New pre release with some more lhr fixes/adjustments. Reset device will now run each hour.

TeamBlackMiner_1_30_cuda_11_5_pre.7z
https://www.virustotal.com/gui/file/75a6af9d5a854e60629877562c0301782fb2e047989871c06548f62fe07a8908?nocache=1


3080ti;

core: 0 default
mem: (+1000) 10500mhz
pl: 70% power

lhr-unlock [1] = 81 MHASH stable for 1  hour.
lhr-unlock [1]  --tweak [1] = 85 MHASH stable for 8-15 minutes.

3060:

lhr-unlock [1] = 32 MHASH stable for 1  hour.
lhr-unlock [40] = 35 MHASH stable for 1  hour.

---

Increase the --xintensity to 284 (old default) to get back the 2MH/s or more. This miner works best on high diff and high intensity, but some pools works better on low intensity.

Here is v1.29 been running on --xintensity 144 for ~2.5 days
Old testrig with broken cards.



[moderator's note: consecutive posts merged]

normal launch config is working here.

Set ETC_WALLET=(your wallet)

:loop
   TBMiner --algo etchash --hostname etc.2miners.com --port 1010 --wallet %ETC_WALLET% --worker-name %computername%
   timeout 10
goto :loop

---

v1.30
1. NVIDIA partial LHR unlock for all models. enabled with (--lhr-unlock)
2. Removed vardiff in the display when pool has static diff.
3. Reduced memory usage.
4. Added support for more pools.

https://github.com/sp-hash/TeamBlackMiner/releases/


pre-release. Linux probobly tomorrow.

Unlock device 1 and 2 in the rig

--lhr-unlock [0,1,1,0,0,0]

Higher number give a faster speed (but can trigger LHR mode)

--lhr-unlock [0,2,10,0,0,0]

Another way to do a more agressive unlock (add tweak 1,2,3,4...)

--lhr-unlock [0,1,1,0,0,0] --tweak [0,1,1,0,0,0]

This version doesn't have a LHR detector but will reset the device every 5 hours just in case.

TeamBlackMiner_1_30_cuda_11_5_pre.7z
https://www.virustotal.com/gui/file/75a6af9d5a854e60629877562c0301782fb2e047989871c06548f62fe07a8908?nocache=1

[moderator's note: consecutive posts merged]

My code unlocks without the need to lock the clock.


LHR was created by NVIDIA to reduce the hashrate, so creating a miner that breaks LHR is not an easy task. There are different ways to do it. T-rex way is working good, but failing to unlock on higher power. That meens that around ~90 will be the limit. TBM is not happy with 90, and we try to do 100MHASH++. But we miss is LHR detection and Auto-tune, so more work is needed.


https://github.com/sp-hash/TeamBlackMiner/issues

We have 153 solved issues, and fix the issues as they come in. Miners want the features of the other miners, so we fix it for them. Our changelog is long, and we have releases every 5-7 days. We are new in the market and don't have all the features of the other miners yet, but it's coming. Our miner is getting better and better for each release.


https://github.com/sp-hash/TeamBlackMiner/blob/main/OPTIONS.md

The correct parameter is called

--fan-speed-min  (or -f)

--lock-mclock is only working for the supported clockspeeds and not for overclocking. Use another program for that.

We don't have a discord, but you can reach us on here, on github, and reddit.

Your card is abit better than mine, but powerlmit 65 instead of 75 seems to help. But 75 is needed to do 100+ stable.

I guess my core clock and power settings where too high for the t-rex miner, been running it for a while and it recovered up to ~70 ish. But tbm without LHR code is doing 69MHASH. Perhaps I need --lock-cclock to make it work? Please share your settings..

I am running

core +0 (default setting) trottles abit
mem 10700 (+1350)
tdp 75%
p0 state

The T-Rex unlocker isn't working on the latest driver for the 3080ti





Tbm is +80% faster on the 3080ti with the latest drivers. LOL

Not on the 497.09 driver. T-rex is only doing 40MHASH  

No you can keep them in the same config. f.ex rig with 4 cards,  Amd in device 0 and 3 nvidia cards --xintensity [1024,4096,4096,4096]

Run with --list-devices to see the correct order.

The AMD intensity needs to be 1/4th of the nvidia intensity. so 448 is the same as 1792.

To reduce dag validation error you can try --dagintensity 1 ,  or with a low number. This will slow down the dag memory buffer creation, but is stable on higher clocks

 

LHR cards are a pain. Take forever to find the perfect settings, but here is a preview of v1.30

Just some debug output we forgot to remove. Don't worry about it.

Not 100 in v1.30, but stable ~80-85 on all drivers. The problem left is that each rig need manual tuning to get it right. Our code is also missing detect LHR, so our only option is to reset the device every 10-30minutes or more, and hope the unlock will stay unlocked.

No This might be a small bug. I think MPH have a static diff. Will check it.


So I reproduced this on etheremine when testing the new LHR unlock for 3080ti on maximum oc. I reverted a change in the code and it seems towork now. Will include the fix in 1.30.

The LHR2 unlock with the latest drivers (windows 497.09) will have 3 or more modes.

3080ti

LHR1: 72MHASH
LHR2: 75MHASH
LHR3; 78MHASH
LHR4: not yet stable.

The current beta version will reset the unlock every 5 minutes. Because we haven't made a way to detect LHR yet.

Enable with --lhr-unlock [0,2,3,0]

Power around 230 watt.

Will be released in v1.30. Note that older drivers can give better results, but we want to try to do it on the latest driver.

You can try to run with --xintensity 284 (the default in 1.28)

Which pool? AMD or nvidia?

All our testrigs run fine all night with 1.29.

did you run the linux build?

---

Tweak is normally used with a high intensity setting.

Getting closer to a 90-100MHASH unlock on 3080ti LHR2

Hashrates fluctate, but stable without triggering the LHR for ~5 minutes at 100% tdp.


v1.29 cuda 11.5 build with the following parameters:

--log --lhr-unlock 1 -U [ 0 ]

Not yet production ready but you can check if it unlocks. When the unlocking fails, hashrates fall to 50-60..

(driver 496.49)


when testing start on lower tdp like 70, or 80, and increase



Beta testers with LHR cards wanted

v1.29
1. Display difficulty change in the miner window.
2. Use ETH+ZIL shares when calculating pool speed and shares per minute.
3. Stratum change to avoid rejected shares on ezil.me and hiveon.net pools.
4. The default --xintensity changed to 144 on NVIDIA to reduce stale/rejected on hiveon, crazypool, flexpool, etheremine and binance pools.

https://github.com/sp-hash/TeamBlackMiner/releases/tag/

TeamBlackMiner_1_29_cuda_11_5.7z
https://www.virustotal.com/gui/file/652cbdb1ba76f4a1364eaaa0141e01da0a93b52f39999ff99f43995dc2a8070b?nocache=1

TeamBlackMiner_1_29_cuda_11_4.7z
https://www.virustotal.com/gui/file/3a58add075188ff9bb4a8eddb7e4f108967484b49f2cb239ccde4fedc4a90767?nocache=1

TeamBlackMiner_1_29_cuda_11_2.7z
https://www.virustotal.com/gui/file/b38b349cf8068f5615e322609bbb0c4a885741ccd6b0bc4172f927098171d74d?nocache=1

TeamBlackMiner_1_29_Ubuntu_18_04_Cuda_11_5.tar.gz
https://www.virustotal.com/gui/file/127a3958d0965fa93d70fef4ec525cae05546aa2ebcbfe7a52a064f99c078d72?nocache=1

TeamBlackMiner_1_29_Ubuntu_18_04_Cuda_11_5.tar.gz
https://www.virustotal.com/gui/file/cae029810937a5c53af1943cac0b15ae297e2e46937f9b71b6e016872d6ea97a?nocache=1

Users are reporting good results with --xintensity 60 latest drivers. You might need to add a few watts as well to reach 30.

1.28 have multiple issues reported, so you should rollback to 1.27  (but reduced GPU solution time is not one of the reported issues)
I have recieved more LHR cards this week and busy with the lhr2 unlock, but we will try to get a patch out today or tomorrow.