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

[UnclWish]

it's an idea, but as a workaround. I really want to have the hashrate just stable, with no tricks involved.
Once i'm done with the new CPU version i go back to this problem as top priority.

Hey JCE

I figured out a manual fix to get those rx 580 cards to hash at max speeds on Heavy and be stable, any time i see one that has lost the max speed, i pause/unpause 4 times each (8 button pushes) in quick succession and the card comes back to maximum speed.  May be something like this can be coded into the miner?

Test it on RX 580 8G - miner launched 5+ hours. 1st Gpu runs at full speed, 2nd speed drops. Tried your method on 2nd GPU - speed didn't changes.
EDIT: Yes, sometimes it works... But not everytime...

[1714605608]

1714605608

[1714605608]

1714605608

[JCE-Miner]

I've a new idea to workaround this, and using the hybrid as a side-effect is a failure, hence why I stayed on the experimental version until I succedded or gave up. Need a new cycle of trial-and-error.

Online is the 0.33g CPU Windows (linux to come next)

CN-FreeHaven fork
MoneroOcean feature "algo_perf" available
Updated doc

[UnclWish]

I've a new idea to workaround this, and using the hybrid as a side-effect is a failure, hence why I stayed on the experimental version until I succedded or gave up. Need a new cycle of trial-and-error.

Online is the 0.33g CPU Windows (linux to come next)

CN-FreeHaven fork
MoneroOcean feature "algo_perf" available
Updated doc

Good! And think please about lowering power consumption on v8 algo...

[UselessGuru]

I am trying to include your miner into www.multipoolminer.io.

There is an issue with the CPU version.

I start the miner by command line and retrieve the process id of the miner (so I can later kill the process to switch to another miner):

Code:

jce_cn_cpu_miner64.exe --no-gpu --auto --variation 15 -o stratum+tcp://cryptonightv8.eu.nicehash.com:3367 -u *********************************.multipoolminer -p x --nicehash --mport 4003 --no-warmup --low

Now here's where the problem starts.
JCEMiner for some reason spawns another process 'Attrib.exe' with a different process id which is doing the actual mining. Why this fake name?

How can I start your miner so that the miner stops when I kill the parent process?

[JCE-Miner]

this is the normal behavior of JCE from day one, take a look at the doc i explain it.
the child process should suicide when the parent one is killed, but i admit i didn't test this at every release, there may be a regression, i'll re-check.

this is the same for the gpu version.

edit: while it still works with normal run, the --elevate and --elevate-and-quit modes no longer suicide the child process, you're right there's a regression out there, i'll fix it asap, it's quite easy and reproducible.

[OfficialS_Face]

I don't know if you guys can help me but I currently have a problem with my rig where I can't get the second GPU to get past 590 h/s

Rig is as follows:
2x 290
1x 390
1x470
FX 6300
8GB DDR3
Currently using :jce_cn_gpu_miner.033b3
Tested:jce_cn_gpu_miner.033b4 (same perf)

drivers: 18.9.3 (have tried 18.5.1 but it made no difference)


I have tried different intensity's and switching GPU's around but that's either made the hashrate worse or no change. Below is my current config

Code:

"cpu_threads_conf" : 
[ 
     { "cpu_architecture" : "bulldozer", "affine_to_cpu" : 0, "use_cache" : true },
     { "cpu_architecture" : "bulldozer", "affine_to_cpu" : 0, "use_cache" : true },
     { "cpu_architecture" : "bulldozer", "affine_to_cpu" : 0, "use_cache" : true },
     { "cpu_architecture" : "bulldozer", "affine_to_cpu" : 0, "use_cache" : true },
],

"gpu_threads_conf" : 
[ 
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 0, "multi_hash":800 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 0, "multi_hash":800 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 1, "multi_hash":448 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 1, "multi_hash":448 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 2, "multi_hash":560 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 2, "multi_hash":560 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 3, "multi_hash":800 },
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4,"index" : 3, "multi_hash":800 },

],

Hashing speed:

R9 390= 782-714 h/s
R9 290= 556-540 h/s
R9 290= 720-699 h/s
RX 470= 840-820 h/s

Total Speed:2930-3118 h/s

Mining: stellite

P.S are my RX 470 speeds okay ? it seems that it should be a bit faster

[sergneo]

I've a new idea to workaround this, and using the hybrid as a side-effect is a failure, hence why I stayed on the experimental version until I succedded or gave up. Need a new cycle of trial-and-error.

Online is the 0.33g CPU Windows (linux to come next)

CN-FreeHaven fork
MoneroOcean feature "algo_perf" available
Updated doc

there were critical errors in the jsonstak logs
here's the log after 2 minutes of work

   {"version":"jce/0.33g/cpu","hashrate":{"threads":[[21.0,21.0,21.0],[21.4,21.4,21.4]],"total":[42.4,42.4,42.4],"highest":42.9},"results":{"diff_current":500,"shares_good":6,"shares_total":6,"avg_time":257202032.0,"hashes_total":4689,"best":[106400,2201,2078,994,845,558,0,0,0,0],"error_log":[]},"connection":{"pool": "xmr.pool.minergate.com:45700","uptime":1543212192,"ping":79,"error_log":[]}}

[Boss_One]

I'm become crazy , i've 3 PC and ALL Crash, with 14.4 driver and 6950HD, i've allready post my config.
Could u help me ? i'm very glad to fee you my hash

I must use this 14.4 driver with your miner ?  i'm the only one who have problem with 6950.

You must have Windows 8.1+. What config you're using? What amount of video memory your cards have? And what algo you what to mine?
[/quote]

Yes i'm on W8.1
My 6950 got 2Gb ram
I'm mining V8 cryptonight the lastest fork ...

So, it's hash a little now, my config parameters sucks.

I'm testing on 33B6 and with this "jce_cn_gpu_miner64.exe   Experimental boost ~2% for HD6000/7000"

I've try a lot of parameters, but get not more than  100/113 h/s  

It's crash when i up "multi_hash" value more than 240


That's the best result i can get with this config


"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash":240},
    
    
],

if i try Worksize 16/32 i can't UP than 240 thread.

Did u have some values i can try ? because i've  try a lot of combination but can't reach more hashrate, driver crash.

i didn't understand whats that 2 values (alpha" : 64, "beta" :  8 ) ?

The most curious it was, if i didn't OC my card like @910/1250 driver crash instantanetly when i start miner.
in 800/1250 or 830/1250 driver crash. With all same values in config.txt . Wtf hell !!!  

I've try on each card on different PC (don't have a rig) 3 different PC.
Thx for helping.

[UnclWish]

Yes i'm on W8.1
My 6950 got 2Gb ram
I'm mining V8 cryptonight the lastest fork ...

So, it's hash a little now, my config parameters sucks.

I'm testing on 33B6 and with this "jce_cn_gpu_miner64.exe   Experimental boost ~2% for HD6000/7000"

I've try a lot of parameters, but get not more than  100/113 h/s  

It's crash when i up "multi_hash" value more than 240


That's the best result i can get with this config


"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash":240},
    
    
],

if i try Worksize 16/32 i can't UP than 240 thread.

Did u have some values i can try ? because i've  try a lot of combination but can't reach more hashrate, driver crash.

i didn't understand whats that 2 values (alpha" : 64, "beta" :  8 ) ?

The most curious it was, if i didn't OC my card like @910/1250 driver crash instantanetly when i start miner.
in 800/1250 or 830/1250 driver crash. With all same values in config.txt . Wtf hell !!!  

I've try on each card on different PC (don't have a rig) 3 different PC.
Thx for helping.

Try to use double threads this way:

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 224},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 224},     
],

also try this config:

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 464},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 464},     
],

or

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 448},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 448},     
],

or

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 432},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 432},     
],

[Boss_One]

Try to use double threads this way:

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 224},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 224},     
],

also try this config:

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 464},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 464},     
],

or

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 448},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 448},     
],

or

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 432},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 432},     
],

Double thread 224   => 113 h/s same as 1 Thread
Double thread 464 => PC Freez

Double thread 448 => Driver crash
Double thread 432 => Driver crash

What's wrong ?  (same result on a VTX HD6950 and Gigabyte HD6950)

i've already try 2 thread before asking on forum, same result as 1 thread ? Is it normal with this card ?

[UnclWish]

It's strange... Can you provide screenshots of 1st and 2nd tabs GPU-Z? And GPU Caps Viewer - OpenCL tab...
Looks like for OpenCL this card have less than 2Gb vmemory...

try this set:

"gpu_threads_conf" :
[
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 240},
     { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "index" : 0, "multi_hash": 240},    
],

wait about 10 minutes to look real speed. At start speed maybe low.

[JCE-Miner]

I rushed the 33g to provide the MoneroOcean new feature, and it shows...

so far three regressions/bugs
* Stak JSON timers
* Shadow sub-process suicide
* FreeHaven wallets defaults to... Monero algo

I'm rebuilding it and will re-release.

Looks like for OpenCL this card have less than 2Gb vmemory...

Did he set the env variables?

Code:

set GPU_MAX_HEAP_SIZE=100
set GPU_USE_SYNC_OBJECTS=1
set GPU_MAX_ALLOC_PERCENT=100
set GPU_SINGLE_ALLOC_PERCENT=100

[sergey777]

У мeня msi HD6950 2Gb, win 7, дpaйвepa 13.12, выдaeт нa xmrig-amd-2.8.6 250 h/s. Hacтpoйки: 1 пoтoк, worksize 64, multi_hash 832.
Cкoлькo дoлжнa дaвaть h/s нa мaйнepe JCE? Ecть ли cмыcл cтaвить Win 10?, yж oчeнь мнoгo нyжнoгo xлaмa и кoшeлькoв нa этoм pигe....

Boss_One пoпpoбyйтe дpaйвepa 13.12

[JCE-Miner]

Minimum is Windows 8.1 x64 for JCE GPU
For CPU, that's ok down to Vista32

[sergey777]

Дык пoэтoмy я и cпpaшивaю, ecть ли cмыcл cтaвить win 10 и иcпoльзoвaть Baш мaйнep? У мeня 250 h/s, нa Baшeм бoльшe?

[UnclWish]

Дык пoэтoмy я и cпpaшивaю, ecть ли cмыcл cтaвить win 10 и иcпoльзoвaть Baш мaйнep? У мeня 250 h/s, нa Baшeм бoльшe?

He знaю. Boзмoжнo и нeт. Taм вoн y чeлoвeкa бoльшe 113 нe пoлyчaeтcя нa 6950... Haдo выяcнить, cкoлькo JCE нa 6950 дoлжeн выдaвaть...

JCE, how much h/s your miner must give on HD6950 ? Did you test this card?

[fenomenyaa]

Could someone share proper settings and estimated h/s for 5704GB -580  8GB for heaven algo?  

[JCE-Miner]

I've one remaining HD6950 but not on a rig, on a cheap PC whose PSU cannot afford real mining so I tested it works, but not for speed.
I could unplug it and put it on a real rig to do real tests but not immediately, lack of time.

The minning speed provided come from a ticket in the Github support.

Ok I tested the miner now, best results I achieved is 270 h/s per core on 6990.
So that is 2x 270 h/s = ~ 540 h/s total for 6990 (which is 2 gpu cores of 6970s in it)
Settings used : 32 worksize and 864 intensity.

That's about HD6990 but that's a dual HD6970, the results on a HD6950 should be around 240.

[UnclWish]

That's about HD6990 but that's a dual HD6970, the results on a HD6950 should be around 240.

Thanks.
Method to restore maximum heavy algo speed after speed drop often works. Just quick pressing 8 times number of GPU on which speed droped can restore it! It's strange that quick pause\unpause GPU 4 times in row can do that...

[JCE-Miner]

I need to make it the normal way, not needing a manual tweak such this one. I knew my opencl was good, but the sync is glitchy, and the hybrid didn't help.

Online is the 0.33g CPU Windows re-released undercover with the said regression fixed.

* With normal mode, kill the parent, the child attrib suicides, and vice-versa.
* With --elevate, four processes are started, two jce and two attrib, one of each normal, one of each elevated. Kill any, the three other suicide.
* With --elevate-and-quit, the parent suicide at the beginning, hence the name, so the attrib remains in all cases, on purpose. Don't use this mode if you want to grab the parent process-id to kill JCE from an external tool.