SRBMiner Cryptonight AMD GPU Miner V1.9.3 - native algo switching

[heavyarms1912]

forget about cn-gpu crap. It is much better on nvidia. And will be easy for fpga/asic if have any volume

Yup. it's much better on Nvidia but floating point math is not efficient on FPGAs and ASICs.

Are you joking? I've read about fpgas doing scientific floating points math years ago )) It is what they are designed for ))

never said they can't do. read; efficiency.

EDIT:  new FPGAs do come with loads of DSP blocks so perhaps that gap is quite narrow now.

EDIT2: https://medium.com/@crypto_ryo/cryptonight-gpu-fpga-proof-pow-algorithm-based-on-floating-point-instructions-92524debf8e8

[1714052711]

1714052711

[1714052711]

1714052711

[1714052711]

1714052711

[1714052711]

1714052711

[1714052711]

1714052711

[doktor83]

When is cn/wow implementation expected to be done?

I can't tell exactly, but new version will be released in max 10 days, all  new algos added.

EDIT2: https://medium.com/@crypto_ryo/cryptonight-gpu-fpga-proof-pow-algorithm-based-on-floating-point-instructions-92524debf8e8

that article was written by fireice so it shouldn't be taken seriously 

[katar14]

algo heavyX? when??

[thenextdinosaur]

algo heavyX? when??

He literally answered above your message.

[heavyarms1912]

When is cn/wow implementation expected to be done?

I can't tell exactly, but new version will be released in max 10 days, all  new algos added.

EDIT2: https://medium.com/@crypto_ryo/cryptonight-gpu-fpga-proof-pow-algorithm-based-on-floating-point-instructions-92524debf8e8

that article was written by fireice so it shouldn't be taken seriously  

I never trust one source but on paper fp math is slower on FPGAs unless they have lots of DSP slices.  
Now that's the question how many is available and for how much.  Efficiency wise they would be order of magnitude better than GPUs.

I got a source that mentions on the FPGAs capable of doing 10 TFLOPs but looks pretty expensive.  10 TFLOPs is what an RTX 2080 does.
https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/pt/stratix-10-product-table.pdf

https://www.digikey.com/product-detail/en/intel/DK-SOC-1SSX-L-A/DK-SOC-1SSX-L-A-ND/9863681
and a 280K cell FPGA is priced at $7.5k

[katar14]

algo heavyX? when??

He literally answered above your message.

ok

[thenextdinosaur]

One issue that I've been getting since earlier versions is that, while mining on a pool with multi-algo support (MO), sometimes the miner will just close without any apparent reason. The command prompt window just closes. When checking the log, nothing abnormal was happening. Sometimes the last line on the log is an algo change, sometimes is submitting a share. Miner just closes. Any idea of what may cause that or what else can I do to test more and provide more feedback?

[doktor83]

I know about this issue, and i probably fixed it (khm, i really hope), so it will be available in the next version soon.

[SpceGhst]

One issue that I've been getting since earlier versions is that, while mining on a pool with multi-algo support (MO), sometimes the miner will just close without any apparent reason. The command prompt window just closes. When checking the log, nothing abnormal was happening. Sometimes the last line on the log is an algo change, sometimes is submitting a share. Miner just closes. Any idea of what may cause that or what else can I do to test more and provide more feedback?

I was having the same issue without using profit switching, just a single coin.  Tracked the problem down to the AMD driver crashing.  The only solution I found was to lower the intensity setting by 10-20%.  Yes, you get a lower hash rate, but the program doesn’t crash either.

[doktor83]

One issue that I've been getting since earlier versions is that, while mining on a pool with multi-algo support (MO), sometimes the miner will just close without any apparent reason. The command prompt window just closes. When checking the log, nothing abnormal was happening. Sometimes the last line on the log is an algo change, sometimes is submitting a share. Miner just closes. Any idea of what may cause that or what else can I do to test more and provide more feedback?

I was having the same issue without using profit switching, just a single coin.  Tracked the problem down to the AMD driver crashing.  The only solution I found was to lower the intensity setting by 10-20%.  Yes, you get a lower hash rate, but the program doesn’t crash either.

This is different, yes driver can crash if you push the cards too much, that can happen with any miner.
Also try to lower memory clock or give the gpu a little bit more voltage to the core, and maybe you wont have to lower the intensity at all.

[t0lsty]

one of vega`s rig gives me such error after 10-20 minutes :

"Pool rejected result 0x000xxxx  (invalid job id) "

when I lower the intensity  - it works few hours, but then again this error.

it`s on turtle algo. but have same on stellite with newer srb

Whan can be a problem ??

[Csouza]

What about hashrate on Vega56@64, RX580 4GB and RX570 4GB on CNGPU algo ?

SRB Miner 1.7.8
                                                    Core   Mem       HS
GPU0 [7]   Radeon RX 570 Series   1200   1850      620
GPU1 [5]   Radeon RX 570 Series   1200   1850      608
GPU2 [6]   Radeon RX 570 Series   1200   1850      618
GPU3 [1]   Radeon RX 570 Series   1200   1850      652
GPU4 [3]   Radeon RX 570 Series   1200   1850      645
GPU5 [4]   Radeon RX 570 Series   1200   1850      607


GPU0 [1]   Radeon RX 580           1150   2050      710
GPU1 [4]   Radeon RX 580           1150   2050      710
GPU2 [9]   Radeon RX Vega         1450   1000      1572
GPU3 [19]   Radeon RX Vega       1450   1000      1580
GPU4 [12]   Radeon RX Vega       1450   1000      1572
GPU5 [16]   Radeon RX Vega       1450   1000      1576

[ManDoneKTM]

What about hashrate on Vega56@64, RX580 4GB and RX570 4GB on CNGPU algo ?

SRB Miner 1.7.8
                                                    Core   Mem       HS
GPU0 [7]   Radeon RX 570 Series   1200   1850      620
GPU1 [5]   Radeon RX 570 Series   1200   1850      608
GPU2 [6]   Radeon RX 570 Series   1200   1850      618
GPU3 [1]   Radeon RX 570 Series   1200   1850      652
GPU4 [3]   Radeon RX 570 Series   1200   1850      645
GPU5 [4]   Radeon RX 570 Series   1200   1850      607


GPU0 [1]   Radeon RX 580           1150   2050      710
GPU1 [4]   Radeon RX 580           1150   2050      710
GPU2 [9]   Radeon RX Vega         1450   1000      1572
GPU3 [19]   Radeon RX Vega       1450   1000      1580
GPU4 [12]   Radeon RX Vega       1450   1000      1572
GPU5 [16]   Radeon RX Vega       1450   1000      1576

GPU0 :    2534 H/s [T: 49c, RPM: 2100, CC: 1260 MHz, MC: 2100 MHz][BUS:7]

[Csouza]

What about hashrate on Vega56@64, RX580 4GB and RX570 4GB on CNGPU algo ?

SRB Miner 1.7.8
                                                    Core   Mem       HS
GPU0 [7]   Radeon RX 570 Series   1200   1850      620
GPU1 [5]   Radeon RX 570 Series   1200   1850      608
GPU2 [6]   Radeon RX 570 Series   1200   1850      618
GPU3 [1]   Radeon RX 570 Series   1200   1850      652
GPU4 [3]   Radeon RX 570 Series   1200   1850      645
GPU5 [4]   Radeon RX 570 Series   1200   1850      607


GPU0 [1]   Radeon RX 580           1150   2050      710
GPU1 [4]   Radeon RX 580           1150   2050      710
GPU2 [9]   Radeon RX Vega         1450   1000      1572
GPU3 [19]   Radeon RX Vega       1450   1000      1580
GPU4 [12]   Radeon RX Vega       1450   1000      1572
GPU5 [16]   Radeon RX Vega       1450   1000      1576

GPU0 :    2534 H/s [T: 49c, RPM: 2100, CC: 1260 MHz, MC: 2100 MHz][BUS:7]

What VGA is?

[doktor83]

What about hashrate on Vega56@64, RX580 4GB and RX570 4GB on CNGPU algo ?

SRB Miner 1.7.8
                                                    Core   Mem       HS
GPU0 [7]   Radeon RX 570 Series   1200   1850      620
GPU1 [5]   Radeon RX 570 Series   1200   1850      608
GPU2 [6]   Radeon RX 570 Series   1200   1850      618
GPU3 [1]   Radeon RX 570 Series   1200   1850      652
GPU4 [3]   Radeon RX 570 Series   1200   1850      645
GPU5 [4]   Radeon RX 570 Series   1200   1850      607


GPU0 [1]   Radeon RX 580           1150   2050      710
GPU1 [4]   Radeon RX 580           1150   2050      710
GPU2 [9]   Radeon RX Vega         1450   1000      1572
GPU3 [19]   Radeon RX Vega       1450   1000      1580
GPU4 [12]   Radeon RX Vega       1450   1000      1572
GPU5 [16]   Radeon RX Vega       1450   1000      1576

You are probably using xmr stak settings here. well don't. Try out auto setup.
Vega56 @ 1450/900 = 1750khs on 110/32/2 F0

[ManDoneKTM]

What about hashrate on Vega56@64, RX580 4GB and RX570 4GB on CNGPU algo ?

SRB Miner 1.7.8
                                                    Core   Mem       HS
GPU0 [7]   Radeon RX 570 Series   1200   1850      620
GPU1 [5]   Radeon RX 570 Series   1200   1850      608
GPU2 [6]   Radeon RX 570 Series   1200   1850      618
GPU3 [1]   Radeon RX 570 Series   1200   1850      652
GPU4 [3]   Radeon RX 570 Series   1200   1850      645
GPU5 [4]   Radeon RX 570 Series   1200   1850      607


GPU0 [1]   Radeon RX 580           1150   2050      710
GPU1 [4]   Radeon RX 580           1150   2050      710
GPU2 [9]   Radeon RX Vega         1450   1000      1572
GPU3 [19]   Radeon RX Vega       1450   1000      1580
GPU4 [12]   Radeon RX Vega       1450   1000      1572
GPU5 [16]   Radeon RX Vega       1450   1000      1576

GPU0 :    2534 H/s [T: 49c, RPM: 2100, CC: 1260 MHz, MC: 2100 MHz][BUS:7]

What VGA is?

RX 580 8GB , own timmings and straps bios .

[ivomm]

One issue that I've been getting since earlier versions is that, while mining on a pool with multi-algo support (MO), sometimes the miner will just close without any apparent reason. The command prompt window just closes. When checking the log, nothing abnormal was happening. Sometimes the last line on the log is an algo change, sometimes is submitting a share. Miner just closes. Any idea of what may cause that or what else can I do to test more and provide more feedback?

I confirm this issue for V8. My cards are running fine with 1.7.0. After I upgraded to 1.7.8 they work several hours and this happens. This is not related to a card crash, the voltage etc. settings remain intact and after restart everything works OK.

[doktor83]

V1.7.9
- Added new Monero algo 'Cryptonight R'
- Added algo 'Wownero'
- Added algo 'Conceal'
- Added algo 'Zelerius'
- Added algo 'Xcash'
- Replaced 'Mox' algo with new 'Mox V8' algo
- Added parameter 'maxnosharesent' in cmd
- Added parameter 'randomnonce' in cmd
- Added parameter 'enablecoinforking' in cmd
- Added algo switching info in json and on html stats page
- Lot of bug fixes

+ Monero will fork to a new algo 'Cryptonight R' soon. You can use for cryptonight_type 'normalv4', 'cryptonight_r' or just 'monero'.
+ Wownero already forked to a custom version of 'Cryptonight R'
+ Xcash can be mined by using 'xcash' or 'heavyx' for cryptonight_type
+ Conceal and Zelerius will fork too
+ Mox algo is now replaced with the new Mox V8 algo, so if you want to mine it until the fork, use previous version of SRBMiner

+ There was already a built in watchdog that restarted miner if no share was sent to the pool for 20 minutes. A new parameter was added so user can set this time : --maxnosharesent value in seconds (min. 1 min , max. 120 min)
+ Parameter --randomnonce creates totally random nonces for every gpu thread. This is like the 'i'm feeling lucky' button on Google
+ Json API and stats page now contains some info about algoswitching (if used)

+ Parameter --enablecoinforking allows miner to switch to an algo if the fork is predefined in the miner. If enabled it will auto switch for MONERO, CONCEAL AND ZELERIUS in this version (sorry Mox).
Don't use this parameter on a pool that does algoswitching, because it can think that wow is cn/r

How will this look for Monero fork:

start.bat
.
.
.
SRBMiner-CN.exe --enablecoinforking --config Config\config-normalv8.txt --pools pools.txt

Explained :
1. Start mining with normalv8 algo, because Monero still uses that algo
2. Add --enablecoinforking parameter so miner knows you want to use the built-in (predefined) fork informations
3. When pool sends a job for the new algo, miner will auto switch to the new algo, that must be defined in algos.txt
4. By default the new Monero algo 'normalv4' is already defined in algos.txt, so if you won't remove it everything will be fine

+ A big reorganisation happened in the miner code, so hopefully the misterious 'crashing out of nowhere' is fixed. I really hope.

Please guys who have this random crashing issue try the new version and report back.

[GKumaran]

Thank for the new version Doc !! 

Can u explain --randomnonce for complete noobs like us, u mentioned its  Im lucky button, so does that mean we should use it?

P.S: That Coinforking was a great parameter in time saving

[GKumaran]

New algorithm Hashrates:
Run on 6 Vega 64 in 18.6.1, and averaged to single card numbers
All run using default SRB config files
All run in Same overclock setting -> stable
So lower power and higher HR can also be acheived

Algorithm	Hashrate	Power	Notes
MoX	4195	215	4500 HR and 190 watt before fork is this drop extected? or cause of some 'old_mode'?
GPU	1800	230	Intersting to note this algo isnt even using Memory like other CN variants
Zelerius	2635	210
Conceal	4050	195	Power usage isnt stable, keeps fluctuating widely
Wownero	1970	210
Cryptonight R	1965	210	Future XMR : Same clocks -> 2.25% drop in HR
NormalV8	2010	210	Current XMR : Same clocks
NormalV7	2120	180	Algo before current XMR : Same clocks
Xcash	1020	205