Bitcoin Forum
May 20, 2022, 10:31:55 AM *
News: Latest Bitcoin Core release: 23.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1] 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 ... 146
  Print  
Author Topic: [ANN] TeamRedMiner 0.9.4.6 - Ethash/FiroPow/Kawpow/Etchash/Autolykos2 and More  (Read 205723 times)
todxx
Member
**
Offline Offline

Activity: 175
Merit: 76


View Profile
October 30, 2018, 06:07:01 AM
Last edit: May 05, 2022, 10:26:43 PM by todxx
Merited by scryptr (10), rednoW (10), frodocooper (7), CjMapope (5), suchmoon (4), Ursul0 (3), heavyarms1912 (2), alucard20724 (1), PIOUPIOU99 (1), RuMiner (1), N2DCRYPT (1)
 #1

teamredminer v0.9.4.6

This is an optimized miner for AMD GPUs and Xilinx FPGAs created by todxx and kerney666.

This is a public beta release and only available through the download links below

We've chosen to release v0.9.4.6 as a public beta to gather more user feedback without triggering automatic updates in mining distros. The main reason is that we still have some outlier configurations that don't respond as expected to the new R-mode. When it works as intended, which is maybe 95-97% of all rigs we've tested on, things generally run very well. If you don't use the new R-mode, or you're a Windows miner, you still have an upside running this new version for the "smooth power" rewrite described below.

This rewrite is targeting ethash family algorithms. There are two distinct parts:
  • Smooth power transitions. This affects all gpus, and both windows and linux. This is mainly a (theoretical) stability feature that improves power draw on a very small scale level, making sure the gpu activity and resulting power draw is as smooth as possible, avoiding psu slew rate spikes. The nice side effect is that it in most cases also improves hashrate slightly, on all gpus.
  • R-mode. This is our most advanced mode work so far. It is linux only, like the previous C-mode for Radeon VIIs. R-mode is applicable for all non-Polaris gpus. It also needs special linux kernel boot parameters set, like the Radeon VII C-mode. R-mode will be used by default when the necessary linux kernel parameters are set. A huge difference between the new R-mode and previous B/C-modes is that the R-mode does not degrade over time as the dag grows. The efficiency improvements on many gpus are significant. Biggest winners are 5700XT/5700/5600XT. Please(!) read the R-mode tuning guide for more details and before switching to R-mode.

Download links:

HiveOS test instructions

You need a modern Hive image to run the new version in R-mode. Running on older linux kernels will not end well, we experienced hard crashes in early testing. Our testing was mainly performed on reimaged installations of 0.6-212, 0.6-216, and 0.6-217. You're free to test whatever version you're currently running, but be prepared for hard hangs potentially requiring power cycling. Older kernel versions might also seem to run ok but then result in a higher-than-usual amount of stale pool shares.

To simplify miner upgrade, we've taken the time to build a custom miner package for Hive. Set up a new flight sheet as follows:
  • Create new flight sheet.
  • Select ETH and your wallet of choice, then some dummy pool.
  • In the miner search textbox, type "custom", then click "Show all", and select "Custom".
  • Copy/paste https://dl.teamredminer.com/beta/trm_beta-0.9.4.6.tgz as the installation url. The miner name will be derived automatically.
  • Depending on your pool, choose a Wallet and worker template. %WAL%.%WORKER_NAME% is usually a good choice.
  • Type your pool url(s) in the "Pool URL" field. This must be done - the pool you selected in step 2 will not be used.
  • In "Extra config arguments", add "--kernel_vm_mode=RR" and any other TRM arguments you want to add.

The --kernel_vm_mode=RR argument runs the miner's integrated support for setting the necessary kernel boot parameters using a bundled script. The argument value "RR" means "(R)eboot if necessary and set (R)-mode". You can keep the argument in the flight sheet. This way the R-mode arguments will be restored automatically after a Hive update.

MMP OS test instructions

For MMP, simply download the miner package and overwrite the existing teamredminer binary. MMP should run with the amd opencl driver 21.40.2 or 21.50.2, 20.40 has not worked well in our tests. Procedure:

Step 1: upgrade miner binary
  • Set up TRM 0.9.4.2 (or some other version) running on the rig.
  • Open up a shell on your MMP rig using the remote console or a ssh session.
  • Type or copy/paste the following commands:
Code:
sudo su -
cd /opt/mmp/miners/teamredminer
wget https://dl.teamredminer.com/beta/teamredminer-v0.9.4.6-linux.tgz
tar xvzf teamredminer-v0.9.4.6-linux.tgz
mv teamredminer teamredminer.old
mv teamredminer-v0.9.4.6-linux/teamredminer teamredminer
rm -rf teamredminer-v0.9.4.6-linux
  • Optional: restart the miner, check logs to verify the version running is 0.9.4.6.

Step 2: enable R-mode support
  • Disable the "System watchdog" for your rig.
  • Open up a shell on your MMP rig using the remote console or a ssh session.
  • Run "mmp update".
  • Run "mmp upgrade" (necessary to get the enable-trm-boost command).
  • Let the system reboot.
  • After reboot, open a new shell.
  • Run "enable-trm-boost". Confirm that you see "ENABLE_VMSIZE[TRM BOOST] is set to enabled but its currently disabled - enabling! [WARN]", if not rerun "enable-trm-boost" again.
  • Reboot with the "srb" command or "sudo reboot".
  • R-mode should now be available and used by default on supported gpus.

Other distros

We might include more distro-specific descriptions here further on, but the general way of testing the new version is to loosely follow the MMP OS manual upgrade path:
  • Set up an older version of TRM mining, for example v0.9.4.2.
  • Drop to a shell/terminal on your rig and locate the directory where the TRM binary is stored.
  • Download the beta version using
Code:
wget https://dl.teamredminer.com/beta/teamredminer-v0.9.4.6-linux.tgz
  • Unpack the downloaded file:
Code:
tar xvzf teamredminer-v0.9.4.6-linux.tgz
  • Move the current TRM binary out of the way:
Code:
mv teamredminer teamredminer.old
  • Move the downloaded binary:
Code:
mv teamredminer-v0.9.4.6-linux/teamredminer teamredminer
  • Manually edit /etc/default/grub and add the necessary kernel boot parameters (see the R-mode tuning guide).
  • Change your tuning before rebooting. We recommend lowering core clk -200 MHz on all gpus except the Radeon VII family.
  • Reboot. You should now be mining in R-mode. Increase core clk slowly again until you hit your old hashrate.


Download is available in the github releases section.

TRM official website: https://www.teamredminer.com

TRM discord server: https://discord.gg/RGykKqB

Below is a list of mining operating systems and management software that have built-in support for teamredminer:

This miner supports a range of algorithms.  Please see the list below for details.
The miner is configured via command line only, please run with the --help option to print a short help message for how to use the command line options.

GPU Support

GPUs supported and tested:
  • Navi - RX 5700(XT)/5600(XT)/5500(XT) for supported algos in the table below.
  • Big Navi - RX 6900XT, RX 6800(XT), RX 6700(XT), RX6600XT - same support as for Navi.
  • Vega - RX Vega 64/56, Vega FE, Radeon VII (Vega 2)
  • Polaris - RX 580/480/570/470/560/460/550
  • Fiji - R9 Fury/Fury X/Nano, MI8 (supported but with very limited testing).
  • Tonga/Antigua - R9 285/285X/380/380X, W7100, S7150 (beta support from 0.8.2. Only ethash+kawpow available.)

Supported GPU algorithms and their respective dev fees:
GPU AlgorithmFee
Ethash on Polaris GPUs0.75%
Ethash on all other GPUs1.0%
Kawpow2.0%
Verthash2.0%
Autolykos22.0%
Ton1.0%
Nimiq2.5%
Cryptonight R2.5%
Cryptonight v8 upx22.5%
Cryptonight v8 turtle2.5%
Cryptonight v8 half2.5%
Cryptonight v8 double2.5%
Cryptonight v8 rwz2.5%
Cryptonight v82.5%
Cryptonight heavy2.5%
Cryptonight haven2.5%
Cryptonight saber2.5%
Cryptonight conceal2.5%
Chukwa-512 (Turtlecoin)2.5%
Chukwa-1024 (Turtlecoin)2.5%
x16r2.5%
x16rv22.5%
x16s2.5%
x16rt2.5%
MTP2.5%
Cuckatoo312.5%
Cuckarood292.5%
Lyra2rev32.5%
Lyra2z3.0%
Phi23.0%

Some algorithms are not supported on some GPU architectures and/or drivers.  Below is the compatiblity table:
Navi
Vega
Polaris
Fiji
Tonga
Ethash
Y
Y
Y
Y
Y
Kawpow
Y
Y
Y
Y
Y
Verthash
Y
Y
Y
Y
N
Autolykos2
Y
Y
Y
Y
Y
Firopow
Y
Y
Y
Y
Y
Ton
Y
Y
Y
Y
Y
Nimiq
Y
Y
Y
Y
N
Cryptonight R
N
L
L
L
N
Cryptonight v8 upx2
N
L
L
L
N
Cryptonight v8 turtle
N
L
L
L
N
Cryptonight v8 half
N
L
L
L
N
Cryptonight v8 double
N
L
L
L
N
Cryptonight v8 rwz
N
L
L
L
N
Cryptonight v8
N
L
L
L
N
Cryptonight heavy
N
L
L
L
N
Cryptonight haven
N
L
L
L
N
Cryptonight saber
N
L
L
L
N
Cryptonight conceal
N
L
L
L
N
Chukwa-512
N
L
L
L
N
Chukwa-1024
N
L
L
L
N
x16r
N
Y
Y
Y
N
x16rv2
N
Y
Y
Y
N
x16s
N
Y
Y
Y
N
x16rt
N
Y
Y
Y
N
MTP
Y
Y
Y
Y
N
Cuckatoo31
N
Y
Y
Y
N
Cuckarood29
N
Y
Y
Y
N
Lyra2rev3
N
L
L
L
N
Lyra2z
N
L
L
L
N
Phi2
N
L
L
L
N

Support legend:
  • Y = Supported
  • N = Not supported
  • L = Limited support: algos are supported on windows and linux with amdgpu-pro drivers, not supported on ROCm drivers.

FPGA Support

FPGA Devices supported and tested in Linux (Windows is not currently supported):
  • Xilinx Varium C1100
  • SQRL Forest Kitten 33 - performance limited by product design, see guide for details
  • Xilinx/TUL/Osprey U50C/ECU50

Supported FPGA algorithms and their respective dev fees:
FPGA AlgorithmFee
Ethash4.0%


The miner reports GPU hash rates every 30 seconds.  These are the full GPU hash rates before dev fee deduction (your pool hashrate will be slightly lower).

The miner includes a read-only api based on the sgminer-5.5 API.  Both the json and text formats are supported.  For more details, we refer to the sgminer api documentation.
The miner also includes a Claymore miner compatible API with support for a subset of the API.

For reporting bugs and/or for features requests, please open an issue on this project's github issue tracker.

For example command lines please see the batch/shell scripts in the miner download packages.
Miner command line options are available here


Release Notes
v0.9.4.6
Changes
  • GPU:  R-mode introduced. See separate documentation. Applicable for Vegas/VIIs/Navi10/Big Navi.
  • GPU:  Smooth power transitions for ethash family algos.
  • GPU:  Added support for multiple --api_listen and --api_listen2 endpoints (up to 8 total).
  • FPGA: Fixed crash bug with > 16 fpgas running in a single miner instance.
v0.9.4.2
Changes
  • GPU:  Fixed eth+ton issue that could lead to a higher eth shares stale rate. Primarily for large gpus (6800/6900XT) with aggressive tuning.
  • GPU:  Internal split of binaries helping some Vegas and Navis with crash issues on ethash from 0.9.2 and forward.
  • FPGA: Added initial support for TUL TH53/TH55.
v0.9.4.1
  • Replaced by 0.9.4.2 due to bug with dual mining device selection.
v0.9.4
Changes
  • GPU:  Navi/Big Navi rewrite for eth+ton, increased hashrates on both algos and more stable setup.
  • GPU:  Navi/Big Navi eth+ton rigs with stale eth issues should be fixed.
  • GPU:  Pool outage for dual algo now results in eth mining only instead of pausing. This will reduce crashes.
  • GPU:  Fixed a potential deadlock when mining eth+ton. Rigs that have gotten strange "crashes" should upgrade, especially when coupled with a network or pool outage.
  • GPU:  Added --dual_tuner_step and --dual_tuner_period to configure the dual tuner accuracy.
  • GPU:  Fixed race bug for ethash where gpus could accidentally build a dag for epoch 0 at startup.
  • GPU:  RaveOS fix for running the TRM wss proxy for Nimiq and certain TON pools.
v0.9.3
Changes
  • GPU:  Added Polaris support for dual ETH+TON (full eth + 600-750 MH/s ton per gpu). All AMD gpus now supported.
  • GPU:  Updated DUAL_ETH_MINING.txt with information for all AMD gpus.
  • GPU:  Added support for TON Pool (ton-pool.com). See TON_MINING.txt for details.
  • GPU:  Fixed bug that could cause stale shares on gpus disabled for dual mining.
v0.9.2.1
Changes
  • GPU:  Fixed a critical bug for Polaris and Vegas on Windows and Linux amdcl2, only reaching 50% poolside hashrate.
  • GPU:  Added support for selecting dual algo devices using a "-d x,y,z,..." argument inside the --ton .. -ton_end clause.
v0.9.2
Changes
  • GPU:  Added TON algo for single algo mining on all gpu generations (see TON_MINING.txt).
  • GPU:  Added dual ETH+TON mining for Navi and Big Navi gpus (see DUAL_ETH_MINING.txt). Vega and Polaris upcoming shortly.
  • GPU:  Added dual mining tuner based on scoring weights (see --dual_tuner_weights).
  • GPU:  Faster initial ethash tuning on startup.
  • FPGA: Added --fpga_max_jtag_mhz for limiting the JTAG communication frequency used.
  • FPGA: Fixed DNA for vu35p - now matches Vivado hardware manager and NextJTAG.
v0.9.1
Changes
  • FPGA: Updated FPGA_GUIDE.txt with new devices, voltage tuning, and more.
  • FPGA: Added U50C/ECU50 FPGA suppport.
  • FPGA: Added custom SC firmware for C1100 and U50C.
  • FPGA: Added TRM SC firmware programming support, see --fpga_update_fw.
  • FPGA: Added voltage control for C1100 and U50C (using new firmware), see --fpga_vcc_int, etc options.
  • FPGA: Added support for additional FK33 sensors (hbm voltage, current sensors).
  • FPGA: Added support for DNAs in the --fpga_devices option.
  • FPGA: Added --fpga_allow_unsafe option for disabling safety limits.
  • FPGA: Added timeout for ethash DAG gen on FPGAs.
  • GPU:  Added --eth_ignore_abort_fail to disable intensity adjustment due to failed aborts.
  • GPU:  Added better handling of dead gpu logging in corner cases that previously didn't mention a specific gpu.
v0.9.0
Changes
  • Added initial FPGA ethash support.  See FPGA_GUIDE.txt for details.
v0.8.7
Changes
  • General: added offline benchmark mode for (almost) all algos (see --benchmark).
  • Network: improved situations with the "Dev pool failed to connect." error message appearing.     Users in China should preferably run with "--dev_location=cn".
  • Network: added experimental support for DNS-over-HTTPS (see --dns_https and --dns_https_sni).
v0.8.6.3
Changes
  • Autolykos2: emergency release to support larger pad size from block 614400.  Some GPUs (mainly Vegas) will need higher core clocks to achieve the same performance as previous versions due to larger pads breaking some previous optimizations.  More optimizations to come in future releases.
  • Autolykos2: added argument --autolykos_ignore_diff for certain pools that aren't compatible with the ERGO reference miner pool implementation.
  • Progpow: fixed wrong --help text for --prog_micro_tune.
v0.8.6.2
Changes
  • Firopow/MTP: added a synthetic algo 'mtp_firopow' that will use the correct algo given the system time and shut down the miner at the time of the Firo fork on Oct 26.
  • Ethash/progpow: split the ethash and progpow algos into separate binaries, mainly for some Vegas that ran into stability issues going from 0.8.5 to 0.8.6.
  • Autolykos2: added support for Tonga.
  • Advanced SSL usage: changed the default SSL behavior to not provide a SNI hostname during handshake. See the updated --help section on the -o argument for how to force a SNI hostname being sent.
v0.8.6
Changes
  • Firopow: new algo added for Firo's upcoming fork (see start_firo.sh/bat). Only testnet mining is available before the fork, see start_firo_testnet.sh/bat.
  • Firopow/kawpow: rewrote tuning guide (see KAWPOW_FIROPOW_TUNING.txt).
  • Firopow/kawpow: slight hashrate improvements, mainly from choosing full gpu tuning by default and adding a micro-tuning mechanism for Polaris gpus (see new argument --prog_micro_tune).
  • Autolykos2: added high score support and display of submitted share difficulty.
  • General: fixed Radeon VII support on recent win drivers. NOTE: HBCC must be DISABLED for all Vegas.
  • General: fixed broken fan control for Navis on some driver versions.
  • General: improved Windows compute mode enable with automatic elevation (with --uac) and restart of gpus.
v0.8.5
Changes
  • General: added Windows clocks/voltage/built-in timings control (beta functionality). See section in --help and CLOCKS_VOLTAGE_CONTROL.txt.
  • General: fixed some issues with Windows fan control, mostly for Big Navi gpus.
  • General: added mem temp limits (stop/resume), see --mem_temp_limit and --mem_temp_resume.
  • Ethash: added support for forcing ethash pool hashrate reports (see --eth_hash_report).
  • Ethash: fixed hashrate reports for Crazypool when using failover pools.
  • Autolykos2: added extranonce subscription support for e.g. Nicehash.
v0.8.4
Changes
  • General: Added fix for failure to load GPU kernels on newer drivers, e.g. Big Navi on Adrenalin 21.6.x.
  • Emergency patch: fixed Haven mining generating mostly hw errs after the recent hard fork.
v0.8.3
Changes
  • General: added Autolykos2 algo for ERGO (very good target for Vega 56/64, see the AUTOLYKOS_TUNING.txt guide).
  • General: improved and simplified dual ZIL mining for ethash/kawpow/verthash/autolykos2 (see the new DUAL_ZIL_MINING.txt guide).
v0.8.2.1
Changes
  • Bugfix: fixed windows issue with intermittently reading zero gpu stats and blasting fans (bug in 0.8.2).
  • Verthash: added --verthash_max_uploads=N to control the upload of the verthash table to gpus at startup. Some chipsets get issues running > 4 gpus concurrently. Typical error is that multiple gpus die immediately at startup when running the full rig, but work fine if you only run 3-4 gpus.
  • Verthash: fixed support for 550 2GB gpus.
v0.8.2
Changes
  • General: added Verthash algo for Vertcoin.
  • General: added Tonga and Antigua support for ethash and kawpow.
  • General: extended support to 24 gpus (previously max 16 gpus possible).
  • General: better handling of driver issues related to gpu clocks/temps stats.
  • General: now showing sensor power in stats output.
  • General: fixed device name parsing in ROCm 4.1 and newer win drivers.
  • API: added sensor power, jnct temp and mem temp to API output.
  • Ethash: added --eth_micro_delay for weaker psus (see help section for details).
  • Ethash: fixed watchdog DEAD gpus on small gpus building kawpow DAG.
  • Kawpow: now working fully with MiningPoolHub, regardless of bad seedhash values.
v0.8.1.1
Changes
  • Patched v0.8.1 with new devices ids for 6700XT support.
v0.8.1
Changes
  • General: basic Big Navi support added (ethash, kawpow, mtp, nimiq supported algos).     See the ETHASH_TUNING_GUIDE.txt for a short section on Big Navis.
         Driver support missing on Windows for fan control and mem temp.
  • Ethash: display gpus now use A-mode by default to prevent failed allocation and low hashrates.
  • Stats: handling of negative temperatures under Linux.
  • Eth+zil: added new pool strategy 'min_epoch' for dual pool connection switching (see start_zil_eth.bat/.sh example script).
v0.8.0
Notes

Biggest release in a long while with rewritten ethash kernels and new mining modes for all gpu types!

Users are highly(!) recommended to take a few minutes to read the 0.7-to-0.8 migration guide and the new ethash tuning guide. Key highlights:
  • Polaris: Efficiency and slight hashrate increase. B-mode reintroduced for added hash. B-mode must be enabled with --eth_aggr_mode or --eth_config=Bxxx.
  • Vega 56/64: greatly improved base kernel for efficiency. New B-mode that can shave off additional 1-2W on top of the A-mode kernel. B-mode must be enabled manually with --eth_config (--eth_aggr_mode does not apply). Tuning numbers have changed - do NOT keep your old static --eth_config values.
  • Radeon VII: huge boost with its new C-mode but requires a special Linux setup. Can now do 100 MH/s on most air cooled VIIs. See tuning guide.
  • 5700/5700XT: can shave off as much as 8-9W(!) of power using the new B-mode and dropping core clk+voltage. B-mode now the default mining mode. Unless you retune your core clk+voltage you will see a tiny power draw increase instead and not benefit from the upgrade, so read the migration guide.
  • 5600XT: new B-mode has a much smaller effect. A-mode remains the default mining mode. See new tuning guide for more details.
  • The dag cache is NOT compatible with the new B/C-modes. ETH+ZIL switchers have to choose between caching the epoch 0 dag and using the new mining modes.
  • Ethash 4GB kernels NOT rewritten in this release, performance remains the same as in 0.7.x.
  • See the migration guide for hashrate and power draw comparisons between 0.7.21 and 0.8.0.
Changes
  • Ethash: VII kernel rewrite and new C-mode with boost feature (see guide).
  • Ethash: Navi kernel slight optimization and new B-mode for lower core clock and power.
  • Ethash: Vega kernel rewrite and new B-mode for lower core clock and power.
  • Ethash: Polaris kernel rewrite and new B-mode for slightly higher perf.
  • Ethash: added share processing timeout and default for Binance pool (see --pool_share_limit_ms).
  • Claymore API: fixed leak that stopped serving requests after 32k api calls.
  • Claymore API: added password support (see --cm_api_password).
  • Logging: added log rotation support (see --log_rotate).
  • Logging: log files now contain the miner welcome message so the version is stored.
  • Kawpow: now mining ok at MiningPoolHub even though their seedhash is broken.
  • Fan control: added min/max fan speed range (see --fan_control).
  • General: added argument to turn off duplicate pci bus id filtering (see --allow_dup_bus_ids).
v0.7.22
Notes
  • NAVI KERNEL REWRITE! Over the last month we've been working on new kernels. The first one released is for Navi. The main feature is lower power consumption, hashrate will remain about the same but depends somewhat on clocks. NOTE: please let the miner retune any existing --eth_config arguments for all Navi gpus, the new values will be significantly lower.
  • Claymore API now supported, meaning you can use EthMan to monitor TRM rigs.
Changes
  • Ethash: Kernel rewrite for Navi.  Should now be more stable and use less power.  Vega/Polaris still in the works.
  • General: Slightly reworked init procedure again to address some rigs running better on <= 0.7.18 than >= 0.7.19.
  • General: Added Claymore compatible API, see the --cm_api_listen option.
v0.7.21
Notes

Quick release that addresses situations where Ethash with capped DAG on 4GBs would crash after 5-10 mins when running at 4078-4080MB.
Changes
  • Ethash: improved probability for high (4078-4080MB) ethash 4GB capped allocation running stable over time.
v0.7.20
Notes

ETH is now approaching the start of the 4GB death with the upcoming switch to epoch 382. TRM was the first miner to add support for mining with 4GB gpus with a partial DAG, here called "extended mining". However, it has always required manual configuration. This release adds a default conservative cap at 4072MB for DAG storage on 4GB gpus for out-of-the-box execution of the miner. For maximum performance, miners are still advised to manually tune their rig(s) with the --eth_4g_max_alloc argument. Most rigs runs stable with a higher value than 4072, which makes a significant difference for upcoming ETH epochs.

For more detailed instructions, our Ethash 4GB guide has also been updated and is available at https://github.com/todxx/teamredminer/blob/master/doc/ETHASH_4GB_HOWTO.txt
Changes
  • Ethash: added default capped DAG allocation for 4GBs at 4072MB (see --eth_4g_max_alloc).
  • Ethash: bugfix for crashes using --eth_dag_cache on 4GB gpus.
v0.7.19
Notes
  • Ethash solo-miner share high-score list added, use --high_score to enable.
  • Ethash single buffer dag for a small power save. Only available on recent drivers and should in theory always be an improvement. It is not verified 100% stable yet and therefore not the default in this version. Enable with --eth_dag_buf=A to test.
  • Miner gpu initialization procedure rewritten. The init procedure should now hopefully be much more stable, especially on Navis in general and even more so for 5600(XT)s.
Changes
  • General: added adjustable ratio support for quota and load_balance pool strategies (see --pool_ratio).
  • General: rewrote the miner init procedure to be more stable in general.
  • General: fixed issue where output blocking could cause mining work to stop.
  • General: reduced main binary size.
  • Ethash: reintroduced single DAG buffer support for recent drivers allowing large single allocations (see --eth_dag_buf).
  • Ethash: added high score list of the 15 highest value shares found since start (see --high_score).
  • Ethash: small improvements in keeping the gpu busy over epoch switches.
  • Ethash: fixed bug that would cause deadlocks in rare occasions.
  • Ethash: bugfix - dag cache wasn't enabled for 4GB gpus.
v0.7.18
Notes
  • Support for the ETC fork added. Run with "-a etchash" for easiest support, see "--eth_variant_mode" in the help for more info.
  • ETH+ZIL miners: add "--eth_dag_cache=0" to use a second dag cache and prebuild epoch 0 immediately at miner start.
  • Ethash miners relying on ramp-up and stagger for stability: the default mode of these tweaks has been changed to off. They now need to be enabled with --eth_ramp_up and --eth_stagger. While helping a number of rigs, these tweaks caused increased instability of others, making the choice of a good default mode difficult.
Changes
  • Added etchash support (see algo etchash and --eth_variant_mode).
  • Added dag cache support, mostly intended for eth+zil mining (see --eth_dag_cache).
  • Fixed hex char parsing in enable/disable submenu, can now work with >= 10 gpus.
  • Changed the default for ethash ramp-up and staggering to false (see --eth_ramp_up and --eth_stagger).
  • Added check for broken win setups with duplicated opencl platforms and/or devices.
v0.7.17
Changes
  • BETA: Added fan control, see --help section or USAGE.txt. Enable with --fan_control.
  • Fixed deadlock bug that could happen when using multiple pools with the failover strategy.
  • Added option for not sending stale shares (see --no_stale_submit).
  • Added watchdog check for early gpu init hangs.
  • Added example watchdog scripts for win/linux.
v0.7.16
Notes

BETA RELEASE - small fixes for Turtlecoin's new algo chukwa2.
Changes
  • Chukwa2: fixed kernels loading for Radeon VII gpus.
  • Chukwa2: fixed mem footprint for 2GB gpus.
v0.7.16b
Notes

BETA RELEASE - adds Turtlecoin's new algo chukwa2. Not tested as rigorously as other versions. Only use for the Turtlecoin fork.
NOTE: Navi gpus are NOT supported on chukwa2 for now. Support will be added if there's sufficient interest.
Changes
  • Added algo trtl_chukwa2 for the Turtlecoin fork on 2020-10-22.
v0.7.15
Notes

Small bugfix release.
Changes
  • Ethash: fixed bug for ZIL (epoch 0) mining on 4GB gpus (bug appeared in 0.7.14).
  • Ethash: reverted to v0.7.10 kernels for stability purposes except for ROCm-based rigs.
v0.7.14
Notes

This release wraps up our work for 4GB gpus mining ethash. Linux 4GB are verified to mine up to and including epoch 381 on drivers 18.50/19.30/19.50/20.10. After that, we have added support for extended mining from epoch 382 and up. Extended mining is a trade-off between hashrate and being able to mine at all, gpus will lose additional hashrate for every extra epoch. Please see the Linux section in the ETHASH_4GB_HOWTO.txt for more details and instructions.
Changes
  • Ethash: verified Linux support for 4GB gpus up to and including epoch 381.
  • Ethash: added support for extended 4GB mining from epoch 382 and up.
  • Kawpow: cleaned up false hw errs for some block heights.
v0.7.13
Notes
  • Windows user with 4GB rigs - see the new guide ETHASH_4GB_HOWTO.txt for instructions.
Changes
  • Added better default Windows dag allocation strategy on 4GB gpus.
  • Added experimental support for resetting gpus before mining (Windows only).
  • Added argument for manual adjustment of 4GB dag allocation (see --eth_4g_alloc_adjust=X,Y,Z,...).
  • Added protection against broken dev servers with additional random server selection mechanism.
v0.7.12
Notes
  • Bugfix release: v0.7.11 could crash with rejected ethash shares.
Changes
  • Fixed print format bug on ethash rejected shares.
v0.7.11
Notes
  • Ethash additions: printing share diff in GH (always enabled), hashrate watchdog (see --eth_hashwatch).
  • Better debug support: --long_timestamps, --pool_debug.
  • We've found certain motherboard/bios combinations that due to kernel and/or driver bugs have issues with the first or last gpu in rigs. If you see excessive hw errors reported on a single gpu, and switching around gpus still always generates errors in the first or last gpu, try using --eth_dag_alloc_patch.
Changes
  • Added DAG allocation patch under linux for certain mobo/bios combinations (see --eth_dag_alloc_patch).
  • Added ethash hashrate watchdog (see --eth_hashwatch).
  • Added ethash abort mechanism for Vegas on ROCm (see --eth_direct_abort). Not useful in other contexts.
  • Added pool traffic debug (use --pool_debug).
  • Added print of share difficulty for ethash family algos. Unit is always GH.
  • Added microsec timestamp resolution in logs (use --long_timestamps).
  • Regression bug: argon2/chukwa was broken in 0.7.10, now working again.
  • Turned off compute mode checks for non-Polaris gpus.
v0.7.10
Notes
  • Ethash 4GB rigs, especially on win, should use --eth_alloc_epoch=N with N being the max epoch that the gpus can allocate. This will avoid DAG reallocation issues.
  • For Navi rigs having issues with eth+zil or Nicehash mining with frequent DAG switching, try using --eth_dag_slowdown=9.
Changes
  • Added default log filename (trm_<algo>_<yyyymmdd_hhmmss>.log)
  • Added ethash forced initial allocated epoch ( --eth_alloc_epoch=N). Note: HIGHLY recommended for 4GB rigs.
  • Added ethash family DAG build slowdown configuration (--eth_dag_slowdown=N, default value 4).
  • Added ethash family DAG build staggering across gpus (disable with --eth_no_stagger).
  • Added ethash family intensity ramp-up (disable with --eth_no_ramp_up).
  • Added option for forcing dev fee region (see --dev_location).
  • Added MTP for Navi gpus.
  • Added MTP ramp-up after pad rebuild.
  • Fixed error printouts on failed watchdog script execution on Linux.
  • Stats: now shows more clocks and temps, and also adds gpu state in hashrate prints (see --short_stats to disable).
  • Added checks for and enabling compute mode and disabling crossfire on Windows (see --enable_compute).
v0.7.9
Changes
  • Fixes for mixed rig mining on Linux.
  • Added --eth_epoch argument for easier epoch testing.
  • Added --eth_aggr_mode for automatic aggressive 'B' mode on Polaris 8GB gpus.
  • Added --watchdog_disable argument.
v0.7.8
Changes
  • Upgrade for the upcoming Haven hard fork (July 20, 2020).
  • Tiny Nimiq optimizations (1-2% max, mostly Vega and Navi).
  • Fixed CPU usage issue when redirecting stdin to /dev/null.
  • Fixed pool name returned by API for Nimiq.
v0.7.7
Changes
  • Added support for Nimiq Polaris/Vega (dumb mode only).
  • Integrated a Nimiq node.js network proxy into the miner.
  • Fixed Nimiq bug that could lose shares, especially against lower vardiff pools.
  • Fixed Nimiq bug that could cause duplicate shares on startup for low-diff pools.
  • Fixed regression bug for ethash Nicehash, correct stratum mode now used again.
  • Added support for Nimiq Navi.
  • Added support for icemining Nimiq solo mining mode (set -p m=solo).
v0.7.6
Changes
  • Fixed broken keyboard input in tmux+screen sessions (e.g. Hive OS).
  • Added support for 5500(xt).
  • Fixed Linux watchdog support for hard driver crashes (script was not executed).
  • Fixed kawpow nicehash extranonce support.
v0.7.5
Changes
  • Fixed broken optimizations for kawpow in 0.7.4.
v0.7.4
Changes
  • Increased ethash support on 4GB GPUs up to epoch 380-383
  • Implemented split ethash dag buffers for 8GB GPUs to support DAGs over 4GB
  • Kawpow optimizations (Navi +2.25%, Vega +1.25%, Polaris +0.25%)
  • Added gpu enable/disable API support.
  • Windows TDR detection/handling/warning.
  • Monitor detection on Windows/Linux with intensity adjustment.
  • Fix for ethash pool hashrate reporting stopping after network outage.
v0.7.3
Changes
  • Emergency patch for 4GB cards to handle a few more ETC epochs.
    A more complete patch is coming out shortly.
v0.7.2
Changes
  • Fixed kawpow dag build DEAD gpu issue on windows Adrenalin 2020 drivers.
  • Fixed Navi 5600(xt) support on windows.
  • Fixed mining on Vegas on older amdgpu-pro drivers.
  • Fixed ADL reporting of stats on windows for newer cards.
v0.7.1
Changes
  • Fixed issue with VII ethash/kawpow on windows.
v0.7.0
Changes
  • Added kawpow algo for Ravencoin.
  • Added Navi support for kawpow and ethash.
  • Changed device ordering to be pcie bus based by default, added --opencl_order option.
  • Fixed issue with --list devices not working without other args.
  • Reformatted help message to hopefully make it easier to read.
  • Added multipool example scripts.
  • Removed ssl/tls server name verification (was re-added with TLS SNI fix)
  • Fixed an unhandled signal bug that would cause rare crashes.
  • Fixed multi-pool API bug.
1653042715
Hero Member
*
Offline Offline

Posts: 1653042715

View Profile Personal Message (Offline)

Ignore
1653042715
Reply with quote  #2

1653042715
Report to moderator
1653042715
Hero Member
*
Offline Offline

Posts: 1653042715

View Profile Personal Message (Offline)

Ignore
1653042715
Reply with quote  #2

1653042715
Report to moderator
1653042715
Hero Member
*
Offline Offline

Posts: 1653042715

View Profile Personal Message (Offline)

Ignore
1653042715
Reply with quote  #2

1653042715
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
kerney666
Member
**
Offline Offline

Activity: 658
Merit: 86


View Profile
October 30, 2018, 06:07:29 AM
Last edit: November 27, 2018, 10:44:11 PM by kerney666
Merited by scryptr (10), dragonmike (10), frodocooper (7)
 #2

This is the tuning guide included in the release, posting it here as well for people to read before they download.

A noteworthy fact not mentioned below is that we also asked two forum members not affiliated with us in any way, dragonmike and N2DCRYPT, to verify the miner hashrates for us on their rigs in an early test release. If someone asks, I'm sure they will present their own benchmarks. Imho, their results were in line with our own testing presented below.


Team Red Miner CNv8 Tuning

Background

This miner is more lean than other CN miners. This can translate into either increased hashrate, lower power draw, or both, or none of the above. Your mileage may vary and is highly dependent on mem straps and clocks. If running at the same hashrate as other CN miners, you can expect your power draw to decrease between -5-20W per gpu depending on gpu model and clocks.

To enable a higher hashrate, this miner generally requires a higher _core clk_ than normally used in CN mining. Unless you go nuts, your power draw will most probably be lower than that of other miners at lower clocks in the end anyway. This is the only way to achieve a > 2200 h/s read on a Vega 64.

There are fewer controls in this miner than the standard CN miner. It always runs with two threads, and you control the intensity per thread in the range 1-16. Moreover, you can choose between two modes, + and -. The mode will most often not have any effect on the end result, but it never hurts to try it.

Tuning guide

[Windows drivers]
For some Polaris cards, the good ol' blockchain driver works fine. However, the one driver that seems to be a good fit across the board is 18.6.1, and that's the driver we have used in all our Windows tests.

[Windows swap space]
You should to set up your swap space to be at least the sum of all GPU memory you intend to use when mining. Typically, for a 4GB card this is 3.5GB and for an 8GB it's 7.5GB. Playing it safe is recommended, i.e. rather add the full memory size of all your GPUs and set the swap to the total sum or more.

[Linux drivers]
You need the amdgpu-pro 18.30 drivers for your Vegas under linux to reach the max possible hashrate. Also, please note that this release does not include ROCm support for CNv8, it will be included in an upcoming release instead.

[Polaris cards (470-580)]
The standard configuration is 8+8 for all of these cards. 8+6 or 7+7 might give the same optimal hashrate, and 9+9 can, especially under linux, give a better result for 480/580. For some cards, 16+14 is the best choice but also increases the probability of stale shares. You must have good mem straps to reach a good hashrate. Normally e.g. the Pimp My Straps function in SRB Polaris Bios Editor is sufficient. For mem clk, boosting it as much as possible while avoiding mem errors is a good thing. The core clk should generally end up between 1230-1270. For 580s, a boosted core clk to 1300 can push the hashrate to 1100 h/s while still staying at a reasonable power draw. We have seen few Polaris cards not being able to reach 1020-1030 h/s with this miner when the proper mem straps are in place.

[Baffin and Lexa Pro cards (550-560)]
From v0.3.8, this miner has now been better optimized for these smaller cards. The major additions are that the '+' mode has been optimized and a 'L' prefix mode designed for the smallest Lexa Pro cards has been added. Some rules of thumb when you optimize your rigs:

  • We'd expect 4+4 and 4+3 to be the only interesting configs for 4GB cards.
  • For Lexa Pro cards with 8 CUs, prefix your config with 'L', i.e. L4+3.
  • The 'L' prefix is designed for Lexa Pro, but can also work well for Baffin with 10 CUs.
  • Many 2GB Lexa Pro can't do L4+3 under Win, only L3+3. For max performance you should try Linux and L4+3.
  • For an overkill full range test, you should try all of 4+4,4+3,3+3,3+2,2+2 in four modes: X+Y,LX+Y,X-Y,LX-Y

[Vega cards]
The Vegas can end up anywhere from 1900-2270 depending on if it's a 56 or 64, a reference card or not and your choice of clocks. The interesting configurations to try for your Vega are 14+14, 14-14, 15+15, 16+14. You can also try 16+15, 15+14, etc. The mem clk is very important, and you should aim for as high as possible while keeping your rig stable. If you have a Vega 64, please don't be shy to try a higher core clk. The 16+14 configuration will often not show its true capability before hitting 1500 core clk. Your power draw should still stay reasonable (as in lower than other miners at more standard clocks). For a lower core clk around 1408, some cards do best with 16+14, others with 15+15, some with 14+14, YMMV.

[Older cards]
We're sorry, we only support 470-580, 550/560 and Vega cards.

Benchmark results

For most Polaris cards below, one-click Pimp My Straps in SRB Polaris Bios Editor has been used for mem straps.

[6 x Rx 470 8GB (Samsung mem) rig]
8+8, 1250/900 cclk, 2000/900 mclk, 6105 h/s, total rig 685W

[Rx 560 4GB]
4+4, 1230/900 cclk, 2000/900 mclk, 540 h/s, unknown power draw

[Rx 570 8GB (Samsung mem)]
8+8, 1270/900 cclk, 2100/900 mclk, 1030 h/s @ ~100W at wall

[Rx 580 8GB (Hynix mem)]
8+8, 1250/900 cclk, 2000/900 mclk, 1029 h/s @ ~105W at wall

[Vega 56 reference card (56 bios, ppt mod)]
16+14, 1413/880 cclk, 935/880 mclk, ~2000 h/s @ ~197W at wall

[Vega 64 liquid cooling]
15+15, 1408/880 cclk, 1100/880 mclk, ~2100 h/s @ ~190W(?) at wall
16+14, 1560/925 cclk, 1100/880 mclk, ~2270 h/s @ ~210W(?) at wall

Example command lines

[Windows, one Vega 64 and one Rx580]
teamredminer.exe -a cnv8 -o stratum+tcp://pool.somexmrpool.com:7777 -u <wallet> -p MyWorker -d 0,1 --cn_config 16+14,8+8

[Linux, rig of 6 x Rx480 with different settings, not very realistic]
./teamredminer -a cnv8 -o stratum+tcp://pool.somexmrpool.com:7777 -u <wallet> -p MyWorker -d 0,1,2,3,4,5 --cn_config 8+8,7+7,9+9,8+8,8+8,9+9
rubine2323
Member
**
Offline Offline

Activity: 84
Merit: 10


View Profile
October 30, 2018, 07:05:28 AM
 #3

Great to see some work from you again!

Ive just tested it real quick on my Vega and 470 Rig.
The power consumption is amazing!

Vega 56 2016H/s @ 164 Watts
RX 470 1008H/s @ 126 Watts

Crazy stuff. Ill have to do some more tests.

Unfortunately i dont have any merit points left  Angry
fluxy12
Jr. Member
*
Offline Offline

Activity: 145
Merit: 1


View Profile
October 30, 2018, 07:16:01 AM
 #4

Hello tested on windows 10 with 6*RX580, it doesn't launch unfortunatly. i tryed without Antivirus and without OC.

I got the message "failed to list OpenCL devices".

Can it be a driver problem ? i have blockchain drivers.

Thank you
rubine2323
Member
**
Offline Offline

Activity: 84
Merit: 10


View Profile
October 30, 2018, 07:31:40 AM
 #5

Hello tested on windows 10 with 6*RX580, it doesn't launch unfortunatly. i tryed without Antivirus and without OC.

I got the message "failed to list OpenCL devices".

Can it be a driver problem ? i have blockchain drivers.

Thank you


Add to your batch file:

--platform=1
todxx
Member
**
Offline Offline

Activity: 175
Merit: 76


View Profile
October 30, 2018, 08:28:15 AM
 #6

Great to see some work from you again!

Ive just tested it real quick on my Vega and 470 Rig.
The power consumption is amazing!

Vega 56 2016H/s @ 164 Watts
RX 470 1008H/s @ 126 Watts

Crazy stuff. Ill have to do some more tests.

Unfortunately i dont have any merit points left  Angry

Thanks rubine!  It's been quite the push to get this release out.  Thanks for giving it a spin so quickly.
Your RX 470 power is a bit higher than what we've been seeing, what clocks are you running on it?

Hello tested on windows 10 with 6*RX580, it doesn't launch unfortunatly. i tryed without Antivirus and without OC.

I got the message "failed to list OpenCL devices".

Can it be a driver problem ? i have blockchain drivers.

Thank you


Hi flux,
As rubine mentioned, it's likely that you have more than one opencl platform, and the miner uses the first platform by default.
If adding the --platform=1 option that rubine suggested doesn't work, could you run clinfo on the command line and post the output?
dragonmike
Hero Member
*****
Offline Offline

Activity: 1274
Merit: 556



View Profile
October 30, 2018, 09:27:53 AM
 #7

I tested this last night on two rigs (RX570's and Vega 56's).
Great work indeed. It's hands down the fastest and least power-consuming CN miner around.

For everyone's benefit, I got most hashrate using --cn_config=7+7 on the 570's and 16+14 on the Vegas so far, for virtually no increase in power draw. Net net a gain of about 4% hashrate at 7% less power than SRBminer.

Well done.

EDIT: looking forward to you releasing the same sort of improvements on the other CN variants (heavy/heaven/fast etc) and other algos that are grossly unoptimised for AMD hardware (looking at you x16r)!  Wink
netto7
Jr. Member
*
Offline Offline

Activity: 252
Merit: 4


View Profile
October 30, 2018, 09:53:21 AM
 #8

have plan to add more algo?
stolarzz
Newbie
*
Offline Offline

Activity: 111
Merit: 0


View Profile
October 30, 2018, 10:05:08 AM
 #9

820kh/s for Nitro+ 570 4Gb it's optimal?
Is there a possibility to hash 1k+ with this card?
dragonmike
Hero Member
*****
Offline Offline

Activity: 1274
Merit: 556



View Profile
October 30, 2018, 10:08:21 AM
 #10

820kh/s for Nitro+ 570 4Gb it's optimal?
Is there a possibility to hash 1k+ with this card?
Have you modded your mem straps?
If not, do it. You should be getting over 1k easily with this card.
stolarzz
Newbie
*
Offline Offline

Activity: 111
Merit: 0


View Profile
October 30, 2018, 10:16:06 AM
 #11

820kh/s for Nitro+ 570 4Gb it's optimal?
Is there a possibility to hash 1k+ with this card?
Have you modded your mem straps?
If not, do it. You should be getting over 1k easily with this card.
They were modded long time ago for eth the card is getting 30+mh avarage there, is there any difference for cryptonight?
dragonmike
Hero Member
*****
Offline Offline

Activity: 1274
Merit: 556



View Profile
October 30, 2018, 10:17:21 AM
 #12

820kh/s for Nitro+ 570 4Gb it's optimal?
Is there a possibility to hash 1k+ with this card?
Have you modded your mem straps?
If not, do it. You should be getting over 1k easily with this card.
They were modded long time ago for eth the card is getting 30+mh avarage there, is there any difference for cryptonight?
Not much.
Have you tried using --cn_config=7+7 ?
stolarzz
Newbie
*
Offline Offline

Activity: 111
Merit: 0


View Profile
October 30, 2018, 10:19:49 AM
 #13

Quote
Have you tried using --cn_config=7+7 ?
Yup there is no change at all. All my AMD's are 4GB and never get more than 820kh with cryptonight Sad
dragonmike
Hero Member
*****
Offline Offline

Activity: 1274
Merit: 556



View Profile
October 30, 2018, 10:27:15 AM
Last edit: October 30, 2018, 10:51:57 AM by dragonmike
 #14

what clocks?


EDIT: sidenote... for the lulz I increased clock speeds on the Vega56's to 1590... Hashrate went up to 2150 h/s!  Grin
But power went up quite massively as well (over 225W per GPU).
rubine2323
Member
**
Offline Offline

Activity: 84
Merit: 10


View Profile
October 30, 2018, 10:52:33 AM
 #15

Great to see some work from you again!

Ive just tested it real quick on my Vega and 470 Rig.
The power consumption is amazing!

Vega 56 2016H/s @ 164 Watts
RX 470 1008H/s @ 126 Watts

Crazy stuff. Ill have to do some more tests.

Unfortunately i dont have any merit points left  Angry

Thanks rubine!  It's been quite the push to get this release out.  Thanks for giving it a spin so quickly.
Your RX 470 power is a bit higher than what we've been seeing, what clocks are you running on it?

Hello tested on windows 10 with 6*RX580, it doesn't launch unfortunatly. i tryed without Antivirus and without OC.

I got the message "failed to list OpenCL devices".

Can it be a driver problem ? i have blockchain drivers.

Thank you


Hi flux,
As rubine mentioned, it's likely that you have more than one opencl platform, and the miner uses the first platform by default.
If adding the --platform=1 option that rubine suggested doesn't work, could you run clinfo on the command line and post the output?

Current clk at 1250@900, 2000@900. My bios mod may differ a bit.
vmozara
Member
**
Offline Offline

Activity: 190
Merit: 59


View Profile
October 30, 2018, 11:00:32 AM
 #16

this looks promising.

Especially in a sea of viruses and trojan miners that we had here. I wish i had any lefover merit to give to you Smiley

I tested at my home PC and now i will proceed to update my rigs. It seems your dev fee is high but it is well deserved.

Test at home PC: Adrenalin 18.10.2, win 10 1809, vega 64 custom cooling, power table 85% from vega guide (unfortunate vega), core 1550/900 mem 1150/900

All consumption below is entire system consumption from the wall

JCE miner: 2080H/s @ 300W
SRBminer: 2130H/s @ 300W
TeamRedMiner: 2300H/s @ 285W

 Shocked

I will provide here results from optimized Vega rigs later on.

Is there any special driver version you recommend for Vega?

Thanks!
N2DCRYPT
Jr. Member
*
Offline Offline

Activity: 148
Merit: 5


View Profile
October 30, 2018, 11:06:04 AM
 #17

Posting to confirm that the advertised hashrates and power draws for this miner are legitimate. Quite an impressive release from kerney666 and todxx; personally, I've had issues switching over rigs to v8 because of the increased power draw, and this miner has alleviated those problems, while giving increased hashrate!

My Vega 64s are currently working at about 2100 H/s and Vega 56's are running 2000 H/s with lower power draw and higher hashrates than I was running on SRBMiner.
dafuzz_rsa
Newbie
*
Offline Offline

Activity: 40
Merit: 0


View Profile
October 30, 2018, 11:06:22 AM
 #18

Absolutely amazing results. Well done!
Looking forward to seeing what else you guys have to offer
N2DCRYPT
Jr. Member
*
Offline Offline

Activity: 148
Merit: 5


View Profile
October 30, 2018, 11:08:48 AM
 #19


Is there any special driver version you recommend for Vega?


Vmozara, the Adrenalin 18.6.1 drivers should work well; and that's what I'm running on all of mine--anything older than that tends to have problems with v8.
todxx
Member
**
Offline Offline

Activity: 175
Merit: 76


View Profile
October 30, 2018, 11:54:04 AM
 #20

I tested this last night on two rigs (RX570's and Vega 56's).
Great work indeed. It's hands down the fastest and least power-consuming CN miner around.

For everyone's benefit, I got most hashrate using --cn_config=7+7 on the 570's and 16+14 on the Vegas so far, for virtually no increase in power draw. Net net a gain of about 4% hashrate at 7% less power than SRBminer.

Well done.

EDIT: looking forward to you releasing the same sort of improvements on the other CN variants (heavy/heaven/fast etc) and other algos that are grossly unoptimised for AMD hardware (looking at you x16r)!  Wink

Thanks for helping with the early testing dragonmike!  It really helps to have some outside validation of our numbers as a bit of a sanity check before a release Smiley
As for CN variants and other new algos, we will be adding more, but it may be a little while.  This last push to get CNv8 from the drawing board and into production has taken over a month of full time work on both our parts and we both need a bit of a breather after this one.  That said, adding CN variants now that we have the general CN infrastructure in place should be much easier.  We've also done some very initial work on x16r and it looks like there's plenty of room for improvement, but at this point we can't make any promises on what we'll be able to release and when.

what clocks?


EDIT: sidenote... for the lulz I increased clock speeds on the Vega56's to 1590... Hashrate went up to 2150 h/s!  Grin
But power went up quite massively as well (over 225W per GPU).

Haha, I just spent a couple hours pushing my Vega FE too.  I managed to hit just over 2400h/s running at 1610MHz core and 1195MHz mem clocks!  Grin  The card was pulling about 300W alone so not very useful, but I love seeing that 2400h/s readout!

Quote
Have you tried using --cn_config=7+7 ?
Yup there is no change at all. All my AMD's are 4GB and never get more than 820kh with cryptonight Sad

Do you know what type of memory is on your card?
I have some MSI RX470 mining edition cards that have Micron memory and so far I can only get them to about 850h/s with the best straps I could find.

this looks promising.

Especially in a sea of viruses and trojan miners that we had here. I wish i had any lefover merit to give to you Smiley

I tested at my home PC and now i will proceed to update my rigs. It seems your dev fee is high but it is well deserved.

Test at home PC: Adrenalin 18.10.2, win 10 1809, vega 64 custom cooling, power table 85% from vega guide (unfortunate vega), core 1550/900 mem 1150/900

All consumption below is entire system consumption from the wall

JCE miner: 2080H/s @ 300W
SRBminer: 2130H/s @ 300W
TeamRedMiner: 2300H/s @ 285W

 Shocked

I will provide here results from optimized Vega rigs later on.

Is there any special driver version you recommend for Vega?

Thanks!

Thanks vmozara!  Nice to see someone else out there pushing the higher 2300+ hashrates  Smiley
As for the driver version, we have done most of our testing on 18.6.1, so I would consider that the safest bet.  I believe kerney saw some issues with Vegas running the old blockchain drivers, so I would avoid those for Vegas.

Posting to confirm that the advertised hashrates and power draws for this miner are legitimate. Quite an impressive release from kerney666 and todxx; personally, I've had issues switching over rigs to v8 because of the increased power draw, and this miner has alleviated those problems, while giving increased hashrate!

My Vega 64s are currently working at about 2100 H/s and Vega 56's are running 2000 H/s with lower power draw and higher hashrates than I was running on SRBMiner.

Thanks for helping with the early testing N2DCRYPT!  We really appreciate you taking the time to kick the tires before the release!
Also I'm glad we could help get your rigs back into their old power profiles Smiley

Absolutely amazing results. Well done!
Looking forward to seeing what else you guys have to offer

Thanks dafuzz_rsa!  We're looking forward to cranking out some new algos after we take a little bit of a breather Smiley
Pages: [1] 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 ... 146
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!