T-Rex 0.26.8 ETHW, ETC (+dual mining) GPU miner - page 46.

ExEric3

full member

Activity: 176

Merit: 100

Quote from: trexminer on November 21, 2020, 08:21:59 AM

Yes I can see the memtweak is off, thanks. However, I'm very surprised about your 1080 being slower than my 1070Ti on epoch 383. I'm getting 24.5Mh/s with PL 90W, CORE 0, MEM +1000. DAG generation time is ~8.8s, without memory tweaks

Because u are compareing two different memory architectures. GDDR5x isnt GDDR5 and vice versa. My results are slower because I used default clocks. I dont want burn my house on this Asus Dragon Grin

When I put MEM + 1000 will get generation time 10s and hashrate about 26.5Mh/s - Epoch 383. But still I will not get results like 33Mh/s with enabled mem tweak. You can nicelly see it on Nicehash graphs when someone order ETC hashrate how it all drops since it have different DAG size like ETH currently.

trexminer

member

Activity: 283

Merit: 63

Quote from: ExEric3 on November 21, 2020, 08:01:05 AM

How I typed from start. Mem tweak doesnt work anymore. Now you can see its due DAG size - specified EPOCH number. Another fact is Mem tweak just works for GDDR5x memory. Your 1070Ti is just GDDR5. But its there small hashrate drop but not significant like with mem tweak switch.

Also look on that generating times for DAG which proove mem tweak is off. (8 secs with mem tweak and 12 secs without him)

Quote

20201121 13:45:23 WARN: BENCHMARK MODE (ethash)
20201121 13:45:23 WARN: EPOCH 383
20201121 13:45:23 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20
20201121 13:45:26 GPU #0: generating DAG 3.99 GB for epoch 383 ...
20201121 13:45:38 GPU #0: DAG generated [time: 12377 ms], memory left: 3.58 GB
20201121 13:45:55 GPU #0: using kernel #2
20201121 13:45:56 Total: 21.47 MH/s
20201121 13:45:58 Total: 21.48 MH/s

20201121 13:46:40 WARN: BENCHMARK MODE (ethash)
20201121 13:46:40 WARN: EPOCH 384
20201121 13:46:40 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20
20201121 13:46:43 GPU #0: generating DAG 4.00 GB for epoch 384 ...
20201121 13:46:55 GPU #0: DAG generated [time: 12389 ms], memory left: 3.57 GB
20201121 13:47:13 GPU #0: using kernel #5
20201121 13:47:14 Total: 20.87 MH/s
20201121 13:47:16 Total: 20.88 MH/s

Yes I can see the memtweak is off, thanks. However, I'm very surprised about your 1080 being slower than my 1070Ti on epoch 383. I'm getting 24.5Mh/s with PL 90W, CORE 0, MEM +1000. DAG generation time is ~8.8s, without memory tweaks

ExEric3

full member

Activity: 176

Merit: 100

Quote from: trexminer on November 21, 2020, 07:23:29 AM

Hmm, very curious indeed. I did a benchmark with my 1070Ti with --mt 5 on Linux and didn't notice a hashrate decrease for 383/384 epoch threshold. Could you try benchmarking without --mt?
UPD: What OS are you using? Win 7 by any chance?

How I typed from start. Mem tweak doesnt work anymore. Now you can see its due DAG size - specified EPOCH number. Another fact is Mem tweak just works for GDDR5x memory. Your 1070Ti is just GDDR5. But its there small hashrate drop but not significant like with mem tweak switch.

Also look on that generating times for DAG which proove mem tweak is off. (8 secs with mem tweak and 12 secs without him)

Quote

20201121 13:45:23 WARN: BENCHMARK MODE (ethash)
20201121 13:45:23 WARN: EPOCH 383
20201121 13:45:23 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20
20201121 13:45:26 GPU #0: generating DAG 3.99 GB for epoch 383 ...
20201121 13:45:38 GPU #0: DAG generated [time: 12377 ms], memory left: 3.58 GB
20201121 13:45:55 GPU #0: using kernel #2
20201121 13:45:56 Total: 21.47 MH/s
20201121 13:45:58 Total: 21.48 MH/s

20201121 13:46:40 WARN: BENCHMARK MODE (ethash)
20201121 13:46:40 WARN: EPOCH 384
20201121 13:46:40 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20
20201121 13:46:43 GPU #0: generating DAG 4.00 GB for epoch 384 ...
20201121 13:46:55 GPU #0: DAG generated [time: 12389 ms], memory left: 3.57 GB
20201121 13:47:13 GPU #0: using kernel #5
20201121 13:47:14 Total: 20.87 MH/s
20201121 13:47:16 Total: 20.88 MH/s

trexminer

member

Activity: 283

Merit: 63

Quote from: ExEric3 on November 21, 2020, 05:51:22 AM

So christmas gift will get miners - looks like mem tweak works just to 3.99GB DAG size:

Hmm, very curious indeed. I did a benchmark with my 1070Ti with --mt 5 on Linux and didn't notice a hashrate decrease for 383/384 epoch threshold. Could you try benchmarking without --mt?
UPD: What OS are you using? Win 7 by any chance?

ExEric3

full member

Activity: 176

Merit: 100

Quote from: trexminer on November 21, 2020, 04:13:18 AM

I reckon this info is only relevant for 4GB cards where the GPU can't hold the entire DAG. 1080Tis have plenty of memory, so it must be something different. You can do some experiments with "t-rex -a ethash --benchmark --benchmark-epoch NNN" and see how it behaves for different epochs

So christmas gift will get miners - looks like mem tweak works just to 3.99GB DAG size:

Quote

20201121 11:30:24 WARN: BENCHMARK MODE (ethash)
20201121 11:30:24 WARN: EPOCH 383
20201121 11:31:10 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20, mtweak 5
20201121 11:31:13 GPU #0: generating DAG 3.99 GB for epoch 383 ...
20201121 11:31:20 GPU #0: DAG generated [time: 8008 ms], memory left: 3.28 GB
20201121 11:31:37 GPU #0: using kernel #2
20201121 11:31:38 Total: 33.39 MH/s
20201121 11:31:40 Total: 33.39 MH/s

20201121 11:32:32 WARN: BENCHMARK MODE (ethash)
20201121 11:32:32 WARN: EPOCH 384
20201121 11:33:02 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20, mtweak 5
20201121 11:33:05 GPU #0: generating DAG 4.00 GB for epoch 384 ...
20201121 11:33:13 GPU #0: DAG generated [time: 7936 ms], memory left: 3.27 GB
20201121 11:33:30 GPU #0: using kernel #5
20201121 11:33:31 Total: 20.89 MH/s
20201121 11:33:33 Total: 20.88 MH/s

20201121 11:34:52 WARN: BENCHMARK MODE (ethash)
20201121 11:34:52 WARN: EPOCH 385
20201121 11:35:22 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20, mtweak 5
20201121 11:35:25 GPU #0: generating DAG 4.01 GB for epoch 385 ...
20201121 11:35:33 GPU #0: DAG generated [time: 7950 ms], memory left: 3.27 GB
20201121 11:35:51 GPU #0: using kernel #5
20201121 11:35:54 Total: 20.72 MH/s
20201121 11:35:56 Total: 20.72 MH/s

20201121 11:37:05 WARN: BENCHMARK MODE (ethash)
20201121 11:37:05 WARN: EPOCH 390
20201121 11:37:36 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20, mtweak 5
20201121 11:37:38 GPU #0: generating DAG 4.05 GB for epoch 390 ...
20201121 11:37:46 GPU #0: DAG generated [time: 8030 ms], memory left: 3.22 GB
20201121 11:38:04 GPU #0: using kernel #5
20201121 11:38:05 Total: 20.66 MH/s
20201121 11:38:07 Total: 20.66 MH/s

20201121 11:38:58 WARN: BENCHMARK MODE (ethash)
20201121 11:38:58 WARN: EPOCH 395
20201121 11:39:29 WARN: GPU #0(000100): ASUS GeForce GTX 1080, intensity set to 20, mtweak 5
20201121 11:39:32 GPU #0: generating DAG 4.09 GB for epoch 395 ...
20201121 11:39:39 GPU #0: DAG generated [time: 8111 ms], memory left: 3.24 GB
20201121 11:39:57 GPU #0: using kernel #5
20201121 11:39:58 Total: 20.40 MH/s
20201121 11:40:00 Total: 20.41 MH/s

trexminer

member

Activity: 283

Merit: 63

I reckon this info is only relevant for 4GB cards where the GPU can't hold the entire DAG. 1080Tis have plenty of memory, so it must be something different. You can do some experiments with "t-rex -a ethash --benchmark --benchmark-epoch NNN" and see how it behaves for different epochs

ExEric3

full member

Activity: 176

Merit: 100

Quote from: trexminer on November 21, 2020, 03:37:10 AM

Wow, that difference in hashrate can't be explained by a not functioning memory tweak. "--mt" is only supposed to boost your hashrate by a couple of MH/s per card, there's definitely something else going on. I've seen similar reports elsewhere it's a common problem regardless of what miner is used. I'm not aware of a solution for this issue, unfortunately.

I think its this (Mem tweak was designed for 4GB?) - yeah page aimed on RX 580 but similiar behaviour Im seeing now on nVidia:
https://2miners.com/blog/how-to-mine-ethereum-and-ethereum-classic-on-4gb-gpus/

What u think?

trexminer

member

Activity: 283

Merit: 63

Quote from: ExEric3 on November 21, 2020, 02:45:28 AM

Are you sure? It causing different epoch. Also on ETC GPU clock is 2100MHz and MCU 40% usage. On ETH GPU clock is 1500MHz MCU 100% usage. Totaly real usage during mining.

ETC-mtweak 1
20201121 07:32:17 Mining at ethash.mine.zergpool.com:9999
20201121 07:32:17 GPU #1: EVGA GTX 1080 Ti - 18.20 MH/s, [T:49C, P:144W, F:55%, E:128kH/W], 2/2 R:0%
20201121 07:32:17 GPU #3: EVGA GTX 1080 Ti - 18.33 MH/s, [T:51C, P:146W, F:55%, E:127kH/W], 2/2 R:0%
20201121 07:32:17 Shares/min: 4.048 (Avr. 1.524), Avr.Power: 286W, Avr.Efficiency: 128kH/W
20201121 07:32:17 Uptime: 2 mins 37 secs | Algo: ethash | T-Rex v0.18.11

ETH-mtweak 1
20201121 07:33:52 Mining at ethash.mine.zergpool.com:9999
20201121 07:33:52 GPU #1: EVGA GTX 1080 Ti - 42.85 MH/s, [T:46C, P:146W, F:54%, E:330kH/W], 1/1 R:0%
20201121 07:33:52 GPU #3: EVGA GTX 1080 Ti - 42.14 MH/s, [T:47C, P:143W, F:55%, E:329kH/W], 3/3 R:0%
20201121 07:33:52 Shares/min: 23.706 (Avr. 5.614), Avr.Power: 258W, Avr.Efficiency: 329kH/W
20201121 07:33:52 Uptime: 42 secs | Algo: ethash | T-Rex v0.18.11

Wow, that difference in hashrate can't be explained by a not functioning memory tweak. "--mt" is only supposed to boost your hashrate by a couple of MH/s per card, there's definitely something else going on. I've seen similar reports elsewhere it's a common problem regardless of what miner is used. I'm not aware of a solution for this issue, unfortunately.

ExEric3

full member

Activity: 176

Merit: 100

Quote from: trexminer on November 16, 2020, 03:49:44 AM

Quote from: ExEric3 on November 15, 2020, 01:49:24 PM

Memtweak doesnt work on ETC anymore with 1080Ti?

It does

Are you sure? It causing different epoch. Also on ETC GPU clock is 2100MHz and MCU 40% usage. On ETH GPU clock is 1500MHz MCU 100% usage. Totaly real usage during mining.

ETC-mtweak 1
20201121 07:32:17 Mining at ethash.mine.zergpool.com:9999
20201121 07:32:17 GPU #1: EVGA GTX 1080 Ti - 18.20 MH/s, [T:49C, P:144W, F:55%, E:128kH/W], 2/2 R:0%
20201121 07:32:17 GPU #3: EVGA GTX 1080 Ti - 18.33 MH/s, [T:51C, P:146W, F:55%, E:127kH/W], 2/2 R:0%
20201121 07:32:17 Shares/min: 4.048 (Avr. 1.524), Avr.Power: 286W, Avr.Efficiency: 128kH/W
20201121 07:32:17 Uptime: 2 mins 37 secs | Algo: ethash | T-Rex v0.18.11

ETH-mtweak 1
20201121 07:33:52 Mining at ethash.mine.zergpool.com:9999
20201121 07:33:52 GPU #1: EVGA GTX 1080 Ti - 42.85 MH/s, [T:46C, P:146W, F:54%, E:330kH/W], 1/1 R:0%
20201121 07:33:52 GPU #3: EVGA GTX 1080 Ti - 42.14 MH/s, [T:47C, P:143W, F:55%, E:329kH/W], 3/3 R:0%
20201121 07:33:52 Shares/min: 23.706 (Avr. 5.614), Avr.Power: 258W, Avr.Efficiency: 329kH/W
20201121 07:33:52 Uptime: 42 secs | Algo: ethash | T-Rex v0.18.11

trexminer

member

Activity: 283

Merit: 63

T-Rex 0.18.11

Bug fixes:
* (Windows) Miner is not displaying GPU power consumption for some video driver versions
* "can't pair device with KEY" error (see https://github.com/trexminer/T-Rex/issues/89)
* Display GPU names when NVML library can't be loaded or disabled

trexminer

member

Activity: 283

Merit: 63

T-Rex 0.18.10

* Add optional --no-strict-ssl parameter to disable certificate validation for SSL connections
* Change --fork-at parameter syntax to the following:
Epoch number: =epoch: (eg: --fork-at etchash=epoch:390 <-- use this one for upcoming ETC hardfork).
Block number: =block: (eg: --fork-at x16rv2=block:6526421).
Time: =time:. Time must be set in UTC+0.

trexminer

member

Activity: 283

Merit: 63

Quote from: Joatle on November 18, 2020, 07:41:48 AM

Thanks, that stoped that unvalid problem, but also this came now:

20201118 13:38:13 GPU #6: generating DAG 3.88 GB (96.41% of full size) for epoch 387 ...
20201118 13:38:13 WARN: GPU #0: not enough memory to hold the full DAG for epoch 387.
20201118 13:38:13 WARN: GPU #0: effective hashrate will be ~9.3% of its usual value

So i guess i'll just power down untill the fork and etchash is usable

Yep, 4GB cards will be back in business after the fork

Joatle

newbie

Activity: 2

Merit: 0

Quote from: trexminer on November 18, 2020, 07:34:51 AM

You seem to be mining etchash, whereas ETC is not going to use it until after the fork. Use ethash for now.

Thanks, that stoped that unvalid problem, but also this came now:

20201118 13:38:13 GPU #6: generating DAG 3.88 GB (96.41% of full size) for epoch 387 ...
20201118 13:38:13 WARN: GPU #0: not enough memory to hold the full DAG for epoch 387.
20201118 13:38:13 WARN: GPU #0: effective hashrate will be ~9.3% of its usual value

So i guess i'll just power down untill the fork and etchash is usable

trexminer

member

Activity: 283

Merit: 63

Quote from: Joatle on November 18, 2020, 06:44:04 AM

Hi!

I'm running 7 x 1050 ti's on t-rex-v0.18.9-cuda10.0 on simpleminer OS, and i keep getting invalid shares only..
Any help?

20201118 12:20:22 .......... miner is working at a speed of 90.28 MH/s
20201118 12:20:25 etchash epoch: 193, block: 11631267, diff: 0.994 = 4.00 Gh
20201118 12:20:30 etchash epoch: 193, block: 11631268, diff: 0.994 = 4.00 Gh
20201118 12:20:43 [FAIL] 0/1 - Provided PoW solution is invalid!, 39ms ... GPU #2
20201118 12:20:50 etchash epoch: 193, block: 11631269, diff: 0.994 = 4.00 Gh
20201118 12:21:05 etchash epoch: 193, block: 11631270, diff: 0.994 = 4.00 Gh
20201118 12:21:07 [FAIL] 0/2 - Provided PoW solution is invalid!, 57ms ... GPU #4
20201118 12:21:31 etchash epoch: 193, block: 11631271, diff: 0.994 = 4.00 Gh
20201118 12:21:31 [FAIL] 0/3 - Provided PoW solution is invalid!, 39ms ... GPU #4
20201118 12:21:40 etchash epoch: 193, block: 11631272, diff: 0.994 = 4.00 Gh
20201118 12:21:48 [FAIL] 0/4 - Provided PoW solution is invalid!, 38ms ... GPU #6

You seem to be mining etchash, whereas ETC is not going to use it until after the fork. Use ethash for now.

Joatle

newbie

Activity: 2

Merit: 0

Hi!

I'm running 7 x 1050 ti's on t-rex-v0.18.9-cuda10.0 on simpleminer OS, and i keep getting invalid shares only..
Any help?

20201118 12:20:22 .......... miner is working at a speed of 90.28 MH/s
20201118 12:20:25 etchash epoch: 193, block: 11631267, diff: 0.994 = 4.00 Gh
20201118 12:20:30 etchash epoch: 193, block: 11631268, diff: 0.994 = 4.00 Gh
20201118 12:20:43 [FAIL] 0/1 - Provided PoW solution is invalid!, 39ms ... GPU #2
20201118 12:20:50 etchash epoch: 193, block: 11631269, diff: 0.994 = 4.00 Gh
20201118 12:21:05 etchash epoch: 193, block: 11631270, diff: 0.994 = 4.00 Gh
20201118 12:21:07 [FAIL] 0/2 - Provided PoW solution is invalid!, 57ms ... GPU #4
20201118 12:21:31 etchash epoch: 193, block: 11631271, diff: 0.994 = 4.00 Gh
20201118 12:21:31 [FAIL] 0/3 - Provided PoW solution is invalid!, 39ms ... GPU #4
20201118 12:21:40 etchash epoch: 193, block: 11631272, diff: 0.994 = 4.00 Gh
20201118 12:21:48 [FAIL] 0/4 - Provided PoW solution is invalid!, 38ms ... GPU #6

trexminer

member

Activity: 283

Merit: 63

Quote from: pitersan on November 17, 2020, 03:36:00 PM

HELP

PL 50

Looks like an overclock issue. Try dialing it down, or select the kernel manually with "--kernel N" (N from 1 to 5), and see which one doesn't crash the miner.

pitersan

member

Activity: 827

Merit: 14

cuda 11
457.30-desktop-win10-64bit-international-dch-whql
Gut?

pitersan

member

Activity: 827

Merit: 14

HELP

PL 50

trexminer

member

Activity: 283

Merit: 63

0.18.9

* (ethash) Increase pool side hashrate by 1-2% by reducing the amount of stale shares the miner drops before sending to the pool
* (ethash, kawpow, progpow) Set worker name to "%hostname%" if not specified

Bug fixes:
* (ethash) Low pool side hashrate when mining ETH+ZIL due to incorrect difficulty parsing
* (ethash, kawpow, progpow) Miner crashes on low RAM multi-GPU rigs during DAG rebuild

trexminer

member

Activity: 283

Merit: 63

Quote from: ExEric3 on November 15, 2020, 01:49:24 PM

Memtweak doesnt work on ETC anymore with 1080Ti?

It does

Topic: T-Rex 0.26.8 ETHW, ETC (+dual mining) GPU miner - page 46. (Read 174982 times)