I’m running EVGA GeForce GTX 1660 SC Ultra with PhoenixMiner 5.4c, hoping someone can point me in the right direction for ideal settings to mine Ethereum.
So far my best stable settings appear to be +125 GPU +900 MEM 75% PL which is giving me 57C @ 70W, 26.7 MH/s, which is pretty close to the stock results in that other thread.
I am getting an error message in my log files: “Unknown OpenCL driver version! Hashrate and stale shares may suffer”
Likewise when I attempt to run -straps I get “hwmc GPU1: unable to set straps - unsupported GPU 218410DE”
Is it the “recommende” or the “optional” driver version? I had an issue with the opencl error on a 3060ti because I downloaded the optional one instead of the recommended, more stable version.
It was using the default driver that Windows 10 chose to install, v 432.00
I went to nvidia.com for the latest drivers, and found a “game ready” and a “studio” driver, but no “recommended” or “optional”. I tried both game ready and studio, however both versions gave me errors from the -cclock +125 -mclock +900 lines in the .bat file and would only give me 21.4 MH/s
I then tried downloading the latest drivers from evga.com, however I had the same problems with those.
Finally I reverted back to the 432.00 drivers I started with, and am back at 26.7 MH/s, however it’s now using 98 watts @ 64C and is unstable, it crashed as I’m typing this.
Your error “Unknown OpenCL driver version” results from your already crashed GPU.
I would knock that Core GPU clock to 100, take the mem clock to 850 then lower the PL until the core clock on the display of Afterburner can’t maintain 1500. You definitely don’t want the clock to fall below 1300 as it will drop like a rock if the power level is again lowered. Like I said, shoot for 1500.
This should give you a stable run with reduced temperature. You can set a fan curve or manual fan speed such that you get close to 80% fan.
Once you have a stable run, then you can raise your core gpu back to 125 and test for stability. If not, return to 100. Do the same separately with the mem clock. Raise to 900 and see if stable. If not, return to 850. Leave the PL as low as possible to get 1300-1500 core speed on the display.
You have a 1660, not a SUPER and NOT a TI. I get 30MH/s on a 1660TI so 26 MH/s on a 1600 sounds about right. You want stability most of all, and each card is different. pushing limits will cause crashes.
Other errors you may receive would include “not able to read temperature on gpu #xxx”, “memory crash, not able to access memory on gpu #xxx” These are crashed Gpu errors, most of which require a complete reboot rather than just restarting the miner.
Phoenix miner also allows an intensity setting and kernels 1-4 that can be tested or forced. with the -mi which can be 0-14 and the -nvKernel . I use <0> which is auto. The choices are: Type of Nvidia kernel: 0 auto (default), 1 old (v1), 2 newer (v2), 3 latest (v3).
I followed your suggestions and was able to take the PL all the way down to 53% which appears to be the lower limit in Afterburner. At this level the GPU clock hovers around 1545-1590, with -cclock +125 -mclock +900.
I tried various -mi and -nvKernel settings, however none of them had positive results so I’ve just removed them from the .bat file.
Also, despite a couple cold reboots I’m still getting that “Unknown OpenCL driver version” error, although the very next line in the log files reads “OpenCL platform: OpenCL 2.1 AMD-APP (3075.13)”
I’m getting 26.7 Mh/s at 56C, 70W now.
One other question, in Afterburner Power Limit is linked by default to Temp Limit, dropping the PL to 53% lowered the TL to 65C, should I leave that linked?
I typically unlink the PL and TEMP settings by breaking the chain icon. If the temp limit is too low, it will throttle performance. 56c or lower is where you want to be but the temp limit should be higher like 85C. You won’t be getting there because you will set the fan curve or manual speed such that the fans run between 60-85% all the time. Control the temp with the fans.
Only good cards can run continuously at 900 mclock. Many can run with cclock at 125 but they are more stable around 100 cclock.
Now for the good part. The Driver you talk about so much and are having problems with cause changes in Phoenix Miner. Notice the CLI (dos screen) when phoenix starts up. What CUDA version does it load? The 1660 will only run with CUDA 10, and won’t run CUDA 11. I found by trying several driver versions that Phoenix Miner will run CUDA 10 with driver version 442.71 or older and everything should be fine. Installing a newer driver version like 461.40 will force Phoenix Miner into CUDA 11 mode and will not run your 1660’s.
The opposite is true of 3000 series gpu cards. They must have the latest drivers because they have to run in CUDA 11.1
To get older drivers, open a browser, and google nvidia driver download.
Next, even though it looks like you are on the correct page, you’re not…so click the drivers TAB at the top of the page. Select ALL nVidia drivers. Geforce drivers are for 16xx series cards, RTX are for 20xx series cards but that’s not important.
Scroll down the page to the list where the top line says “beta and older drivers”. On the new page, select Geforce, 1600 series, GTX 1660, and SEARCH.
You now have a list of ALL the older drivers. I prefer WHQL drivers and not BETA drivers.
It seems to be running stable now so I don’t really want to alter it at the moment. It’s loading CUDA 10.
I added -powlim -47 to the .bat file and am not using Afterburner at all, the speed is holding steady at 26.7 MH/s and the GPU is stable at 54C. I haven’t specified anything in the .bat for the fan speed or temp limit. I decided not to use Afterburner to control it so that I can have it reboot and start mining automatically in case of power interruption.
It’s going on nearly 24 hours continuous at +125/+900, if I encounter any instability I’ll keep your thoughts in mind and dial it back a bit.
Thanks, glad you can make it work without Afterburner. There is an afterburner setting to restart with windows which only works if the lock is open on the SAVE SETTINGs button. It will restart using the same settings that were running when the pc rebooted.
Those are good values, and if stable, you’ve solved your problem.