Topic: T-Rex 0.26.8 ETHW, ETC (+dual mining) GPU miner - page 10. (Read 174690 times)

Hi, i am interested which ports and pools are used for dev fee. Need that info to open access for them. Anyone can help me?

Hi. Can you paste the settings for 3060ti?thanks

any chance of adding cortex algo im sick of the other miners not working properly many thanks pleaseeeeeeeee.

yes correct

Thanks for the help,  for the 3070's I just did core clock -500 as I didn't lock it at a specific voltage as the 3080's.

So instead of what I had run without the --lock-cv and without --cclock    But with --lock-cclock 1095,1075,1095,1075 ?


t-rex.exe -a ethash -o stratum+tcp://eu1.ethermine.org:4444 -u myaddress -p x -w rig0 --pl 63,55,63,55 --fan 85,60,75,65 --lock-cclock 1095,1075,1095,1075 --mclock 1450,1350,1450,1350

I don't use fixed voltage on my cards, but apparently that's okay, just test to see, why don't you fix the core of the 3070s too?

--lock-cclock 1095,1075,1095,1075

I've been mining for a while with T-rex on win 10 and always used MSI Afterburn for the OC settings of my cards. Now I would like to try the OC settings being done by T-Rex,  As from what I can understand from the help page this is what I put together. I understand how to put the easy settings (fan, power limit, etc) in the file , just the undervolting of my 3080's is something I'm not sure if they are correct....

Can somebody verify if I wrote it correct ?


So I have 4 cards running(non LHR), two rtx 3080's that use mostly the same settings and two rtx3070's with small difference in settings.  gpu0 is a 3080, gpu1 is a 3070, gpu2 is a 3080 and gpu3 is a 3070.

Only the 3080's are undervolted , locked voltage at 712mv at 1095mhz, and this is where I'm not sure if I have it correct in the Bat file.

Thanks



t-rex.exe -a ethash -o stratum+tcp://eu1.ethermine.org:4444 -u myaddress -p x -w rig0 --pl 63,55,63,55 --fan 85,60,75,65 --lock-cclock 1095,0,1095,0 --lock-cv 712,0,712,0 --cclock 0,-500,0,-500 --mclock 1450,1350,1450,1350

So I am mining Ethash on a 1080 TI with 11 Gb under linux and I notice that with only relatively minor variation whenever a share is submitted it is around 52ms, give or take only 5ms almost every time.

If there is a particularly "unlucky" block and it takes a lot longer for one of the miners to find a share and move it on to the next block, I never see a time to find share of, say, 120ms.

I haven't run the stats, but I would say the distribution of times does not reflect what I would expect for the solve time, you would think that all guesses are equal up until a block is solved, so they shouldn't be so heavily clustered as this - what's up here and what does this tell us about maybe optimising the algorithm?

I have 1070, 3070ti and 3080ti, they all show fan speed... I had some problems with not showing consumption, and clock, I had to format windows and reinstall everything again...

my 1070 doesn't show memory temperature, neither in the trex nor in the nicehash, but it's because it doesn't have a sensor in the memory...

Is T-rex capable of sending a notification when hashing rate drops below a certain value?

Sometimes my miners stay down for hours and I do not know about it...

--pl 65 --lock-cclock 1050 --mclock 1200 --lhr-tune 74.7 --lhr-autotune-step-size 0.1 --lhr-autotune-interval 30

Thanks for the prompt reply, I will try that if I have to.  It would be neat to experiment with Linux, and there's supposed to be things one can do with LHR cards that work better in Linux I could experiment with as well.  I've been meaning to do so but never got around to it, and the Thanksgiving holiday may be a good time to do it.

I think I should learn though what's needed to run T-Rex besides a OS install and the Nvidia driver program installing an appropriate driver without errors though.  I am trying to install the latest CUDA toolkit for Windows 7 64 bit, version 10.2, to see if that fixes things.  But any suggestions or background on what's supposed to be there which maybe most people don't need to manually install on Win 10 64 bit would be appreciated.

***

Update - after installing the Cuda toolkit version 10.2 and rebooting and seeing no difference, I went ahead and upgraded the server to Windows 10 64 bit and it is working ok with T-Rex now.  Surprisingly, I have to say, as it normally requires manual installation of drivers for Windows 7 (it originally was designed to run on mid '00s era Microsoft server OSs).  Anyway, I may try out Linux over the holidays but I just wanted it to mine something for now and if Windows 10 works I'll go with that.  All I have to deal with now that's unusual is when I boot I hear a voice that says "Luke, I am your father" and since the old server doesn't have any sound drivers or speakers I think that's kind of odd but it works ok otherwise.  May the force be with you all!

Why not grab a usb 3.0 16+gb memstick and slap any of the linux mining OSes on it.  Should be no issue to support the older hardware.  Dont even have to blank the drive (disconnect it if you want).  Most of the linux based mining OSes will mine on just about anything!

I am trying to get T-Rex miner running on Win 7 64 bit old computer /w one 3060 TI card.  The 3060 TI has worked well with T-Rex in a newer computer running Windows 10 64 bit, but I got a new and faster card for it and so I want to keep the older card (the 3060 TI) running in my older computer.

The older computer is a converted UNIX server which officially supports Windows 7 64 bit professional which I have on it now. I just put a fresh install of Win 7 64 bit professional on it and installed a couple of Microsoft security hotfixes that the Nvidia driver installer required, and the Nvidia driver installation program completed successfully and the card seems to work, that is it displays an image and I can change resolution on it, and Windows 7's device manager and GPU-Z recognizes it is installed and gives it the 3060 TI name.  

However, I can't run T-Rex miner without getting errors, namely:

20211122 16:33:20 ERROR: Can't start T-Rex, can't initialize CUDA engine, cuda exception: CUDA_ERROR_UNKNOWN. Is NVIDIA driver installed?

I am installing now all the updates I can for Windows 7 64 bit from wsusoffline dot net which may have newer DirectX files and whatever else I don't know now how to describe, and that may fix it.  In any case the 3060 TI should be supported per a press release at drivers.softpedia.com/blog/nvidia-adds-support-for-its-geforce-rtx-3060-ti-gpu-get-geforce-457-51-531680.shtml   (sorry if the link doesn't work, it seems this board truncates URLs it considers suspicious, so I suppose it should be possible to run T-Rex on it with the proper drivers, updates, etc as well.

But for now, can anyone tell me what I *should* have installed on a Win 7 64 bit computer besides Win 7 and the latest updates (keeping in mind Microsoft's updates stopped early last year) and of course Nvidia's drivers, like the latest Nvidia drivers for Windows 7 64 bit, to run T-Rex miner?  I want to start with the recommended or at least a known working configuration and troubleshoot from there.

I know someone may suggest I try to upgrade to Windows 10 but that isn't supported on the older computer.  It might work or at least boot but if it didn't both run ok as a Win 10 instal and fix the CUDA issue it would give me more complications to try to fix.

Guys, if I need to lower -i parameter for just one card in my 6 cards rig, how do I do that leaving others at default intensity?

almost the same thing happened for me, but on the Palit 3080 ti video card with 2 power connectors, then I changed the batch file a little and brought two different pcie branches to the riser and to the power connectors, after which it became normal (~90mhs)
in the example below, the second card with lhr and batch file settings (driver 472.12)
--lhr-tune 0,74,0,0 --lhr-autotune-step-size 0.1 --validate-shares --send-stales --kernel 0 --pstate p0 --lock-cv 650,0,725,725 --lock-cclock 1005,1125,1110,1110 -i 25,0,25,25 --mclock 800,1000,980,820 --fan t:60[0-90], tm:90[0-90], tm:90[0-90], tm:90[0-90] --temperature-limit 65 --temperature-start 55 --gpu-init-mode 0 --dag-build-mode 2 --pl 54,0,74,102

Hi. Can you paste the settings for 3070ti?

Thanks for the reply rafamonteiroo,

I try youre settings but not stable on my gpu .I cant mclock more from 500 , becouse over this number i cant lock cclock. If i set mclock 1200 or more core clock jump araund like crazy in the range 900-1300mhz.

make 82mhs       pl 78 --lock-cclock 1185 --mclock 500 --lhr-tune 73.7 --lhr-autotune-step-size 0.1 --lhr-autotune-interval 30

windows nv driver 472.12


https://imgur.com/UlFD5tM

mine is 3x8pin... https://en.colorful.cn/product_show.aspx?mid=102&id=1850

make 90mhs

https://i.imgur.com/G29bedY.png

--pl 78 --lock-cclock 1185 --mclock 1200 --lhr-tune 74.2 --lhr-autotune-step-size 0.1 --lhr-autotune-interval 30

windows nv driver 472.12

I want to ask about msi rtx3080ti suprim x ... I see that you reach a speed of 90Mhsh but for me the GPU core lock is impossible.I just want to add that gpu has 3x8pin power cables and obviously this is also very important for LHR.I tested all available drivers under windows 10 and an average score greater than 83Mhsh could not do. I will be happy if someone who has a similar model with 3x8pin power cables shared whether and how he maked a score of 90Mhsh with t-rex 0.24.7.