- Can you set specific OC via batch or config file with t-rex.?
- Is there a log option since I get softcrashes or the software randomly quits on me every once in a while.?

Thank you in advance for any advice here.

Hi,

is there a log option with T-rex?

My rig is crashing after couple of hours operating fine.

Also can you set the OC settings by the miner batch file itself?

Hey,

I have a rig with mixed GTX 1xxx cards. Today I have put 2 RTX 3070 into that rig. But T-Rex wont run. No matter if I try to run only some cards (like all the 1070 or 1080...), or all together, or only the old ones, or just the 3070. Even if I only put one 3070 in there or switch the sockets, there is always the same error.

Tried also with 0.19.5 and 0.19.11, with cuda 10.0 and 11.1., with admin and without, with and without the --no-nvml tag and with older 457.51 driver and the newer 461.40. Its a Win 10 pro 2004 version. The error I get is:

With both cards enabled:
> 20210215 22:06:58 ERROR: Can't start miner, can't pair device with KEY 000e00 [GPU #6, dom:0, bus:e, slot:0], failed to get handler of GPU nvml_id=350e75db36d696834d658e638e62f489: error code 999

Just one card:
> 20210215 22:29:32 ERROR: Can't start miner, can't pair device with KEY 000d00 [GPU #5, dom:0, bus:d, slot:0], failed to get handler of GPU nvml_id=8d88412b26e66b4ca70c342a6a9bbf2e: error code 999

All other rigs are fine with RTX 3080 and 3090 only cards. But I never mixed different generations. I also have not tested if I get the same error when I mix RTX 2xxx with the 3070. But maybe you already know what this can be. Other miners like the G** and L** dont have this problem and running fine, but I dont want to use them. So I think its some bug or whatever in your miner when mixing up generations.

Can you solve this somehow? Thanks!

EDIT: I just tested the 3070 card(s) in an RTX 2xxx system with Win 10 Pro 2004 and the newest 2H20 including older and newest nVidia drivers and got the same error:
> 20210216 10:32:57 ERROR: Can't start miner, can't pair device with KEY 000400 [GPU #1, dom:0, bus:4, slot:0], failed to get handler of GPU nvml_id=8d88412b26e66b4ca70c342a6a9bbf2e: error code 999

It doesn't matter if I try to start the 3070 alone or any other card, or all together, T-Rex wont run and just give this error. But all other miners have no problem in any combination. Is it possible that your miner doesn't know what to do with a 3070? In some different systems I have no problem with 3080 and 3090 cards, its just that 3070 cards... Strange

Tell us more about the rig. 
Are cards on risers or all on mobo?

Windows or linux?

hi newest version has more updated miners so will be faster in some algos also a few new algos have been added since January so your also missing out on them eg. x16r on zergpool is doing very well atm
Best Regards
Nemo

back off intensity until you can run your cards at 120 watts

Try 80% power/+100 core/+500 mem

Stability of a non-riser trio rig can be a bit better - fewer parts to go bad - but COOLING of the middle card in such a rig is a major problem.
You can work around that somewhat by using a "shorty" card as the third card, like a Zotac mini or the Gigabyte ITX models, but then you get to deal with the junk fans on the shorty cards (EVGA makes short cards with good fans but I don't think they go above a 1060 on them).

 I built quite a few such rigs during my Folding days, as Folding@Home needs a lot of CPU support and a lot of PCI-E bandwidth to get good PPD (hashrate) out of the cards so riser rigs were NOT a viable option.
 In theory, RIBBON-type x16 risers would have worked, but those tend to be a NIGHTMARE to work with and also tend to be very short.