Rakly3
Newbie
 Offline
Activity: 45
Merit: 0
|
 |
June 13, 2019, 11:08:24 PM |
|
Well, it should really only happen when we restart our servers, which isn't often at all. So, there's something else cutting the connection for you. In some cases there are home routers that cut connections, in other cases many countries/jurisdictions don't like neither SSL traffic nor mining traffic. May I ask where you're located? You can PM me if you want to. Also, I'd love to know what version of TRM you're running? We have noticed a higher degree of short-lived inbound connections lately, but can't explain why.
people start autotune, notice they didn't set their own wallet and stop the miner? Just a thought. There's an easy way to test if this is the case. set a minername in the batchfiles that come with the miner. If they forgot to change the wallet, likely they didn't change the miner id either. |
|
|
|
|
Rakly3
Newbie
Offline
Activity: 45
Merit: 0
|
|
June 13, 2019, 11:37:35 PM |
|
I have the same problem on 4 rigs with 4 various Vega 56 cards (ref with samsun g and asus/powercolor with hynix, with timings and without timings ). The card runs for an hour or two and then the hash rate drops to zero or to 1400 h/s with a DEAD message. I have observed in several rigs, but for some reason problems always with the first PCI slot (PCIe 3:0.0). If you do not stick the card into the first PCI slot (PCI 16x), then the problem disappears. Now I just do not use 16x pci slot  Hi, interesting... for me it is GPU4 and it is located at PCIex16 slot too. LnkSta: Speed 8GT/s, Width x16, TrErr- Train- SlotClk+ DLActive- BWMgmt- ABWMgmt- I have the same problem on 4 rigs with 4 various Vega 56 cards (ref with samsun g and asus/powercolor with hynix, with timings and without timings ). The card runs for an hour or two and then the hash rate drops to zero or to 1400 h/s with a DEAD message. I have observed in several rigs, but for some reason problems always with the first PCI slot (PCIe 3:0.0). If you do not stick the card into the first PCI slot (PCI 16x), then the problem disappears. Now I just do not use 16x pci slot All primary x16 slots on any motherboard always are uplink. What hardware do you guys have? (only for the vega rigs) H110 3da with latest bios (F25), Celeron G3900 (Skylake) all (2) cores enabled. igfx enabled as primary. and a vega 56 in every pcie slot (stock bios, but we've already established bios, settings and algo don't matter). Single channel & single bank 4Gb memstick Crucial. For most of us TRM 4.4 seems to work normally although a teeny tiny bit slower, but it's still better than a whole card not running  . 5.2 only just released, so can't say yet. |
|
|
|
|
kamisama233
Newbie
Offline
Activity: 42
Merit: 0
|
|
June 14, 2019, 12:09:39 AM |
|
GPU 2: detected DEAD (13:00.0)
completely different card & bus now :/
I'm still using the stock bios btw. Different clockspeeds or voltages don't seem to impact it.
Timing level don't seem to matter either (driver setting, no mod) But I havn't tested for that specifically yet.
Hi! Even though I’m not replying to every message I’m always reading everything. I’m currently testing a few things and have gone through our full commit history from 0.4.4 and forward. We have a few small bug fixes on the way out, but it’s impossible to tell if any of those are the underlying issue here. I’ll reach out to a few of you in pm to see if you can run some test builds. Would love to nail this, if possible. hi, Dev, the network connection problem always cause the rig crash, [2019-06-13 21:52:22] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.205.58. [2019-06-13 21:52:30] Stats Uptime: 0 days, 05:58:08 [2019-06-13 21:52:30] GPU 0 [48C, fan 49%] cnr: 533.0 h/s, avg 530.7 h/s, pool 504.0 h/s a:203 r:0 hw:0 [2019-06-13 21:52:30] GPU 1 [49C, fan 39%] cnr: 539.2 h/s, avg 536.6 h/s, pool 525.9 h/s a:216 r:0 hw:0 [2019-06-13 21:52:30] GPU 2 [47C, fan 39%] cnr: 538.3 h/s, avg 533.8 h/s, pool 527.7 h/s a:212 r:0 hw:0 [2019-06-13 21:52:30] GPU 3 [52C, fan 39%] cnr: 543.6 h/s, avg 540.3 h/s, pool 551.6 h/s a:225 r:0 hw:0 [2019-06-13 21:52:30] GPU 4 [53C, fan 49%] cnr: 516.1 h/s, avg 513.2 h/s, pool 456.6 h/s a:182 r:0 hw:0 [2019-06-13 21:52:30] GPU 5 [45C, fan 0%] cnr: 520.5 h/s, avg 517.9 h/s, pool 542.8 h/s a:215 r:0 hw:0 [2019-06-13 21:52:30] Total cnr: 3.191kh/s, avg 3.173kh/s, pool 3.109kh/s a:1253 r:0 hw:0 [2019-06-13 21:52:32] Pool asia.cryptonight-hub.miningpoolhub.com connect timed-out to address 172.105.205.58. [2019-06-13 21:52:32] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.210.117. [2019-06-13 21:52:42] Pool asia.cryptonight-hub.miningpoolhub.com connect timed-out to address 172.105.210.117. [2019-06-13 21:52:42] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 139.162.46.14. [2019-06-13 21:52:52] Pool asia.cryptonight-hub.miningpoolhub.com connect timed-out to address 139.162.46.14. after several retries, success connect to the pool again [2019-06-13 21:52:52] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.197.212. [2019-06-13 21:52:52] Pool asia.cryptonight-hub.miningpoolhub.com successfully connected to address 172.105.197.212. [2019-06-13 21:52:53] Pool asia.cryptonight-hub.miningpoolhub.com login succeeded.(1106 ms) [2019-06-13 21:52:53] Pool asia.cryptonight-hub.miningpoolhub.com received new job. (job_id: 212828560336492) [2019-06-13 21:52:53] Pool asia.cryptonight-hub.miningpoolhub.com set difficulty to 54564 but this make the rig crash, and restart, is there any way to prevent this, thanks [2019-06-13 21:55:44] Initializing GPU 0. [2019-06-13 21:55:44] Watchdog thread starting. [2019-06-13 21:55:44] API initialized on 0.0.0.0:522 [2019-06-13 21:55:44] Runtime Command Keys: h - help, s - stats, e - enable gpu, d - disable gpu, q - quit [2019-06-13 21:55:44] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.210.117. [2019-06-13 21:55:44] Dev pool connected and ready. [2019-06-13 21:55:44] Initializing GPU 1. [2019-06-13 21:55:45] Initializing GPU 2. [2019-06-13 21:55:46] Initializing GPU 3. [2019-06-13 21:55:47] Miner load at 0%, hashrates are not at max speed yet. [2019-06-13 21:55:47] Initializing GPU 4. [2019-06-13 21:55:48] Initializing GPU 5. |
|
|
|
|
lebuawu2
Jr. Member
Offline
Activity: 176
Merit: 2
|
|
June 14, 2019, 12:31:13 AM |
|
Well, it should really only happen when we restart our servers, which isn't often at all. So, there's something else cutting the connection for you. In some cases there are home routers that cut connections, in other cases many countries/jurisdictions don't like neither SSL traffic nor mining traffic. May I ask where you're located? You can PM me if you want to. Also, I'd love to know what version of TRM you're running? We have noticed a higher degree of short-lived inbound connections lately, but can't explain why.
Hi Kerney, I already PM you, please check your message. Thanks. |
|
|
|
|
netgeek3
Newbie
Offline
Activity: 12
Merit: 0
|
|
June 14, 2019, 02:20:52 PM
Last edit: June 14, 2019, 02:44:05 PM by netgeek3 |
|
I have the same problem on 4 rigs with 4 various Vega 56 cards (ref with samsun g and asus/powercolor with hynix, with timings and without timings ). The card runs for an hour or two and then the hash rate drops to zero or to 1400 h/s with a DEAD message. I have observed in several rigs, but for some reason problems always with the first PCI slot (PCIe 3:0.0). If you do not stick the card into the first PCI slot (PCI 16x), then the problem disappears. Now I just do not use 16x pci slot I am using all 1x slots and am having the same behavior. I am seriously thinking that this is a driver compatibility issue with Win10 for a couple of reasons: 1) It is always (in my case) the first card enumerated on the PCIe bus 2) It doesn't matter what motherboard I'm using (have tried on 7 different) 3) Starting ANY other applications on the same system slows down ONLY the affected GPU, and it shouldn't affect anything, historically. 4) In digging through logs and in the rare cases of a BSOD, there are consistently "stuck thread" issues 5) This phenomenon is not isolated to TRM - I am having the same issue with SRB, Phoenix, Claymore and others I did not have this problem with earlier builds, but since the latest Windows and AMD updates, I can't get rid of it. I have tried rolling back through every iteration all the way back to Blockchain drivers without success. I should mention that I am using a mix of V64 and V56 cards in my rigs, but all have Samsung memory. |
|
|
|
|
SigiSinatra
Newbie
Offline
Activity: 50
Merit: 0
|
|
June 14, 2019, 02:30:24 PM |
|
The best, the fastest and most stable AMD miner ever.
It works perfect on all supported algos.
|
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
June 14, 2019, 06:07:58 PM |
|
I have the same problem on 4 rigs with 4 various Vega 56 cards (ref with samsun g and asus/powercolor with hynix, with timings and without timings ). The card runs for an hour or two and then the hash rate drops to zero or to 1400 h/s with a DEAD message. I have observed in several rigs, but for some reason problems always with the first PCI slot (PCIe 3:0.0). If you do not stick the card into the first PCI slot (PCI 16x), then the problem disappears. Now I just do not use 16x pci slot I am using all 1x slots and am having the same behavior. I am seriously thinking that this is a driver compatibility issue with Win10 for a couple of reasons: 1) It is always (in my case) the first card enumerated on the PCIe bus 2) It doesn't matter what motherboard I'm using (have tried on 7 different) 3) Starting ANY other applications on the same system slows down ONLY the affected GPU, and it shouldn't affect anything, historically. 4) In digging through logs and in the rare cases of a BSOD, there are consistently "stuck thread" issues 5) This phenomenon is not isolated to TRM - I am having the same issue with SRB, Phoenix, Claymore and others I did not have this problem with earlier builds, but since the latest Windows and AMD updates, I can't get rid of it. I have tried rolling back through every iteration all the way back to Blockchain drivers without success. I should mention that I am using a mix of V64 and V56 cards in my rigs, but all have Samsung memory. Hi! Just curious, did you have time to test 0.5.2 yet? It has a first wave of changes that only aim to stabilize the miner. No real guarantees, only some changes that theoretically should be nicer to the hardware. Otherwise, I agree, we’re dealing with some very delicate issues, and the software/driver/hardware interaction here is not all hunky dory. |
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
June 14, 2019, 06:08:44 PM |
|
The best, the fastest and most stable AMD miner ever.
It works perfect on all supported algos.
Thanks for the kind words, much appreciated! |
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
June 14, 2019, 06:17:02 PM |
|
GPU 2: detected DEAD (13:00.0)
completely different card & bus now :/
I'm still using the stock bios btw. Different clockspeeds or voltages don't seem to impact it.
Timing level don't seem to matter either (driver setting, no mod) But I havn't tested for that specifically yet.
Hi! Even though I’m not replying to every message I’m always reading everything. I’m currently testing a few things and have gone through our full commit history from 0.4.4 and forward. We have a few small bug fixes on the way out, but it’s impossible to tell if any of those are the underlying issue here. I’ll reach out to a few of you in pm to see if you can run some test builds. Would love to nail this, if possible. hi, Dev, the network connection problem always cause the rig crash, [2019-06-13 21:52:22] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.205.58. [2019-06-13 21:52:30] Stats Uptime: 0 days, 05:58:08 [2019-06-13 21:52:30] GPU 0 [48C, fan 49%] cnr: 533.0 h/s, avg 530.7 h/s, pool 504.0 h/s a:203 r:0 hw:0 [2019-06-13 21:52:30] GPU 1 [49C, fan 39%] cnr: 539.2 h/s, avg 536.6 h/s, pool 525.9 h/s a:216 r:0 hw:0 [2019-06-13 21:52:30] GPU 2 [47C, fan 39%] cnr: 538.3 h/s, avg 533.8 h/s, pool 527.7 h/s a:212 r:0 hw:0 [2019-06-13 21:52:30] GPU 3 [52C, fan 39%] cnr: 543.6 h/s, avg 540.3 h/s, pool 551.6 h/s a:225 r:0 hw:0 [2019-06-13 21:52:30] GPU 4 [53C, fan 49%] cnr: 516.1 h/s, avg 513.2 h/s, pool 456.6 h/s a:182 r:0 hw:0 [2019-06-13 21:52:30] GPU 5 [45C, fan 0%] cnr: 520.5 h/s, avg 517.9 h/s, pool 542.8 h/s a:215 r:0 hw:0 [2019-06-13 21:52:30] Total cnr: 3.191kh/s, avg 3.173kh/s, pool 3.109kh/s a:1253 r:0 hw:0 [2019-06-13 21:52:32] Pool asia.cryptonight-hub.miningpoolhub.com connect timed-out to address 172.105.205.58. [2019-06-13 21:52:32] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.210.117. [2019-06-13 21:52:42] Pool asia.cryptonight-hub.miningpoolhub.com connect timed-out to address 172.105.210.117. [2019-06-13 21:52:42] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 139.162.46.14. [2019-06-13 21:52:52] Pool asia.cryptonight-hub.miningpoolhub.com connect timed-out to address 139.162.46.14. after several retries, success connect to the pool again [2019-06-13 21:52:52] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.197.212. [2019-06-13 21:52:52] Pool asia.cryptonight-hub.miningpoolhub.com successfully connected to address 172.105.197.212. [2019-06-13 21:52:53] Pool asia.cryptonight-hub.miningpoolhub.com login succeeded.(1106 ms) [2019-06-13 21:52:53] Pool asia.cryptonight-hub.miningpoolhub.com received new job. (job_id: 212828560336492) [2019-06-13 21:52:53] Pool asia.cryptonight-hub.miningpoolhub.com set difficulty to 54564 but this make the rig crash, and restart, is there any way to prevent this, thanks [2019-06-13 21:55:44] Initializing GPU 0. [2019-06-13 21:55:44] Watchdog thread starting. [2019-06-13 21:55:44] API initialized on 0.0.0.0:522 [2019-06-13 21:55:44] Runtime Command Keys: h - help, s - stats, e - enable gpu, d - disable gpu, q - quit [2019-06-13 21:55:44] Pool asia.cryptonight-hub.miningpoolhub.com connecting to address 172.105.210.117. [2019-06-13 21:55:44] Dev pool connected and ready. [2019-06-13 21:55:44] Initializing GPU 1. [2019-06-13 21:55:45] Initializing GPU 2. [2019-06-13 21:55:46] Initializing GPU 3. [2019-06-13 21:55:47] Miner load at 0%, hashrates are not at max speed yet. [2019-06-13 21:55:47] Initializing GPU 4. [2019-06-13 21:55:48] Initializing GPU 5. Hmm, the intention is that the miner should use the same ramp-up logic as we use at start-up when a pool has been disconnected, this in order to not kill the PSU by going from 0-100% load on all GPUs at the same time. I can't say I have tested it lately though, and I can't say for sure that this is the issue here. It's a good guess though, no real reason why the rig would croak otherwise. I will do some tests. |
|
|
|
|
Rakly3
Newbie
Offline
Activity: 45
Merit: 0
|
|
June 14, 2019, 06:40:13 PM
Last edit: June 15, 2019, 02:31:41 AM by Rakly3 |
|
Hi! Just curious, did you have time to test 0.5.2 yet? It has a first wave of changes that only aim to stabilize the miner. No real guarantees, only some changes that theoretically should be nicer to the hardware. Otherwise, I agree, we’re dealing with some very delicate issues, and the software/driver/hardware interaction here is not all hunky dory.
Still running! https://i.ibb.co/FgjLTWJ/Untitled.pngedit: 1d 4hours on 5.2 and still no dead GPU. I guess we can call it fixed! (?) |
|
|
|
|
HughHashner
Newbie
Offline
Activity: 5
Merit: 0
|
|
June 15, 2019, 02:36:01 AM |
|
I have 1 Vega 56 that was previously crashing 2-3x per day while my other Vegas chugged along. Since flipping to 0.5.2 yesterday I have not had a single crash.
No changes to OC or mem timings.
Cheers toddxx!!
|
|
|
|
|
Beruang123
Newbie
Offline
Activity: 10
Merit: 0
|
|
June 15, 2019, 04:54:21 AM |
|
I have the same problem on 4 rigs with 4 various Vega 56 cards (ref with samsun g and asus/powercolor with hynix, with timings and without timings ). The card runs for an hour or two and then the hash rate drops to zero or to 1400 h/s with a DEAD message. I have observed in several rigs, but for some reason problems always with the first PCI slot (PCIe 3:0.0). If you do not stick the card into the first PCI slot (PCI 16x), then the problem disappears. Now I just do not use 16x pci slot I am using all 1x slots and am having the same behavior. I am seriously thinking that this is a driver compatibility issue with Win10 for a couple of reasons: 1) It is always (in my case) the first card enumerated on the PCIe bus 2) It doesn't matter what motherboard I'm using (have tried on 7 different) 3) Starting ANY other applications on the same system slows down ONLY the affected GPU, and it shouldn't affect anything, historically. 4) In digging through logs and in the rare cases of a BSOD, there are consistently "stuck thread" issues 5) This phenomenon is not isolated to TRM - I am having the same issue with SRB, Phoenix, Claymore and others I did not have this problem with earlier builds, but since the latest Windows and AMD updates, I can't get rid of it. I have tried rolling back through every iteration all the way back to Blockchain drivers without success. I should mention that I am using a mix of V64 and V56 cards in my rigs, but all have Samsung memory. Hi! Just curious, did you have time to test 0.5.2 yet? It has a first wave of changes that only aim to stabilize the miner. No real guarantees, only some changes that theoretically should be nicer to the hardware. Otherwise, I agree, we’re dealing with some very delicate issues, and the software/driver/hardware interaction here is not all hunky dory. Apparently the new release didnot fix the issue.. still have the same error https://imgur.com/a/GraVN7H?desktop=1 |
|
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
June 15, 2019, 07:46:45 AM |
|
I have the same problem on 4 rigs with 4 various Vega 56 cards (ref with samsun g and asus/powercolor with hynix, with timings and without timings ). The card runs for an hour or two and then the hash rate drops to zero or to 1400 h/s with a DEAD message. I have observed in several rigs, but for some reason problems always with the first PCI slot (PCIe 3:0.0). If you do not stick the card into the first PCI slot (PCI 16x), then the problem disappears. Now I just do not use 16x pci slot I am using all 1x slots and am having the same behavior. I am seriously thinking that this is a driver compatibility issue with Win10 for a couple of reasons: 1) It is always (in my case) the first card enumerated on the PCIe bus 2) It doesn't matter what motherboard I'm using (have tried on 7 different) 3) Starting ANY other applications on the same system slows down ONLY the affected GPU, and it shouldn't affect anything, historically. 4) In digging through logs and in the rare cases of a BSOD, there are consistently "stuck thread" issues 5) This phenomenon is not isolated to TRM - I am having the same issue with SRB, Phoenix, Claymore and others I did not have this problem with earlier builds, but since the latest Windows and AMD updates, I can't get rid of it. I have tried rolling back through every iteration all the way back to Blockchain drivers without success. I should mention that I am using a mix of V64 and V56 cards in my rigs, but all have Samsung memory. Hi! Just curious, did you have time to test 0.5.2 yet? It has a first wave of changes that only aim to stabilize the miner. No real guarantees, only some changes that theoretically should be nicer to the hardware. Otherwise, I agree, we’re dealing with some very delicate issues, and the software/driver/hardware interaction here is not all hunky dory. Apparently the new release didnot fix the issue.. still have the same error  Ah interesting! You see, I didn’t make the same small stabilizing adjustments in the main 4MB variant kernels. Obviously I should have, my mistake, I simply forgot. Will try fix them as well very soon and will let you know when a release is available. |
|
|
|
|
todxx (OP)
Member
Offline
Activity: 176
Merit: 76
|
|
June 15, 2019, 09:25:08 AM |
|
I have 1 Vega 56 that was previously crashing 2-3x per day while my other Vegas chugged along. Since flipping to 0.5.2 yesterday I have not had a single crash.
No changes to OC or mem timings.
Cheers toddxx!!
Glad to hear it's running smoother You should send your praise in Kerney666's direction though, he's been the one fighting these issues over the last few days. |
|
|
|
|
HughHashner
Newbie
Offline
Activity: 5
Merit: 0
|
|
June 15, 2019, 01:20:50 PM |
|
I have 1 Vega 56 that was previously crashing 2-3x per day while my other Vegas chugged along. Since flipping to 0.5.2 yesterday I have not had a single crash.
No changes to OC or mem timings.
Cheers toddxx!!
Glad to hear it's running smoother You should send your praise in Kerney666's direction though, he's been the one fighting these issues over the last few days. Much love to Kerney666, too! |
|
|
|
|
pigfrown
Jr. Member
Offline
Activity: 47
Merit: 1
|
|
June 15, 2019, 04:37:42 PM |
|
I saw in the release notes that Tonga support has been added, so I tried the latest version (0.5.2) with a Tonga card (r9 380 4GB) but get the following error on startup:
Failed to initialise device idx 3 (-13)
Is this card supported? Or is extra config required for Tonga?
08:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tonga PRO [Radeon R9 285/380] (rev f1)
|
Bitrated user: pigfrown.
|
|
|
|
Iamtutut
|
|
June 15, 2019, 05:17:07 PM |
|
I saw in the release notes that Tonga support has been added, so I tried the latest version (0.5.2) with a Tonga card (r9 380 4GB) but get the following error on startup:
Failed to initialise device idx 3 (-13)
Is this card supported? Or is extra config required for Tonga?
08:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tonga PRO [Radeon R9 285/380] (rev f1)
I have the same issue on all CN algos except turtlecoin when I launch TRM while any web browser is opened (be Firefox, brave, Edge). |
|
|
|
|
pigfrown
Jr. Member
Offline
Activity: 47
Merit: 1
|
|
June 15, 2019, 06:04:04 PM |
|
I saw in the release notes that Tonga support has been added, so I tried the latest version (0.5.2) with a Tonga card (r9 380 4GB) but get the following error on startup:
Failed to initialise device idx 3 (-13)
Is this card supported? Or is extra config required for Tonga?
08:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tonga PRO [Radeon R9 285/380] (rev f1)
I have the same issue on all CN algos except turtlecoin when I launch TRM while any web browser is opened (be Firefox, brave, Edge). I get this error on a linux box with no GUI. You are probably getting that error because your browser(s) are using the GPU? Try turning off "hardware acceleration" in chrome/brave. |
Bitrated user: pigfrown.
|
|
|
kerney666
Member
Offline
Activity: 658
Merit: 86
|
|
June 15, 2019, 07:21:55 PM |
|
I saw in the release notes that Tonga support has been added, so I tried the latest version (0.5.2) with a Tonga card (r9 380 4GB) but get the following error on startup:
Failed to initialise device idx 3 (-13)
Is this card supported? Or is extra config required for Tonga?
08:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tonga PRO [Radeon R9 285/380] (rev f1)
I have the same issue on all CN algos except turtlecoin when I launch TRM while any web browser is opened (be Firefox, brave, Edge). I get this error on a linux box with no GUI. You are probably getting that error because your browser(s) are using the GPU? Try turning off "hardware acceleration" in chrome/brave. Hi! Yes, the web browser error is because the browser has stolen some amount of gpu vram and is using the gpu. It's logical that it still works for turtle (and upx2) since those pads are so tiny, they don't really use that much vram compared to the 2MB/4MB variants. For the Tonga error, man I don't know. The error is when it tries to "build" the program, which indicates that it the driver doesn't accept our binary with the kernels. Neither me nor todxx have a Tonga to test with unfortunately. Can you tell me (1) your amdgpu-pro version, (2) give me an output from "./teamredminer --list_devices", (3) the full log when you start the miner until you get that error message (preferably only running for that card)? PM is of course fine if you don't want to paste here. Thanks, K |
|
|
|
|
pigfrown
Jr. Member
Offline
Activity: 47
Merit: 1
|
|
June 15, 2019, 08:13:08 PM |
|
I saw in the release notes that Tonga support has been added, so I tried the latest version (0.5.2) with a Tonga card (r9 380 4GB) but get the following error on startup:
Failed to initialise device idx 3 (-13)
Is this card supported? Or is extra config required for Tonga?
08:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tonga PRO [Radeon R9 285/380] (rev f1)
I have the same issue on all CN algos except turtlecoin when I launch TRM while any web browser is opened (be Firefox, brave, Edge). I get this error on a linux box with no GUI. You are probably getting that error because your browser(s) are using the GPU? Try turning off "hardware acceleration" in chrome/brave. Hi! Yes, the web browser error is because the browser has stolen some amount of gpu vram and is using the gpu. It's logical that it still works for turtle (and upx2) since those pads are so tiny, they don't really use that much vram compared to the 2MB/4MB variants. For the Tonga error, man I don't know. The error is when it tries to "build" the program, which indicates that it the driver doesn't accept our binary with the kernels. Neither me nor todxx have a Tonga to test with unfortunately. Can you tell me (1) your amdgpu-pro version, (2) give me an output from "./teamredminer --list_devices", (3) the full log when you start the miner until you get that error message (preferably only running for that card)? PM is of course fine if you don't want to paste here. Thanks, K amdgpu-pro version: amdgpu-pro-core/unknown,now 17.40-483984 all It does seem to detect the card. ./teamredminer --list_devices
Team Red Miner version 0.5.2
[2019-06-15 20:58:51] Auto-detected AMD OpenCL platform 0
[2019-06-15 20:58:53] Detected 5 devices, listed in OpenCL order:
[2019-06-15 20:58:53] Miner Platform OpenCL BusId Name Model Nr CUs
[2019-06-15 20:58:53] ----- -------- ------ -------- ------------- ------------------------- ------
[2019-06-15 20:58:53] 0 0 0 01:00.0 Ellesmere Radeon RX 580 Series 36
[2019-06-15 20:58:53] 1 0 1 03:00.0 Ellesmere Radeon RX 580 Series 36
[2019-06-15 20:58:53] 2 0 2 04:00.0 Ellesmere Radeon RX 570 Series 32
[2019-06-15 20:58:53] 3 0 3 06:00.0 Ellesmere Radeon RX 570 Series 32
[2019-06-15 20:58:53] 4 0 4 08:00.0 Tonga AMD Radeon (TM) R9 380 Se 28
[2019-06-15 20:58:53] Successful clean shutdown.
This is the error: ./teamredminer -d 4 --debug -a cnr <snip>
Team Red Miner version 0.5.2
[2019-06-15 21:11:14] Auto-detected AMD OpenCL platform 0
[2019-06-15 21:11:15] Initializing GPU 0.
[2019-06-15 21:11:15] Failed to initialize device idx 4 (-13)
[2019-06-15 21:11:15] Successful clean shutdown.
Thanks |
Bitrated user: pigfrown.
|
|
|
|
