Thanks for your reply. Your example shows me maybe doing something wrong way with my rig, or you got lucky with yours. Probably the first. Now around the holidays, I should have extra time to work on it and see what I'm doing wrong.

Of course, everyone is very welcome to share their thoughts on what I wrote above, and I'm open to any ideas on how to resolve my issues, but my main idea with the post was to remind the devs about their promise a while ago ( when released V5.9) for big changes in the miner this month. The month and the year are almost gone.

I would say it works somehow, but there are still a lot of bugs needed fixing.

My problems with PM started back in May or June when I try to run RX 6700 XT in HiveOS. First, I had to run an older version of PM to make it works. Then, I found -clKernel 0 is fixing this issue, but the GPUs which are running with it, are producing a lot of invalid shares, and the hashrate is unstable. Since then, I added some RX6600's to that rig, and PM had a few updates, but I never got my RX 6x00 cards working with -clKernel 1. Anyway I believe, this issue is on the HiveOS side (I try to contact them, but no answer from them yet) because PM requires current new drivers, but Hive sticks with old ones. I think PM devs need to work with Hive to resolve it.

Now back to my Windows rigs.
When I run any of my RX 6x00 GPUs on my testing PC as a single card with current PM, I do not have any issues. Frequencies and voltages changes as expected  -clKernel 1 works fine, and cards are singing.
All my problems are starting after moving the cards to a mixed rig. Unfortunately, I don't have enough GPUs and free space to combine only cards of the same type.
After swapping GPUs around my rigs, the current situation is:
Running Windows 10 Pro 21H2 with AMD driver 21.8.2 minimal required for RX 6600 XT. Driver 21.8.1 is more unstable, and higher versions do not make any better.
The RX 6600 XT now is running as expected, but in the past, on a different rig with driver  21.10.2 I had to run it with -clKernel 0 together with some other BigNavi cards.
The problem now is an RX 5500 XT which must be run with -clKernel 0 and it does not accept the commands -cclock and -cvddc from the miner unless I do not change OC settings in the AMD software before I start the miner. This makes the card jump power usage from 50-52 Wats to over 80 Wats with no changes on the hashrate.
Another issue shows up on this rig after I added the RX 6600 all Polaris cards, or randomly some of them are unable to init straps and set straps. For more details, please see parts of my log file posted below.

P.S. Also on this rig, I currently have a couple of 4G Nvidia cards, mining RVN.

RX 6600 not mining with phoenixminer.
gminer works. please update your code.

In my setting, every 30 seconds, "main Eth: Accepted shares (# of shares)" is recorded in the log. Does the number include shares found during "DevFee" or not? 

Be extremely careful and immediately delete anything that is not yet published by the Phoenixminer team itself.So far we are up only to 5.8c and in their official website which you can find in the first page of this thread 5.9b is not yet released.Probably you have got some malware and you need to delete it right away.

If you made a typo and mean 5.7b you should upgrade to latest version 5.8c and use MSI Afterburner to change memory clock,core clock and fan speed,these are all needed to be modified in order to get the max from your card.Still if you have a 5.9b definitely is malware and you need to delete it right away.

Official link for the official miner is here

https://phoenixminer.info/downloads/

Hi,

unfortunately I have a problem with all phoenixminers over version v5.4c. From phoenixminer-v5.5 I have the problem that the program tells me very quickly that a GPU fault is allegedly detected.

GPU fault detected: 146 0x0ff8c40c for process PhoenixMiner pid 9762 thread PhoenixMiner pid 9762 [ 33.172465] amdgpu 0000:03:00.0: amdgpu: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x00105DFF [ 33.178320] amdgpu 0000:03:00.0: amdgpu: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x080C400C [ 33.184106] amdgpu 0000:03:00.0: amdgpu: VM fault (0x0c, vmid 4, pasid 32769) at page 1072639, read from 'TC3' (0x54433300) (196) [ 33.243697] amdgpu 0000:03:00.0: amdgpu: GPU fault detected: 146 0x0ff8c40c for process PhoenixMiner pid 9762 thread PhoenixMiner pid 9762 [ 33.249757] amdgpu 0000:03:00.0: amdgpu: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x00105DFF [ 33.255801] amdgpu 0000:03:00.0: amdgpu: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x080C400C [ 33.261813] amdgpu 0000:03:00.0: amdgpu: VM fault (0x0c, vmid 4, pasid 32769) at page 1072639, read from 'TC3' (0x54433300) (196)



The RIG Data

6x Powercolor RX 580 8GB
OS : simplemining
Kernel: 5.10.60-sm1#021
Driver: amd20.40r5.11.24
Auto Update: v1319
CPU: Intel(R) Pentium(R) CPU G3260 @ 3.30GHz
RAM: 4 GB
Drive: Samsung SSD 850 120GB
Motherboard: ASRock H81 Pro BTC R2.0
Motherboard BIOS: P1.20 07/22/2014

Startconfig:
-pool ssl://eu1.ethermine.org:5555 -wal walletadress.W1 -rmode 2

The same system without changing anything runs flawlessly with the phoenixminer-v5.4c at around 190 MH / s.

Thanks for any help.

Thanks for the answer.
Do you get at least 30 MH/s? I can't go over -mclock 1890 MHz, and my hashrate stays under 29.50 MH/s.

Hi everyone,
After a couple of weeks of waiting for shipment to arrive and after a couple of days of working I got these results for my XFX RX 6600 SWFT 210 :


It is a kind of a disappointment for me because I heard these GPUs should be, the best power-efficient on the current market, doing 30Mh/s for less than 50 Wats. Unfortunately, so far, this is not true. My Gigabyte RX 6600 XT is more power-efficient. 

Is anyone has better results for AMD RX 6600? Please share.
Thank you.

This is SCAM! Someone please report.

The card that works fine will work fine by itself or when in a rig with other cards.
The cards that don't work as expected behave the same if they're in a rig by themselves or with other cards.
The same cards fail regardless if I specify the voltages together or individually (eg. -cvddc 800  or -cvddc 800,800,800).

I have reinstalled drivers, reset all the settings for all cards in radeon software befopre running the miner, etc etc.
It feels almost like 2 of the 3 cards have some kind of different hardware on them that regulates voltage and isn't very well supported by phoenix, but that seems pretty unlikely given they're all the same brand and it's a new card...

Hello, I have 3 x MSI rx6600s (non-xt) cards which are behaving strangely when I use the -cvddc command. Running under windows 10.

One of the cards will happily change voltage via phoenixminer using e.g. -cvddc 750 or -cvddc 800 or whatever. Works fine using 750, pulls about 55w which is ok.

The other 2 cards automatically crash the driver (and the card won't generate a DAG) whenever I try to set core voltage via commandline for some reason - regardless of what voltage I set. Even if it's -cvddc 800 or 900 or 1000 or 1100 or whatever (more than it should need), the 2 cards won't change voltage via phoenixminer.

Both these cards WORK FINE when you don't set voltage via phoenixminer, and you can change their voltage via the radeon software no problems.
The cards behave the same regardless of which PCI slot they're in, or whether or not they use a riser.
Same behaviour in phoenixminer 5.8b or 5.8c.
Same behaviour using either AMD drivers that support 6600 non-xt cards (21.10.2 or 21.10.3)

Does anyone have any suggestions for this issue? (other than "install linux"). Is this an issue with the driver, miner, or cards? (although if this was happening on one card I would think perhaps it's the card, but 2 out of 3 seems unlikely).

Thanks in advance for any assistance.

-hstats <n>
   Level of hardware monitoring: 0 - temperature and fan speed only; 1
   - temperature, fan speed, and power; 2 - full (include core/memory
   clocks, voltages, P-states). The default is 1.

Log says it can only do relative values.

ReleaseNotes.txt:
  - Implemented optimized kernels for AMD RX6600XT cards
  - Added support for the latest AMD Windows drivers (21.10.2, 21.10.1, 21.9.2, and 21.9.1).
  - Added support for the latest AMD Linux drivers 21.30
  - Implemented full hardware control for AMD RX6900/6800/6700 cards under Linux.
    Note that with these cards under Linux you need to specify relative core voltage:
    e.g. -cclock -50 will set the core voltage to be 50 mV under the default value

Apparently it also counts for -cvddc

I use absolute value for all of my 5700XT cards and 6600. Is this special for 6600xt? The API help says "Set GPU core voltage in mV (0 for default)" and doesn't mention the relative value. May it works also.

-cvddc -100 - Apparently you have to set it to relative values.

    It shouldn't be a problem to run 5700XT and 6600 on the same rig with 21.10.2. The latest drivers will have problems only with very old cards like Baffin or Gfx804 (RX 460/550/560), or anything older than Polaris. Polaris cards (RX 470/480/570/580/590), Vega, RX5x00, and RX6x00 cards will work fine with the latest drivers. We haven't tested with 6600 yet but the we managed to order a few of them so we will be able to confirm that they work in a week or so.