JCE-GPU 0.30a out* No change on the CPU part, if you mine with no GPU, ignore
* ARQ coin added
* Should be more compatible with APU (but not tested)
* Light perf increase ~0.8% depending on the GPU
* Per-GPU hashrate
* Detection of Stale shares
* Fixed path problem, you can run JCE from any directory
* Added per-gpu stats in the JSON
* Fixed bad message in case of invalid configuration
Thanks. Now will test it... Speed is a little more... But "job not found" still happens. And looks like it happens so often as in previous version... And sometimes "invalid nonce. is miner not compatible with nicehash?" rejects share too, but not so often. Tested on heavy algo on nicehash pool. |
|
|
|
Try to lower "multi_hash":1200 in 16 decrements, so first try "multi_hash":1184
I rebooted and even with "multi_hash":944 which was stable it doesnt compile. i think i found the bug : the miner doesn't supported to be launched from a different dir. said differently, its current dir must be the path where the .exe is if you call the .bat or exe from a different dir, it fails. i'll fix this in next revision. the .bat is in te same dir, sadly. Also giving the absolute path to config instead of a relative doesnt change the behaviour. This is my current start-batchfile: setx GPU_FORCE_64BIT_PTR 1
setx GPU_MAX_HEAP_SIZE 100
setx GPU_MAX_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_MAX_SINGLE_ALLOC_PERCENT 100
W:
cd W:\Miner\JCE CN GPU Miner
jce_cn_gpu_miner64.exe -c "W:\Miner\JCE CN GPU Miner\confgpu.txt" -o loki.miner.rocks:5555 -u L6B6Dkz7VmLSMwEJgWeUUxHFpNeCShavBMcZ6Fid6xynbVCxD2brTsAasXviwYHKdz5i1ULS2rm2gGQqA759BdkEK4Mxzty.15000 -p w=windorf --no-cpu
PAUSE
If path is long-name you must set it in quotes: W: cd "W:\Miner\JCE CN GPU Miner" Strangely i dont need to. It changes directory fine. I think you only need to use quotes in the command line if your path contains a space, but in a batch file its fine. windows lol. Anyways: still compile error for the opencl kernel :/ Try to set algo manul by --variation option. Give you confgpu.txt to see there. What cards do you have? Drivers? One RX 580 8GB, Adrenalin 18.5.2, Windows 10 x64 Setting variation manually didnt change anything :/ "cpu_threads_conf" :
[
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 0, "use_cache" : true, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 2, "use_cache" : false, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 4, "use_cache" : true, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 6, "use_cache" : false, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 8, "use_cache" : true, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 10, "use_cache" : false, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 12, "use_cache" : true, },
{ "cpu_architecture" : "ryzen", "affine_to_cpu" : 14, "use_cache" : false, },
],
"gpu_threads_conf" :
[
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash":832 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash":832 },
]
From cmd.exe run jce_cn_gpu_miner64.exe --probe and place here results. |
|
|
|
Try to lower "multi_hash":1200 in 16 decrements, so first try "multi_hash":1184
I rebooted and even with "multi_hash":944 which was stable it doesnt compile. i think i found the bug : the miner doesn't supported to be launched from a different dir. said differently, its current dir must be the path where the .exe is if you call the .bat or exe from a different dir, it fails. i'll fix this in next revision. the .bat is in te same dir, sadly. Also giving the absolute path to config instead of a relative doesnt change the behaviour. This is my current start-batchfile: setx GPU_FORCE_64BIT_PTR 1
setx GPU_MAX_HEAP_SIZE 100
setx GPU_MAX_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_MAX_SINGLE_ALLOC_PERCENT 100
W:
cd W:\Miner\JCE CN GPU Miner
jce_cn_gpu_miner64.exe -c "W:\Miner\JCE CN GPU Miner\confgpu.txt" -o loki.miner.rocks:5555 -u L6B6Dkz7VmLSMwEJgWeUUxHFpNeCShavBMcZ6Fid6xynbVCxD2brTsAasXviwYHKdz5i1ULS2rm2gGQqA759BdkEK4Mxzty.15000 -p w=windorf --no-cpu
PAUSE
If path is long-name you must set it in quotes: W: cd "W:\Miner\JCE CN GPU Miner" Strangely i dont need to. It changes directory fine. I think you only need to use quotes in the command line if your path contains a space, but in a batch file its fine. windows lol. Anyways: still compile error for the opencl kernel :/ Try to set algo manul by --variation option. Give you confgpu.txt to see there. What cards do you have? Drivers? |
|
|
|
Try to lower "multi_hash":1200 in 16 decrements, so first try "multi_hash":1184
I rebooted and even with "multi_hash":944 which was stable it doesnt compile. i think i found the bug : the miner doesn't supported to be launched from a different dir. said differently, its current dir must be the path where the .exe is if you call the .bat or exe from a different dir, it fails. i'll fix this in next revision. the .bat is in te same dir, sadly. Also giving the absolute path to config instead of a relative doesnt change the behaviour. This is my current start-batchfile: setx GPU_FORCE_64BIT_PTR 1
setx GPU_MAX_HEAP_SIZE 100
setx GPU_MAX_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_MAX_SINGLE_ALLOC_PERCENT 100
W:
cd W:\Miner\JCE CN GPU Miner
jce_cn_gpu_miner64.exe -c "W:\Miner\JCE CN GPU Miner\confgpu.txt" -o loki.miner.rocks:5555 -u L6B6Dkz7VmLSMwEJgWeUUxHFpNeCShavBMcZ6Fid6xynbVCxD2brTsAasXviwYHKdz5i1ULS2rm2gGQqA759BdkEK4Mxzty.15000 -p w=windorf --no-cpu
PAUSE
If path is long-name you must set it in quotes: W: cd "W:\Miner\JCE CN GPU Miner" |
|
|
|
Effective hashrate : i observed the same thing with Claymore 10+ (but not 9.x), a pool hashrate a lot lower (~10%) than the displayed one. Even fees deduced. Curiously, this problem occurs only on close source miners, on open ones, where you can check in code there's no cheat, the reported rate is 100% the same (minus fees).
JCE does no tweak, and you can double check pool side.
This is why I counter-argument with Unclewish about Claymore 11.3 being faster than JCE on HD7800 : the displayed number, maybe. The displayed hashrate on my HD7850 rig was 2200 on Claymore 11.3, and 1950 pool side. With JCE, it's 2080 both side. Even after 24h.
The answer for reduced hashrate on the poolside is here: "Job not found" ok i get it, and that's because JCE is a GPU miner based on CPU miner netcode. A round of CPU is so fast that the job never timeout during mining. A GPU round is often 1 full second, a lot more than the ping. So you can get a stale share. I'll fix that in next version. No need to fix on PU version, only GPU.
All GPU miners looses effective speed on rejected shares because of job was gone while GPU findes a share. It's a lucky thing. And pool difficulty. On high pool diffilulty rejected share have big amount, and speed lowers significant. That's why on the same miner different peoples often have different poolside effective results. |
|
|
|
|
Author, too often miner finds rejected shares "Job not found" on heavy and v7, ssl and nossl... nicehash...
|
|
|
|
Ohhh. For my RX 580 8Gb finded new best intensity for heavy algo: { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":832 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":832 },
Also finded good parameters for cn-v7 algo for 8Gb cards: { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":1696 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":1696 },
Also miner often gives rejected shares "Job not found"... |
|
|
|
I really cannot use a multiple smaller than 16, it would invalidate all my optims. I did a lot of data grouping by chuncks of 16, because that's 4x4. Going down to 2x2 would be impracticable. About cached compile, that's possible, but more complicated than what Stak does. And please, switch to the SRB topic for the SRB tips, it's not a rant, no problem to compare with SRB, even for me, but it's out of place and makes the topic harder to follow.  Yes, I got it. But think, how make more fine tuning... |
|
|
|
Max multihash for heavy algo for 8Gb cards is about 960. With it's near 7.5Gb video memory usage. You both talking about different algo's. One about heavy, other about normal v7. For heavy algo optimal vaules for 580 8Gb: { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":944 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":944 },
Also good this: { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":896 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":896 },
For normal v7: { "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":1152 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 8, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 1, "multi_hash":1152 },
|
|
|
|
Wow, so many comments, thanks all
Multiple of 16 : that's a technical constraint due to how is my code. I cannot work with smaller multiples.
Cached pre-compile : not that simple, on JCE since the OpenCL code is generated, pre-optimized and injected on the fly, it's not just a per-GPU like for Stak. Not impossible to cache, but not trivial. I know that on a 8-gpu system it takes a long time.
Multihash... You must think about smaller multiples. It can make faster speed in many cases... Pre-compiling... Also need something to do - it's to long wait every time... |
|
|
|
XMR-Stak is not fastest cryptonight miner. You must try SRB miner. It faster XMR-Stak or XMRig.
I tried them all my friend, SRB for me does give a higher hashrate displayed on the rig, but on the pool the hashrate is lower, also SRB is higher about 1-3 % so not worth the hassle considering the fee You just didn't know optimal settings for SRB miner for your cards. It's faster significunt. Expeccially on heavy. On normal v7 srb 1.6.0 can give max speed with the right .srb kernel file. For heavy 1.6.1 can give max speed with the right kernel. On the pools side all right. I checked SRB on few pools, all right. Many forgot that often pool take fee. Can you share you SRB config/kernel for V7 on RX570/580 4/8GB? I would love to try it out Ok. I didn't have 4Gb 570/580 card. There is .srb kernel bin file for rx 580 8Gb: https://yadi.sk/d/1QFcuG-g3YVNWgThis file needs to be copied to "cache" subfolder of SRB miner folder. Than you need 1.6.0 version of miner. On 1.6.1 speed on v7 is lower. In config: { "id" : 0, "intensity" : 116.0, "worksize" : 8, "threads" : 2, "persistent_memory" : false}, It's for 580 8Gb card without connected monitor (more available video memory). Maked and tested on Windows 10 x64 1803 with AMD 18.6.1 driver. This settings require about 7.2Gb of free video memory. P.S. Kernel bin from link woudn't work on 1.6.1 version. doctor changed cn kernels on 1.6.1 version... Thanks for the link, just tried and in fact its a better kernel for RX580 8GB. Where did you get it, is it possible to get one for 4GB RX570/580, i only have 2 8GB cards  You can try to make these kernel bins yourself. Just try to run 4Gb cards with the next parameters: { "id" : 0, "intensity" : 58.0, "worksize" : 8, "threads" : 2, "persistent_memory" : false, "fragments": 43650010}, Maybe new kernel will be faster. |
|
|
|
XMR-Stak is not fastest cryptonight miner. You must try SRB miner. It faster XMR-Stak or XMRig.
I tried them all my friend, SRB for me does give a higher hashrate displayed on the rig, but on the pool the hashrate is lower, also SRB is higher about 1-3 % so not worth the hassle considering the fee You just didn't know optimal settings for SRB miner for your cards. It's faster significunt. Expeccially on heavy. On normal v7 srb 1.6.0 can give max speed with the right .srb kernel file. For heavy 1.6.1 can give max speed with the right kernel. On the pools side all right. I checked SRB on few pools, all right. Many forgot that often pool take fee. Can you share you SRB config/kernel for V7 on RX570/580 4/8GB? I would love to try it out Ok. I didn't have 4Gb 570/580 card. There is .srb kernel bin file for rx 580 8Gb: https://yadi.sk/d/1QFcuG-g3YVNWgThis file needs to be copied to "cache" subfolder of SRB miner folder. Than you need 1.6.0 version of miner. On 1.6.1 speed on v7 is lower. In config: { "id" : 0, "intensity" : 116.0, "worksize" : 8, "threads" : 2, "persistent_memory" : false}, It's for 580 8Gb card without connected monitor (more available video memory). Maked and tested on Windows 10 x64 1803 with AMD 18.6.1 driver. This settings require about 7.2Gb of free video memory. P.S. Kernel bin from link woudn't work on 1.6.1 version. doctor changed cn kernels on 1.6.1 version... |
|
|
|
Vega 64s
confgpu.txt
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash":1920 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 4, "delta" : 4, "epsilon" : 4, "zeta" : 4, "index" : 0, "multi_hash":1920 },
(etc.)
12:08:34 | Hashrate GPU Thread 0: 1019.38 h/s
12:08:34 | Hashrate GPU Thread 1: 1042.35 h/s
12:08:34 | Hashrate GPU Thread 2: 1022.37 h/s
12:08:34 | Hashrate GPU Thread 3: 1035.88 h/s
12:08:34 | Hashrate GPU Thread 4: 1026.74 h/s
12:08:34 | Hashrate GPU Thread 5: 1029.77 h/s
12:08:34 | Hashrate GPU Thread 6: 1030.88 h/s
12:08:34 | Hashrate GPU Thread 7: 1024.28 h/s
12:08:34 | Hashrate GPU Thread 8: 1020.92 h/s
12:08:34 | Hashrate GPU Thread 9: 1031.15 h/s
12:08:34 | Hashrate GPU Thread 10: 1007.70 h/s
12:08:34 | Hashrate GPU Thread 11: 1027.57 h/s
12:08:34 | Hashrate GPU Thread 12: 1019.47 h/s
12:08:34 | Hashrate GPU Thread 13: 1027.29 h/s
12:08:34 | Total: 14365.70 h/s - Max: 14388.12 h/s
It's need results from other miners like SRB and Cast XMR to compare with. And point algo and amount of video memory please. |
|
|
|
XMR-Stak is not fastest cryptonight miner. You must try SRB miner. It faster XMR-Stak or XMRig.
I tried them all my friend, SRB for me does give a higher hashrate displayed on the rig, but on the pool the hashrate is lower, also SRB is higher about 1-3 % so not worth the hassle considering the fee You just didn't know optimal settings for SRB miner for your cards. It's faster significunt. Expeccially on heavy. On normal v7 srb 1.6.0 can give max speed with the right .srb kernel file. For heavy 1.6.1 can give max speed with the right kernel. On the pools side all right. I checked SRB on few pools, all right. Many forgot that often pool take fee. |
|
|
|
First of all, Thanks for the great work @JCE-Miner , i believe in a while this will be the fastest miner  Now here are the results of my test on one of my rigs. Setup : 6 Cards Asus ROG Strix RX570 4GB Hynix - Index 0 Sapphire Pulse RX580 4GB Hynix - Index 3 Sapphire Pulse RX580 4GB Hynix - Index 4 MSI RX550 LP 2GB Elpida - Index 1 MSI RX550 LP 2GB Elpida - Index 2 Asus RX550 LP 2GB Hynix - Index 5 All flashed with custom bios Mining Monero (XMR) start.bat command line for miner: jce_cn_gpu_miner64.exe -c confgpu.txt --no-cpu --any --forever --variation %FORK% --low -o %POOL%:%PORT% -u %WALLET% -p %PASSWORD% %SSL% %*
confgpu.txt: "gpu_threads_conf" :
[
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 0, "multi_hash":912 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 0, "multi_hash":912 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 1, "multi_hash":432 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 1, "multi_hash":432 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 2, "multi_hash":432 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 2, "multi_hash":432 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 3, "multi_hash":912 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 3, "multi_hash":912 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 4, "multi_hash":912 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 4, "multi_hash":912 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 5, "multi_hash":432 },
{ "mode" : "GPU", "worksize" : 8, "alpha" : 64, "beta" : 16, "gamma" : 8, "delta" : 8, "epsilon" : 8, "zeta" : 8, "index" : 5, "multi_hash":432 },
] After playing with config values for about an hour , looks like these are the best i came up with. the issue is the a very long time it takes to restart the miner because of opencl payload compilation Result after 12 hours running: 08:19:36 | Accepted by the pool.
08:19:37 | Hashrate GPU Thread 0: 476.87 h/s
08:19:37 | Hashrate GPU Thread 1: 472.91 h/s
08:19:37 | Hashrate GPU Thread 2: 258.28 h/s
08:19:37 | Hashrate GPU Thread 3: 259.26 h/s
08:19:37 | Hashrate GPU Thread 4: 249.72 h/s
08:19:37 | Hashrate GPU Thread 5: 249.91 h/s
08:19:37 | Hashrate GPU Thread 6: 502.95 h/s
08:19:37 | Hashrate GPU Thread 7: 502.85 h/s
08:19:37 | Hashrate GPU Thread 8: 487.18 h/s
08:19:37 | Hashrate GPU Thread 9: 486.15 h/s
08:19:37 | Hashrate GPU Thread 10: 262.35 h/s
08:19:37 | Hashrate GPU Thread 11: 262.57 h/s
08:19:37 | Total: 4470.93 h/s - Max: 4479.91 h/s XMR Stak result:
HASHRATE REPORT - AMD
| ID | 10s | 60s | 15m | ID | 10s | 60s | 15m |
| 0 | 471.2 | 471.9 | (na) | 1 | 471.2 | 470.5 | (na) |
| 2 | 248.9 | 248.0 | (na) | 3 | 251.1 | 247.7 | (na) |
| 4 | 243.4 | 241.6 | (na) | 5 | 238.8 | 239.5 | (na) |
| 6 | 493.0 | 493.8 | (na) | 7 | 493.3 | 492.5 | (na) |
| 8 | 477.3 | 477.7 | (na) | 9 | 477.1 | 476.6 | (na) |
| 10 | 255.4 | 255.7 | (na) | 11 | 251.9 | 254.3 | (na) |
Totals (AMD): 4372.5 4369.8 0.0 H/s
-----------------------------------------------------------------
Totals (ALL): 4372.5 4369.8 0.0 H/s
Highest: 4376.5 H/s
-----------------------------------------------------------------
Rig was running very stable for over 12 hours. Very impressive RX550 performance, over 500h/s, great job on this one! Now couple of remarks: 1. Takes a long time to generate the opencl payload for the cards, i know that this is just the first beta but would be a really good idea to implement cache so need to generate only once and then just use the pre-compiled binary and load into the cards, like xmr-stak does. 2. When reporting the share value, its always the same and is equal to a static diff i set to mine at, here is the log: 08:19:45 | GPU Thread 1 Lane 14 finds a Share, value 27060
08:19:45 | Accepted by the pool.
08:19:47 | GPU Thread 6 Lane 138 finds a Share, value 27060
08:19:47 | Accepted by the pool.
08:19:55 | GPU Thread 2 Lane 69 finds a Share, value 27060
08:19:56 | Accepted by the pool.
08:19:56 | GPU Thread 11 Lane 380 finds a Share, value 27060
08:19:56 | GPU Thread 11 Lane 390 finds a Share, value 27060
08:19:56 | Accepted by the pool.
08:19:56 | Accepted by the pool.
08:19:56 | GPU Thread 10 Lane 254 finds a Share, value 27060 as you can see value is always 27060 Otherwise, great job!!! hope there will be more optimizations and another release soon for even more hashrate XMR-Stak is not fastest cryptonight miner. You must try SRB miner. It faster XMR-Stak or XMRig. |
|
|
|
|
Heavy algo didn't grows up with time... At least about 10-15 min didn't added speed. So on RX 580 heavy algo max 1000 h/s. On SRB it's up to 1120 h/s.
|
|
|
|
thanks for the feedback your results are suprising on 270X on my hd7870 i really was far above Claymore, and Claymore is easy to configure. against srb, well i probably didn't configure it at max but i couldn't reach the 540 i get on JCE. i'll retry with your exact same settings. to disable opencl, parameter --no-gpu also a very important thing, i let opencl allocate resource progressively, so the max hashrate is reached after a few minutes. at start, jce is often at 80% speed. Claymore punch the card to the max immediately. i'll add a message when the miner starts so people are not disapointed. SRB miner latest versions works not so good on older cards like 2xx and 3xx. I tried to wait some time and yes: speed grows. On 270X i got 545-546 h/s. But Claymore's 11.3 550 h/s still faster and without fee... I think if you make multihash not devidable by 16 to make it possible to set it to 460, your miner can achive Claymore's speed... Now will try heavy algo on 580... P.S. You must add saving gpu kernels to not compile it every time like it SRB miner maked. |
|
|
|
|
What Windows 10 build and AMD driver version?
|
|
|
|
|
Thanks for gpu update!
Tested new gpu part of miner...
What can I say... It's very good! Very!
Now my results. Tested on 270X 4Gb and RX 580 8Gb.
On all my cards best alpha is 64. 128 always lower speed, especcially on 270X.
Other greeks have too small affect...
CN-v7
270X - 528 h/s (m.hash - 464), Claymore 11.3 - 550 h/s and NO FEE (-h 460 -dmem 1)
RX580 - 970-975 h/s (m.hash - 1152), Claymore 11.3 - 967 h/s and NO FEE (-h 1152 -dmem 1), SRB miner 1.6.0 - 1025 h/s devfee 0.85% (intensity - 117)
CN-Heavy
270X - 412 h/s (m.hash - 448), Claymore 11.3 - no algo support, SRB miner 1.6.0 - 410-420 h/s devfee 0.85% (intensity - 27-28)
RX580 - 990 h/s (m.hash - 944), Claymore 11.3 - no algo support, SRB miner 1.6.1 - 1120 h/s devfee 0.85% (intensity - 59.5)
I didn't compare with XMRig and Stak because they are slower on my cards than Claymore or SRB. GGS is often between Claymore and SRB...
And another one thing: on RX 580 miner is stable. But on 270X 4Gb it often hangs computer till restart. Claymore 11.3 is very stable for it. So as SRB.
On Claymore and SRB it's more fine tuning of intensity. Your miner supports only multihash devided by 16. It's impossible to set multihash 460 for 270X, but it will be more speedy than 464, as I can see on Claymore.
All cards have modded bioses. And of course all tests was made on the same overclock and voltage settings. Drivers 18.6.1 Windows 10 x64 1803.
Good work! But it's still many improvements need to do.
P.S. Please make an option to disable detecttion of OpenCl devices so miner will be only CPU miner. OpenCL detection can affect on other miners if they allready running during your miner startup.
|
|
|
|
Does anyone can use high intensiity 100-110+ on 1.6.1 CN-v7 algo without speed drops?
No one tries CN-v7 on 1.6.1. on 8Gb cards? With intensity higher 100-110? I use 118 intensity dual threads on 8gb cards CN-V7 I'm talking about 1.6.1 version. You post screen from 1.6.0. On 1.6.0 I can use 110+ intensity with 2 threds without problems too. But on 1.6.1 I have speed drops to 0 if intensity higher 99. I also feel there is some changes for intensity in CN-V7, normally with previous version my intensity 72 got stable hash rate, now with 1.6.1 got up and down hash rate. Intensity 72 in 1.6.1 is especcial!!! It got speed to jump down to 875 and up to 1250. But average is still 925. On 1.6.0 with intensity 108 I has 985 on this card. But on 1.6.1 with intensity 108 I have speed 980 and drops to 0 so average speed is 600-700... I never could go intensity 110 and 2 threads (that would mean you are using around 7GB vram). I even cant go over 65 with 2 threads cause video driver hangs instantly (blockchain drivers). So i am curious which drivers, clocks and straps are you guys using. I have about 12x580 8g and a few more 570 8g, 470 8g cards and none of them does this crazy hash jumping you described in 1.6.1. So i really don't know what to think. What's the problem to use about 7Gb vram? I used 116-117 intensity on 1.6.0 version with 2 threads on RX 580 8Gb CN-v7 algo. This used 7.3Gb vram and give me about 1kh/s speed without drops and serious jumps. I can't get this speed with any other intensity... I use latest AMD drivers 18.6.1, all works normal on 18.5.2. On 1.6.1 intensity higher 100 causes speed jumps from 0 to 980. Other intensities on 1.6.1 (54,56,72,80,90,99) give for my card max 965 h/s. |
|
|
|
|
Show Posts
