files in miner directory via API.
Add functionality to keep your failover pools separately in a config file. May be helpful if you need to load pools from file. (t-rex -c pools.json -a x16r <your parameters>)
Remove assert on NULL rpath in Linux.
Fix average hashrate reporting when using large -N values.
Added algo: x22i
Note: if the miner crashes on x22i, decrease your overclock, the algorithm is quite “unstable”
Added algos: sha256t, timetravel
Replace “–exit-on-fail-shares” parameter with “–reconnect-on-fail-shares”
–temperature-start: When a GPU cools down to this temperature the miner starts using it again
–exit-on-fail-shares: Now replaced with –reconnect-on-fail-shares
-r (–retries): if set to 0, miner exits immediately upon the first network failure
–cpu-priority: Set process priority (default: 2) 0 idle, 2 normal to 5 highest
Allow C-style comments in config files (see config_example). Can be used to disable failover pools temporarily etc.
A few cosmetic changes.
New API handler “config”.
Some miner parameters can now be changed and saved on the fly.
Please see the help file for additional details.
“bad weak ptr” error
Share rejects were not displayed (Suprnova, Aikapool, possibly some others)
Command line arguments did not always override config file
Speed improvements: there’s ~1% improvement in x16 family of algorithms on some configurations.
Add support for Volta (Titan V) and Turing (20xx) cards – not optimised yet. Download CUDA 10 version to run on 20xx cards.
–temperature-limit: When a GPU reaches the temperature limit (in Celsius) the miner stops using it and keeps mining with the rest of GPUs
–exit-on-fail-shares: Forces miner to immediately exit on N successively rejected shares (default: 10)
Command line arguments now take precedence over config files. If an argument is set in both places, the value from cmd line overrides the config one.
New HTTP API handler that allows you to control the miner via a web browser on the fly. By default is accessible at
Stability fixes, especially on x16r/s and bitcore
CUDA version reported by the miner is now the version that the miner has been built against rather than the one installed on user’s machine
Redirection of console output to stdout on Windows with no buffering in place.
Add miner efficiency (Hashes per Watt) to console output and API
Print “Submitting found block!” message when a block is found and return this info in API
Speed improvements: x16r/s, x17, c11, bitcore, bcd, sonoa – 1-2%
Add functionality for setting failover pools in cmd line by specifying -o, -u and -p arguments multiple times.
New cmd line arguments:
–gpu-report-interval: controls how often GPU report stats are displayed (by default every 5th share).
–no-color: disable coloured output
–time-limit: maximum time to mine before exiting the program.
–quiet: disable GPU stats output
Add ability to change averaging hashrate window, enable protocol dump mode, and change GPU report interval via API or web browser on the fly (without stopping the miner) – see the help file.
Invalid shares on x16r
Long log path issue on Linux
“extranonce subscribe timed out” on suprnova pools
Showing configured pools upon start up
Displaying colour representation of hashorders for x16r/s
Miner uptime is displayed as part of GPU stats report
Speed improvements: bcd – 5%
Cosmetic change: the miner will not show duplicate hashorder messages for x16r/s, bitcore, and they are no longer labeled as WARN
To disable API altogether set –api-bind-http / –api-bind-telnet to 0.
HTTP API: new control command “shutdown” to quit the miner (read help file for details)
New algo: bcd
x16r/s, bitcore, renesis, hmq1725 – 2-3%
x17, c11, sonoa – 1-3%
New algo: hmq1725
Improved x16r/s, x17, sonoa, c11, bitcore, renesis – 5-15% depending on algo and cards
Bug fix: “hashrate-avr” parameter is ignored when using json config
New algos: balloon, polytimos, skunk
Improved bitcore, c11, sonoa, x16r/s, x17 by a few percent. Renesis, phi – about 3-5%
New algos: bitcore, sonoa. Dropped phi2 support.
Slight improvements (few %) of all algorithms except lyra2z.
Added extra shares submit functionality. Reduced the number of rejects on pool side (both resulting in increased poolside hashrate).
Fix for multiple scenarious where the miner used to freeze.
Fix “doesn’t validate on CPU” errors for phi/renesis algos.
Display pool latency when submitting shares.
Support decimal intensity values (e.g. 21.3)
Highlight accepted/rejected shares using different colours (Windows version)
Display GPU full name/temperature/power consumption/fan details.
New command line option -P (stratum protocol dump).
-N option now uses seconds rather than samples as a unit of measurement. E.g. -N 30 means a sliding window of 30 seconds is used to calculate average hashrate.
Added support for Compute Capability 5.0 cards (e.g. 750, 750ti).
Added GPU full name/temperature/power consumption/fan details.
Added current pool details (name, difficulty etc)
Faster renesis, hsr, x16r, x16s, x17, c11. Few % improvement across all of them.
0.5.6 (for those upgrading from 0.5.1)
Added new algos: x16r, x16s, renesis.
Faster hsr, x17.
JSON API – return active pool information
Added command line parameter “-N” – average hashrate display (number of samples in sliding window)
Miner displays a notification in the console if the new version is available
Miner hanging on “Graceful shutdown”
Config file – miner wouldn’t start if no intensity specified
Telnet API (ccminer compatible) – inconsistent newline character
Hung issue (after “ApiServer: stopped listening” and connection loss)
Immediate reconnect upon connection loss
Incorrect initialisation when invalid command line argument supplied (now the miner won’t start in that case)
Faster phi2, lyra2z, x17, c11.
Added new algo HSR.
Added API (HTTP JSON [preferable] and telnet ccminer compatible) – pending integration with monitoring software (ethmonitoring, awesome miner etc.)
Multiple bug fixes, new config options, failover pools.
JSON config file support. Multiple bugfixes.
Miner source code is rewritten from scratch and no longer based on ccminer.
Faster Lyra2z, Tribus and slightly faster Phi1612.
Warning: If using Lyra2z in this version you will likely need to lower your overclock settings and probably also increase the power limit.
Added Lyra2z and Tribus.
Initial release. Phi1612 added.
This post is also available in: Русский (Russian)