Gateless Gate Sharp 1.3.8: 30Mh/s (Ethash) on RX 480!

[zawawa]

What about Pitcairns low speed in 1.1.2?
Why Pitcairn sensors didn't work?
And why enabled Pitcairn card disables hardware sensors on all cards.

Please see my post above. Problems with sensors are actually driver problems, and there is not much I can do about it at this point. Sadly, it seems that AMD's support for GCN1 cards is faltering... I will probably able to fix this problem when I update the codes for hardware management, but that would take weeks, if not months.

Understand. Sensors can wait. Please check the speed issue on Pitcairns with 1.1.2.

I suggested in a post above that you should check the number of threads.

I tried 1 thread, but it gives even lower speed than 2...
I will try more...

Did you update the display driver recently?
Besides the default number of threads, there shouldn't be any differences between 1.1.1 and 1.1.2 in terms of the CryptoNight kernel.

[1714768210]

1714768210

[1714768210]

1714768210

[1714768210]

1714768210

[1714768210]

1714768210

[1714768210]

1714768210

[1714768210]

1714768210

[UnclWish]

What about Pitcairns low speed in 1.1.2?
Why Pitcairn sensors didn't work?
And why enabled Pitcairn card disables hardware sensors on all cards.

Please see my post above. Problems with sensors are actually driver problems, and there is not much I can do about it at this point. Sadly, it seems that AMD's support for GCN1 cards is faltering... I will probably able to fix this problem when I update the codes for hardware management, but that would take weeks, if not months.

Understand. Sensors can wait. Please check the speed issue on Pitcairns with 1.1.2.

I suggested in a post above that you should check the number of threads.

I tried 1 thread, but it gives even lower speed than 2...
I will try more...

Did you update the display driver recently?
Besides the default number of threads, there shouldn't be any differences between 1.1.1 and 1.1.2 in terms of the CryptoNight kernel.

Yes. Driver 17.11.4 beta. In 1.1.1 with 2 threads about 495 h/s. In 1.1.2 with the same settings and driver about 400. 1 thread gives 325-350 h/s.
I will check it again right now...

P.S. On RX 580 with 2 threads on 1.1.2 speed the same as in 1.1.1. Default # of threads from 2 to 1 not for all cards?

[pokeytex]

@zawawa - is dual mining of lbry active now on nvidia pascal cards?

[UnclWish]

Spend several hours to find normal parameters for Pitcairn GPU in 1.1.2.
But unlucky ((((
With one thread max I can reach is about 380-400 h/s with Intensity 1152-1280. With intensity 512 speed is 225 h/s. With 1024 - 350 h/s
With 2 threads max Intensity is 512. Higher gives errors on shares. Speed not higher than 350-375 h/s.
It's something wrong in 1.1.2 with GCN1 cards I think. Forced to use 1.1.1 for normal speed on Pitcarin chip.

And some notices:
1. After pressing "Stop" miner dosn't release videomemory and if press "Start" speeds can be other or may start to give 100% rejected shares, than after closing GGS and starting him again.
2. After starting GGS with enabled auto-start function, speed on some GPU may be not maximum. But if press "Stop" and "Start" gives max speed. Maybe without auto-start it works another way. I didn't check it. Mabe it have some link with notice from p.1.

[zawawa]

Spend several hours to find normal parameters for Pitcairn GPU in 1.1.2.
But unlucky ((((
With one thread max I can reach is about 380-400 h/s with Intensity 1152-1280. With intensity 512 speed is 225 h/s. With 1024 - 350 h/s
With 2 threads max Intensity is 512. Higher gives errors on shares. Speed not higher than 350-375 h/s.
It's something wrong in 1.1.2 with GCN1 cards I think. Forced to use 1.1.1 for normal speed on Pitcarin chip.

And some notices:
1. After pressing "Stop" miner dosn't release videomemory and if press "Start" speeds can be other or may start to give 100% rejected shares, than after closing GGS and starting him again.
2. After starting GGS with enabled auto-start function, speed on some GPU may be not maximum. But if press "Stop" and "Start" gives max speed. Maybe without auto-start it works another way. I didn't check it. Mabe it have some link with notice from p.1.

I wouldn't recommend 1.1.1, though. Bug fixes in 1.1.2 are too important to ignore.

[zawawa]

@zawawa - is dual mining of lbry active now on nvidia pascal cards?

It is not available on any platform yet. I am finalizing dual-mining with Pascal now.
After that, I will try Lbry one more time.

[UnclWish]

I wouldn't recommend 1.1.1, though. Bug fixes in 1.1.2 are too important to ignore.

But losing more than 100 h/s of cryptonight speed is substantial difference...
I don't know what else I can try...

EDIT:
I noticed that in 1.1.2 after start in log writes "Built options: -O1  -IKernels -DWORKSIZE=8"
But on 1.1.1 it writes "Built options: -O5  -IKernels -DWORKSIZE=8"

I don't know what it does, but I finded in OpenCLCryptoNightMiner.cs
string 109:
                String buildOptions = (Device.Vendor == "AMD"    ? "-O1 " : //"-O1 " :
on 1.1.1 it was
                String buildOptions = (Device.Vendor == "AMD"    ? "-O5 " : //"-O1 " :

As I understand - in 1.1.1 parameter -O was 5 for AMD cards and 1 for others.
But in 1.1.2 you make it equal.

[zawawa]

That's actually an excellent catch!
The "-O1" build option is generally more preferable because excessive optimizations tend to cause instability.
I will fix this in the next version.

[UnclWish]

That's actually an excellent catch!
The "-O1" build option is generally more preferable because excessive optimizations tend to cause instability.
I will fix this in the next version.

Thanks! But could you be more specifical?
O1 - means less optimization
O5 - more optimization
Right?
Are there other variants of O option?
Strange then that O1 didn't work normal for Pitcairn, but O5 works... Although for RX 580 the difference is not noticeable.

[zawawa]

zawawa, take a look into Pitcairn problem with cryptonight, please. I think there is a lot of peoples who has 270/370 cards.
But I was forced to return to 1.1.1 version. 1.1.2 gives too low speed.

And yet another wishlist:
1. Average speed for hour/minute/from start
2. Total finded/rejected shares.
3. Individual settings for each thread.
4. Individual algo for each GPU.
5. Info about current difficulty, dag #
6. Possibility to running multiple instances of GSS.
7. Possibility to start miner via command line.

See, it is not that hard to add more information on the Dashboard tab page, but I don't want to stuff too much in there. I got to come up with a good idea...

Actually, I found a pretty good solution to this dilemma around screen real estate.
I will implement it when I find time...

[zawawa]

That's actually an excellent catch!
The "-O1" build option is generally more preferable because excessive optimizations tend to cause instability.
I will fix this in the next version.

Thanks! But could you be more specifical?
O1 - means less optimization
O5 - more optimization
Right?
Strange then that O1 didn't work normal for Pitcairn, but O5 works... Although for RX 580 the difference is not noticeable.

There is logic behind these differences, but you really don't want to know. Trust me.

[UnclWish]

That's actually an excellent catch!
The "-O1" build option is generally more preferable because excessive optimizations tend to cause instability.
I will fix this in the next version.

Thanks! But could you be more specifical?
O1 - means less optimization
O5 - more optimization
Right?
Strange then that O1 didn't work normal for Pitcairn, but O5 works... Although for RX 580 the difference is not noticeable.

There is logic behind these differences, but you really don't want to know. Trust me.

Ok, of course. You're the brain! I'm only test and notice a little mistakes in your awesome work!

[seth6]

That's actually an excellent catch!
The "-O1" build option is generally more preferable because excessive optimizations tend to cause instability.
I will fix this in the next version.

Hi, Zawawa.
First of all, thank you. You do a good job. Better stability and more hashrate , than other miners - for me at least.
I use v.1.1.1 because in ver. 1.1.2 I have the same problem with my 280x GPU. Hashrate drop from 560 to 300 h/s and other GPUs lost some 35 h/s with the same settings and drivers.(blockchain drivers)
R9 290 / 390bios - 830h dro to 795h
R9 Fury / 400 mem straps  on 500 MHz - 950h drop to 910

I have tried to run your miner only on 2 GPUs - without 280x, but it seems that miner start build new cryptonight kernel with erroneous values and hashrate go crazy ( 3h/s or 1300h/s and all shares rejected) -the same happend,if I change number of threads to 1 or worksize to 4. After that I have to reboot and enabling all GPUs to back to normal.

Other issue is that sometimes appear a lot of rejected low diff shares, mining on Dwarfpool.  XMRig solved this problem,  take a look at this
 https://github.com/xmrig/xmrig/issues/153

Small bug: when the number of accepted shares exceeds 1000, miner stops showing percentage of accepted shares.

And finally, it would be great if you could add  CPU miner and overclock, voltage and fan speed settings into the miner. Blockchain drivers and Msi AB does not work very well together.

Sorry for my google english.

[ceasarmadini]

Could it mine sumokoin?

[FFI2013]

Could someone who's running 390's and 270's care to share their settings it's been awhile since I ran mine been using rx cards and 1070's but I figured with bitcoin being so high I'd bust out my r9 cards but I'm having trouble finding good settings for xmr thanks

[UnclWish]

Could someone who's running 390's and 270's care to share their settings it's been awhile since I ran mine been using rx cards and 1070's but I figured with bitcoin being so high I'd bust out my r9 cards but I'm having trouble finding good settings for xmr thanks

For these cards use 1.1.1. version until zawawa updates 1.1.2.
For 390 didn't know. For 270 - try Threads 2/Intensity 480/LocalWorkSize 8.

[seth6]

That's actually an excellent catch!
The "-O1" build option is generally more preferable because excessive optimizations tend to cause instability.
I will fix this in the next version.

Hi, Zawawa.
First of all, thank you. You do a good job. Better stability and more hashrate , than other miners - for me at least.
I use v.1.1.1 because in ver. 1.1.2 I have the same problem with my 280x GPU. Hashrate drop from 560 to 300 h/s and other GPUs lost some 35 h/s with the same settings and drivers.(blockchain drivers)
R9 290 / 390bios - 830h dro to 795h
R9 Fury / 400 mem straps  on 500 MHz - 950h drop to 910

I have tried to run your miner only on 2 GPUs - without 280x, but it seems that miner start build new cryptonight kernel with erroneous values and hashrate go crazy ( 3h/s or 1300h/s and all shares rejected) -the same happend,if I change number of threads to 1 or worksize to 4. After that I have to reboot and enabling all GPUs to back to normal.

Other issue is that sometimes appear a lot of rejected low diff shares, mining on Dwarfpool.  XMRig solved this problem,  take a look at this
 https://github.com/xmrig/xmrig/issues/153

Small bug: when the number of accepted shares exceeds 1000, miner stops showing percentage of accepted shares.

And finally, it would be great if you could add  CPU miner and overclock, voltage and fan speed settings into the miner. Blockchain drivers and Msi AB does not work very well together.

Sorry for my google english.

- a lot of rejected shares start when receive duplicate job




2017-12-04 20:56:19.4196 [15] Device #0 submitted a share.
2017-12-04 20:56:19.5517 [17] Share accepted.
2017-12-04 20:56:21.2264 [18] Device #2: 411,84 h/s
2017-12-04 20:56:21.4555 [6] Device #2: 412,36 h/s
2017-12-04 20:56:21.5215 [14] Device #1: 272,38 h/s
2017-12-04 20:56:26.9557 [15] Device #0: 464,86 h/s
2017-12-04 20:56:28.7094 [13] Device #1: 272,47 h/s
2017-12-04 20:56:29.8789 [16] Device #0: 463,81 h/s
2017-12-04 20:56:31.8326 [18] Device #2: 411,84 h/s
2017-12-04 20:56:32.0477 [6] Device #2: 412,36 h/s
2017-12-04 20:56:33.0281 [14] Device #1: 272,43 h/s
2017-12-04 20:56:34.3126 [17] Received new job: 673208636092022       -----------------DUPLICATE JOB
2017-12-04 20:56:35.5972 [17] Received new job: 673208636092022
2017-12-04 20:56:37.4169 [16] Device #0 submitted a share.
2017-12-04 20:56:37.4869 [17] Share accepted.
2017-12-04 20:56:38.2672 [15] Device #0: 462,61 h/s
2017-12-04 20:56:40.2140 [13] Device #1: 272,71 h/s
2017-12-04 20:56:41.1824 [16] Device #0: 463,32 h/s
2017-12-04 20:56:42.4379 [18] Device #2: 411,84 h/s
2017-12-04 20:56:42.6390 [6] Device #2: 412,50 h/s
2017-12-04 20:56:44.5317 [14] Device #1: 272,65 h/s
2017-12-04 20:56:47.8211 [14] Device #1 submitted a share.
2017-12-04 20:56:47.9221 [17] Share rejected: Low difficulty share
2017-12-04 20:56:49.5728 [15] Device #0: 462,12 h/s
2017-12-04 20:56:51.7216 [13] Device #1: 272,27 h/s
2017-12-04 20:56:52.4899 [16] Device #0: 463,56 h/s
2017-12-04 20:56:53.0452 [18] Device #2: 411,57 h/s
2017-12-04 20:56:53.2302 [6] Device #2: 412,55 h/s
2017-12-04 20:56:56.0404 [14] Device #1: 272,51 h/s
2017-12-04 20:57:00.8813 [15] Device #0: 462,04 h/s
2017-12-04 20:57:03.2253 [13] Device #1: 272,98 h/s
2017-12-04 20:57:03.6514 [18] Device #2: 411,84 h/s
2017-12-04 20:57:03.7955 [16] Device #0: 462,82 h/s
2017-12-04 20:57:03.8215 [6] Device #2: 412,40 h/s
2017-12-04 20:57:07.5440 [14] Device #1: 272,81 h/s
2017-12-04 20:57:08.3603 [6] Device #2 submitted a share.
2017-12-04 20:57:08.4614 [17] Share rejected: Low difficulty share
2017-12-04 20:57:12.1939 [15] Device #0: 462,48 h/s
2017-12-04 20:57:14.2567 [18] Device #2: 411,84 h/s
2017-12-04 20:57:14.4118 [6] Device #2: 412,54 h/s
2017-12-04 20:57:14.7329 [13] Device #1: 272,47 h/s
2017-12-04 20:57:15.0990 [16] Device #0: 462,58 h/s
2017-12-04 20:57:19.0516 [14] Device #1: 272,32 h/s
2017-12-04 20:57:20.3201 [18] Device #2 submitted a share.
2017-12-04 20:57:20.4202 [17] Share rejected: Low difficulty share
2017-12-04 20:57:23.4954 [15] Device #0: 462,76 h/s
2017-12-04 20:57:24.8770 [18] Device #2: 410,22 h/s
2017-12-04 20:57:25.0090 [6] Device #2: 411,18 h/s
2017-12-04 20:57:26.2385 [13] Device #1: 272,58 h/s
2017-12-04 20:57:26.4116 [16] Device #0: 462,33 h/s
2017-12-04 20:57:28.0513 [6] Device #2 submitted a share.
2017-12-04 20:57:28.1893 [17] Share rejected: Low difficulty share
2017-12-04 20:57:30.5563 [14] Device #1: 272,63 h/s
2017-12-04 20:57:34.8020 [15] Device #0: 462,54 h/s
2017-12-04 20:57:35.5803 [18] Device #2: 405,96 h/s
2017-12-04 20:57:35.6833 [6] Device #2: 407,47 h/s
2017-12-04 20:57:37.7201 [16] Device #0: 462,09 h/s
2017-12-04 20:57:37.7482 [13] Device #1: 272,19 h/s
2017-12-04 20:57:40.4562 [15] Device #0 submitted a share.
2017-12-04 20:57:40.5623 [17] Share rejected: Low difficulty share
2017-12-04 20:57:42.0699 [14] Device #1: 271,70 h/s
2017-12-04 20:57:42.0699 [14] Device #1 submitted a share.
2017-12-04 20:57:42.1689 [17] Share rejected: Low difficulty share
2017-12-04 20:57:46.1125 [15] Device #0: 463,21 h/s
2017-12-04 20:57:46.4336 [18] Device #2: 399,44 h/s
2017-12-04 20:57:46.4907 [6] Device #2: 401,34 h/s
2017-12-04 20:57:49.0267 [16] Device #0: 461,60 h/s
2017-12-04 20:57:49.2578 [13] Device #1: 272,90 h/s
2017-12-04 20:57:51.9359 [14] Device #1 submitted a share.
2017-12-04 20:57:52.0779 [17] Share rejected: Low difficulty share
2017-12-04 20:57:53.5785 [14] Device #1: 272,74 h/s
2017-12-04 20:57:57.4081 [15] Device #0: 464,79 h/s
2017-12-04 20:57:57.4461 [18] Device #2: 394,90 h/s
2017-12-04 20:57:57.4631 [6] Device #2: 395,93 h/s
2017-12-04 20:58:00.3322 [16] Device #0: 463,31 h/s
2017-12-04 20:58:00.7664 [13] Device #1: 272,49 h/s
2017-12-04 20:58:05.0841 [14] Device #1: 272,42 h/s
2017-12-04 20:58:08.5415 [6] Device #2: 393,52 h/s
2017-12-04 20:58:08.5415 [18] Device #2: 393,40 h/s
2017-12-04 20:58:08.7196 [15] Device #0: 462,93 h/s
2017-12-04 20:58:11.6368 [16] Device #0: 463,31 h/s
2017-12-04 20:58:11.7178 [18] Device #2 submitted a share.
2017-12-04 20:58:11.8429 [17] Share rejected: Low difficulty share
2017-12-04 20:58:12.2710 [13] Device #1: 272,73 h/s
2017-12-04 20:58:16.5928 [14] Device #1: 272,70 h/s
2017-12-04 20:58:19.6550 [18] Device #2: 393,16 h/s
2017-12-04 20:58:19.6550 [6] Device #2: 393,16 h/s
2017-12-04 20:58:20.0242 [15] Device #0: 462,16 h/s
2017-12-04 20:58:22.9433 [16] Device #0: 463,56 h/s
2017-12-04 20:58:23.7777 [13] Device #1: 272,27 h/s
2017-12-04 20:58:28.0984 [14] Device #1: 272,45 h/s
2017-12-04 20:58:30.7645 [6] Device #2: 393,15 h/s
2017-12-04 20:58:30.7645 [18] Device #2: 393,15 h/s
2017-12-04 20:58:30.7645 [6] Device #2 submitted a share.
2017-12-04 20:58:30.8685 [17] Share rejected: Low difficulty share
2017-12-04 20:58:31.3347 [15] Device #0: 462,22 h/s
2017-12-04 20:58:34.2509 [16] Device #0: 463,31 h/s
2017-12-04 20:58:35.2843 [13] Device #1: 272,81 h/s
2017-12-04 20:58:35.5264 [6] Device #2 submitted a share.
2017-12-04 20:58:35.6374 [17] Share rejected: Low difficulty share
2017-12-04 20:58:36.3177 [14] Device #1 submitted a share.
2017-12-04 20:58:36.4217 [17] Share rejected: Low difficulty share
2017-12-04 20:58:38.0224 [16] Device #0 submitted a share.
2017-12-04 20:58:38.1244 [17] Share rejected: Low difficulty share
2017-12-04 20:58:38.7007 [6] Device #2 submitted a share.
2017-12-04 20:58:38.7007 [6] Device #2 submitted a share.
2017-12-04 20:58:38.8037 [17] Share rejected: Low difficulty share
2017-12-04 20:58:38.9077 [17] Share rejected: Low difficulty share
2017-12-04 20:58:39.6030 [14] Device #1: 272,90 h/s
2017-12-04 20:58:41.8749 [6] Device #2: 392,91 h/s
2017-12-04 20:58:41.8749 [18] Device #2: 392,91 h/s
2017-12-04 20:58:42.6463 [15] Device #0: 461,98 h/s
2017-12-04 20:58:43.6687 [16] Device #0 submitted a share.
2017-12-04 20:58:43.7727 [17] Share rejected: Low difficulty share
2017-12-04 20:58:45.5534 [16] Device #0: 462,86 h/s
2017-12-04 20:58:46.7939 [13] Device #1: 272,44 h/s
2017-12-04 20:58:51.1107 [14] Device #1: 272,33 h/s
2017-12-04 20:58:52.9854 [18] Device #2: 393,16 h/s
2017-12-04 20:58:52.9854 [6] Device #2: 393,16 h/s
2017-12-04 20:58:53.9498 [15] Device #0: 462,25 h/s
2017-12-04 20:58:56.8660 [16] Device #0: 462,08 h/s
2017-12-04 20:58:58.2975 [13] Device #1: 272,64 h/s
2017-12-04 20:59:00.3774 [17] Received new job: 172417954239062
2017-12-04 20:59:02.6143 [14] Device #1: 272,75 h/s
2017-12-04 20:59:04.0959 [18] Device #2: 393,16 h/s
2017-12-04 20:59:04.0959 [6] Device #2: 393,16 h/s
2017-12-04 20:59:05.2583 [15] Device #0: 462,59 h/s
2017-12-04 20:59:05.9016 [14] Device #1 submitted a share.
2017-12-04 20:59:06.0156 [17] Share accepted.

[blevotron]

And that's great because we dont need that readability when we mine ))

[UnclWish]

- a lot of rejected shares start when receive duplicate job

It may be a pool problem.

[seth6]

- a lot of rejected shares start when receive duplicate job

It may be a pool problem.

Yes, it is a pool problem. Only XMRig solved this pool issue  disconnecting from pool if duplicated job received and  establishing a new connection.
Easy solution - Zawawa please fix this.