SRBMiner-MULTI GPU & CPU Miner 0.9.4 - page 49.

SCF0

newbie

Activity: 9

Merit: 0

Quote from: doktor83 on August 22, 2020, 12:35:20 AM

Hi, sorry i missed your message.
1.9.4 is just a quick patch for Haven, and yes i probably screw up some things.. Im not maintaining the CN version anymore, just wanted to give an option for XHV miners who want to continue using SRBMiner.
For all the rest algorithms just use 1.9.3 version.

No worries.

That's what I assumed as well. The question is, since it was just a quick patch, how much work would it be for you to get it working on those older GCN cards again? Some of them would still be viable in XHV, especially with cheap power.

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: SCF0 on August 20, 2020, 06:45:50 AM

doktor83,

sorry for bugging once again, but please respond to the issue about SRBMiner-CN 1.9.4 I reported here https://bitcointalksearch.org/topic/m.54900572

SRBMiner-CN 1.9.4 is still the fastest option on some CN altcoins that are somewhat viable on older hardware so it would be sad to lose this option.

Hi, sorry i missed your message.
1.9.4 is just a quick patch for Haven, and yes i probably screw up some things.. Im not maintaining the CN version anymore, just wanted to give an option for XHV miners who want to continue using SRBMiner.
For all the rest algorithms just use 1.9.3 version.

yadacoin.io

newbie

Activity: 198

Merit: 0

Looks like a great miner. We might be interested in integrating our coin with your miner.

SCF0

newbie

Activity: 9

Merit: 0

doktor83,

sorry for bugging once again, but please respond to the issue about SRBMiner-CN 1.9.4 I reported here https://bitcointalksearch.org/topic/m.54900572

SRBMiner-CN 1.9.4 is still the fastest option on some CN altcoins that are somewhat viable on older hardware so it would be sad to lose this option.

VictorGT

hero member

Activity: 748

Merit: 523

Great. I will like a lot to test beta version if you need betatesters Cheesy

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: VictorGT on August 07, 2020, 08:48:51 AM

Hi doktor83,

Scala (XLA) changes mining algorithm a week ago. No more Defyx, now is Panthera algo.

May you want to update your miner?. SRBminer was really faster than official miner on Defyx Cheesy

.

Next release will contain the change (but its ~ same speed as xlarig unfortunately) and also a ton of new stuff. Just need a few more weeks Cool

VictorGT

hero member

Activity: 748

Merit: 523

Hi doktor83,

Scala (XLA) changes mining algorithm a week ago. No more Defyx, now is Panthera algo.

May you want to update your miner?. SRBminer was really faster than official miner on Defyx Cheesy

.

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: yadacoin.io on August 06, 2020, 12:53:15 PM

Quote from: doktor83 on August 06, 2020, 02:01:19 AM

Quote from: yadacoin.io on August 06, 2020, 01:09:41 AM

YadaCoin may be interested in integrating our algo into your miner. We're a RandomX variant. Is it possible for our coin to be integrated into your main release?

Hi, can you give me a link to your github so i can see the changes you made with RX ? Also is there a stratum pool for your coin ?

Our changes:
https://github.com/tevador/RandomX/compare/7567cef4c6192fb5356bbdd7db802be77be0439b...575d33f8620557e0b2cd0069d09a19a43f4aaede#diff-9e23bd2bad82d05cb017e4933a8c7b66

We have not implemented a stratum pool yet. Is it required? If so, we can make it happen.

If you want to attract more miners then yes, and my miner doesn't support solo mining, only stratum

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: BoozyTalking on August 06, 2020, 08:22:16 AM

I will not argue with you about allocated or not in huge(large) pages, but XMRigCC hashing 2x times faster then SRBminer on this processors, so looks like it use huge pages.

This is how log looks on Intel Celeron without enough L3 cache size:

Code:

* ABOUT XMRigCC/2.7.0 gcc/10.1.0 (RELEASE)
* LIBS libuv/1.38.0 OpenSSL/1.1.1g hwloc/2.2.0
* HUGE PAGES permission granted
* CPU Intel(R) Celeron(R) CPU J1900 @ 1.99GHz (1) x64 -AES
L2:2.0 MB L3:0.0 MB 4C/4T NUMA:1
* DONATE 1%
* ASSEMBLY auto:none
* POOL #1 ********* algo rx/0
* COMMANDS hashrate, pause, resume, quit
[2020-08-06 17:10:29.950] CC feature is disabled.
[2020-08-06 17:10:30.777] net use daemon ************ TLSv1.2 *************
[2020-08-06 17:10:30.780] net fingerprint (SHA-256): "***************************"
[2020-08-06 17:10:30.781] net new job from ******** target 000009375d366e6c algo rx/0 height 119187
[2020-08-06 17:10:30.784] rx init dataset algo rx/0 (4 threads) seed 4181a493b397a733...
[2020-08-06 17:10:30.791] rx allocated 2336 MB (2080+256) huge pages 0% 0/1168 +JIT (4 ms)
[2020-08-06 17:10:54.601] rx dataset ready (23806 ms)
[2020-08-06 17:10:54.604] cpu use profile rx/0 (2 threads) scratchpad 2048 KB
[2020-08-06 17:10:54.647] cpu READY threads 2/2 (2) huge pages 0% 0/2 memory 4096 KB (40 ms)

Maybe hashes 2x faster but still the dataset is not allocated in huge pages. ( huge pages 0% 0/1168 )

Look how that line should look when dataset is allocated successfully in HP mem.

yadacoin.io

newbie

Activity: 198

Merit: 0

Quote from: doktor83 on August 06, 2020, 02:01:19 AM

Quote from: yadacoin.io on August 06, 2020, 01:09:41 AM

YadaCoin may be interested in integrating our algo into your miner. We're a RandomX variant. Is it possible for our coin to be integrated into your main release?

Hi, can you give me a link to your github so i can see the changes you made with RX ? Also is there a stratum pool for your coin ?

Our changes:
https://github.com/tevador/RandomX/compare/7567cef4c6192fb5356bbdd7db802be77be0439b...575d33f8620557e0b2cd0069d09a19a43f4aaede#diff-9e23bd2bad82d05cb017e4933a8c7b66

We have not implemented a stratum pool yet. Is it required? If so, we can make it happen.

BoozyTalking

newbie

Activity: 315

Merit: 0

I will not argue with you about allocated or not in huge(large) pages, but XMRigCC hashing 2x times faster then SRBminer on this processors, so looks like it use huge pages.

This is how log looks on Intel Celeron without enough L3 cache size:

Code:

* ABOUT XMRigCC/2.7.0 gcc/10.1.0 (RELEASE)
* LIBS libuv/1.38.0 OpenSSL/1.1.1g hwloc/2.2.0
* HUGE PAGES permission granted
* CPU Intel(R) Celeron(R) CPU J1900 @ 1.99GHz (1) x64 -AES
L2:2.0 MB L3:0.0 MB 4C/4T NUMA:1
* DONATE 1%
* ASSEMBLY auto:none
* POOL #1 ********* algo rx/0
* COMMANDS hashrate, pause, resume, quit
[2020-08-06 17:10:29.950] CC feature is disabled.
[2020-08-06 17:10:30.777] net use daemon ************ TLSv1.2 *************
[2020-08-06 17:10:30.780] net fingerprint (SHA-256): "***************************"
[2020-08-06 17:10:30.781] net new job from ******** target 000009375d366e6c algo rx/0 height 119187
[2020-08-06 17:10:30.784] rx init dataset algo rx/0 (4 threads) seed 4181a493b397a733...
[2020-08-06 17:10:30.791] rx allocated 2336 MB (2080+256) huge pages 0% 0/1168 +JIT (4 ms)
[2020-08-06 17:10:54.601] rx dataset ready (23806 ms)
[2020-08-06 17:10:54.604] cpu use profile rx/0 (2 threads) scratchpad 2048 KB
[2020-08-06 17:10:54.647] cpu READY threads 2/2 (2) huge pages 0% 0/2 memory 4096 KB (40 ms)

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: yadacoin.io on August 06, 2020, 01:09:41 AM

YadaCoin may be interested in integrating our algo into your miner. We're a RandomX variant. Is it possible for our coin to be integrated into your main release?

Hi, can you give me a link to your github so i can see the changes you made with RX ? Also is there a stratum pool for your coin ?

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: BoozyTalking on August 06, 2020, 12:52:55 AM

If you think i'm wrong about miner, just for example XMRigCC:

Intel core i5:

Code:

* ABOUT XMRigCC/2.7.0 gcc/10.1.0 (RELEASE)
* LIBS libuv/1.38.0 OpenSSL/1.1.1g hwloc/2.2.0
* HUGE PAGES permission granted
* CPU Intel(R) Core(TM) i5-4670 CPU @ 3.40GHz (1) x64 AES
L2:1.0 MB L3:6.0 MB 4C/4T NUMA:1
* DONATE 1%
* ASSEMBLY auto:intel
* POOL #1 ********** algo rx/0
* COMMANDS hashrate, pause, resume, quit
[2020-08-06 08:44:58.286] CC feature is disabled.
[2020-08-06 08:44:59.202] net use daemon ********* TLSv1.2 ***********
[2020-08-06 08:44:59.206] net fingerprint (SHA-256): "f4fa3fae42315e4ad55027451ececc9768d2b7403e5a8ef859dd1ab83df9a152"
[2020-08-06 08:44:59.208] net new job from ******** target 0000054899c39bd5 algo rx/0 height 119123
[2020-08-06 08:44:59.211] msr service WinRing0_1_2_0 is already exists
[2020-08-06 08:44:59.215] rx init dataset algo rx/0 (4 threads) seed 4181a493b397a733...
[2020-08-06 08:44:59.685] rx allocated 2336 MB (2080+256) [b]huge pages 0% 0/1168[/b] +JIT (468 ms)
[2020-08-06 08:45:07.523] rx dataset ready (7836 ms)
[2020-08-06 08:45:07.524] cpu use profile * (1 thread) scratchpad 2048 KB
[2020-08-06 08:45:07.526] cpu READY threads 1/1 (1) huge pages 100% 1/1 memory 2048 KB (1 ms)

and even old Intel Xeon E5520 (which even not support intel optimization):

Code:

* ABOUT XMRigCC/2.7.0 gcc/10.1.0 (RELEASE)
* LIBS libuv/1.38.0 OpenSSL/1.1.1g hwloc/2.2.0
* HUGE PAGES permission granted
* CPU Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (1) x64 -AES
L2:1.0 MB L3:8.0 MB 4C/8T NUMA:1
* DONATE 1%
* ASSEMBLY auto:none
* POOL #1 ********* algo rx/0
* COMMANDS hashrate, pause, resume, quit
[2020-08-06 08:40:56.331] CC feature is disabled.
[2020-08-06 08:40:57.269] net use daemon ******** TLSv1.2 ***********
[2020-08-06 08:40:57.271] net fingerprint (SHA-256): "f4fa3fae42315e4ad55027451ececc9768d2b7403e5a8ef859dd1ab83df9a152"
[2020-08-06 08:40:57.272] net new job from ********** target 0000054899c39bd5 algo rx/0 height 119123
[2020-08-06 08:40:57.282] rx init dataset algo rx/0 (8 threads) seed 4181a493b397a733...
[2020-08-06 08:40:57.375] rx allocated 2336 MB (2080+256) [b]huge pages 0% 0/1168[/b] +JIT (90 ms)
[2020-08-06 08:41:10.746] rx dataset ready (13367 ms)
[2020-08-06 08:41:10.748] cpu use profile rx/0 (2 threads) scratchpad 2048 KB
[2020-08-06 08:41:10.800] cpu READY threads 2/2 (2) huge pages 100% 2/2 memory 4096 KB (50 ms)

Sorry but you are wrong, you can see in the log you pasted that xmrig also isn't allocating the dataset in huge pages ( huge pages 0% 0/1168 ). It should be huge pages 100% 1168/1168.
Like you see lower there : READY threads 2/2 (2) huge pages 100% 2/2 that the scratchpad for the threads was allocated in HP.

yadacoin.io

newbie

Activity: 198

Merit: 0

YadaCoin may be interested in integrating our algo into your miner. We're a RandomX variant. Is it possible for our coin to be integrated into your main release?

BoozyTalking

newbie

Activity: 315

Merit: 0

If you think i'm wrong about miner, just for example XMRigCC:

Intel core i5:

Code:

* ABOUT XMRigCC/2.7.0 gcc/10.1.0 (RELEASE)
* LIBS libuv/1.38.0 OpenSSL/1.1.1g hwloc/2.2.0
* HUGE PAGES permission granted
* CPU Intel(R) Core(TM) i5-4670 CPU @ 3.40GHz (1) x64 AES
L2:1.0 MB L3:6.0 MB 4C/4T NUMA:1
* DONATE 1%
* ASSEMBLY auto:intel
* POOL #1 ********** algo rx/0
* COMMANDS hashrate, pause, resume, quit
[2020-08-06 08:44:58.286] CC feature is disabled.
[2020-08-06 08:44:59.202] net use daemon ********* TLSv1.2 ***********
[2020-08-06 08:44:59.206] net fingerprint (SHA-256): "f4fa3fae42315e4ad55027451ececc9768d2b7403e5a8ef859dd1ab83df9a152"
[2020-08-06 08:44:59.208] net new job from ******** target 0000054899c39bd5 algo rx/0 height 119123
[2020-08-06 08:44:59.211] msr service WinRing0_1_2_0 is already exists
[2020-08-06 08:44:59.215] rx init dataset algo rx/0 (4 threads) seed 4181a493b397a733...
[2020-08-06 08:44:59.685] rx allocated 2336 MB (2080+256) huge pages 0% 0/1168 +JIT (468 ms)
[2020-08-06 08:45:07.523] rx dataset ready (7836 ms)
[2020-08-06 08:45:07.524] cpu use profile * (1 thread) scratchpad 2048 KB
[2020-08-06 08:45:07.526] cpu READY threads 1/1 (1) huge pages 100% 1/1 memory 2048 KB (1 ms)

and even old Intel Xeon E5520 (which even not support intel optimization):

Code:

* ABOUT XMRigCC/2.7.0 gcc/10.1.0 (RELEASE)
* LIBS libuv/1.38.0 OpenSSL/1.1.1g hwloc/2.2.0
* HUGE PAGES permission granted
* CPU Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (1) x64 -AES
L2:1.0 MB L3:8.0 MB 4C/8T NUMA:1
* DONATE 1%
* ASSEMBLY auto:none
* POOL #1 ********* algo rx/0
* COMMANDS hashrate, pause, resume, quit
[2020-08-06 08:40:56.331] CC feature is disabled.
[2020-08-06 08:40:57.269] net use daemon ******** TLSv1.2 ***********
[2020-08-06 08:40:57.271] net fingerprint (SHA-256): "f4fa3fae42315e4ad55027451ececc9768d2b7403e5a8ef859dd1ab83df9a152"
[2020-08-06 08:40:57.272] net new job from ********** target 0000054899c39bd5 algo rx/0 height 119123
[2020-08-06 08:40:57.282] rx init dataset algo rx/0 (8 threads) seed 4181a493b397a733...
[2020-08-06 08:40:57.375] rx allocated 2336 MB (2080+256) huge pages 0% 0/1168 +JIT (90 ms)
[2020-08-06 08:41:10.746] rx dataset ready (13367 ms)
[2020-08-06 08:41:10.748] cpu use profile rx/0 (2 threads) scratchpad 2048 KB
[2020-08-06 08:41:10.800] cpu READY threads 2/2 (2) huge pages 100% 2/2 memory 4096 KB (50 ms)

SCF0

newbie

Activity: 9

Merit: 0

After upgrading to SRBMiner-CN 1.9.4 I get CL_COMPILE_PROGRAM_FAILURE on a Radeon 290 and the miner exits afterwards. With 1.9.3 everything worked fine. I tried to mine Haven but when I try to run other algorithms I get the same error on 1.9.4. I use the 20.4.2 WHQL driver on Windows 10 if that matters.

Oh and I replied here as the update was announced here. I wonder if discussion regarding SRBMiner-CN should rather go to its own thread.

BoozyTalking

newbie

Activity: 315

Merit: 0

I have 8Gb RAM, try run it right after computer restart - same result.

doktor83

hero member

Activity: 2548

Merit: 626

Quote from: BoozyTalking on July 23, 2020, 03:05:27 AM

Hi, Doktor83! Will you plan to fix RandomX Huge/Large pages allocation for Intel I5 processors windows 7?
Coz it not work.

There is nothing to fix specifically for that processor or win7, miner uses windows mechanisms to allocate mem in huge pages, so it should work same with every cpu

First make sure you have enough RAM (4gb for example), and that you tried to run miner after computer restart, so that there are still free pages available for miner to allocate.

BoozyTalking

newbie

Activity: 315

Merit: 0

Hi, Doktor83! Will you plan to fix RandomX Huge/Large pages allocation for Intel I5 processors windows 7?
Coz it not work.

Code:

CPU0: Intel(R) Core(TM) i5-4670 CPU @ 3.40GHz [L3:6144 KB][L2:256 KB][L1:32 KB]
--------------------------------------------------------------------------------
Algorithm : randomx
Gpu mining : disabled
Cpu mining : enabled
Huge-pages : enabled
HW-Aes : available
Startup monitor : enabled

[2020-07-23 11:01:07] Initialising dataset & cache & vm

Quote

[2020-07-23 11:01:07] Couldn't allocate 2129920 kB in large-page memory. Using slower allocation

Code:

[2020-07-23 11:01:15] Dataset & Cache & VM initialized in 7709 ms

Acex29

member

Activity: 182

Merit: 20

Quote from: doktor83 on July 16, 2020, 02:43:12 PM

SRBMiner-CN 1.9.4

Haven algorithm updated and ready for the fork. You can start using it now, it will auto-switch after fork

https://mega.nz/folder/qVIgxAwB#kKmgCDICmQwbdVvMb-tAag

cool, great update

Topic: SRBMiner-MULTI GPU & CPU Miner 0.9.4 - page 49. (Read 45097 times)