Topic: Gateless Gate Sharp 1.3.8: 30Mh/s (Ethash) on RX 480! - page 68. (Read 214431 times)

Everything that is necessary for this feature is already in place.
I am not sure about enabling it for alpha versions, though...
I have a few more hurdles to go over before the miner is "feature-complete."
We will see how that goes.

I didn't do anything to optimize Lyra2REv2 yet.
I know for a fact that pre-calc with CPU is possible for the first rounds, though.
I will get to it when I have time.

Gateless Gate Sharp 1.2.11 alpha
https://github.com/zawawawa/GatelessGateSharp/releases/tag/v1.2.11-alpha

* Addressed compatibility issues with yiimp-based pools.
* Made stratum-related codes more robust.
* Adjusted default settings for RX 470/480/570/580.
* Switched from intensity to raw intensity for NeoScrypt.
* Enhanced the automatic upgrade feature.

This should fix all the reported pool-related problems.
Please do let me know if they still persist.

Any idea to get 30mh/s or more with a Rx580 in Lyra2Rev with Gateless Gate Sharp as is possible with another miner?

Worksize, intensity...

Or wait for a new version?

Downgraded to 1.2.9 and number of shares in Gui is not correct. See the status bar against Gui:

https://i.imgur.com/DwhWMkE.png

Exactly same problem here with pool.Bsod.pw. Downgrading...



1.2.10:

...

2018-01-28 22:41:14.2152 [16] Device #2 submitted Share #13 to pool.bsod.pw as RQL48SX2P3m1ySmsyFFexNNKFiJiQ9e4e9.RXNITROMQ0.
2018-01-28 22:41:14.5339 [15] Share #12 accepted.
2018-01-28 22:41:14.7889 [16] Device #2: 884,41 Kh/s (NeoScrypt)
2018-01-28 22:41:14.8848 [15] Share #13 accepted.
2018-01-28 22:41:24.7909 [16] Device #2: 900,33 Kh/s (NeoScrypt)
2018-01-28 22:41:34.7973 [16] Device #2: 899,49 Kh/s (NeoScrypt)
2018-01-28 22:41:44.7997 [16] Device #2: 897,40 Kh/s (NeoScrypt)
2018-01-28 22:41:54.8027 [16] Device #2: 898,04 Kh/s (NeoScrypt)
2018-01-28 22:42:01.8441 [15] Received new job: 7e
2018-01-28 22:42:04.8299 [16] Device #2: 902,64 Kh/s (NeoScrypt)
2018-01-28 22:42:09.9310 [10] Running garbage collection...
2018-01-28 22:42:09.9310 [10] Memory used before collection: 46MB
2018-01-28 22:42:09.9565 [10] Memory used before collection: 12MB
2018-01-28 22:42:14.8569 [16] Device #2: 902,66 Kh/s (NeoScrypt)
2018-01-28 22:42:23.4647 [1] Switching to the DEVFEE mode...
2018-01-28 22:42:24.8840 [16] Device #2: 900,22 Kh/s (NeoScrypt)
2018-01-28 22:42:26.2380 [16] Restarting miner thread...
2018-01-28 22:42:27.4606 [1] Stopping miners...
2018-01-28 22:42:27.5470 [16] Exception in miner thread: Stratum server failed to send a new job.   em GatelessGateSharp.OpenCLNeoScryptMiner.MinerThread()
2018-01-28 22:42:27.5746 [1] Stopped miners.
2018-01-28 22:42:38.5971 [1] Stopping miners...
2018-01-28 22:42:38.6335 [1] Stopped miners.
2018-01-28 22:42:38.6480 [1] Launching NeoScrypt miners for DEVFEE...
2018-01-28 22:42:39.9507 [20] Miner thread for Device #2 started.
2018-01-28 22:42:39.9593 [20] Loaded Kernels\neoscrypt.cl for Device #2.
2018-01-28 22:42:41.4645 [1] Stopping miners...
2018-01-28 22:42:44.8357 [20] Built neoscrypt program for Device #2.
2018-01-28 22:42:44.8357 [20] Build options: -O5 -legacy -IKernels -DWORKSIZE=256
2018-01-28 22:42:44.8793 [1] Stopped miners.
2018-01-28 22:42:54.9307 [1] Stopping miners...
2018-01-28 22:42:54.9845 [1] Stopped miners.
2018-01-28 22:42:54.9965 [1] Launching NeoScrypt miners for DEVFEE...
2018-01-28 22:42:56.4951 [20] Miner thread for Device #2 started.
2018-01-28 22:42:56.5026 [20] Loaded Kernels\neoscrypt.cl for Device #2.
2018-01-28 22:42:57.4622 [1] Stopping miners...
2018-01-28 22:43:01.4841 [20] Built neoscrypt program for Device #2.
2018-01-28 22:43:01.4841 [20] Build options: -O5 -legacy -IKernels -DWORKSIZE=256
2018-01-28 22:43:01.5132 [1] Stopped miners.
2018-01-28 22:43:11.5371 [1] Stopping miners...
2018-01-28 22:43:11.5715 [1] Stopped miners.
2018-01-28 22:43:11.5890 [1] Launching NeoScrypt miners for DEVFEE...
2018-01-28 22:43:12.9924 [20] Miner thread for Device #2 started.
2018-01-28 22:43:12.9989 [20] Loaded Kernels\neoscrypt.cl for Device #2.
2018-01-28 22:43:13.4632 [1] Stopping miners...
2018-01-28 22:43:17.9529 [20] Built neoscrypt program for Device #2.
2018-01-28 22:43:17.9529 [20] Build options: -O5 -legacy -IKernels -DWORKSIZE=256
2018-01-28 22:43:17.9956 [1] Stopped miners.
2018-01-28 22:43:28.0267 [1] Stopping miners...
2018-01-28 22:43:28.0796 [1] Stopped miners.
2018-01-28 22:43:28.0915 [1] Launching NeoScrypt miners for DEVFEE...
2018-01-28 22:43:29.4734 [20] Miner thread for Device #2 started.
2018-01-28 22:43:29.4810 [20] Loaded Kernels\neoscrypt.cl for Device #2.
2018-01-28 22:43:30.2591 [1] Switching back from the DEVFEE mode...
2018-01-28 22:43:34.7274 [20] Built neoscrypt program for Device #2.
2018-01-28 22:43:34.7274 [20] Build options: -O5 -legacy -IKernels -DWORKSIZE=256
2018-01-28 22:43:44.7576 [20] Device #2: 900,42 Kh/s (NeoScrypt)
2018-01-28 22:43:54.7726 [20] Device #2: 899,57 Kh/s (NeoScrypt)
2018-01-28 22:44:04.7843 [20] Device #2: 893,86 Kh/s (NeoScrypt)
2018-01-28 22:44:14.7944 [20] Device #2: 901,79 Kh/s (NeoScrypt)
2018-01-28 22:44:24.8033 [20] Device #2: 899,74 Kh/s (NeoScrypt)
2018-01-28 22:44:34.8135 [20] Device #2: 897,45 Kh/s (NeoScrypt)
2018-01-28 22:44:44.8322 [20] Device #2: 898,16 Kh/s (NeoScrypt)
2018-01-28 22:44:54.8449 [20] Device #2: 891,87 Kh/s (NeoScrypt)
2018-01-28 22:45:04.8669 [20] Device #2: 898,01 Kh/s (NeoScrypt)
2018-01-28 22:45:14.8865 [20] Device #2: 899,04 Kh/s (NeoScrypt)
2018-01-28 22:45:24.9138 [20] Device #2: 896,59 Kh/s (NeoScrypt)
2018-01-28 22:45:34.9307 [20] Device #2: 901,32 Kh/s (NeoScrypt)
2018-01-28 22:45:44.9454 [20] Device #2: 897,82 Kh/s (NeoScrypt)
2018-01-28 22:45:54.9626 [20] Device #2: 902,09 Kh/s (NeoScrypt)
2018-01-28 22:46:04.9801 [20] Device #2: 898,35 Kh/s (NeoScrypt)
2018-01-28 22:46:14.9901 [20] Device #2: 901,94 Kh/s (NeoScrypt)
2018-01-28 22:46:25.0078 [20] Device #2: 900,17 Kh/s (NeoScrypt)


Never comes to send any share.


Another problem in 1.2.9 aplha, at 4k resolution, not able to see Local Work size dua GUI design error .
http://tinypic.com/r/w003zo/9

Can you tell the overclock/memory/voltage settings. Which rom did you use and what is your memory thype of RX580s. I can't get more that 770h/s for my RX580s and they draw 115W each.

Ok, I'm understand. But how much CPU usage will be after 24 hours? Or more? If it allways growing than it may by very high. Maybe you add an option to schedule restart of GGS after selected period of time?

7-8% isn't bad at all given the fact that GGS is written in C#.
GGS'es CPU usage wouldn't be the same as other miners written in C++.
It is all about making a trade-off between efficiency and shorter development turnaround times.

What about CPU usage? After 5-7 hours it allready 7-8% - average from jumps 0 to 15%.

Great,  thank you.. will test as soon at release...

No, I don't speak Italian. lol I am of Japanese descent. I lived in the United States for a long time, though.

As for the DEVFEE, I need at least 10 times as much as I am making right now for my wife and I to live off of it.
My forte is actually kernel optimizations, and I haven't been able to do much of that with GGS up until now.
The miner is definitely getting more stable, so hopefully I get to optimizations sooner than later to increase a cash flow.

The stability issue with yiimp-based pools and NeoScrypt is fixed, I think.
I will push a new release shortly.

I don't think those "jumps" are problematic. I could use raw intensity instead of intensity to get rid of them, but that is not urgent.
As for Pircairn, the problem is really with the driver and there is not much I could do about it at this point...

This issue appears not allways. It can happens or not...

Yup, i got that too.. from 900 h/s suddenly jump to 1000 h/s even 1,200 h/s for 1 second and drop to 600 h/s

And again meet bug with jumping speed. After 1,3 hour speed on all GPU's start to jump fron 0 to 1000 h/s and then stabilize on normal speed.
And it's need to fix overclock system hangs on Pitcairn. So as binary kernel slow down on 4Gb Pitcairn.

P.S. Yes, CPU usage is growing with time. At start it's only 0-1-2%. It's good. But after 1 hour it's allready 0-4%. And jumps continues to grow.

So chaotic is our Cryptoland... I will implement a workaround.

Confirm this problem on some pools,  zpool and mine,   in both tested   Neoscrypt,  after some time workers disappear at pools, but app still works and load GPU 100%

Yeah, i think adding a user preferred/auto restart function would be cool too.
Btw, can add a function to change font size too?
Sorry, old man can't see very well .