|
PhoenixMiner (OP)
|
 |
January 20, 2019, 08:23:13 AM |
|
What is the best driver for ethereum? Im using 18.6.1
In terms of hashrate there is no big difference. We found that 18.5.1 are the best as far as stability goes but YMMV. i use Adrenalin 2019 Edition 18.12.2 and rx580 8gb but in this new 4.X ver i can't use -tt and -tmax and everything about fan control is not working
in ver 3.5 i set this and worked perfectly
-tt 47
-tmax 52
-fanmin 10
-fanmax 55
but now i use the same commend but when gpu go to 59 degree fan start on 24% and after it's cooling they be shutdown and again this Repeat 59start .........
This is most probably driver related. Downgrade to 18.5.1 or similar drivers. There are hardware control bugs in the latest AMD drivers. We will try to think of some workaround but the for now the best solution is to stay away from the latest drivers. Noticed the same thing with the latest version and some RX580 under Linux.
Had -tt set to 68 and one GPU kept creeping up way past 80 degrees and the fan was idling at 23%.
-tt is not sported under Linux yet. Sadly 4.1c does not support 19.1.1, as it gives the Unknown OpenCL version error and suffers a hashrate loss because of it.
-clkernel 3 mode does not even work with it. ( clSetKernelArg (-49) )
I have 19.1.1 installed and latest version of miner works normal. RX 580 8 Gb Windows 10. That`s strange... Well, I guess I can go back to 18.12.3 until we get a fix.
2019.01.19:21:21:33.279: main Phoenix Miner 4.1c Windows/msvc - Release build
2019.01.19:21:21:33.279: main Cmd line: -pool @@@
2019.01.19:21:21:33.279: main No CUDA driver found
2019.01.19:21:21:35.924: main OpenCL driver version: 19.1.1
2019.01.19:21:21:35.924: main Unknown OpenCL driver version! Hashrate and stale shares may suffer
2019.01.19:21:21:35.924: main OpenCL platform: OpenCL 2.0 AMD-APP (2348.3)
2019.01.19:21:21:35.924: main Ignoring old OpenCL driver(s), using driver version 19.1.1
2019.01.19:21:21:35.924: main Available GPUs for mining:
2019.01.19:21:21:35.924: main GPU1: Radeon (TM) RX 480 Graphics (pcie 1), OpenCL 2.0, 8 GB VRAM, 36 CUs
2019.01.19:21:21:36.002: main ADL library initialized In your log we can see that you have two versions of the drivers mixed up. The 19.1.1 are recognized properly but there are others too (probably "leftovers" from the previous driver version). You should use DDU to completely remove the drivers and then install 18.5.1 or other drivers besides 18.12.x and 19.1.x, which seems to cause problems. I see totally minuscule reported hasharete improvement with "3" kernels, wonder if there would be any real hashrate improvement.
The real hashrate improvement will be the same in the long run. However make sure to check your power consumption as it may increase in some cases, negating the benefit of the slightly higher hashrate. |
|
|
|
|
|
|
|
|
|
|
|
|
|
The forum was founded in 2009 by Satoshi and Sirius. It replaced a
SourceForge forum.
|
|
|
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
|
|
|
|
|
singlass
Jr. Member
 Offline
Activity: 65
Merit: 1
|
|
January 20, 2019, 05:14:47 PM |
|
finally !! thank you for the -clkernel 2 i used to be forced to use the generic -clkernel 0 which was instable & produced a lotta stale shares
|
|
|
|
|
kevyong
Newbie
Offline
Activity: 6
Merit: 0
|
|
January 21, 2019, 06:35:19 AM |
|
Still no Yocoin for devfee to avoid switching DAGs while mining Yocoin?
Thank you!
|
|
|
|
|
gploj
Newbie
Offline
Activity: 3
Merit: 0
|
|
January 21, 2019, 09:56:38 AM |
|
hello,
how write comand line for turbo mode ?
On this rig must have only 8Gb cards ? I have 8x 8Gb and 1x 4Gb is this problem ?
Tnx
|
|
|
|
|
bategojko74
Member
Offline
Activity: 220
Merit: 12
|
|
January 21, 2019, 12:04:50 PM |
|
hello,
how write comand line for turbo mode ?
On this rig must have only 8Gb cards ? I have 8x 8Gb and 1x 4Gb is this problem ?
Tnx
clkernel -3 recognizes the GPU RAM size and switches automatically the cards with 4GB to clkernel -1. Sou you will not have a problem by simply passing clkernel -3. If you like your 4GB card to use clkernel -2 you have to pass clkernel 333333332 if your 4GB card is with index 9, otherwise you have to put the '2' on a position equal to the 4GB card index. You can find the index of the GPU in PM's log when it initially lists the available GPUs. |
|
|
|
|
ardeneast
Newbie
Offline
Activity: 1
Merit: 0
|
|
January 21, 2019, 01:48:08 PM |
|
Hello I use Phoenix since last spring. At first, I used it for Nice Hash and did not experience any problems, an excellent miner. Since last summer I switched to ethermine pool and everything worked fine. But from the end of autumn I constantly try to return to Nice Hash, but the miner constantly stops working. it shows an error after allocating DAG buffer. (this is visible on the screenshot of the page) In the topic such a problem is indicated, but it says that the problem lies in the small swap file. The size of my page file is more than 30 GB per 6 1080ti video cards. At the same time, the phoenix works great on the ezermine pool. And only if I switch to Nice hash - then this problem occurs. What should I do so that the phoenix could work for Nice Hash. My system is Windows 7. Video cards on this rig are 5 pieces of 1080 ti, power supply units 2 pcs corsair HX1200. RAM 4 GB, Video card driver 417.71, PhoenixMiner 4.1c https://i.gyazo.com/760b8fcdd8b9a7cea52bf11cee0b50f4.png |
|
|
|
|
gploj
Newbie
Offline
Activity: 3
Merit: 0
|
|
January 21, 2019, 01:55:19 PM |
|
hello,
how write comand line for turbo mode ?
On this rig must have only 8Gb cards ? I have 8x 8Gb and 1x 4Gb is this problem ?
Tnx
clkernel -3 recognizes the GPU RAM size and switches automatically the cards with 4GB to clkernel -1. Sou you will not have a problem by simply passing clkernel -3. If you like your 4GB card to use clkernel -2 you have to pass clkernel 333333332 if your 4GB card is with index 9, otherwise you have to put the '2' on a position equal to the 4GB card index. You can find the index of the GPU in PM's log when it initially lists the available GPUs. tnx for your answer. I test it. Power consuption is the same ? tnx |
|
|
|
|
carlosmonaco
Newbie
Offline
Activity: 105
Merit: 0
|
|
January 21, 2019, 06:13:14 PM |
|
Hello ,
What are the best setting for VEGA 56 and 64 for ETH ?
|
|
|
|
|
|
PhoenixMiner (OP)
|
|
January 22, 2019, 08:00:45 AM |
|
Hello
I use Phoenix since last spring. At first, I used it for Nice Hash and did not experience any problems, an excellent miner.
Since last summer I switched to ethermine pool and everything worked fine. But from the end of autumn I constantly try to return to Nice Hash, but the miner constantly stops working. it shows an error after allocating DAG buffer. (this is visible on the screenshot of the page) In the topic such a problem is indicated, but it says that the problem lies in the small swap file. The size of my page file is more than 30 GB per 6 1080ti video cards. At the same time, the phoenix works great on the ezermine pool. And only if I switch to Nice hash - then this problem occurs. What should I do so that the phoenix could work for Nice Hash.
My system is Windows 7. Video cards on this rig are 5 pieces of 1080 ti, power supply units 2 pcs corsair HX1200. RAM 4 GB, Video card driver 417.71, PhoenixMiner 4.1c
The page file should be enough for six cards. Instability during DAG generation could indicate too much overclocking - try to lower the memory overclock a little (especially if the crashes occur on the same GPU). Although your PSUs seem more than enough for the system, you can also try the option -gser 1 (or bigger value, if 1 doesn't help). Finally, you can also try to set -gpureset 0 . If nothing helps, please cut the last few hundred rows from the log file before the crash and send it to us (via PM for example) to see what is going on. Hello ,
What are the best setting for VEGA 56 and 64 for ETH ?
Depends on the card but basically you have to experiment a little - overclock the memory as much as it can without losing stability and lower the core clock as much as you can without affecting the hashrate too much (the temperatures and the power consumption depend mostly on the core clock). Also, make sure that the temperature is kept no more than 60-64 C because the memory is stacked on the chip and can get quite hot even the reported temperature is relatively low. |
|
|
|
|
cadj92
Newbie
Offline
Activity: 3
Merit: 0
|
|
January 22, 2019, 12:43:19 PM |
|
Sadly 4.1c does not support 19.1.1, as it gives the Unknown OpenCL version error and suffers a hashrate loss because of it.
-clkernel 3 mode does not even work with it. ( clSetKernelArg (-49) )
I have 19.1.1 installed and latest version of miner works normal. RX 580 8 Gb Windows 10. That`s strange... Well, I guess I can go back to 18.12.3 until we get a fix.
2019.01.19:21:21:33.279: main Phoenix Miner 4.1c Windows/msvc - Release build
2019.01.19:21:21:33.279: main Cmd line: -pool @@@
2019.01.19:21:21:33.279: main No CUDA driver found
2019.01.19:21:21:35.924: main OpenCL driver version: 19.1.1
2019.01.19:21:21:35.924: main Unknown OpenCL driver version! Hashrate and stale shares may suffer
2019.01.19:21:21:35.924: main OpenCL platform: OpenCL 2.0 AMD-APP (2348.3)
2019.01.19:21:21:35.924: main Ignoring old OpenCL driver(s), using driver version 19.1.1
2019.01.19:21:21:35.924: main Available GPUs for mining:
2019.01.19:21:21:35.924: main GPU1: Radeon (TM) RX 480 Graphics (pcie 1), OpenCL 2.0, 8 GB VRAM, 36 CUs
2019.01.19:21:21:36.002: main ADL library initialized In your log we can see that you have two versions of the drivers mixed up. The 19.1.1 are recognized properly but there are others too (probably "leftovers" from the previous driver version). You should use DDU to completely remove the drivers and then install 18.5.1 or other drivers besides 18.12.x and 19.1.x, which seems to cause problems. Thanks, I always use DDU, but it seem the update to 1809 switched back Window's auto-driver-download option. Switching it off and installing a clean 19.1.1 after a DDU solved the problem. |
|
|
|
|
|
VasilyS
|
|
January 22, 2019, 03:43:14 PM |
|
Today I launched Phoenix miner 4.1c in dual mode with blake2s on three R9 290. They recently gave about 23 mh/s on Ether in solo. Unlike the RX series, there is no support in AMD drivers for a large DAG size for old cards, so their speed, which was once 30-31 max, has seriously decreased. Got a fantastic speed in the miner window (45 mh/s). Blake2s showed 1300 mh/s, but there were no shares at all.
I was very pleased, but after checking on the pool (Ethermine) it turned out that these are only numbers in the miner's console. The real speed remained 22-23 mh/s as it was earlier.
|
|
|
|
|
VasilyS
|
|
January 22, 2019, 08:03:37 PM |
|
It seems to me that your last version doesn't support R9 290 cards while dual mining (Ether and Blake2s algo). I tested different kernels and there are a lot of errors. I switched back to Claymore.
|
|
|
|
Bigdrago
Newbie
Offline
Activity: 312
Merit: 0
|
|
January 22, 2019, 10:23:49 PM |
|
I am using 4.0B without any issues.
Does this only affect if the miner crash and have to restart?
|
|
|
|
|
ALuckyGuy
Newbie
Offline
Activity: 2
Merit: 0
|
|
January 23, 2019, 01:06:30 PM |
|
Just wanted to post a little thank you for your continued work and the new turbo core.
Using 4.1c with the 19.1.1 Adrenalin drivers and my RX580... stable all week and hitting just slightly over 32 MH/s.
|
|
|
|
|
|
PhoenixMiner (OP)
|
|
January 24, 2019, 07:22:39 AM |
|
Today I launched Phoenix miner 4.1c in dual mode with blake2s on three R9 290. They recently gave about 23 mh/s on Ether in solo. Unlike the RX series, there is no support in AMD drivers for a large DAG size for old cards, so their speed, which was once 30-31 max, has seriously decreased. Got a fantastic speed in the miner window (45 mh/s). Blake2s showed 1300 mh/s, but there were no shares at all.
I was very pleased, but after checking on the pool (Ethermine) it turned out that these are only numbers in the miner's console. The real speed remained 22-23 mh/s as it was earlier.
We have tested extensively with R9 390, which is virtually the same as 290. Most probably your drivers aren't recognized properly. Please check the logs - you will see a yellow warning message "Unknown OpenCL driver version! Hashrate and stale shares may suffer". Here is the list of supported drivers for Windows: 15.12, 17.7.2, 17.8-BETA (Blockchain drivers), 17.8.x, 17.9.x, 17.10.x, 17.11.x, 17.12.x (avoid these, some bugs), 18.x.x, 19.1.1 If you are using one of the supported drivers but the problem still persists, please send us the first few pages from the log file to diagnose the problem. I am using 4.0B without any issues.
Does this only affect if the miner crash and have to restart?
The DAG issue was already solved in 4.0b (it supports DAGs over 235). So, you don't need to upgrade unless you want some of the other new features. Just wanted to post a little thank you for your continued work and the new turbo core.
Using 4.1c with the 19.1.1 Adrenalin drivers and my RX580... stable all week and hitting just slightly over 32 MH/s.
Thank you, always nice to hear!  |
|
|
|
|
ledgerdary
Newbie
Offline
Activity: 7
Merit: 0
|
|
January 24, 2019, 09:52:24 AM |
|
Which driver would you recommend for Vega 56s? And is there perhaps any other parameter that you suggest I vary in particular to increase the hash rate?
Thanks in advance for your reply. Phoenix, keep up the good work. I really like your miner!
|
|
|
|
|
Bigdrago
Newbie
Offline
Activity: 312
Merit: 0
|
|
January 24, 2019, 12:40:58 PM |
|
Getting errors with clkernel 3
gpu 7: clenqueueCopyBuffer (-4)
Thread(s) not responsding. Restarting
gpu 9: clenqueueCopyBuffer (-4)
gpu 8: clenqueueCopyBuffer (-4)
gpu 10: clenqueueCopyBuffer (-4)
gpu 11: clenqueueCopyBuffer (-4)
gpu 13: clenqueueCopyBuffer (-4)
gpu 12: clenqueueCopyBuffer (-4)
I have 3 identical rigs (13x radeon 570 8gb). Rig 1 is working (speed rised from 408 to 411). Rig 2 working after restart. Rig 3 is not working, have to use clkernel 1.
|
|
|
|
|
Xoide
Newbie
Offline
Activity: 2
Merit: 0
|
|
January 24, 2019, 01:39:28 PM |
|
Hello! I am new here but have been tweaking my hynix cards for about a year now and just got a perfectly stable rx580 hynix bios that is currently running 29,5/MHS - 31/MHS depending on mem clocks, if you set mem to 2100 and core to 1150 you should see about 29 - 30 MHS. Power consumtion is around 90 to 95 for each card but you can tweak this more if you want to, i am happy with those settings though. Just reply to my inbox if you want the bios and i will send it to you. Kindest regards Xoide Smiley
EDIT: The cards is ASUS ROG-STRIX-RX580-O8G-Gaming
|
|
|
|
|
twotwosix
Newbie
Offline
Activity: 118
Merit: 0
|
|
January 24, 2019, 04:48:58 PM |
|
Getting errors with clkernel 3
gpu 7: clenqueueCopyBuffer (-4)
Thread(s) not responsding. Restarting
gpu 9: clenqueueCopyBuffer (-4)
gpu 8: clenqueueCopyBuffer (-4)
gpu 10: clenqueueCopyBuffer (-4)
gpu 11: clenqueueCopyBuffer (-4)
gpu 13: clenqueueCopyBuffer (-4)
gpu 12: clenqueueCopyBuffer (-4)
I have 3 identical rigs (13x radeon 570 8gb). Rig 1 is working (speed rised from 408 to 411). Rig 2 working after restart. Rig 3 is not working, have to use clkernel 1.
What menory brand? |
|
|
|
|
Bigdrago
Newbie
Offline
Activity: 312
Merit: 0
|
|
January 24, 2019, 04:49:55 PM |
|
Getting errors with clkernel 3
gpu 7: clenqueueCopyBuffer (-4)
Thread(s) not responsding. Restarting
gpu 9: clenqueueCopyBuffer (-4)
gpu 8: clenqueueCopyBuffer (-4)
gpu 10: clenqueueCopyBuffer (-4)
gpu 11: clenqueueCopyBuffer (-4)
gpu 13: clenqueueCopyBuffer (-4)
gpu 12: clenqueueCopyBuffer (-4)
I have 3 identical rigs (13x radeon 570 8gb). Rig 1 is working (speed rised from 408 to 411). Rig 2 working after restart. Rig 3 is not working, have to use clkernel 1.
What menory brand? I think that rig have samsung micron |
|
|
|
|
|
