 Show Posts
|
|
Pages: [1] 2 »
|
|
I have couple question regarding using the TRM.
OS - Windows 10
AMD Driver version - 21.3.2
Mobo - Asrock BTC Pro v2.0
GPU - RX5700XT from different vendors
TRM version - 0.9.0
Some background here. I have 6 GPUs. If I only use 5 I can set the primary display using IGP on the mobo (in BIOS). But if I use 6 GPUs, the mobo wont post if I try to use IGP as primary display. So I am forced to use 5700XT for primary display when using 6 GPUs.
If I don't connect a display or dummy HDMI dongle, when I start TRM, the GPU 0, used as primary display, shows very low hashrate like below.
GPU 0 [33C, fan 43%] ethash: 1.115Mh/s, avg 1.032Mh/s, pool 0.000 h/s a:0 r:0 hw:0
Once I connect a display or use HDMI dongle, the GPU will start working but with two caveats. Is monitor or HDMI dongle required?
1) Its hashrate is slightly lower than other GPUs - 56.2Mh (Primary GPU) vs 56.8Mh (Other GPU) in my case. Is this the correct behavior?
2) By default, TRM allocates 7GB vram for GPU 0. If I tried to allocate more vram using command option, the GPU 0 will be detected as dead soon after I start mining. This is correct behavior also?
|
|
|
|
RX 6600 not mining with phoenixminer.
gminer works. please update your code.
It does. I am using Windows 10/PM 5.8c and RX6000 is running well. I would say it works somehow, but there are still a lot of bugs needed fixing. I use RX6000 (non-XT) with WIndows10 21H2, AMD driver 21.10.2, PM 5.8C in a rig with five RX5700(XT) and one RX6600 without a single issue. 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. Everyone's situation could be different. In my case, I don't use Afterburner. I only use PM command line for OC settings. Below is what in config.txt -pool ssl://us1.ethermine.org:5555 -pool2 ssl://us2.ethermine.org:5555 -wal 0xxxxxxxxxxxxxxxxxxxxxxxx.yyyy -rmode 0 -log 1 -logdir C:\Users\Username\Desktop\PhoenixMiner\log -logfile PhoenixMiner.log -logsmaxsize 50 -gswin 30 -gsi 0 -hstats 2 -wdog 1 -wdtimeout 45 -tt xx.xx.xx.xx.xx -ttmem xx.xx.xx.xx.xx -cvddc xx.xx.xx.xx.xx -cclock xx.xx.xx.xx.xx -mclock xx.xx.xx.xx.xx -tmaxmem 96 |
|
|
|
RX 6600 not mining with phoenixminer.
gminer works. please update your code.
It does. I am using Windows 10/PM 5.8c and RX6000 is running well. 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. I use RX6000 (non-XT) with WIndows10 21H2, AMD driver 21.10.2, PM 5.8C in a rig with five RX5700(XT) and one RX6600 without a single issue. |
|
|
|
RX 6600 not mining with phoenixminer.
gminer works. please update your code.
It does. I am using Windows 10/PM 5.8c and RX6000 is running well. |
|
|
|
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?
I found answer for myself - No, the accepted shares doesn't include DevFee share. I did this by analyzing a log file running for 24 hours and counted all numbers of ETH share and DevFee share. |
|
|
|
|
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?
|
|
|
|
I can't modify the overclock settings of my RX 6600 with PhoenixMiner 5.9b in windows. Using AMD driver 21.11.2. It's my oc settings: -fanmin 50 -fanmax 50 -mclock 945 -cvddc 735 -mvddc 1150 -cclock 900 -amd But nothing changed. (except Core clock) What should I do?  System: Windows 10 20H2 B85 / G1840 / 4G RAM MSI MECH RX 6600 x 8 (Every card can work) AMD driver 21.11.2 Run in admin privilege Thanks. 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/5.9b has been released as Beta which you can download from https://phoenixminer.info/downloads/beta/. I am using 5.8c with following command line settings for 6600 and it works well. I am using AMD 21.10.2 -fanmin 15 -cvddc 900 -cclock 1100 -mclock 1900 I think you need to use double frenquency for mclock in WIndows. |
|
|
|
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.
If it runs without issue with PM v5.4c why you want to change? My understanding is that most latest change were for later GPUs. |
|
|
|
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.
I haven't noticed anyone reports that they can get 30mh for 6600 Non-XT. I think PM shows the highest number. Mine is about 29.4Mh from the miner. Most people say 28.9 to 29.1Mh. I set mclock to 1900 in config and the GPU actually runs at 1890Mhz. I tried using Radeon software for OC and set memory clock to 1900 maximum, the GPU still runs at 1890. I guess that it is the limit for this GPU. |
|
|
|
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 : 2021.10.30:02:20:54.046: main Available GPUs for mining:
2021.10.30:02:20:54.046: main GPU0: AMD Radeon RX 6600 (pcie 3), OpenCL 2.0, 8 GB VRAM, 28 CUs
2021.10.30:02:20:54.046: main GPU0: 42C 67% 55W
GPU0: cclock 1233 MHz, cvddc 656 mV, mclock 1890 MHz, Tj 48C, 536 kH/J
GPUs power: 55.0 W; 536 kH/J
2021.10.30:02:20:54.046: main Current -gt 15
2021.10.30:02:20:54.046: main Eth: Accepted shares 628 (3 stales), rejected shares 0 (0 stales)
2021.10.30:02:20:54.046: main Eth: Incorrect shares 0 (0.00%), est. stales percentage 0.48%
2021.10.30:02:20:54.046: main Eth: Maximum difficulty of found share: 6034.1 GH (!)
2021.10.30:02:20:54.046: main Eth: Average speed (5 min): 29.463 MH/s
2021.10.30:02:20:54.046: main Eth: Effective speed: 28.06 MH/s; at pool: 28.06 MH/s
2021.10.30:02:20:54.046: main
2021.10.30:02:20:54.061: eths Eth: Send: {"id":5,"jsonrpc":"2.0","method":"eth_getWork","params":[]}
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. I have exact Type of 6600 and PM reports 55W. I am using Windows. I saw one reported with Reddit that their GPU reports 44W with HiveOS. May be different brand. |
|
|
|
The new version is finally ready. You can download PhoenixMiner 5.8c from here:
This is SCAM! Someone please report. Already did |
|
|
|
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...
That behavior really seems to be strange. I only have one 6600 among others and it works fine with command line option for hardware settings. We will wait to see if others have clue on it. |
|
|
|
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.
If you only use one of the two cards without others, will it work? And when you use PM to specify -cvddv, do you specify one value or three values such as -cvddc 800,800,800? Note that space is not allows between the numbers. It causes PM to crash when I tried it with 5700XT. |
|
|
|
The miner log has more information like Tmem which I think is important if you want to monitor the GPU remotely. The doc says that GPU temp can be returned. How about Tmem shown in the miner log?
-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. I am asking about Jason based API. Not the command options. |
|
|
|
|
The miner log has more information like Tmem which I think is important if you want to monitor the GPU remotely. The doc says that GPU temp can be returned. How about Tmem shown in the miner log?
|
|
|
|
-cvddc -100 - Apparently you have to set it to relative values.
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. In my case, 6600 non-xt, PM takes the absolute value. The log shows below. I have "-cvddc 900" in config.txt set GPU clocks to 1100 MHz (Vddc 900 mV)What the log shows if you use absolute value for 6600xt? 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 Alright, you are using Linux and I am using Windows. |
|
|
|
-cvddc -100 - Apparently you have to set it to relative values.
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. In my case, 6600 non-xt, PM takes the absolute value. The log shows below. I have "-cvddc 900" in config.txt set GPU clocks to 1100 MHz (Vddc 900 mV)What the log shows if you use absolute value for 6600xt? |
|
|
|
-cvddc -100 - Apparently you have to set it to relative values.
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. |
|
|
|
|
Just want to report back with the status of using 5.8b/c for AMD 6600 (non-XT). When using 5.7c, I had 32% stales and 1.11% incorrect shares in 24 hours. With 5.8b, both are reduced to 0.12% and 0%. AMD driver is 21.10.2 on Windows 10. Thank you, PhoenixMiner!
|
|
|
|
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.
It is good to know and thank you for the advise. Normally I don't use AMD driver when it is released immediately as I had bad experience with 21.4.1 and 21.5.1 with 5700XT. I will try it out soon. |
|
|
|
|
