Pages:
Author

Topic: [XMR] JCE Miner Cryptonight/forks, now with GPU! - page 57. (Read 90814 times)

newbie
Activity: 296
Merit: 0
Nice miner,

Have support yet for apus?

I got a ryzen 2200 with vega 8 Smiley

Thanks
member
Activity: 350
Merit: 22
Also, an overclocking capability will open sea of troubles, problems, compatibility issues and questions for you.

Thanks for the arguments, that's the point.
Also, JCE claims in the doc that it changes nothing on your computer, except the Huge page config (for cpu) and the log if enabled. I don't want to add GPU config that can harm your card if badly used.

It will add more problems and support requests than it adds value to the miner. And why not use overdriveNTool? it's just a perfect tool with GUI and command line mode. I use it on my rigs. I start overdriveNTool then immediately JCE and all is good. Also the new param --bus-order removes the complex matching of overdriveNTool order versus legacy JCE order.

Impressive job of diagnostic about the hashrate!
I'll rewrite my .bat to handle that correctly. But you could have checked the wallet as it gets logged as understood when the miner starts.

Quote
"0:17:46",
A test during 17 minutes is far too low. The difference starts huge (sometimes it's 2x lower or 2x bigger) then it converge.
Also:
Quote
400015
Very high, even if I know Nicehash has such a fixed diff. Mathematically, the higher is the pool diff, the higher is the probabilistic variation. Because you have less shares, so the https://en.wikipedia.org/wiki/Law_of_large_numbers takes more time to apply. That's very, very normal.
Quote
"shares": 29,
this is mathematically hard to be considered a Large Number. Mine 1000 shares and you'll see the effective hashrate will match. Or mine on a pool with a low Diff.

Quote
JCE miner is absolutely faster ,just tried it out last night and the results is huge compare to xmrig that I was using before,thanks
thanks man. JCE is arguably the best on CPU, thanks to its hundreds on internal assemblies for each possible CPU combination, and its exclusive features. I'm the only dev so far to use asm for CPU.
On GPU it depends on the card, but JCE has the good property of sending to your pool the same hashrate as displayed, unlike some other miners I tested before, Claymore 10+ being the wors, while 9.7 was the best  Undecided
newbie
Activity: 26
Merit: 1
Finally get it working!!!!!! Cheesy

Before I was modifying the start.bat in the wallet section putting address.diff
apparently the WALLET = blabla doesn't take the .diff...
So running the thingy from command line it worked perfectly or adding the .diff
at the end of the bat file like %WALLET%.diff!

Thanks guys!
newbie
Activity: 54
Merit: 0
Ive just tried with 25000 and it works fine as well.
At this point, I have no clue about what can be wrong in your setup.
newbie
Activity: 26
Merit: 1
Okay man! Thanks for your feedback. I've tried 25k and 50k without no luck...it goes up to 350k super fast...
I'll try with higher values of diff to see if something changes!

The usual minimum value is about 5000.
25000 isn't a low difficulty so it should have been accepted by the pool.

But didn't you write it like this: wallet.25K instead of wallet.25000 ?
Only numbers (with numerical characters) are accepted.

I'm getting shares accepted by the pool, problem is that diff keeps increasing.
And of course I write 25000, i put here 25k cause i'm lazy :p

I'll try right now! Cheesy
newbie
Activity: 54
Merit: 0
Okay man! Thanks for your feedback. I've tried 25k and 50k without no luck...it goes up to 350k super fast...
I'll try with higher values of diff to see if something changes!

The usual minimum value is about 5000.
25000 isn't a low difficulty so it should have been accepted by the pool.

But didn't you write it like this: wallet.25K instead of wallet.25000 ?
Only numbers (with numerical characters) are accepted.
newbie
Activity: 26
Merit: 1
newbie
Activity: 54
Merit: 0
I know! This pool does support it.
I'm here:
https://bxc.mine2gether.com

I'm giving it a try. I've only mined for ~15 minutes so far with a CPU and one GPU but the fixed diff works fine to me.

Perhaps the diff you typed was ignored because the value is too low.

My actual successful test is with 60000 with this command line:

jce_cn_gpu_miner64.exe -c configCN_6C.txt -o ctl.pool.mine2gether.com:4444 -u cxorzSavc8QRe6ryCXZ7dZPzjEQG7VFRkEuTdzYc9HLFZ11J8G7MzRMgjDvKppu13s2vPcECs2fRu5o RJ8GVjPrG12xBUyVY7.60000 -p PC40 --variation 3

And here is the result's log:
Code:
Analyzing Processors topology...
AMD Ryzen 7 1700 Eight-Core Processor
Assembly codename: ryzen
  SSE2          : Yes
  SSE3          : Yes
  SSE4          : Yes
  AES           : Yes
  AVX           : Yes
  AVX2          : Yes
Found CPU 0, with:
  L1 Cache:    32 KB, shared with CPU 1
  L2 Cache:   512 KB, shared with CPU 1
  L3 Cache:  8192 KB, shared with CPU 1, 2, 3, 4, 5, 6, 7
Found CPU 1, with:
  L1 Cache:    32 KB, shared with CPU 0
  L2 Cache:   512 KB, shared with CPU 0
  L3 Cache:  8192 KB, shared with CPU 0, 2, 3, 4, 5, 6, 7
Found CPU 2, with:
  L1 Cache:    32 KB, shared with CPU 3
  L2 Cache:   512 KB, shared with CPU 3
  L3 Cache:  8192 KB, shared with CPU 0, 1, 3, 4, 5, 6, 7
Found CPU 3, with:
  L1 Cache:    32 KB, shared with CPU 2
  L2 Cache:   512 KB, shared with CPU 2
  L3 Cache:  8192 KB, shared with CPU 0, 1, 2, 4, 5, 6, 7
Found CPU 4, with:
  L1 Cache:    32 KB, shared with CPU 5
  L2 Cache:   512 KB, shared with CPU 5
  L3 Cache:  8192 KB, shared with CPU 0, 1, 2, 3, 5, 6, 7
Found CPU 5, with:
  L1 Cache:    32 KB, shared with CPU 4
  L2 Cache:   512 KB, shared with CPU 4
  L3 Cache:  8192 KB, shared with CPU 0, 1, 2, 3, 4, 6, 7
Found CPU 6, with:
  L1 Cache:    32 KB, shared with CPU 7
  L2 Cache:   512 KB, shared with CPU 7
  L3 Cache:  8192 KB, shared with CPU 0, 1, 2, 3, 4, 5, 7
Found CPU 7, with:
  L1 Cache:    32 KB, shared with CPU 6
  L2 Cache:   512 KB, shared with CPU 6
  L3 Cache:  8192 KB, shared with CPU 0, 1, 2, 3, 4, 5, 6
Found CPU 8, with:
  L1 Cache:    32 KB, shared with CPU 9
  L2 Cache:   512 KB, shared with CPU 9
  L3 Cache:  8192 KB, shared with CPU 9, 10, 11, 12, 13, 14, 15
Found CPU 9, with:
  L1 Cache:    32 KB, shared with CPU 8
  L2 Cache:   512 KB, shared with CPU 8
  L3 Cache:  8192 KB, shared with CPU 8, 10, 11, 12, 13, 14, 15
Found CPU 10, with:
  L1 Cache:    32 KB, shared with CPU 11
  L2 Cache:   512 KB, shared with CPU 11
  L3 Cache:  8192 KB, shared with CPU 8, 9, 11, 12, 13, 14, 15
Found CPU 11, with:
  L1 Cache:    32 KB, shared with CPU 10
  L2 Cache:   512 KB, shared with CPU 10
  L3 Cache:  8192 KB, shared with CPU 8, 9, 10, 12, 13, 14, 15
Found CPU 12, with:
  L1 Cache:    32 KB, shared with CPU 13
  L2 Cache:   512 KB, shared with CPU 13
  L3 Cache:  8192 KB, shared with CPU 8, 9, 10, 11, 13, 14, 15
Found CPU 13, with:
  L1 Cache:    32 KB, shared with CPU 12
  L2 Cache:   512 KB, shared with CPU 12
  L3 Cache:  8192 KB, shared with CPU 8, 9, 10, 11, 12, 14, 15
Found CPU 14, with:
  L1 Cache:    32 KB, shared with CPU 15
  L2 Cache:   512 KB, shared with CPU 15
  L3 Cache:  8192 KB, shared with CPU 8, 9, 10, 11, 12, 13, 15
Found CPU 15, with:
  L1 Cache:    32 KB, shared with CPU 14
  L2 Cache:   512 KB, shared with CPU 14
  L3 Cache:  8192 KB, shared with CPU 8, 9, 10, 11, 12, 13, 14
Detecting OpenCL-capable GPUs...
Found GPU 0, with:
  Vendor:                         AMD
  Processor:                Ellesmere
  Device:                       0b:00
  Compute-Units:                   32
  Cache Memory:                 16 KB
  Local Memory:                 32 KB
  Global Memory:              8192 MB
  Addressing:                 64-bits
Found GPU 1, with:
  Vendor:                         AMD
  Processor:                Ellesmere
  Device:                       09:00
  Compute-Units:                   32
  Cache Memory:                 16 KB
  Local Memory:                 32 KB
  Global Memory:              4096 MB
  Addressing:                 64-bits
Found GPU 2, with:
  Vendor:                         AMD
  Processor:                Ellesmere
  Device:                       0a:00
  Compute-Units:                   36
  Cache Memory:                 16 KB
  Local Memory:                 32 KB
  Global Memory:              4096 MB
  Addressing:                 64-bits
HTTP Local Server on port 15263

Preparing 10 Mining Threads...

+-- Thread 0 config -----------------------+
| Run on CPU:             1                |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 1 config -----------------------+
| Run on CPU:             3                |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 2 config -----------------------+
| Run on CPU:             5                |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 3 config -----------------------+
| Run on CPU:             7                |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 4 config -----------------------+
| Run on CPU:             9                |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 5 config -----------------------+
| Run on CPU:             11               |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 6 config -----------------------+
| Run on CPU:             13               |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 7 config -----------------------+
| Run on CPU:             15               |
| Use cache:              yes              |
| Multi-hash:             no               |
| Assembly module:        ryzen            |
+------------------------------------------+

+-- Thread 8 config -----------------------+
| Run on GPU:                      1       |
| Multi-hash:                    976       |
| Worksize:                        8       |
| Factor Alpha                    64       |
| Factor Beta                      8       |
+------------------------------------------+

+-- Thread 9 config -----------------------+
| Run on GPU:                      1       |
| Multi-hash:                    976       |
| Worksize:                        8       |
| Factor Alpha                    64       |
| Factor Beta                      8       |
+------------------------------------------+

Cryptonight Variation: Cryptonight V7 fork of April-2018

Low intensity.

Starting CPU Mining thread 0, affinity: CPU 1
Thread 0 successfully bound to CPU 1
Allocated shared Large Page at: 0000017baa600000
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 1 of NUMA node 0 at: 0000017baa800000

Starting CPU Mining thread 1, affinity: CPU 3
Thread 1 successfully bound to CPU 3
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 3 of NUMA node 0 at: 0000017baaa00000

Starting CPU Mining thread 2, affinity: CPU 5
Thread 2 successfully bound to CPU 5
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 5 of NUMA node 0 at: 0000017baac00000

Starting CPU Mining thread 3, affinity: CPU 7
Thread 3 successfully bound to CPU 7
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 7 of NUMA node 0 at: 0000017baae00000

Starting CPU Mining thread 4, affinity: CPU 9
Thread 4 successfully bound to CPU 9
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 9 of NUMA node 0 at: 0000017bab000000

Starting CPU Mining thread 5, affinity: CPU 11
Thread 5 successfully bound to CPU 11
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 11 of NUMA node 0 at: 0000017bab200000

Starting CPU Mining thread 6, affinity: CPU 13
Thread 6 successfully bound to CPU 13
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 13 of NUMA node 0 at: 0000017bab400000

Starting CPU Mining thread 7, affinity: CPU 15
Thread 7 successfully bound to CPU 15
Allocated 2MB Cached Large Page Scratchpad Buffer for CPU 15 of NUMA node 0 at: 0000017bab600000

Starting GPU Mining thread 8, on GPU 1
Created OpenCL Context for GPU 1 at 0000017baa037710
Created OpenCL Thread 8 Command-Queue for GPU 1 at 0000017baa05ce70
Scratchpad Allocation success for OpenCL Thread 8
Allocating big 1952MB scratchpad for OpenCL Thread 8...
Compiling kernels of OpenCL Thread 8...
Kernels of OpenCL Thread 8 compiled.

Starting GPU Mining thread 9, on GPU 1
Created OpenCL Thread 9 Command-Queue for GPU 1 at 0000017baa05c930
Scratchpad Allocation success for OpenCL Thread 9
Allocating big 1952MB scratchpad for OpenCL Thread 9...
Compiling kernels of OpenCL Thread 9...
Kernels of OpenCL Thread 9 compiled.
Connecting to mining pool ctl.pool.mine2gether.com:4444 ...
12:34:33 | Bitcedi (BXC) Mining session starts!

12:34:33 | GPU Compute allocation starts at 80% and reaches 100% after ~5min,
12:34:33 | during this time, the hashrate may be unstable and inconsistent.
12:34:33 | Let the miner warm-up if you're tuning for performance.
12:34:33 | Connected to pool. Now logging in...
12:34:33 | Successfuly logged as cxorzSavc8QRe6ryCXZ7dZPzjEQG7VFRkEuTdzYc9HLFZ11J8G7MzRMgjDvKppu13s2vPcECs2fRu5oRJ8GVjPrG12xBUyVY7.60000
12:34:33 | Pool changes Difficulty to 60000.
12:35:04 | Hashrate CPU Thread 0: 75.35 h/s
12:35:04 | Hashrate CPU Thread 1: 74.32 h/s
12:35:04 | Hashrate CPU Thread 2: 51.89 h/s
12:35:04 | Hashrate CPU Thread 3: 75.86 h/s
12:35:04 | Hashrate CPU Thread 4: 83.29 h/s
12:35:04 | Hashrate CPU Thread 5: 83.12 h/s
12:35:04 | Hashrate CPU Thread 6: 83.49 h/s
12:35:04 | Hashrate CPU Thread 7: 83.35 h/s - Total CPUs: 610.62 h/s
12:35:04 | Hashrate GPU Thread 8: 520.64 h/s
12:35:04 | Hashrate GPU Thread 9: 515.19 h/s - Total GPU 1: 1035.82 h/s
12:35:04 | Total: 1646.44 h/s - Max: 1646.44 h/s
12:35:08 | GPU 1 Thread 8 Lane 939 finds a Share, value 60000
12:35:08 | Accepted by the pool in 75 ms.
12:35:12 | GPU 1: Temp: 48C - Fan:  0% -- Shares: Good: 1 Bad: 0
12:35:13 | GPU 1 Thread 9 Lane 303 finds a Share, value 60000
12:35:13 | Accepted by the pool in 71 ms.
12:35:20 | CPU Thread 6 finds a Share, value 60000
12:35:20 | Accepted by the pool in 77 ms.
12:35:22 | GPU 1 Thread 8 Lane 354 finds a Share, value 60000
12:35:22 | Accepted by the pool in 74 ms.
12:35:36 | Pool sends a new Job.
12:36:09 | GPU 1 Thread 8 Lane 329 finds a Share, value 60000
12:36:09 | Accepted by the pool in 75 ms.
12:36:14 | CPU Thread 2 finds a Share, value 60000
12:36:14 | Accepted by the pool in 76 ms.
12:36:48 | GPU 1 Thread 8 Lane 809 finds a Share, value 60000
12:36:48 | Accepted by the pool in 82 ms.
12:36:57 | CPU Thread 7 finds a Share, value 60000
12:36:57 | Accepted by the pool in 98 ms.
12:37:10 | GPU 1 Thread 9 Lane 518 finds a Share, value 60000
12:37:10 | Accepted by the pool in 72 ms.
12:38:28 | GPU 1 Thread 9 Lane 186 finds a Share, value 60000
12:38:28 | Accepted by the pool in 84 ms.
12:38:31 | CPU Thread 1 finds a Share, value 60000
12:38:31 | Accepted by the pool in 77 ms.
12:38:36 | GPU 1 Thread 9 Lane 796 finds a Share, value 60000
12:38:36 | Accepted by the pool in 81 ms.
12:38:37 | Hashrate CPU Thread 0: 81.75 h/s
12:38:37 | Hashrate CPU Thread 1: 81.95 h/s
12:38:37 | Hashrate CPU Thread 2: 81.93 h/s
12:38:37 | Hashrate CPU Thread 3: 81.85 h/s
12:38:37 | Hashrate CPU Thread 4: 81.24 h/s
12:38:37 | Hashrate CPU Thread 5: 82.56 h/s
12:38:37 | Hashrate CPU Thread 6: 82.34 h/s
12:38:37 | Hashrate CPU Thread 7: 82.82 h/s - Total CPUs: 656.40 h/s
12:38:37 | Hashrate GPU Thread 8: 518.72 h/s
12:38:37 | Hashrate GPU Thread 9: 518.31 h/s - Total GPU 1: 1037.02 h/s
12:38:37 | Total: 1693.42 h/s - Max: 1693.42 h/s
12:38:58 | GPU 1: Temp: 52C - Fan: 19% -- Shares: Good: 8 Bad: 0
12:39:24 | Pool sends a new Job.
12:40:48 | GPU 1 Thread 8 Lane 647 finds a Share, value 60000
12:40:49 | Accepted by the pool in 79 ms.
12:41:11 | GPU 1 Thread 8 Lane 793 finds a Share, value 60000
12:41:11 | Accepted by the pool in 38 ms.
12:41:36 | GPU 1 Thread 8 Lane 590 finds a Share, value 60000
12:41:36 | Accepted by the pool in 39 ms.
12:42:10 | Hashrate CPU Thread 0: 75.99 h/s
12:42:10 | Hashrate CPU Thread 1: 76.42 h/s
12:42:10 | Hashrate CPU Thread 2: 77.06 h/s
12:42:10 | Hashrate CPU Thread 3: 76.55 h/s
12:42:10 | Hashrate CPU Thread 4: 69.60 h/s
12:42:10 | Hashrate CPU Thread 5: 72.40 h/s
12:42:10 | Hashrate CPU Thread 6: 72.01 h/s
12:42:10 | Hashrate CPU Thread 7: 72.88 h/s - Total CPUs: 592.87 h/s
12:42:10 | Hashrate GPU Thread 8: 518.54 h/s
12:42:10 | Hashrate GPU Thread 9: 518.63 h/s - Total GPU 1: 1037.16 h/s
12:42:10 | Total: 1630.03 h/s - Max: 1693.42 h/s
12:42:44 | GPU 1: Temp: 53C - Fan: 17% -- Shares: Good: 11 Bad: 0
12:43:32 | Pool ctl.pool.mine2gether.com:4444
12:43:32 | Reconnections 0
12:43:32 | Currency Bitcedi (BXC)
12:43:32 | Current pool Difficulty 60000
12:43:32 | Accepted Shares 15
12:43:32 | Total Hashes 900000
12:43:32 | Miner uptime 0:08:59
12:43:32 | Effective net hashrate 1669.76 h/s
12:43:32 | Devices results - Shares Accepted/Ignored/Rejected - Net Hashrate
12:43:32 | * CPUs  -  4/0/0 -  445.27 h/s
12:43:32 | * GPU 1 - 11/0/0 - 1224.49 h/s
12:43:41 | GPU 1 Thread 8 Lane 468 finds a Share, value 60000
12:43:41 | Accepted by the pool in 38 ms.
12:44:55 | Pool sends a new Job.
12:45:33 | CPU Thread 7 finds a Share, value 60000
12:45:33 | Accepted by the pool in 64 ms.
12:45:43 | Hashrate CPU Thread 0: 78.59 h/s
12:45:43 | Hashrate CPU Thread 1: 79.03 h/s
12:45:43 | Hashrate CPU Thread 2: 79.24 h/s
12:45:43 | Hashrate CPU Thread 3: 78.80 h/s
12:45:43 | Hashrate CPU Thread 4: 73.84 h/s
12:45:43 | Hashrate CPU Thread 5: 76.10 h/s
12:45:43 | Hashrate CPU Thread 6: 75.65 h/s
12:45:43 | Hashrate CPU Thread 7: 76.49 h/s - Total CPUs: 617.70 h/s
12:45:43 | Hashrate GPU Thread 8: 517.90 h/s
12:45:43 | Hashrate GPU Thread 9: 516.00 h/s - Total GPU 1: 1033.89 h/s
12:45:43 | Total: 1651.59 h/s - Max: 1693.42 h/s
12:46:02 | GPU 1 Thread 8 Lane 834 finds a Share, value 60000
12:46:02 | Accepted by the pool in 72 ms.
12:46:05 | GPU 1 Thread 8 Lane 394 finds a Share, value 60000
12:46:05 | Accepted by the pool in 39 ms.
12:46:19 | GPU 1 Thread 8 Lane 21 finds a Share, value 60000
12:46:19 | Accepted by the pool in 39 ms.
12:46:30 | GPU 1: Temp: 54C - Fan: 17% -- Shares: Good: 15 Bad: 0
12:46:39 | CPU Thread 6 finds a Share, value 60000
12:46:40 | Accepted by the pool in 87 ms.
12:46:55 | GPU 1 Thread 8 Lane 603 finds a Share, value 60000
12:46:55 | Accepted by the pool in 41 ms.
12:47:52 | Pool sends a new Job.
12:48:06 | CPU Thread 3 finds a Share, value 60000
12:48:06 | Accepted by the pool in 39 ms.
12:48:07 | GPU 1 Thread 8 Lane 878 finds a Share, value 60000
12:48:07 | Accepted by the pool in 38 ms.
12:48:16 | CPU Thread 7 finds a Share, value 60000
12:48:16 | Accepted by the pool in 39 ms.
12:48:54 | CPU Thread 4 finds a Share, value 60000
12:48:54 | Accepted by the pool in 39 ms.
12:48:58 | GPU 1 Thread 8 Lane 297 finds a Share, value 60000
12:48:58 | Accepted by the pool in 39 ms.
12:49:16 | Hashrate CPU Thread 0: 75.71 h/s
12:49:16 | Hashrate CPU Thread 1: 76.27 h/s
12:49:16 | Hashrate CPU Thread 2: 76.48 h/s
12:49:16 | Hashrate CPU Thread 3: 71.70 h/s
12:49:16 | Hashrate CPU Thread 4: 76.07 h/s
12:49:16 | Hashrate CPU Thread 5: 77.42 h/s
12:49:16 | Hashrate CPU Thread 6: 77.42 h/s
12:49:16 | Hashrate CPU Thread 7: 77.89 h/s - Total CPUs: 608.91 h/s
12:49:16 | Hashrate GPU Thread 8: 518.54 h/s
12:49:16 | Hashrate GPU Thread 9: 518.54 h/s - Total GPU 1: 1037.07 h/s
12:49:16 | Total: 1645.98 h/s - Max: 1693.42 h/s
12:50:16 | GPU 1: Temp: 53C - Fan: 17% -- Shares: Good: 18 Bad: 0
12:50:43 | Pool sends a new Job.
12:51:03 | GPU 1 Thread 8 Lane 493 finds a Share, value 60000
12:51:03 | Accepted by the pool in 38 ms.
12:51:24 | Pool sends a new Job.
member
Activity: 190
Merit: 59
Did you try another pool, before anything else - I am asking this because for all my rigs effective hashrate on long term is about the same as nominal hashrate (that is the primary reason why I went away from srb and cast miners, as the effective hashrate was 10-20% less than miner shown hashrate)
newbie
Activity: 26
Merit: 1
...fixed diff still not working for me...I really have NO clue why... Sad

Not every pools handle the fixed difficulty feature, and each one that handles it does it in its own way, usually with wallet.diff or wallet+diff in the wallet field.

Which pool are you using?

I know! This pool does support it.
I'm here:
https://bxc.mine2gether.com
newbie
Activity: 54
Merit: 0
...fixed diff still not working for me...I really have NO clue why... Sad

Not every pools handle the fixed difficulty feature, and each one that handles it does it in its own way, usually with wallet.diff or wallet+diff in the wallet field.

Which pool are you using?
newbie
Activity: 26
Merit: 1
I just got the last version 32k2 and it's working super stable (as happened for me with 31f)
BUT fixed diff still not working for me...I really have NO clue why... Sad
Anyone having this issue???
Cheers!
full member
Activity: 1120
Merit: 131
Mined XTL (Stellite) non stop for 30-35hs, the hashrate at the pool was exactly what it's supposed to be, slightly above 2.7KH/s for my setup.
Wish I can find a cheap Vega 64 to play with.
sr. member
Activity: 1484
Merit: 253
I didn't support dicision about not implementing OC/V/Fan controls... If user didn't know about settings to OC/V/Fan controls he didn't use them. I think, that controls like in Claymore is very usefull. Didn't need to use any 3rd tools. Just set for each algo needed settings in bat or cfg file. Mining is dangerous itself, and even without this settings miner make stress to GPU's. Users that didn't know that shoudn't mines at all...
So adding this controls may be very-very usefull for people, who didn't use overdriveNTool like me...
member
Activity: 361
Merit: 16
love that miner...hoped for MOX that i can switch from SRB - but...will wait

other question

i mine TRTL with GPU @8KHs and with Ryzen @2KHs
BUT why the Masari Fork with GPU @7KHs and Ryzen only @800KHs - shouldnt it be a bit more with Ryzen? (i use auto mode)
member
Activity: 784
Merit: 21
JCE miner is absolutely faster ,just tried it out last night and the results is huge compare to xmrig that I was using before,thanks
member
Activity: 190
Merit: 59
I support your decision to keep the overclocking tool out of the miner. This has to be done externally by specialized tool, if one doesn't know how to setup his rig he should not really mine at all. Also, an overclocking capability will open sea of troubles, problems, compatibility issues and questions for you. I don't believe you are paid well enough to handle that.

In my opinion, just give us stable and efficient miner and we will do the rest.  Grin

I never heard of MOX. You should share this information in MOX topic (if it exists) and on reddit so people can avoid shitcoins and scamcoins. Just look at the front page of this forum, there are at least 3 scam miners promoting themselves.  Sad
member
Activity: 350
Merit: 22
* Watchdog has to be inside the miner, and it's safe, so I add it
* Fan/temp could be monitored from outside, but it's safe, so I add it
* Voltage/fan control can be controller from outside with excellent tools (like overdriveNTool) and is dangerous.

If I add this feature, which would be very easy since I already have code for the fan/temp, it could make JCE responsible (yet indirectly) from buring a GPU, and so i'll have to add tons of disclaimers and warnings Use at own risks, i'm nor responsible for ....

So the answer is an explicit No.

I usually follow Claymore choices about miner code, but not this time.

A completely different topic: MOX

I'm done implementing that fork, and I did it by reading the reference code of mox-rig, the official miner, a fork of xmrig. So far so good.
But after having read the code, i'm wondering whether the dev is very very bad, or very malicious. I give details:

* The fork is not done normally with github, but is a copy-paste of the source. So there's no history and no way to compare with original xmrig, at least not with the normal github tool.
* the first commit is already MOXed
* The doc says MOX is CN-v7 with a light change. This is wrong.
* The code says this is like XTL, which is wrong too.
* The old unit tests are present, but disabled in a very strange fashion. The mox unit test does not exists.
* The miner can be configured to mine good old CN-v7 or Arto, but will produce only bad shares in such case.

All in all, i advice not to use MOX, or at least not to mine with the reference miners. The only other one, waiting for my next release, is SRB, which is both 3rd party and officially listed by the mox team. So i say nothing about it. Maybe the mox dev is fair and just ridiculously bad, and there's no threat here, but all those strange things in the same project looks suspicious to me.

I'll still provide my own implementation, and it will be the fastest on CPU, on par with TRTL.
sr. member
Activity: 1484
Merit: 253
JCE did you plan to add overclock/voltage/fan control to GPU miner?
member
Activity: 350
Merit: 22
Hi all, sorry I had very little time those days

Linux: no ETA so far, i don't even consider the Windows gpu version mature yet. but the cpu version is.

Watchdog: for now i prefer delegate the limit setting to the user, since my warmup may fail and mine at 80% even for more than 5mn

for 1G card start with a very very low single-thread config like multi_hash 64 then increase by step of 16. i tested jce fine on hd 6000 and 7000, not 8000 but it should work if well configured, but will probably mine slowly.
Pages:
Jump to: