Topic: T-Rex 0.26.8 ETHW, ETC (+dual mining) GPU miner - page 53. (Read 174976 times)

thanks that worked, i was putting the -d 0 in the wrong location

HiveOS. Algo x16rv2. Intensity 23. 1060 3gb.

Every 5-60 minutes it writes an error:
ERROR: Can't find nonce with device [ID=1, GPU #1], cuda exception in [find_nonces, 179], an illegal memory access was encountered
ERROR: Can't find nonce with device [ID=3, GPU #3], cuda exception in [find_nonces, 179], an illegal memory access was encountered
ERROR: Can't find nonce with device [ID=0, GPU #0], cuda exception in [find_nonces, 179], an illegal memory access was encountered
ERROR: Can't find nonce with device [ID=2, GPU #2], cuda exception in [find_nonces, 179], an illegal memory access was encountered
ERROR: Can't find nonce with device [ID=4, GPU #4], cuda exception in [find_nonces, 179], an illegal memory access was encountered
ERROR: Can't find nonce with device [ID=5, GPU #5], cuda exception in [find_nonces, 179], an illegal memory access was encountered

Everything works without problems on exactly the same rig.
What could be the problem?

just add "-d 0" to select device (gpu) 0
example:
t-rex -a x16rv2 -o stratum+tcp://YOUR POOL -u YOURWALLETADRESS -p x -d 0

im mining with my PC, i have an older GPU thats not compatible with t-rex miner and i have a 1060 that i want to mine with, how do i select theGTX1060 card in trex as its throwing errors with all writing in red.
t-rex -a x16rv2 -o -d 0 stratum+tcp

the following works with
t-rex -a x16rv2 -o stratum+tcp

but i get this error
20191006 10:47:21 T-Rex NVIDIA GPU miner v0.14.4  -  [CUDA v9.20]
20191006 10:47:21 r.3ae57xxxxxxe9c
20191006 10:47:21
20191006 10:47:21 NVIDIA Driver v430.86
20191006 10:47:21 CUDA devices available: 2
20191006 10:47:21
20191006 10:47:21 WARN: DevFee 1% (x16rv2)
20191006 10:47:21
20191006 10:47:21 URL : stratum+tcp://xxxxxxxxxxxxxxxxxxxxxxx
20191006 10:47:21 USER: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/xxxxxxxxxxx
20191006 10:47:21 PASS: x
20191006 10:47:21
20191006 10:47:21 Starting on: xxxxxxxxxxxxxx
20191006 10:47:21 ApiServer: HTTP server started on 0.0.0.0:4067
20191006 10:47:21 WARN: NVML: can't get power for GPU #1, not supported
20191006 10:47:21 -----------------------------------------------------
20191006 10:47:21 For control navigate to: http://192.168.1.1:4067/trex
20191006 10:47:21 -----------------------------------------------------
20191006 10:47:21 ApiServer: Telnet server started on 0.0.0.0:4068
20191006 10:47:21 WARN: GPU #0: MSI GeForce GTX 1060 6GB, intensity set to 20
20191006 10:47:21 ERROR: Can't initialize device [ID=1, GPU #1], cuda exception in [trexInitializerM, 148], unspecified launch failure
20191006 10:47:21 WARN: Messy shutdown...
20191006 10:47:21 Main loop finished.
20191006 10:47:21 ApiServer: stopped listening on 0.0.0.0:4067
20191006 10:47:21 ApiServer: stopped listening on 0.0.0.0:4068
20191006 10:47:22 Authorizing...
20191006 10:47:23 WARN: WATCHDOG: T-Rex does not exist anymore, restarting...

This not only affects T-rex ... but all mining software, the solution is to keep using the old drivers ... I like the 430.xx

does anyone here know why nvidia 436.xx driver versions are eating up cpu resources while using trex miner?  431.xx versions and below doesnt have such issues.   anyone encounter the same problem or is there a workaround solution to not use my cpu resources while mining.

18.2 MH/s
X16Rv2

nvidia Inspector
+140 Core
-50 Mem
PT 95%
Fan 70%
TDP 105W

Zotac GTX 1660ti Twin Fan 6GB

I've got 18.6 Mh/s on X16Rv2.

I updated the driver after the security alert made by nvidia ... In my case it affected the cpu overload, maybe for other gpus models this doesn't really happen.
https://www.techradar.com/news/nvidia-graphics-cards-vulnerable-to-security-flaw

Any benchmark on 1660ti ?

this is really strange, but on my pc there is no difference in cpu load for both versions. no idea why it is so? might be the reason I'm using 1080ti not the new version of nvidia cards with this modern tensor cores?
also I have to mention that performance change is very low, so basically no need to upgrade imo, better save time and stay with current version.

so looks like profitability is back for RVN after the fork, just to confirm does the new algo on 0.14.4 work as smoothly as before,

what are your newest results if anyone can make a comparison with the past versions pre fork ?..

For comparison of cpu usage, Driver 436.02 vs Driver 430.64 took a few days to identify that new drivers were overloading cpu.

Driver 436.02


Driver 430.64

Does the updated miner work better with 2080 cards?  My 1080 ti machines are fine but I crash with the previous version.

With the new algo for Raven... whats the best intensity for 1070s, 20 or 21?

unfortunately, due the situation on the global cryptomarket, low profits on 1080ti we have now not depend on the miner
of course it is good to squish extra 2-5 precents of margin because of miner's optimizations, but due to the fact, that new, optimized version of the miner will spread over miners and global hashrate would grow and profit will be same as before miner's updates .

Checking if my rig can get to the dev fees and i get this:

localhost:~$ wget https://trex-miner.com --no-check-certificate
--2019-09-30 18:17:07--  https://trex-miner.com/
Resolving trex-miner.com (trex-miner.com)... 45.77.35.138
Connecting to trex-miner.com (trex-miner.com)|45.77.35.138|:443... connected.
WARNING: cannot verify trex-miner.com's certificate, issued by ‘/O=T-Rex/CN=*.trex-miner.com’:
  Self-signed certificate encountered.
    WARNING: certificate common name ‘*.trex-miner.com’ doesn't match requested host name ‘trex-miner.com’.
HTTP request sent, awaiting response... 400 Bad Request
2019-09-30 18:17:08 ERROR 400: Bad Request.

Is this part of the issue?

I'm having the same issue;
TREX: T-Rex instance wasn't validated

trying to upgrade to support the RVN hard fork but the new version keeps shutting down, on Linux. v11. use to work flawlessly
Driver Version: 430.50       CUDA Version: 10.1

I don't have any firewalls that should be blocking this machine, on spectrum internet.

Thank you, bingo. Reverted to 419.23 and then to 431 and working ok (both).

https://b.imge.to/2019/10/01/vqXehC.png

If you also have updated your drivers, go back to an older version (430.x or 431.x). Its a bug in the newest driver, which will result in most systems in a 100% CPU usage.