Topic: PhoenixMiner 6.2c: fastest Ethereum/Ethash miner with lowest devfee (Win/Linux) - page 17. (Read 784958 times)

Hi all,

I posted this question couple of days before, but was deleted. I wonder why, since it's fully on-topic, and perhaps a simptom of a problem woth PhoenixMiner (or it's just me, but the question is honest).
Well, the cas is  I got an out of memory problem mining ETC (eth classic) with my 3GB nVidia GTX1060, since a couple of days ago.
Coincidentally with change to epoch #238 (5.9d say's to be good up to epoch #240 and 3gb cards should be able to mine etc until june/22 or so).
Any one with the same issue?

This is the command line:
./PhoenixMiner -pool us1-etc.ethermine.org:4444 -wal 0xblablabla -worker myworker -pass x -coin etc

This is the output:
Phoenix Miner 5.9d Linux/gcc - Release build
--------------------------------------------

CUDA version: 11.0, CUDA runtime: 8.0
NVML library initialized
Nvidia driver version: 470.86
OpenCL driver version: 20.40-1147287
Available GPUs for mining:
GPU1: ASUS NVIDIA GeForce GTX 1060 3GB (pcie 1), CUDA cap. 6.1, 3 GB VRAM, 9 CUs

GPU2: ASUS NVIDIA GeForce GTX 1060 3GB (pcie 2), CUDA cap. 6.1, 3 GB VRAM, 9 CUs

Eth: the pool list contains 1 pool (1 from command-line)
Eth: primary pool: us1-etc.ethermine.org:4444
Starting GPU mining
Eth: Connecting to ethash pool us1-etc.ethermine.org:4444 (proto: EthProxy)
GPU1: 35C 26% 7W, GPU2: 36C 27% 40W
GPUs power: 47.1 W
Eth: Connected to ethash pool us1-etc.ethermine.org:4444 (172.65.245.119)
Listening for CDM remote manager at port 3333 in read-only mode
Eth: New job #e801904e from us1-etc.ethermine.org:4444; diff: 4000MH
GPU1: Starting up... (0)
GPU1: Generating etchash light cache for epoch #238
GPU2: Starting up... (0)
Eth: New job #764e3116 from us1-etc.ethermine.org:4444; diff: 4000MH
Light cache generated in 2.0 s (22.7 MB/s)
GPU2: Allocating DAG (2.87) GB; good for epoch up to #240
GPU2: Generating DAG for epoch #238
GPU1: Allocating DAG (2.87) GB; good for epoch up to #240
CUDA error in CudaProgram.cu:393 : out of memory (2)
GPU1: CUDA memory: 2.95 GB total, 2.87 GB free
GPU1 initMiner error: out of memory
Fatal error detected. Restarting.
GPU2 initMiner error: Unable to initialize CUDA miner
Eth speed: 0.000 MH/s, shares: 0/0/0, time: 0:00
GPUs: 1: 0.000 MH/s (0) 2: 0.000 MH/s (0)

Hope somebody could help, and thanks in advance for any hint,

What usually works with upgrades before, but now, no...

Won't even work with the supplied batch file.

Latest version, on hiveos and windows, my GTX 1070 make only 27mhs. Before it was always above 30mhs.

How can i get with Phoenix 30mhs?

Latest version, on hiveos and windows, my GTX 1070 make only 27mhs. Before it was always above 30mhs.
How can i get with Phoenix 30mhs?

Just copy/paste the command line or config from V5.8 to V5.9, and it should work.

The new beta version is ready. It is mostly a bugfix release, the more significant upgrades are coming next month. You can download PhoenixMiner 6.0a from here:

The new features in this release are:

• Full LHR disable mode -lhrdis   1 - yes (default), 0 - no
• Show the GPU vendor name in the list of GPUs to make it easier to identify the GPUs
• Added support for the latest AMD Linux drivers 21.40.1. There are some bugs in these drivers, particularly the clocks and voltages can't be set properly with older
  cards (RX4x0/RX5x0/Vega/RadeonVII)
• Validated support for the latest AMD Windows drivers up to 21.11.2
• Fixed issues with AMD RX6700XT cards with the latest AMD and Linux drivers
• Fixed issues with AMD Vega and Radeon VII cards on latest Windows and Linux drivers
• Fixed crash with very old Nvidia drivers (3xx.x)
• Other fixes and small improvements

The updated list of known issues (driver incompatibilities, etc.) and workarounds:
AMD Linux driver 21.40.1 has a bug preventing proper setting of clocks and voltages on older cards (RX4x0/RX5x0/Vega/RadeonVII). Given that these drivers are also
usually slower than the older drivers for these cards, we recommend using Linux drivers 20.30 for anything older than RX6000 series.
If you are using Linux drivers 21.40.1 with Radeon VII cards, you need to add the option -fpwm 1 in order to have proper fan control.
AMD Linux drivers 21.40.1 has finally removed the requirement of PCIe atomics but there are problems when you try to mix Polaris (RX4x0/5x0) cards and Vega or newer
cards on the same rig.
Some Nvidia cards will report a lot of stale shares under Windows 11. Using the same driver version under Windows 10 resolves the issue.

I have a potato Asus PCIe2.0 with a RX580 8g (I7 12g ram) (driver fixed)

I'm finally using W10 after 7 was only giving me 1/3 hash I should have for a year (yes, I'm stubborn)

now, I'm getting the 28+hash I should have had instead of 8-9 at least, but...

I'm on 5.9d AND 21.11.3, but still have to use ckernal 0 to get it to work and that gives me abt 7-10% stales!

Is this normal? Even with the low hash before I never had this high a stale rate, and didnt have to use the ck0 until recently (on my old miner version with no NEW win 7 updates or drivers this started happening out of the blue)

Thanks for any ideas or help!!

What does nvidia-smi say abt cuda version?

C:\Windows\system32>nvidia-smi
Tue Jan 25 17:14:28 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 511.23       Driver Version: 511.23       CUDA Version: 11.6     |
|-------------------------------+----------------------+----------------------+

So I managed to sort out my issues with 5.8. Windows was insisting on updating the drivers, I had been relying on the DDU option to prevent driver update but this did not seem to work. So I had to edit the local policy GPEDIT to prevent the AMD cards from getting updated

https://www.howtogeek.com/263851/how-to-prevent-windows-from-automatically-updating-specific-drivers/

Back to being stable

Still no idea why PM 5.9 starts and immediately exits tho'

I have a rig with 6 x 6700s and 6 Nvidia cards. Windows 10 with latest updates

Has been working ok on 5.8 but has started causing me issues, generating errors on PM start:


No idea why, has been working and stable(ish) for months. Maybe windows update.

So, I thought it might be time to update, so updated gfx drivers and trying to update to 5.9d, but just starts and immediately exits 


Does the same on two of my other rigs, using both my config.txt and the default startminer.cmd


Am I missing something stupid?

I have a potato Asus PCIe2.0 with a RX580 8g (I7 12g ram) (driver fixed)

I'm finally using W10 after 7 was only giving me 1/3 hash I should have for a year (yes, I'm stubborn)

now, I'm getting the 28+hash I should have had instead of 8-9 at least, but...

I'm on 5.9d AND 21.11.3, but still have to use ckernal 0 to get it to work and that gives me abt 7-10% stales!

Is this normal? Even with the low hash before I never had this high a stale rate, and didnt have to use the ck0 until recently (on my old miner version with no NEW win 7 updates or drivers this started happening out of the blue)

Thanks for any ideas or help!!

Now I'm but it's 22-23MH trying different configurations now to get better but still the same (using -clkernel 0 by the way to get it to work)

Is compute mode on?

I'm back to mining after stopping for over 2 years, but I only getting max 14Mh/s hash tried modded bios and stock bios. Is this normal?
Edit: Sorry forgot to say it's Rx570 4GB mining ETC

The way to keep it from trying to put a new driver version back in is to go to device manager, go to display adapters, find yours, and then ROLL BACK to the previous driver version.   It won't update the driver again until there is a new version of Windows(going to Windows 10 or 11 22H1 for example).   My other question though, is why PhoenixMiner just breaks with new AMD driver versions.   If you intentionally install anything newer than Radeon Software 21.11.3 for RDNA 2 cards(including my Radeon 6800), the thing comes back with the unknown version issue and it won't mine without -clkernel 0.    So, is that a hardcoded "look for driver version" and then break as a result, because it should TRY to use the proper code path for these newer drivers, but it seems to be doing a fallback if it doesn't recognize the driver version.

I'm back to mining after stopping for over 2 years, but I only getting max 14Mh/s hash tried modded bios and stock bios. Is this normal?
Edit: Sorry forgot to say it's Rx570 4GB mining ETC

I have a potato Asus PCIe2.0 with a RX580 8g (I7 12g ram) (driver fixed)

I'm finally using W10 after 7 was only giving me 1/3 hash I should have for a year (yes, I'm stubborn)

now, I'm getting the 28+hash I should have had instead of 8-9 at least, but...

I'm on 5.9d AND 21.11.3, but still have to use ckernal 0 to get it to work and that gives me abt 7-10% stales!

Is this normal? Even with the low hash before I never had this high a stale rate, and didnt have to use the ck0 until recently (on my old miner version with no NEW win 7 updates or drivers this started happening out of the blue)

Thanks for any ideas or help!!