[XMR] JCE Miner Cryptonight/forks, now with GPU! - page 55.

meihk

newbie

Activity: 12

Merit: 0

Quote from: JCE-Miner on September 15, 2018, 08:01:20 AM

Other miners Heavy implementation may be just under your GPU limit and won't crash, we all use a different implementation, I got some reports of opposite behavior from users previously (JCE being more stable than others) and some like you, it really depends on your rig.

I agree about hard-copding the tweaks in the bios, i did the same for most of my cards. So far i've no more ideas, maybe the precise combinations of your bios + your drivers + your PSU + jce + CN-Heavy is just not compatible. Sad

Quote

when miner reconnected to the pool the hashrate drops to 7200 h/s and never get back to 8300 h/s unless restart the miner.. how do i solve this issue?

Easy way: use the 0.32m i'm about to release with the watchdog, and this way, with a safe limit of ~8200, the miner will close by itself in case it happens. I said close so if you want to restart, put a loop in your .bat script

Code:

:mine
jce_cn_.....
goto :mine

Otherwise i need to restart the warmup in case of long disconnect, and that's code i must do my side.
Note that the 0.32m will start a bit faster than the previous versions, I improved the init sequence.

edit:
0.32m GPU online

Quote

better watchdog
better support of 64+ cpu
CN-Red
Lots of new shitcoins
--no-warmup

I used 0.032m with watchdog enable and make a loop, but when watchdog triggered it says "Press any key to continue . . ." so it is not quitting automatically. tia

JCE-Miner

member

Activity: 350

Merit: 22

It's so rare i do a blind-fix that works...

Can you provide the log at startup, so i see what the autoconfig did? I provided it its own personality Grin

Now i need to make a 0.32n GPU so the CPU subset of GPU version is on par with pure CPU version. As stated in the doc.

The usual blue tag

0.32n CPU is online

Quote

Backported shitcoins from GPU version (Remix, Catalyst, Parsicoin...)
Correct 64+ cpu support on win64 version, now minimum Win64 is Win 7. Vista64 is no longer supported. Vista32 still supported.

wyzdic

newbie

Activity: 46

Merit: 0

Quote from: JCE-Miner on September 15, 2018, 11:06:51 AM

Because Tube use a lot a compute power, while others (CN-v7, Turtle, Mox...) use rather the memory bandwidth.
So, in case of Tube, having more CU is useful, while with older algos, it helps little.

CN-Heavy/Haven is intermediate.

I've, of course, no 64+ cpu system, but thanks to your logs, i read your windows has placed 48 of your CPU on Node 0 (which work well) and 48 other on node 1, which is not accessible by current JCE. I'm fixing it and hope i could then beat xmrstak, but it will be a blind code, with no way to test, except for regressions.

I'm building 0.32n CPU right now.

32n is very very very good, Use auto config, 64 thread. 3735h/s now! Amazing!

Xmr-stak only have 3200-3300h/s.

Code:

    "thread_all": [57.46, 57.46, 64.65, 64.67, 56.60, 56.60, 64.92, 64.67, 55.75, 55.77, 63.98, 64.00, 55.75, 55.75, 64.30, 64.33, 50.35, 50.35, 61.20, 61.22, 50.43, 50.43, 61.20, 61.13, 50.45, 50.35, 61.20, 60.88, 49.96, 50.19, 60.88, 61.13, 56.04, 56.26, 64.33, 64.33, 57.19, 57.18, 64.00, 64.33, 56.88, 57.19, 64.65, 64.33, 57.18, 57.18, 64.33, 64.33, 51.45, 51.20, 61.13, 62.06, 52.07, 51.77, 61.54, 61.74, 51.19, 51.19, 62.06, 61.74, 51.45, 51.49, 61.54, 61.44],
    "total": 3732.57,
    "max": 3735.01

JCE-Miner

member

Activity: 350

Merit: 22

Because Tube use a lot a compute power, while others (CN-v7, Turtle, Mox...) use rather the memory bandwidth.
So, in case of Tube, having more CU is useful, while with older algos, it helps little.

CN-Heavy/Haven is intermediate.

I've, of course, no 64+ cpu system, but thanks to your logs, i read your windows has placed 48 of your CPU on Node 0 (which work well) and 48 other on node 1, which is not accessible by current JCE. I'm fixing it and hope i could then beat xmrstak, but it will be a blind code, with no way to test, except for regressions.

I'm building 0.32n CPU right now.

wyzdic

newbie

Activity: 46

Merit: 0

Quote from: JCE-Miner on September 15, 2018, 10:17:53 AM

i should have called the 0.32m GPU n, because its CPU part is not the same as the pure 0.32m CPU

I'll rebuild a 0.32n CPU for you, with features of the GPU version (the new coins and the CPU 64+)
JCE GPU needs opencl, even without GPU. That's why I provide a special CPU-only version.

Thanks for you help. And i have another question.
Mined bittube, My Vega64 rig have 1600h/s, Vega56 rigs only have 1420h/s. Why are there such big differences?
Driver 18.6.1;
Win10 LTSB;
1408/1100(Vega64 and Vega56 flash to Vega64);
1407/950(Vega56);
All gpus have samsung memory.

JCE-Miner

member

Activity: 350

Merit: 22

i should have called the 0.32m GPU n, because its CPU part is not the same as the pure 0.32m CPU

I'll rebuild a 0.32n CPU for you, with features of the GPU version (the new coins and the CPU 64+)
JCE GPU needs opencl, even without GPU. That's why I provide a special CPU-only version.

wyzdic

newbie

Activity: 46

Merit: 0

The GPU version can't runing with only CPU rig?
I tried use GPU 0.32m with EPYC server, miner report cannot load library opencl.dll. With no-gpu.

wyzdic

newbie

Activity: 46

Merit: 0

Quote from: JCE-Miner on September 15, 2018, 08:01:20 AM

Other miners Heavy implementation may be just under your GPU limit and won't crash, we all use a different implementation, I got some reports of opposite behavior from users previously (JCE being more stable than others) and some like you, it really depends on your rig.

I agree about hard-copding the tweaks in the bios, i did the same for most of my cards. So far i've no more ideas, maybe the precise combinations of your bios + your drivers + your PSU + jce + CN-Heavy is just not compatible. Sad

Quote

when miner reconnected to the pool the hashrate drops to 7200 h/s and never get back to 8300 h/s unless restart the miner.. how do i solve this issue?

Easy way: use the 0.32m i'm about to release with the watchdog, and this way, with a safe limit of ~8200, the miner will close by itself in case it happens. I said close so if you want to restart, put a loop in your .bat script

Code:

:mine
jce_cn_.....
goto :mine

Otherwise i need to restart the warmup in case of long disconnect, and that's code i must do my side.
Note that the 0.32m will start a bit faster than the previous versions, I improved the init sequence.

edit:
0.32m GPU online

Quote

better watchdog
better support of 64+ cpu
CN-Red
Lots of new shitcoins
--no-warmup

0.32m CPU

Code:

Starting CPU Mining thread 0, affinity: CPU 0
Thread 0 successfully bound to CPU 0
Allocated shared Large Page at: 0000017145600000
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 0 of NUMA node 0 at: 0000017145800000

Starting CPU Mining thread 1, affinity: CPU 0
Thread 1 successfully bound to CPU 0
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 0 of NUMA node 0 at: 0000017145a00000

Starting CPU Mining thread 2, affinity: CPU 1
Thread 2 successfully bound to CPU 1
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 1 of NUMA node 0 at: 0000017145c00000

Starting CPU Mining thread 3, affinity: CPU 2
Thread 3 successfully bound to CPU 2
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 2 of NUMA node 0 at: 0000017145e00000

Starting CPU Mining thread 4, affinity: CPU 2
Thread 4 successfully bound to CPU 2
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 2 of NUMA node 0 at: 0000017146000000

Starting CPU Mining thread 5, affinity: CPU 3
Thread 5 successfully bound to CPU 3
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 3 of NUMA node 0 at: 0000017146200000

Starting CPU Mining thread 6, affinity: CPU 4
Thread 6 successfully bound to CPU 4
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 4 of NUMA node 0 at: 0000017146400000

Starting CPU Mining thread 7, affinity: CPU 4
Thread 7 successfully bound to CPU 4
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 4 of NUMA node 0 at: 0000017146600000

Starting CPU Mining thread 8, affinity: CPU 6
Thread 8 successfully bound to CPU 6
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 6 of NUMA node 0 at: 0000017146800000

Starting CPU Mining thread 9, affinity: CPU 6
Thread 9 successfully bound to CPU 6
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 6 of NUMA node 0 at: 0000017146a00000

Starting CPU Mining thread 10, affinity: CPU 7
Thread 10 successfully bound to CPU 7
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 7 of NUMA node 0 at: 0000017146c00000

Starting CPU Mining thread 11, affinity: CPU 8
Thread 11 successfully bound to CPU 8
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 8 of NUMA node 0 at: 0000017146e00000

Starting CPU Mining thread 12, affinity: CPU 8
Thread 12 successfully bound to CPU 8
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 8 of NUMA node 0 at: 0000017147000000

Starting CPU Mining thread 13, affinity: CPU 9
Thread 13 successfully bound to CPU 9
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 9 of NUMA node 0 at: 0000017147200000

Starting CPU Mining thread 14, affinity: CPU 10
Thread 14 successfully bound to CPU 10
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 10 of NUMA node 0 at: 0000017147400000

Starting CPU Mining thread 15, affinity: CPU 10
Thread 15 successfully bound to CPU 10
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 10 of NUMA node 0 at: 0000017147600000

Starting CPU Mining thread 16, affinity: CPU 12
Thread 16 successfully bound to CPU 12
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 12 of NUMA node 1 at: 0000017147800000

Starting CPU Mining thread 17, affinity: CPU 12
Thread 17 successfully bound to CPU 12
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 12 of NUMA node 1 at: 0000017147a00000

Starting CPU Mining thread 18, affinity: CPU 13
Thread 18 successfully bound to CPU 13
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 13 of NUMA node 1 at: 0000017147c00000

Starting CPU Mining thread 19, affinity: CPU 14
Thread 19 successfully bound to CPU 14
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 14 of NUMA node 1 at: 0000017147e00000

Starting CPU Mining thread 20, affinity: CPU 14
Thread 20 successfully bound to CPU 14
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 14 of NUMA node 1 at: 0000017148000000

Starting CPU Mining thread 21, affinity: CPU 15
Thread 21 successfully bound to CPU 15
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 15 of NUMA node 1 at: 0000017148200000

Starting CPU Mining thread 22, affinity: CPU 16
Thread 22 successfully bound to CPU 16
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 16 of NUMA node 1 at: 0000017148400000

Starting CPU Mining thread 23, affinity: CPU 16
Thread 23 successfully bound to CPU 16
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 16 of NUMA node 1 at: 0000017148600000

Starting CPU Mining thread 24, affinity: CPU 18
Thread 24 successfully bound to CPU 18
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 18 of NUMA node 1 at: 0000017148800000

Starting CPU Mining thread 25, affinity: CPU 18
Thread 25 successfully bound to CPU 18
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 18 of NUMA node 1 at: 0000017148a00000

Starting CPU Mining thread 26, affinity: CPU 19
Thread 26 successfully bound to CPU 19
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 19 of NUMA node 1 at: 0000017148c00000

Starting CPU Mining thread 27, affinity: CPU 20
Thread 27 successfully bound to CPU 20
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 20 of NUMA node 1 at: 0000017148e00000

Starting CPU Mining thread 28, affinity: CPU 20
Thread 28 successfully bound to CPU 20
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 20 of NUMA node 1 at: 0000017149000000

Starting CPU Mining thread 29, affinity: CPU 21
Thread 29 successfully bound to CPU 21
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 21 of NUMA node 1 at: 0000017149200000

Starting CPU Mining thread 30, affinity: CPU 22
Thread 30 successfully bound to CPU 22
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 22 of NUMA node 1 at: 0000017149400000

Starting CPU Mining thread 31, affinity: CPU 22
Thread 31 successfully bound to CPU 22
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 22 of NUMA node 1 at: 0000017149600000

Starting CPU Mining thread 32, affinity: CPU 24
Thread 32 successfully bound to CPU 24
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 24 of NUMA node 2 at: 0000017149800000

Starting CPU Mining thread 33, affinity: CPU 24
Thread 33 successfully bound to CPU 24
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 24 of NUMA node 2 at: 0000017149a00000

Starting CPU Mining thread 34, affinity: CPU 25
Thread 34 successfully bound to CPU 25
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 25 of NUMA node 2 at: 0000017149c00000

Starting CPU Mining thread 35, affinity: CPU 26
Thread 35 successfully bound to CPU 26
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 26 of NUMA node 2 at: 0000017149e00000

Starting CPU Mining thread 36, affinity: CPU 26
Thread 36 successfully bound to CPU 26
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 26 of NUMA node 2 at: 000001714a000000

Starting CPU Mining thread 37, affinity: CPU 27
Thread 37 successfully bound to CPU 27
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 27 of NUMA node 2 at: 000001714a200000

Starting CPU Mining thread 38, affinity: CPU 28
Thread 38 successfully bound to CPU 28
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 28 of NUMA node 2 at: 000001714a400000

Starting CPU Mining thread 39, affinity: CPU 28
Thread 39 successfully bound to CPU 28
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 28 of NUMA node 2 at: 000001714a600000

Starting CPU Mining thread 40, affinity: CPU 30
Thread 40 successfully bound to CPU 30
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 30 of NUMA node 2 at: 000001714a800000

Starting CPU Mining thread 41, affinity: CPU 30
Thread 41 successfully bound to CPU 30
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 30 of NUMA node 2 at: 000001714aa00000

Starting CPU Mining thread 42, affinity: CPU 31
Thread 42 successfully bound to CPU 31
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 31 of NUMA node 2 at: 000001714ac00000

Starting CPU Mining thread 43, affinity: CPU 32
Thread 43 successfully bound to CPU 32
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 32 of NUMA node 2 at: 000001714ae00000

Starting CPU Mining thread 44, affinity: CPU 34
Thread 44 successfully bound to CPU 34
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 34 of NUMA node 2 at: 000001714b000000

Starting CPU Mining thread 45, affinity: CPU 36
Thread 45 successfully bound to CPU 36
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 36 of NUMA node 3 at: 000001714b200000

Starting CPU Mining thread 46, affinity: CPU 37
Thread 46 successfully bound to CPU 37
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 37 of NUMA node 3 at: 000001714b400000

Starting CPU Mining thread 47, affinity: CPU 38
Thread 47 successfully bound to CPU 38
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 38 of NUMA node 3 at: 000001714b600000

Starting CPU Mining thread 48, affinity: CPU 40
Thread 48 successfully bound to CPU 40
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 40 of NUMA node 3 at: 000001714b800000

Starting CPU Mining thread 49, affinity: CPU 42
Thread 49 successfully bound to CPU 42
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 42 of NUMA node 3 at: 000001714ba00000

Starting CPU Mining thread 50, affinity: CPU 43
Thread 50 successfully bound to CPU 43
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 43 of NUMA node 3 at: 000001714bc00000

Starting CPU Mining thread 51, affinity: CPU 44
Thread 51 successfully bound to CPU 44
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 44 of NUMA node 3 at: 000001714be00000

Starting CPU Mining thread 52, affinity: CPU 46
Thread 52 successfully bound to CPU 46
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 46 of NUMA node 3 at: 000001714c000000

Starting CPU Mining thread 53, affinity: CPU 48
GetNumaProcessorNode failed for cpu 48, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714c200000

Starting CPU Mining thread 54, affinity: CPU 49
GetNumaProcessorNode failed for cpu 49, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714c400000

Starting CPU Mining thread 55, affinity: CPU 50
GetNumaProcessorNode failed for cpu 50, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714c600000

Starting CPU Mining thread 56, affinity: CPU 52
GetNumaProcessorNode failed for cpu 52, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714c800000

Starting CPU Mining thread 57, affinity: CPU 54
GetNumaProcessorNode failed for cpu 54, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714ca00000

Starting CPU Mining thread 58, affinity: CPU 55
GetNumaProcessorNode failed for cpu 55, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714cc00000

Starting CPU Mining thread 59, affinity: CPU 56
GetNumaProcessorNode failed for cpu 56, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714ce00000

Starting CPU Mining thread 60, affinity: CPU 58
GetNumaProcessorNode failed for cpu 58, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714d000000

Starting CPU Mining thread 61, affinity: CPU 60
GetNumaProcessorNode failed for cpu 60, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714d200000

Starting CPU Mining thread 62, affinity: CPU 61
GetNumaProcessorNode failed for cpu 61, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714d400000

Starting CPU Mining thread 63, affinity: CPU 62
GetNumaProcessorNode failed for cpu 62, error code: 87
Retrying with no NUMA
Allocated 2MB Cached Large Page Scratchpad Buffer at: 000001714d600000

This is my stak config

Quote

"cpu_threads_conf" :
[
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 0 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 2 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 4 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 6 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 8 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 10 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 12 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 14 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 16 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 18 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 20 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 22 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 24 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 26 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 28 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 30 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 32 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 34 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 36 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 38 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 40 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 42 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 44 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 46 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 48 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 50 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 52 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 54 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 56 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 58 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 60 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 62 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 64 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 66 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 68 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 70 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 72 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 74 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 76 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 78 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 80 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 82 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 84 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 86 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 88 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 90 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 92 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 94 },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
{ "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : false },
],

Stak tell me "WARNING: Windows supports only affinity up to 63", but it work. 3200-3300h/s

Quote

[2018-09-15 22:57:00] : WARNING: Windows supports only affinity up to 63.
[2018-09-15 22:57:00] : WARNING setting affinity failed.
[2018-09-15 22:57:00] : Starting 1x thread, affinity: 92.
[2018-09-15 22:57:00] : hwloc: memory pinned
[2018-09-15 22:57:00] : WARNING: Windows supports only affinity up to 63.
[2018-09-15 22:57:00] : WARNING setting affinity failed.
[2018-09-15 22:57:00] : Starting 1x thread, affinity: 94.
[2018-09-15 22:57:00] : hwloc: memory pinned
[2018-09-15 22:57:00] : WARNING: Windows supports only affinity up to 63.
[2018-09-15 22:57:00] : WARNING setting affinity failed.

JCE-Miner

member

Activity: 350

Merit: 22

Other miners Heavy implementation may be just under your GPU limit and won't crash, we all use a different implementation, I got some reports of opposite behavior from users previously (JCE being more stable than others) and some like you, it really depends on your rig.

I agree about hard-copding the tweaks in the bios, i did the same for most of my cards. So far i've no more ideas, maybe the precise combinations of your bios + your drivers + your PSU + jce + CN-Heavy is just not compatible. Sad

Quote

when miner reconnected to the pool the hashrate drops to 7200 h/s and never get back to 8300 h/s unless restart the miner.. how do i solve this issue?

Easy way: use the 0.32m i'm about to release with the watchdog, and this way, with a safe limit of ~8200, the miner will close by itself in case it happens. I said close so if you want to restart, put a loop in your .bat script

Code:

:mine
jce_cn_.....
goto :mine

Otherwise i need to restart the warmup in case of long disconnect, and that's code i must do my side.
Note that the 0.32m will start a bit faster than the previous versions, I improved the init sequence.

edit:
0.32m GPU online

Quote

better watchdog
better support of 64+ cpu
CN-Red
Lots of new shitcoins
--no-warmup

Almody

member

Activity: 259

Merit: 34

Quote from: JCE-Miner on September 15, 2018, 02:04:11 AM

This is not about the config being difficult, but because each algo pull more or less power from the cards, Heavy pulling a bit more than v7, and if you optimized your card for V7 to the absolute optimum power/speed ratio, it may not work for Heavy. I advise to configure your cards to be stable in all cases about power and memory straps. Then it's easy to change the overclock with OverdriveNTTool depending on what you mine.

Power consumption in increasing order:
Cryptonight classic
Cryptonight V7 & al. (Masari...)
Cryptolight & al. (Turtle, Mox...)
Heavy
Bittube

All my gpu's settings are modded in bios and working fine with all other miners and algos ,stable and without errors and stale shares even with jce miner but not with heavy algo! btw i'm using msi afterberner!

meihk

newbie

Activity: 12

Merit: 0

hey JCE,

I have 8pcs rx 580 8gb with the total hash of 8300 h/s "Auto config", i noticed that when miner reconnected to the pool the hashrate drops to 7200 h/s and never get back to 8300 h/s unless restart the miner.. how do i solve this issue?

thank a lot for this apps, really good hash / card.

https://imgur.com/urCTyEU
https://imgur.com/fygr824

JCE-Miner

member

Activity: 350

Merit: 22

This is not about the config being difficult, but because each algo pull more or less power from the cards, Heavy pulling a bit more than v7, and if you optimized your card for V7 to the absolute optimum power/speed ratio, it may not work for Heavy. I advise to configure your cards to be stable in all cases about power and memory straps. Then it's easy to change the overclock with OverdriveNTTool depending on what you mine.

Power consumption in increasing order:
Cryptonight classic
Cryptonight V7 & al. (Masari...)
Cryptolight & al. (Turtle, Mox...)
Heavy
Bittube

Almody

member

Activity: 259

Merit: 34

Quote from: JCE-Miner on September 14, 2018, 12:44:59 PM

Quote

Hmmm, interesting, does RX 580 8Gb needs warmup? And what warmup does exactly?

It tries several timings and balances between threads to find the best, but doesn't always success. Hence why the hashrate vary a lot during this time.

Quote

Hey JCE, you completely ignore my question about rx 5xx series ! I still have a problem with heavy and Saber algos! Can you check my message above pls!

Sorry i skipped your post Sad

You already lowered the multi_hash, try to set worksize to 4 and alpha to 32 sometimes it helps.
Otherwise, reset all your GPU overclock and memory hacks to factory default to see if at least it's stable in such a normal context. If yes, so your GPU had been tuned to the limit of CN-v7 and cannot afford Heavy keeping the same tweaking.

This is why JCE mines the fees in the same algo as the user: because some ultra-fine tuned GPU can be at 100% on V7, would be at 105% on heavy and would crash. I experienced this on my Bonaire, ultra-stable on V7, crash after 5mn on heavy, with exact same overclock and memory straps.

i think i'll make an intermediate 0.32m GPU with some fixes but not all, and maybe the fix for the 96-cpu dual epyc.

Ok, i'll try to find a right way to overclock my gpu's for just a heavy algo but does it mean that every time need to change the algo (back to V7 for exmpl) i will need to change overcklocking settings again? And if so hope you'll make it easier in a new update!

JCE-Miner

member

Activity: 350

Merit: 22

Nope

JCE never changes anything on your system, except the Large Page config for CPU, and when it does, it says so. Nothing under the hood.
I'm talking about how the two threads on the same GPU are timed each against the other. Nothing to do with memory straps.

dingdongtobias

newbie

Activity: 156

Merit: 0

Quote from: JCE-Miner on September 14, 2018, 12:44:59 PM

It tries several timings and balances between threads to find the best, but doesn't always success. Hence why the hashrate vary a lot during this time.

it mean you change memory timings like zawawa's ggs miner ??

JCE-Miner

member

Activity: 350

Merit: 22

Quote

Hmmm, interesting, does RX 580 8Gb needs warmup? And what warmup does exactly?

It tries several timings and balances between threads to find the best, but doesn't always success. Hence why the hashrate vary a lot during this time.

Quote

Hey JCE, you completely ignore my question about rx 5xx series ! I still have a problem with heavy and Saber algos! Can you check my message above pls!

Sorry i skipped your post Sad

You already lowered the multi_hash, try to set worksize to 4 and alpha to 32 sometimes it helps.
Otherwise, reset all your GPU overclock and memory hacks to factory default to see if at least it's stable in such a normal context. If yes, so your GPU had been tuned to the limit of CN-v7 and cannot afford Heavy keeping the same tweaking.

This is why JCE mines the fees in the same algo as the user: because some ultra-fine tuned GPU can be at 100% on V7, would be at 105% on heavy and would crash. I experienced this on my Bonaire, ultra-stable on V7, crash after 5mn on heavy, with exact same overclock and memory straps.

i think i'll make an intermediate 0.32m GPU with some fixes but not all, and maybe the fix for the 96-cpu dual epyc.

Almody

member

Activity: 259

Merit: 34

Quote from: JCE-Miner on September 13, 2018, 12:13:28 PM

Sorry i missed a question about Linux previously:
* The CPU version is almost on par with the Windows one, and I update it regularly (even if less frequently than the Win version)
* The GPU version just doesn't exists. It's not simple as a recompile, it's closely related to the AMD Adrenalin (and older). Hence why Cast and SRB also don't provide a Linux version. But Claymore did, sure.

Epyc: so let's do the opposite, what config does Stak use on your big 96-cpu server? Then i will be able to translate it into JCE format.

Speed: yeah, thanks. This is a major point of my miner from the very beginning, JCE numbers (except for two display bugs now fixed thanks to UnclWish and 1rv1n) are fair and consistant with the pool hashrate, something that never happened for me with Claymore 10/11 and some other.

Monero v8: it's in the 0.32m+ but disabled, waiting for fixes and tests. I did them at same time as Mox. The new algo is heavier than v7, but less that CN-Heavy. I'm not even sure the Monero team won't change some specs before they switch, in about 6 weeks.

I plan to release a 0.32n GPU soon, with Mox, the timestamped and size-limited log file (good idea!), --no-warmup param, and a better watchdog.

Hey JCE, you completely ignore my question about rx 5xx series ! I still have a problem with heavy and Saber algos! Can you check my message above pls!

GKumaran

member

Activity: 204

Merit: 10

Quote from: lebuawu2 on September 13, 2018, 09:21:59 AM

Quote from: GKumaran on September 13, 2018, 08:37:52 AM

Hey JCE !!

Can confirm that JCE hr is marginally better than SRBMiner hr. And this isnt the best part, the internal check of effective hash rate is a god send. The effective hash rate of JCE is very much in line with the pool hr and beats most other miners. Thx mate, its a win for transparency.

Will be better if the API can provide 5m,30m,60m avgerage actual hr.

Btw my vega 64 reference cards rock 2120~ hash/sec per card with following config:

Code:

{ "mode" : "GPU", "worksize" : 16, "alpha" : 128, "beta" : 16, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash":1968 },
{ "mode" : "GPU", "worksize" : 16, "alpha" : 128, "beta" : 16, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash":1968 }

Cheers mate.

what is your cclock and mclock?

Core @ 1472 and mem @ 1100

UnclWish

sr. member

Activity: 1484

Merit: 253

Hmmm, interesting, does RX 580 8Gb needs warmup? And what warmup does exactly?

JCE-Miner

member

Activity: 350

Merit: 22

Sorry i missed a question about Linux previously:
* The CPU version is almost on par with the Windows one, and I update it regularly (even if less frequently than the Win version)
* The GPU version just doesn't exists. It's not simple as a recompile, it's closely related to the AMD Adrenalin (and older). Hence why Cast and SRB also don't provide a Linux version. But Claymore did, sure.

Epyc: so let's do the opposite, what config does Stak use on your big 96-cpu server? Then i will be able to translate it into JCE format.

Speed: yeah, thanks. This is a major point of my miner from the very beginning, JCE numbers (except for two display bugs now fixed thanks to UnclWish and 1rv1n) are fair and consistant with the pool hashrate, something that never happened for me with Claymore 10/11 and some other.

Monero v8: it's in the 0.32m+ but disabled, waiting for fixes and tests. I did them at same time as Mox. The new algo is heavier than v7, but less that CN-Heavy. I'm not even sure the Monero team won't change some specs before they switch, in about 6 weeks.

I plan to release a 0.32n GPU soon, with Mox, the timestamped and size-limited log file (good idea!), --no-warmup param, and a better watchdog.

Topic: [XMR] JCE Miner Cryptonight/forks, now with GPU! - page 55. (Read 90841 times)