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

[sergey777]

JCE-Miner, UnclWish, HardKano cпacибo зa coвeты, пoxoдy пpичинa в биocax, бyдy пpoбoвaть дpyгиe.

[HardKano]

I hope it's a thanks haha I don't understand right now but will use translator soon hahahaha :-)

[UnclWish]

I hope it's a thanks haha I don't understand right now but will use translator soon hahahaha :-)

Yes, he thanks to us. Agreed that problem in bios, will find other.

[HardKano]

Bios of the RX580 ? What is your precise model with manufacturer part number ? Send me your original bios file. Will look if i have the right straps for you. I tuned my gigabyte and Samsung and did job on anorak.tech for others.

[stsocrat]

May be it's time to add better config.txt lines for each type of card at the 1st post? It's the only reason many people can't use jce.
Or you seek to improve --auto?

[cryptoprofitswitcher]

I'm using 1x Vega 56 + 1x Vega 64 and with using 0.33b6 I got 7740 H/s on mining Conceal (Cryptonight Fast)

Since version 0.33b7 until the latest 0.33b10 I only get 5733 H/s on the same configuration.

Drivers: Adrenaline 18.6.1, Windows 10 x64

[JCE-Miner]

Right, i focused so much on Heavy for RX on the latter releases that i got regressions on old cards and old algos, i'm restoring them in next release. speed for old cards (hd7000) is already restored in b10

Thanks for your detailed report!

[djmuler]

Hey

testing your miner it throws this error on start on my 6x RX580 rig

Code:

Starting GPU Thread 4, on GPU 0
Created OpenCL Context for GPU 0 at 000002bbc42e5030
Created OpenCL Thread 4 Command-Queue for GPU 0 at 000002bbc4615e30
Scratchpad Allocation success for OpenCL Thread 4
Allocating big 3328MB scratchpad for OpenCL Thread 4...
Couldn't create OpenCL Thread 4 Scratchpad on GPU 0
Error: CL_INVALID_BUFFER_SIZE

Starting GPU Thread 5, on GPU 0
Created OpenCL Thread 5 Command-Queue for GPU 0 at 000002bbc4615e90
Scratchpad Allocation success for OpenCL Thread 5
Allocating big 3328MB scratchpad for OpenCL Thread 5...
Couldn't create OpenCL Thread 5 Scratchpad on GPU 0
Error: CL_INVALID_BUFFER_SIZE

so GPU 0 is not working...
why?

GPU 0 is working fine with other miners

[JCE-Miner]

It's not about your GPU0 not working, but

Allocating big 3328MB scratchpad for OpenCL Thread 4...
Couldn't create OpenCL Thread 4 Scratchpad on GPU 0
Error: CL_INVALID_BUFFER_SIZE

OpenCL refused to allocate such a big buffer. Are all your GPU RX 8GB cards? The gpu order may be different in jce than in other miner, look at the gpu list displayed in green to see what GPU has what index. If you have a 4G card, it may be the GPU0.

Also, very possible is that your GPU0 is bound to your internal IGP, APU or motherboard GPU. JCE include Intel and AMD IGP/APU into its list, as GPU0. If that's the case, just skip it and number your GPU in the config starting from GPU1.

That choice i made to include Intel IGP in the list was a bad choice, sad.

Online is the 0.33h CPU Windows

Little optim for Zen
FredCoin defaults to CN-Light v7

Online is the 0.33b11 GPU Windows

Restored more speed regressed for old algos/old cards
Far better pool-side efficiency

About the last point, you may experience a peak hashrate on Heavy/forks a bit lower than b9/b10, but the pool-side effective will be a lot better, near 99%, fees deduced, which is the maximum. Same for v8.
This is a point where SRB improved recently while I regressed. I re-benched our both last versions and found we're on par on big RX heavy, pool-side. my b10 and before were slightly below.

The Vega on old algos (non-heavy) should be restored to b6 speed, but i didn't had time to test all combinations of cards and algos.

[alivanich]

I have stable config on b6 version for my rig (mix 550, 560, 470)
"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 1, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 1, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 2, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 2, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 3, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 3, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 4, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 4, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 5, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 5, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 6, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 6, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 7, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 7, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 8, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 8, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 9, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 9, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 10, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 10, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 11, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 11, "multi_hash":944 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 12, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 12, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 13, "multi_hash":432 },
     { "mode" : "GPU", "worksize" : 16, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta":4, "index" : 13, "multi_hash":432 },
],

but b10 crashing with this config.

What do I need to change to adapt to the new version of the program?

[HardKano]

What is the crash message or the symptoms ?

[JCE-Miner]

13 cards, woow...

What coin/algo do you mine?
Does it crash at startup or after some time?
If at startup, at what step?

[HardKano]

13 cards, woow...

What coin/algo do you mine?
Does it crash at startup or after some time?
If at startup, at what step?

Also run --probe to look if there GPUs id are still assign to the right ones for your multihash. If it changed you might try to put a scratch pad from a 8gb card into a 4gb one.

JCE, got 12 :-) mining heavy right now. Always looking for newer projects that seem promising in the future (1-2-3 years)

[alivanich]

13 cards, woow...

What coin/algo do you mine?
Does it crash at startup or after some time?
If at startup, at what step?

most of my rigs have 13amd or 11amd+8nvidia
v8 monero
crash after compiled all kernels
crash = windows blue screen and restart

[HardKano]

Look over Windows updates ? There is a big one this month. After the big reboot, 1-2 small. And run driver easy to update everything. Most of my crashes was don solved with that

[UnclWish]

And look on pagefile size. You need it big size!

Testing heavy algo on b11 now. Speed became the same as in SRB miner... But eff.hashrate impressive for now. Almost 5 hours of test - eff.hashrate is almost equal to speed... Now waiting more time to test...

[Iamtutut]

Mining bittube with the lastest version, I have a weird message when I launch the miner. I just started, and twice in a row this is what I get (I restarted the miner):
19:23:55 | Pool changes Difficulty to 133724.
19:23:55 | GPU 3 Thread 7 Lane -1 finds a Share, value 133724
19:23:55 | Stale Share that may be refused by the pool.
19:23:55 | Rejected by the pool in 33 ms.
19:23:55 | Message from the pool: Invalid job id
19:23:55 | Your first share was rejected, you probably need to change the fork
19:23:55 | with parameter --variation N, try different values for N in [1-18]
19:23:55 | until you find the one that works.
19:23:55 | Pool sends a new Job.
19:24:13 | GPU 2 Thread 5 Lane 325 finds a Share, value 133724
19:24:13 | Accepted by the pool in 90 ms.
19:24:24 | GPU 1 Thread 2 Lane 19 finds a Share, value 133724
19:24:24 | Accepted by the pool in 36 ms.

And all the next shares seem accepted.

[HardKano]

Do you use fork=0 ? It's the auto detect algo setting. If not, do it. If yes, I don't know LoL

[JCE-Miner]

GPU Lane -1 ??
woow, big bug, but i tested a lot and didn't encountered such, i probably missed one specific corner case.
i'll re-release with a hotfix.

otherwise, are the pool-side perf better?


ok found the bug, due to my new sync, the very first hash is an empty one, with value 0 and id 0. the diff being the inverse of the value, it's +inf, sent to the pool and rejected. not bad, just cosmetical, but very ugly.

[JCE-Miner]

And look on pagefile size. You need it big size!

Testing heavy algo on b11 now. Speed became the same as in SRB miner... But eff.hashrate impressive for now. Almost 5 hours of test - eff.hashrate is almost equal to speed... Now waiting more time to test...

that's exactly what i expected: optimal effective, on par with srb. i don't say i'm better, i benched to be at +/- 0.2%, so i claim to be on par.