approx every 30 minutes i receive message about finished precompilation for each gpu. is looks like not ok. just started using tweaking with rx550
It is ok, because this is how CN/R works. (precompiles program every 15 blocks) It's just that previous versions did not display this to the user. is there any way to store logs in separate folder? better if without hardlinks.
Separate folders for config, pools, logs SRBMiner-CN.exe --config Config\config-normalv4.txt --pools Pools\pools-normalv4.txt --logfile Logs\log-v4.txt Of course you need to create those folders yourself inside the miner folder
|
|
|
No windows update in my case. Weird. I wish i could reproduce this but i don't know where to start. I think it is not SRB related if it worked and all of a sudden (nothing changed in batch files) it stops and won't work even after restart.. Really weird.
|
|
|
I used driver fixdag, only this driver give me 9xx h.how virtual memory should i set it, I set 42gb for this case
try number of gpu's x 8gb. So if 6gpu rig then try 48gb. Also i sent you a personal message
|
|
|
i cant seem to get miners stable at --cgputweakprofile 4,4,4,4,4 Only half seem stable (i am not on facility maybe just need to reboot or so i am using Awesome Miner)
i choose low value because default my hashrate is around 1.85Kh/s for CN R, which i think Vega should do atleast 2Kh/s
my startup commando is
--ccryptonighttype normalv4 --cgpuid 0,1,2,3,4 --cgpuworksize 16,16,16,16,16 --cgpuintensity 112,112,112,112,112 --cgputhreads 2,2,2,2,2 --cgputweakprofile 4,4,4,4,4
and use OverdriveNtools
core 1450/900 mem 935/900
are some settings off are am i missing a point/setting here?
The default ~1850hs without tweaks for a Hynix mem Vega seems ok with those clocks in SRB. So what is your hashrate with tweak_profile 4 ? Also are you cooling those Vegas enough ? Hello Dok, They tend to 1950/1980 which seems decent if remain stable I run facility with also alot of Asics running also (around 70x) The miners dont go above 60c most play around 50 to 55 So i would say i cool the good also doing 900mV on mem I get ~2100hs on my Gigabyte V56 Hynix with tweak_profile 10, but i guess i am lucky. Going to implement a parameter than 'weakens' the tweak profiles in next version so maybe you will be able to achieve more hash.
|
|
|
Anyone know a coin that is using the old swap algo? Believe it was renamed "Super Fast".
I know about FuryCoin, it uses that algo
|
|
|
Okay, got it now. Checked one of the logs and i reallt think the BC driver is the guilty one here, or maybe not enough virtual memory is allocated. Sometimes precompilation time is just right (few seconds) : [2019-06-09 22:47:53] Precompiling program [BUS_ID: 7] [2019-06-09 22:47:57] Finished program precompilation [BUS_ID: 7] [2019-06-09 22:47:58] Precompiling program [BUS_ID: 3] [2019-06-09 22:48:03] Finished program precompilation [BUS_ID: 3]
And sometimes is crazy high (few minutes) : [2019-06-11 23:47:56] Precompiling program [BUS_ID: 3] [2019-06-11 23:54:19] Finished program precompilation [BUS_ID: 3]
|
|
|
i cant seem to get miners stable at --cgputweakprofile 4,4,4,4,4 Only half seem stable (i am not on facility maybe just need to reboot or so i am using Awesome Miner)
i choose low value because default my hashrate is around 1.85Kh/s for CN R, which i think Vega should do atleast 2Kh/s
my startup commando is
--ccryptonighttype normalv4 --cgpuid 0,1,2,3,4 --cgpuworksize 16,16,16,16,16 --cgpuintensity 112,112,112,112,112 --cgputhreads 2,2,2,2,2 --cgputweakprofile 4,4,4,4,4
and use OverdriveNtools
core 1450/900 mem 935/900
are some settings off are am i missing a point/setting here?
The default ~1850hs without tweaks for a Hynix mem Vega seems ok with those clocks in SRB. So what is your hashrate with tweak_profile 4 ? Also are you cooling those Vegas enough ?
|
|
|
Thanks Doc. It has been working for weeks. Yesterday it stopped working and I rebooted the box. The miner started to be like that. My other miners are using the same software and same config. All are just fine. SRBMiner-CN.exe --enablecoinforking --config Config\config-litev7.txt --pools pools.txt --logfile %LOGTIME%
Try without the --enablecoinforking parameter, maybe that's causing the issue. Interesting, then i guess it isn't that parameter that causes the issue, but maybe windows pulled some driver update. To be sure that's not the case you could do a quick DDU and driver reinstall.
|
|
|
SRBMiner-CN.exe --enablecoinforking --config Config\config-litev7.txt --pools pools.txt --logfile %LOGTIME%
Try without the --enablecoinforking parameter, maybe that's causing the issue.
|
|
|
I can't open it says the archive is broken, and why is it 45mb ?
|
|
|
Im not surprised you got stale shares, holy cr**, 8 minutes to precompile 15 blocks?! Please upload the logfile somewhere and send me the link in pm
|
|
|
I have the same issue. Tried to reinstall the driver and the same result. The mb looks ok and everything runs and recognized. Any idea what is going on? Hello,
I'm knew to SRB miner and I'm having some difficulty. I followed instructions in the guide and can only manage to get.
Enumerating OpenCL devices, please wait...
Restarting miner...
Enumerating OpenCL devices, please wait...
Restarting miner...
Enumerating OpenCL devices, please wait...
Restarting miner...
Times infinity. I'm using the generic guided set up with V1.8.9 and 8.6.1 for my vegas. I'm not sure what else to add, any advice?
Try reinstalling the drivers using DDU. Otherwise, it might be an indication of a faulty motherboard. Show me your start.bat file
|
|
|
I've got all kinda of polaris cards on my rigs and after your post Dok, I went ahead and tried the 18.6.1 drivers on a single rig. I got them to mine at full speed by having the 'Radeon settings' be installed and manually one by one setting compute mode (so painful).
I can report back that it is the drivers. Blockchain drivers hang (doing so right now even) while 18.6.1 continues to work. Sorry for it taking so long to time to respond but the blockchain drivers did not fail for 4 days nearly! They just failed about 20 minutes prior to reporting this and still are while typing this.
You don't need to go that painful road , don't install the whole AMD 18.6.1 driver suite, just the drivers, and after that run this in an admin elevated cmd prompt : SRBMiner-CN.exe --setcomputemode
It will set every AMD gpu in your system to compute mode. Or you can use the AMD-Compute-Mode-SwitcherAnyways, back to the driver installation. After doing the cleanup with DDU, run the AMD driver installer and wait until it gets to the screen where you can actually install it (it was unpacking data before that to c:\AMD). Close the window, don't install anything. Go to Device Manager , under Display Adapters you should see your GPU's not recognised (because no driver was installed). Right click on any -> update driver -> browse my computer for driver -> and the search location should be C:\AMD\Win10-64Bit-Radeon-Software-Adrenalin-Edition-18.6.1-June13\Packages\Drivers\Display\WT6A_INFThis will install the 18.6.1 drivers without all the other junk. After this patch the driver with Pixel Clock Patcher if needed. And set the GPU's to compute mode with the above described method. I am preparing for the next SRB version a parameter that can set the number of blocks that get precompiled, so maybe it will help someone reduce precompilation time..
|
|
|
@hitmand @ipp2
i switched 18.6.1 drivers on a 470 4/8g mixed rig to the blockchain drivers to see if that's causing some issues with the 'invalid job' and @hitmand's 3 minute / gpu precompile time, but no - all should be fine with BC drivers, so at least now i know its not the driver that is making the problem. --> which vram rx470, mod bios or stock bios.
Can you guys tell me which SRB version are you using? I see @hitmand uses 1.8.8 , have you tried 1.8.9 ? Are you mining monero or some other CN/R coin ?---- > i try 1.8.9 , 1.8.8 and 1.8.6 . Only cause when mining Monero, not cause with MSR before
-- i try "Radeon Software Crimson ReLive Edition Beta for Blockchain Compute Driver for Windows® 10 64-bit" and has issues about 1 hour. - (~ 9xx h/z) i try " 1.8.9....." with 4 steps ....you give. (~ 4xx h/z) My card used elpida and samsung ram. Could you try it help me http://www.mediafire.com/file/78j51aca1le46v5/Ellesmere-samsung.rom/filehttp://www.mediafire.com/file/yqii06b9ajyui35/Ellesmere-elpida.rom/file Vram manufacturer and bios does not matter in this case. I really have no clue why it takes so much time on your rig to precompile, rx 470 4gb should not take more than 15-20 sec.
|
|
|
@hitmand @ipp2
i switched 18.6.1 drivers on a 470 4/8g mixed rig to the blockchain drivers to see if that's causing some issues with the 'invalid job' and @hitmand's 3 minute / gpu precompile time, but no - all should be fine with BC drivers, so at least now i know its not the driver that is making the problem.
Can you guys tell me which SRB version are you using? I see @hitmand uses 1.8.8 , have you tried 1.8.9 ? Are you mining monero or some other CN/R coin ?
|
|
|
I'm having the exact same issue as hitmandn, btw.
Newer drivers give me the 4xx speeds, old blockchainers don't have any issue except for the quite obvious stalling.
Something very odd goes on once it needs to compile, like it loses the block.
I have another rig, vega cards that run through this issue no problem but also using the recommended drivers.
Odd, you are missing a step 100% because the rest of the world can run 470/480/570/580 just fine with 18.6.1. I have some test rigs on w10, 470 4/8g and 18.6.1 .. Steps would be: 1.use ddu to delete drivers, and this is very important 2.install driver 3.patch drivers 4. set compute mode Only these steps are required.
|
|
|
I am looking into fixing this 'invalid job id' issue, my guess is it happens shortly after the precompilation part. For some reason gpu gets stuck working at the old job if newer was received.
@hitmandn something aint right there, a rx470 should need no more than 20 sec to precompile a program, yours are doing it for 3 minutes! What drivers are you using? Have you tried 18.6.1 or the newest adrenaline 2019?
I'm using Driver Fixdag. I tried 18.6.1 but hashrate to slow ( ~4xx ) Probably need to switch to compute mode, you can do that with SRBMiner, also that needs to be done only once How to switch to computer mode with SRBminer, i'll try it. THanks open a cmd prompt with admin privileges and run : SRBMiner-CN.exe --setcomputemode
|
|
|
I am looking into fixing this 'invalid job id' issue, my guess is it happens shortly after the precompilation part. For some reason gpu gets stuck working at the old job if newer was received.
@hitmandn something aint right there, a rx470 should need no more than 20 sec to precompile a program, yours are doing it for 3 minutes! What drivers are you using? Have you tried 18.6.1 or the newest adrenaline 2019?
I'm using Driver Fixdag. I tried 18.6.1 but hashrate to slow ( ~4xx ) Probably need to switch to compute mode, you can do that with SRBMiner, also that needs to be done only once
|
|
|
Hi livada and doctor I have same problem after 1.8.8 - Failed to load EIO.dll One start srb 1.8.8 and i take this problem with amdtweaker. Reboot rig - same. Help me pls. (and i newbie and PM not work)
I dont like reinstall win.
Use the latest 189 version it does not interfere with the eliovp tool
|
|
|
I am looking into fixing this 'invalid job id' issue, my guess is it happens shortly after the precompilation part. For some reason gpu gets stuck working at the old job if newer was received.
@hitmandn something aint right there, a rx470 should need no more than 20 sec to precompile a program, yours are doing it for 3 minutes! What drivers are you using? Have you tried 18.6.1 or the newest adrenaline 2019?
|
|
|
|