Manually, neither don't working.
@todxx
Could it be fix, please?!
Topic: [ANN] TeamRedMiner v0.10.10 - Ironfish/Kaspa/ZIL/Kawpow/Etchash and More - page 34. (Read 211762 times)
November 11, 2020, 02:59:58 AM
What about A and B o/c parameters for Navy GPU's (mean 5700 XT)
Manually, neither don't working.
@todxx
Could it be fix, please?!
Manually, neither don't working.
@todxx
Could it be fix, please?!
November 11, 2020, 02:52:42 AM
I set in my bat file: Config--eth_config=A236, miner start, and works well.
But after few hours miner changed eth_config to A206 , etc.
How to set fixed Config--eth_config=A236 for all time?
UPD:
Job abort failure, decreasing intensity -8 (see if --eth_direct_abort applies to you).
how to disable this decreasing?
But after few hours miner changed eth_config to A206 , etc.
How to set fixed Config--eth_config=A236 for all time?
UPD:
Job abort failure, decreasing intensity -8 (see if --eth_direct_abort applies to you).
how to disable this decreasing?
Well, it's quite unexpected it's happening in the first place. What is your os and driver version? Also, are you running any other gpu work in parallel on the gpu?
Hiveos, amd 20.30 driver, nicehash eth mining. 7gpu all descrease speed from fixed a236 .
I guess I need to run tests on 20.30 for Polaris gpus then. I've never seen the abort mechanism fail on 20.10 and under for Polaris cards.
I still fail to reproduce the issue. However, the miner will give up trying to adjust for failed aborts after a few attemps, so if you're not experiencing issues with stale shares, a patch would be to start the miner with --eth_config=A268. After a while, it will stabilize at A236, which you want. Will test more on this shortly.
Also, are you running cpu mining or a second parallel gpu miner at the same time?
When i set fixed A236, after few hours miner set: A204!! And fix a204 for all time. ( Decrease from 236 to 204 = 32
When i set fixed a255, after few hours miner set: A223, and fix a223 for all time. (Decrease from 255 to 223 = 32.
November 10, 2020, 09:06:11 PM
Love your miner, it’s performing very well most of the time. Still have that weird issue (as others do from what I can see), miner just stops mining without any notice or clues other than no hashing, but miner window is open and shows hashing, but no updates. I can hit the ‘s’ key for stats and that works once, then cmd window freezes, any ideas?
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Issue is still happening always at the same point, just after GPU stats and before Pool stats - see pic.
https://ibb.co/hyvcWfF
Hmm. That _really_ looks like we're hanging on trying to acquire a lock for the pool to read the current stats but get stuck. We did address a pool bug when using multiple pools for failover in v0.7.17, don't know of any other bugs. I see that you have a full log file for the run, would it be possible to DM a link to it it here on bitcointalk or dump it in a DM to me on Discord? I'd like to see if there are any pool related issues noted, and also sort out any underlying issue.
EDIT: also, do you run with a single pool, or in a multi-pool mode with e.g. failover?
Well I didn’t know you added failover pools, so I don’t have any yet. But to give you more info, this is happening with Nicehash, not sure about other pools. I’ll get the log next time it happens, right now things are running smooth so I turned logging off.
November 10, 2020, 08:10:32 PM
Love your miner, it’s performing very well most of the time. Still have that weird issue (as others do from what I can see), miner just stops mining without any notice or clues other than no hashing, but miner window is open and shows hashing, but no updates. I can hit the ‘s’ key for stats and that works once, then cmd window freezes, any ideas?
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Issue is still happening always at the same point, just after GPU stats and before Pool stats - see pic.
https://ibb.co/hyvcWfF
Hmm. That _really_ looks like we're hanging on trying to acquire a lock for the pool to read the current stats but get stuck. We did address a pool bug when using multiple pools for failover in v0.7.17, don't know of any other bugs. I see that you have a full log file for the run, would it be possible to DM a link to it it here on bitcointalk or dump it in a DM to me on Discord? I'd like to see if there are any pool related issues noted, and also sort out any underlying issue.
EDIT: also, do you run with a single pool, or in a multi-pool mode with e.g. failover?
November 10, 2020, 08:06:20 PM
I know that this was requested but it would be great to add cvddc and mvdd(c) to the additinal settings. It is easier to use just one tool than plus overdriven and amdmemtweak, also those two tools sets clock per gpu, other miners like pm and clay you just need to type one value to add it to all cards.
Not 100% decided, but I'm pretty sure we'll do a larger project for clocks/voltages/timings soon. This 4GB craze just needs to calm down a little first.
November 10, 2020, 06:45:56 PM
Love your miner, it’s performing very well most of the time. Still have that weird issue (as others do from what I can see), miner just stops mining without any notice or clues other than no hashing, but miner window is open and shows hashing, but no updates. I can hit the ‘s’ key for stats and that works once, then cmd window freezes, any ideas?
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Issue is still happening always at the same point, just after GPU stats and before Pool stats - see pic.
https://ibb.co/hyvcWfF
November 10, 2020, 10:56:36 AM
I know that this was requested but it would be great to add cvddc and mvdd(c) to the additinal settings. It is easier to use just one tool than plus overdriven and amdmemtweak, also those two tools sets clock per gpu, other miners like pm and clay you just need to type one value to add it to all cards.
November 10, 2020, 10:47:52 AM
Love your miner, it’s performing very well most of the time. Still have that weird issue (as others do from what I can see), miner just stops mining without any notice or clues other than no hashing, but miner window is open and shows hashing, but no updates. I can hit the ‘s’ key for stats and that works once, then cmd window freezes, any ideas?
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
Widows 10, 18.6.1, 6-Vega 64, v0.7.17
EDIT: I’ve update to driver 20.4.2, I’ll see if issue is resolved.
November 10, 2020, 05:46:19 AM
some weird issues with 12x4gb polarises on last version
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
1) I've seen a lot of weird corner case bugs in drivers related to the first or last gpu. I don't think it has anything to do with the last TRM version per se (unless you've verified an earlier version work better?), it's more that time progresses and we keep moving to higher epochs, pushing mem allocation harder and harder on these little 4GBs. Are you on win or linux btw? Win epoch 374 is reaaally pushing it, and you might just need to cap that last gpu. Annoyingly enough, it requires you to enumerate a setting for all gpus, so try adding --eth_4g_max_alloc=374,374,374,374,374,374,374,374,374,374,374,373 to let the first eleven allocate DAG space for epoch 374, and the last one for epoch 373 only. You can also try setting the last to 374 and check, it means it will do a preallocation of the DAG buffers, sometimes that's enough.
2) Key mappings are fixed and will work in the next version.
thanks. manual preallocation per GPU gives the best result:
if I go 3952 on the last one, it will drop to 1.3mh
there's a alpha of PhoenixMiner that can work on 4GB now, but it seems none of the GPUs will have fully allocated DAGs
more here: https://bitcointalksearch.org/topic/m.55557320
November 09, 2020, 11:25:40 PM
Can someone HELP me to determine eth B parameter to encrease hashrate for Navy VGA cards or maybe A parameter?
I am trying to use --eth_config=B(value) command
TNX
I am trying to use --eth_config=B(value) command
TNX
No B-mode available for Navis unf, it will have zero effect. So, you either just let the auto-tune run once to decide some A-value in the optimal range, then you use that from now on, or let the auto-tuner run every time the miner starts.
November 09, 2020, 06:06:50 PM
Hello to everyone!
I read the last page, and I guess I have the same problem. It doesn't give me any more hashrate... I come from PhoenixMiner because it gave me similar hashates.
All my cards are 4GB (3 rx 580 and 1 rx 570). I'm using crimson 18.3.4, windows 10.
Regards
I read the last page, and I guess I have the same problem. It doesn't give me any more hashrate... I come from PhoenixMiner because it gave me similar hashates.
All my cards are 4GB (3 rx 580 and 1 rx 570). I'm using crimson 18.3.4, windows 10.
Regards
You need to bump that driver to a more recent version to have a chance at epoch 374. We've done almost all of our 4GB testing on 20.4.2, so I'd recommend trying that as a first step. Then, some rigs/gpus still won't be able to clear epoch 374. The only remaining option then is to cap the allocation with --eth_4g_max_alloc=373, or use MB and try --eth_4g_max_alloc=4014 and move the 4016 down in steps of maybe -2 at the time until it works. 4014 is almost enough for epoch 374, so that will most probably not work. Any extra MB you can get in there makes a difference in lost hashrate when capping the allocation, so it's worth it.
November 09, 2020, 06:03:05 PM
hello,
most of my rigs (all RX 4G) have the GPU 0 that doesn't work on ETH.
with the --eth_4g_max_alloc=374 parameter, are all GPU not supposed to work, even if it is at a lower hashrate?
win 10 driver 20.5.1
thx
PS: on 6 rigs, only one have all it's graphic cars mining, still on phoenix miner where the others don't work on the miner. Same windows version, same drivers, same cards as other rigs...
most of my rigs (all RX 4G) have the GPU 0 that doesn't work on ETH.
with the --eth_4g_max_alloc=374 parameter, are all GPU not supposed to work, even if it is at a lower hashrate?
win 10 driver 20.5.1
thx
PS: on 6 rigs, only one have all it's graphic cars mining, still on phoenix miner where the others don't work on the miner. Same windows version, same drivers, same cards as other rigs...
Unfortunately, win drivers tend to steal a little more vram on the first gpu, monitor connected or not. You need to dial down the 374 above one step at the time until it works.
November 09, 2020, 06:01:20 PM
Can someone HELP me to determine eth B parameter to encrease hashrate for Navy VGA cards or maybe A parameter?
I am trying to use --eth_config=B(value) command
TNX
I am trying to use --eth_config=B(value) command
TNX
No B-mode available for Navis unf, it will have zero effect. So, you either just let the auto-tune run once to decide some A-value in the optimal range, then you use that from now on, or let the auto-tuner run every time the miner starts.
November 09, 2020, 05:59:57 PM
some weird issues with 12x4gb polarises on last version
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
1) I've seen a lot of weird corner case bugs in drivers related to the first or last gpu. I don't think it has anything to do with the last TRM version per se (unless you've verified an earlier version work better?), it's more that time progresses and we keep moving to higher epochs, pushing mem allocation harder and harder on these little 4GBs. Are you on win or linux btw? Win epoch 374 is reaaally pushing it, and you might just need to cap that last gpu. Annoyingly enough, it requires you to enumerate a setting for all gpus, so try adding --eth_4g_max_alloc=374,374,374,374,374,374,374,374,374,374,374,373 to let the first eleven allocate DAG space for epoch 374, and the last one for epoch 373 only. You can also try setting the last to 374 and check, it means it will do a preallocation of the DAG buffers, sometimes that's enough.
2) Key mappings are fixed and will work in the next version.
November 09, 2020, 02:29:18 PM
some weird issues with 12x4gb polarises on last version
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
Connect a monitor or a dummy (if you have) to it and see if the issue is gone.
November 09, 2020, 12:57:45 PM
some weird issues with 12x4gb polarises on last version
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
last GPU (no matter the PCIE or OpenCL order!) produces 1.1MH, (11 are ok)
1) I'm trying to -d 0,1,2,3,4,5,6,7,8,9,10 - will run 10 devices instead of 11, and -d 0,1,2,3,4,5,6,7,8,9,10,11 WILL indeed run on 11 devices, but the last one will be 1.2MH again
2) runtime menu for disabling the gpu will not work above 10 as well. pressing A will pause dev0 and pressing B will disable dev1
is it possible that the last gpu keeps cross-allocated memory for the rest?
November 09, 2020, 11:39:21 AM
I were use phoenix and with Samsung memory 2060 max or incorrect shares
with TRM I'm able to 2100 memory and all fine .
Thank you developers ,
want build second rig soon but need to buy rig x8 not more than 1000w or close have limited amp
TRM thanks
with TRM I'm able to 2100 memory and all fine .
Thank you developers ,
want build second rig soon but need to buy rig x8 not more than 1000w or close have limited amp
TRM thanks
November 09, 2020, 04:23:05 AM
@gfarr51
I saw Your post but i am on Linux (HiveOS) and ca't use it
I saw Your post but i am on Linux (HiveOS) and ca't use it
November 09, 2020, 04:12:52 AM
Can someone HELP me to determine eth B parameter to encrease hashrate for Navy VGA cards or maybe A parameter?
I am trying to use --eth_config=B(value) command
TNX
I am trying to use --eth_config=B(value) command
TNX
November 07, 2020, 01:24:28 PM
Hello to everyone!
I read the last page, and I guess I have the same problem. It doesn't give me any more hashrate... I come from PhoenixMiner because it gave me similar hashates.
https://ibb.co/TThcYVT
All my cards are 4GB (3 rx 580 and 1 rx 570). I'm using crimson 18.3.4, windows 10.
Regards
I read the last page, and I guess I have the same problem. It doesn't give me any more hashrate... I come from PhoenixMiner because it gave me similar hashates.
https://ibb.co/TThcYVT
All my cards are 4GB (3 rx 580 and 1 rx 570). I'm using crimson 18.3.4, windows 10.
Regards
Jump to: