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

[doktor83]

i got this error . "HW error: BUS_ID_1: 1"
but share still accepted.
what dose it mean?
and what should i do?
thank you.

[2018-05-19 17:24:53] stats: Accepted/Total shares: 2/2

[2018-05-19 17:24:53] stats: Average time to find share: 50 seconds
[2018-05-19 17:24:53] stats: HW error: BUS_ID_1: 1

HW error means result from GPU on BUS ID 1 couldn't be validated, so it's discarded. It was not accepted. You should lower intensity or there is too much overclocking, that causes bad results.
Bus id's are used now for GPU identification cause they are unique, so if it says BUS ID 1, then it's BUS ID 1 in overdriventool, gpuz etc etc. It's easier to identify which card causes problems.

[doktor83]

Good job doktor. Lots of updates.

One of the things that has stopped me from using it over xmr-stak is the web interface when checking for hashrates via browser. In xmr-stak it is very easy to read, it shows errors in detail, like which gpu caused the error or if it was a network error etc. All in a pleasant to the eye interface.

Do this for your and you will have another user

User vitalyoff made a dashboard in HTML+JS, you can check it out here :

https://bitcointalk.org/index.php?topic=3167363.msg36587137#msg36587137

1.Enable API in config.txt
2.Copy the code you need (1,2 or 3 rigs) in notepad, save it with .html extension from the link above
3.Edit this html to fit your needs (Rig frienly name, miner API IP and port)

[livada]

clasic vega 56 powercolor with 64bios  in hevy : gpu=1442mhz 905mv  and mem= 1095mhz 900mv  = 1500+ HR-samsung
saphire nitro+vega 56 with 64 bios in hevy : gpu=1427mhz 920mv  and mem= 1010mhz 920mv  = 1420 HR (and block)-hynix
nitro take more power for smaler HR

Any have  nitro vega 56? overdrieve setup? and HR in hevy?

1420h/s in Heavy is good for hynix memory. It has lower o/c capability and worse timings

now i have 1450-1460HR on hynix vega56/64nitro+: gpu= 1437mhz 920mv   mem=995mhz 905mv

nitro sucks:(. powercolor with samsung have 1600HR with 1452 915mv    and 1100mhz mem on 905mv.

hynix block on 1020mhz:( BAD BaD - and he call this card NITRO. NITRO sucks:(

[rednoW]

clasic vega 56 powercolor with 64bios  in hevy : gpu=1442mhz 905mv  and mem= 1095mhz 900mv  = 1500+ HR-samsung
saphire nitro+vega 56 with 64 bios in hevy : gpu=1427mhz 920mv  and mem= 1010mhz 920mv  = 1420 HR (and block)-hynix
nitro take more power for smaler HR

Any have  nitro vega 56? overdrieve setup? and HR in hevy?

1420h/s in Heavy is good for hynix memory. It has lower o/c capability and worse timings

now i have 1450-1460HR on hynix vega56/64nitro+: gpu= 1437mhz 920mv   mem=995mhz 905mv

nitro sucks:(. powercolor with samsung have 1600HR with 1452 915mv    and 1100mhz mem on 905mv.

hynix block on 1020mhz:( BAD BaD - and he call this card NITRO. NITRO sucks:(

It is a lottery, you can get hynix with any vega56 card except reference ones.
What have you changed to have more hash with lower clocks?

[livada]

clasic vega 56 powercolor with 64bios  in hevy : gpu=1442mhz 905mv  and mem= 1095mhz 900mv  = 1500+ HR-samsung
saphire nitro+vega 56 with 64 bios in hevy : gpu=1427mhz 920mv  and mem= 1010mhz 920mv  = 1420 HR (and block)-hynix
nitro take more power for smaler HR

Any have  nitro vega 56? overdrieve setup? and HR in hevy?

1420h/s in Heavy is good for hynix memory. It has lower o/c capability and worse timings

now i have 1450-1460HR on hynix vega56/64nitro+: gpu= 1437mhz 920mv   mem=995mhz 905mv

nitro sucks:(. powercolor with samsung have 1600HR with 1452 915mv    and 1100mhz mem on 905mv.

hynix block on 1020mhz:( BAD BaD - and he call this card NITRO. NITRO sucks:(

It is a lottery, you can get hynix with any vega56 card except reference ones.
What have you changed to have more hash with lower clocks?

yes . lower clock more hash now i have 1470hr gpu=1437 920mv mem: 1000mhz 905mv.
i try gpu=1437 920mv mem: 1010mhz 905mv. but hr is 1450
and i try gpu=1437 920mv mem: 995mhz 905mv. hr is 1445
best for me is 100mhz 905mv memori
no i go chg gpu freq and voltage

SAPPHIRE VEGA 56 NITRO with 64 bios and hynix memory - algo - HEAVY
https://image.prntscr.com/image/PmnKGQtEStGz5nZwZAvSWA.jpg

POWERCOLOR VEGA 56 with 64 bios and samsung memory - algo HEAVY
https://image.prntscr.com/image/sCQinxq3TiSfkqI3SKDvtA.jpg

[livada]

gpu=1442 920mv mem: 1000mhz 905mv. - DEAD

https://image.prntscr.com/image/Tm7HJ_ZJRjiDH0oZPe5i9w.jpg

if i give more power like 950mv for gpu vor on 1500mhz but i have only 1500hr  and lose more money for power

[rig1313]

R9 390X

using kernel=3

higher hashrate than kernel=1

but

nothing accepted

nothing rejected

only pool sent a new job !

[doktor83]

R9 390X

using kernel=3

higher hashrate than kernel=1

but

nothing accepted

nothing rejected

only pool sent a new job !

kernel 3 and 4 is for old old cards, with 1024 and 512 mb ram.
Altough it should work on every card. Which algo are you trying?

[cryptopavan]

Hi Doktor

Am facing a strange problem, I have two RX 570 4gb 8 card rigs with Samsung memory.

One Rig is running with very good hashing speed. the second Rig am getting an error

Error CL_INVALID_BUFFER_SIZE when creating scratchpad buffer for DeviceID 0 (Thread 0)


Card: RX 570 4GB Samsung memory

Error:  Error CL_INVALID_BUFFER_SIZE when creating scratchpad buffer for DeviceID 0 (Thread 0)
Algo: CN Heavy
Intensity: 59

https://ibb.co/jkFm0o
https://ibb.co/jkFm0o

Can you please help me in this regard.

[rig1313]

R9 390X

using kernel=3

higher hashrate than kernel=1

but

nothing accepted

nothing rejected

only pool sent a new job !

kernel 3 and 4 is for old old cards, with 1024 and 512 mb ram.
Altough it should work on every card. Which algo are you trying?

Heavy Algo

kernel=3
higher hashrate
constant hashrate

on kernel=1

swinging between 550-950 on intensity 61 double threads on

[livada]

Any IDEA how automate ON HBC memory on vega?

ON start rig automatic:
1.disable and enable all card (devcon.exe disable "PCI\VEN_1002&DEV_687F")
2.Run OverdriveNTool.exe (OverdriveNTool.exe -r1 -r2 -r3 -p1profile1 -p2profile1 -p3profile1)
3. start srbminer

but HBC not work and HR is 1300 not 1500 on hevy adn i must stop miner  start srminer  intesity 60 close miner and start miner 56 intensity.

Any IDEA?

[Ballscack]

i use this in config file for v7 normal:

(for vega 64 ref)

"gpu_conf" :
[
   { "id" : 0, "intensity" : 112, "worksize" : 16, "threads" : 2},
        { "id" : 1, "intensity" : 112, "worksize" : 16, "threads" : 2},
        { "id" : 2, "intensity" : 112, "worksize" : 16, "threads" : 2},
        { "id" : 3, "intensity" : 112, "worksize" : 16, "threads" : 2},
        { "id" : 4, "intensity" : 112, "worksize" : 16, "threads" : 2},
        { "id" : 5, "intensity" : 112, "worksize" : 16, "threads" : 2},

]
}

overdrive: p7 1475@960  and p3 1100@980

i get between 2000 and 2040 h/s

driver 18.4.1

[smolsky]

One Rig is running with very good hashing speed. the second Rig am getting an error

Error CL_INVALID_BUFFER_SIZE when creating scratchpad buffer for DeviceID 0 (Thread 0)


Card: RX 570 4GB Samsung memory

Error:  Error CL_INVALID_BUFFER_SIZE when creating scratchpad buffer for DeviceID 0 (Thread 0)
Algo: CN Heavy
Intensity: 59

https://ibb.co/jkFm0o
https://ibb.co/jkFm0o

Can you please help me in this regard.

Hi, you should to increase virtual memory page file, I have 6 gpu's installed, four 470 4gb  and two 480 8 gb -in such case around 31GB swap memory used 

[kirilvvbg]

Guy, am I missing some settings, or in general RX4XX is slower than RX5XX?

[cryptopavan]

One Rig is running with very good hashing speed. the second Rig am getting an error

Error CL_INVALID_BUFFER_SIZE when creating scratchpad buffer for DeviceID 0 (Thread 0)


Card: RX 570 4GB Samsung memory

Error:  Error CL_INVALID_BUFFER_SIZE when creating scratchpad buffer for DeviceID 0 (Thread 0)
Algo: CN Heavy
Intensity: 59

https://ibb.co/jkFm0o
https://ibb.co/jkFm0o

Can you please help me in this regard.

Hi, you should to increase virtual memory page file, I have 6 gpu's installed, four 470 4gb  and two 480 8 gb -in such case around 31GB swap memory used 

Virtual memory already seted as 90 gb still same error.

[rednoW]

Virtual memory already seted as 90 gb still same error.

You are probably trying to run 2 threads with > than 50 intensity on 4gb card. It is not possible

[rednoW]

Guy, am I missing some settings, or in general RX4XX is slower than RX5XX?

It depends on your card (memory, quality of cooling, including vrm and memory cooling, power supply system).
In fact old samsung based rx480 and rx470 cards are very good (not mention power color red dragon crap without memory cooling)
Old hynix and old micron rx480 and rx470 cards are not so good as modern rx580 and rx570 cards with the same memory manufacturer.
Moreover, old rx480 seems can handle only 1100-1140mhz gpu clock @ 900mV but new rx580 can handle 1200-1240mhz with the same voltage. New revision boosted stable gpu clocks by ~100mhz

[christiano88]

Virtual memory already seted as 90 gb still same error.

You are probably trying to run 2 threads with > than 50 intensity on 4gb card. It is not possible

Still some manage to get up to 60/61, how is that possible? I get same issue.

[Kgonla]

Virtual memory already seted as 90 gb still same error.

You are probably trying to run 2 threads with > than 50 intensity on 4gb card. It is not possible

Still some manage to get up to 60/61, how is that possible? I get same issue.

Config for 4 GB
"intensity" : 30, "worksize" : 8, "threads" : 2

[rednoW]

Virtual memory already seted as 90 gb still same error.

You are probably trying to run 2 threads with > than 50 intensity on 4gb card. It is not possible

Still some manage to get up to 60/61, how is that possible? I get same issue.

use 1 thread for 4gb cards