[JCE]Fast & stable CN/v8/Heavy/Tube/XHV miner, CPU+GPU, Vega56 1800+ RX580 1200+

[JCE-Miner]

lebuawu2: thanks for feedback
can you lower significantly your multi-hash  param? it may be related to the config rather than the gpu.

any result on cn-v7 or vega ?

[vmozara]

032t - vega 56 x7, windows 10 1803, adrenalin 18.6.1 or 18.7.1 (dont remember) working ok for me
CN-V7 14280 h/s

[henri2018]

0.32t on Rx 570 4gb, bittube, using 18.5.1, winver 1709, lots of rejected shares, and my miner got banned by miner.rocks.

Backed to 0.32a and all is good.

[JCE-Miner]

use --doublecheck to make bad shares detected before being sent to Pool, to avoid ban.
So far it would be a Ellesmere only bug? Any feedback about cn-v7 and on small rx ?

[henri2018]

0.32t on Rx 570 4gb, bittube, using 18.5.1, winver 1709, lots of rejected shares, and my miner got banned by miner.rocks.

Backed to 0.32a and all is good.

Continue testing:
The above testing was using multi hash 944.

Tried to decrease multi hash to 864, still got rejected shares some times.
Tried again to decrease multi hash to 832, still got rejected shares as well.

Backed to 0.32a with multi hash 944 and all is good, no reject shares.

17:19:21 | Pool sg.bittube.miner.rocks:5555
17:19:21 | Currency Cryptonote
17:19:21 | Current pool Difficulty 160003
17:19:21 | Valid Shares found 3
17:19:21 | Total hashes 480009
17:19:21 | Miner uptime 0:02:40
17:19:21 | Effective net hashrate 3000.06 h/s
17:19:23 | Hashrate GPU Thread 0: 714.35 h/s - Total GPU 0: 714.35 h/s
17:19:23 | Hashrate GPU Thread 1: 719.15 h/s - Total GPU 1: 719.15 h/s
17:19:23 | Hashrate GPU Thread 2: 714.35 h/s - Total GPU 2: 714.35 h/s
17:19:23 | Hashrate GPU Thread 3: 719.15 h/s - Total GPU 3: 719.15 h/s
17:19:23 | Hashrate GPU Thread 4: 720.69 h/s - Total GPU 4: 720.69 h/s
17:19:23 | Hashrate GPU Thread 5: 719.15 h/s - Total GPU 5: 719.15 h/s
17:19:23 | Total: 4306.81 h/s - Max: 4308.57 h/s
17:19:30 | GPU 5 Thread 5 Lane 819 finds a Share, value 160003
17:19:30 | Rejected by the pool.
17:19:30 | Message from the pool: Low difficulty share
17:20:46 | GPU 2 Thread 2 Lane 794 finds a Share, value 160003
17:20:46 | Rejected by the pool.
17:20:46 | Message from the pool: Low difficulty share
17:21:03 | Pool sends a new Job.
17:21:12 | Pool sends a new Job.
17:21:30 | GPU 0 Thread 0 Lane 507 finds a Share, value 160003
17:21:31 | Accepted by the pool.
17:22:08 | GPU 4 Thread 4 Lane 351 finds a Share, value 160003
17:22:08 | Accepted by the pool.

[Iamtutut]

I uploaded the 0.32t (t for test) which is a mix of the 0.32a and c code.
i tested it for hours on all my rigs and got zero bad shares. but i've no Ellesmere.

please tell me if it:
works or not on vega
works or not on other non-Elsmere (rx550, 560...)
works or not on Heavy or v7 or both
if it fails, what card and driver you have
if it fails, retry with a low multihash (to know if i overflow the vram)

that bug i cannot reproduce makes me crazy. it was luck the 32a worked...

Win 10-64 build 1803.
18.6.1 drivers
Mining BITTUBE
3X RX570 4GB with bios mod, config file:

Code:

"gpu_threads_conf" : 
[ 
	{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash": 304}, (display GPU core and mem freq: 1230/2100)
	{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash": 304},
	{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash": 464}, (1200/2020)
	{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash": 464},
	{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 2, "multi_hash": 464}, (1200/2020)
	{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 2, "multi_hash": 464},
],

So far (20mn testing) all's ok no stale no rejected, but raw hashrate is not faster than 0.32a and actual hashrate is low.

[Lermite]

RX 570 8GB and 4GB
Drivers 18.3.4
Win 1607 x64

The 0.32t runs definitely fine on CN Heavy (XHV).

I thought the rejected shares issue might affect only CN Saber but mining Bittube runs fine too, even if the hashrate of my 4BG is still crappy:

Code:

"gpu_threads_conf" : 
[ 
  { "mode": "GPU", "worksize": 4, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 0, "multi_hash": 496 },
  { "mode": "GPU", "worksize": 4, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 0, "multi_hash": 496 },
  { "mode": "GPU", "worksize": 2, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 1, "multi_hash": 480 },
  { "mode": "GPU", "worksize": 2, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 1, "multi_hash": 480 },
]

0: 570 8GB: 839 h/s
1: 570 4GB: 588 h/s

CN Heavy:

Code:

"gpu_threads_conf" : 
[ 
  { "mode": "GPU", "worksize": 4, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 0, "multi_hash": 512 },
  { "mode": "GPU", "worksize": 4, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 0, "multi_hash": 512 },
  { "mode": "GPU", "worksize": 4, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 1, "multi_hash": 480 },
  { "mode": "GPU", "worksize": 4, "alpha": 64, "beta": 8, "gamma": 4, "delta": 4, "epsilon": 4, "zeta": 4, "index": 1, "multi_hash": 480 },
]

0: 570 8GB: 846 h/s
1: 570 4GB: 826 h/s

[JCE-Miner]

Thanks all for the time you take to test.
I reuploaded another test version (same file name) which is still closer to the 0.32a, for those who still get errors.
https://github.com/jceminer/cn_gpu_miner/raw/master/jce_cn_gpu_miner.prototype.032t.zip

@Lermite: why worksize 2?

[dov17]

New 032t, rx552(1170/1930), cn-v7, no rejected.
driver 18.5.1, win10.
Like everything works fine

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 2, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 2, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 3, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 3, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 4, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 4, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 5, "multi_hash" : 464 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 5, "multi_hash" : 464 },

https://imgur.com/a/74zxoKk

[henri2018]

Thanks all for the time you take to test.
I reuploaded another test version (same file name) which is still closer to the 0.32a, for those who still get errors.
https://github.com/jceminer/cn_gpu_miner/raw/master/jce_cn_gpu_miner.prototype.032t.zip

@Lermite: why worksize 2?

Using this latest test version on RX570 4GB, bittube, win 1709, driver 18.6.1, single thread, multi hash 944, now I got exactly same result/hashrate as 0.32a.
No more issue with rejected shares.

Haven't tried yet for heavy/cn-v7.

[JCE-Miner]

Ok, so at least i could go back to Something stable, a warm thank for your tests
I wanted to work on fan/temp and i'm fighting that crap bug...

i'll release another intermediate test version that reintroduce back some optims from 0.32c

[Lermite]

why worksize 2?

Shame on me.
Worksize: 2 was an old attempt to make my 4GB GPU to handle properly the weight of the CN Saber.
Setting it back to 4 increased the hashrate from 588 to 822 h/s on Bittube.

[JCE-Miner]

It looked Strange, JCE perf on Saber are close to Heavy (and to Haven)

I re-uploaded a 3rd t version with some code of 0.32c back. New test on Ellesmere would be welcome again

[Lermite]

Stats per GPU would be nice too, with the numbers and percentage of accepted/rejected/stale share, and the average hashrate over a few minutes.

What it could look like:

Avg Hashrate - Total / Accepted / Rejected / Stale shares
GPU 0: 863 h/s - 748 / 741 (99.06%) / 7 (0.94%) / 22 (2.94%)
GPU 1: 847 h/s - 732 / 718 (98.08%) / 14 (1.91%) / 25 (3.42%)

[rinzlerus]

I used "intensity" : 27, "worksize" : 12, "threads" : 2 on a SRBMiner. What parameters should I use in JSE Miner?

[Lermite]

With the third version of 0.32t:
0: RX 570 8GB
2: RX 570 4GB

CN Saber:

Code:

13:56:41 | Hashrate CPU Thread 3: 67.89 h/s - Total CPUs: 269.15 h/s
13:56:41 | Hashrate GPU Thread 4: 421.03 h/s
13:56:41 | Hashrate GPU Thread 5: 420.11 h/s - Total GPU 0: 841.14 h/s
13:56:41 | Hashrate GPU Thread 6: 411.42 h/s
13:56:41 | Hashrate GPU Thread 7: 412.17 h/s - Total GPU 1: 823.59 h/s

CN Heavy:

Code:

14:00:59 | Hashrate CPU Thread 11: 3.94 h/s - Total CPUs: 330.11 h/s
14:00:59 | Hashrate GPU Thread 12: 415.35 h/s
14:00:59 | Hashrate GPU Thread 13: 432.53 h/s - Total GPU 0: 847.88 h/s
14:00:59 | Hashrate GPU Thread 14: 416.12 h/s
14:00:59 | Hashrate GPU Thread 15: 416.94 h/s - Total GPU 1: 833.05 h/s

And still no rejected share.

[lebuawu2]

It looked Strange, JCE perf on Saber are close to Heavy (and to Haven)

I re-uploaded a 3rd t version with some code of 0.32c back. New test on Ellesmere would be welcome again

RX 580 8GB still got rejected shares, RX Vega 56 running fine without rejected shares but hashrate still lower than SRB.

[JCE-Miner]

I plan to add per GPU info:
* temp
* fan
* broken shares

broken are the shares with bad compute. Info about network-related rejects are irrelevant for a GPU, it's a internet/ping/pool problem.
And the yellow report would include the overral shares found/broken/pool-reject.

Config:
* worksize is the exact same
* intensity in JCE is the multihash parameter in the config, whose value is the same as in Claymore or Stak
* alpha and beta are JCE specific parameters

I strongly advise you fine-tune from scratch, a good SRB config is not always good in JCE, our code is completely separate.
Use the 0.32a, it starts very fast you can tune easily. And you can disable CPU with param --no-cpu

test version : ok thanks a lot, i start to understand what part is broken (even if it doesn't explain why it is broken) and on Ellesmere only.

[henri2018]

It looked Strange, JCE perf on Saber are close to Heavy (and to Haven)

I re-uploaded a 3rd t version with some code of 0.32c back. New test on Ellesmere would be welcome again

Seems that this 3rd version is same as first version. Still no good I think.
Still with RX 570 4GB, Bittube, win 1709, driver 18.6.1, single thread, multi hash 928.

Code:

19:32:27 | Hashrate GPU Thread 4: 812.61 h/s - Total GPU 4: 812.61 h/s
19:32:27 | Hashrate GPU Thread 5: 809.39 h/s - Total GPU 5: 809.39 h/s
19:32:27 | Total: 4849.48 h/s - Max: 4849.48 h/s
19:32:39 | GPU 2 Thread 2 Lane 567 has found an invalid Share.
19:32:51 | GPU 0 Thread 0 Lane 646 has found an invalid Share.
19:33:29 | GPU 2 Thread 2 Lane 306 finds a Share, value 160003
19:33:29 | Accepted by the pool.
19:33:29 | GPU 5 Thread 5 Lane 408 finds a Share, value 160003
19:33:29 | Accepted by the pool.
19:34:07 | GPU 4 Thread 4 Lane 855 has found an invalid Share.
19:34:09 | Pool sends a new Job.
19:34:35 | GPU 2 Thread 2 Lane 372 finds a Share, value 160003
19:34:42 | Accepted by the pool.
19:34:42 | Pool sends a new Job.
19:35:01 | Pool sends a new Job.
19:35:07 | GPU 3 Thread 3 Lane 23 finds a Share, value 160003
19:35:07 | Accepted by the pool.
19:35:29 | Pool sends a new Job.
19:35:37 | GPU 4 Thread 4 Lane 788 has found an invalid Share.
19:35:48 | GPU 0 Thread 0 Lane 522 has found an invalid Share.

[JCE-Miner]

Thanks again, and that's somehow what i would expect.
I re-uploaded a fourth (and I hope, a last) t version with all my optims but the one of the 3rd upload. It should mine good in all cases. If it works, i'll re-release it as a normal version.

Perf on Heavy and al. : for now SRB is better than JCE on RX470/570+ and Vega, since i've none of those cards yet, i cannot blind optimize any more. On small RX550/560 it should be faster.