Author

Topic: [Cast XMR] high speed XMR/CryptoNight miner for RX Vega GPUs (2 KHash/s) - page 113. (Read 206404 times)

member
Activity: 275
Merit: 11
Hi all!

I wonder how does Cast XMR enumerate GPUs? Does it have any relation with OverdriveNTool? For example, OverdriveNTool reports cards in my system in this way:

Code:
0: iGPU
1: Nvidia
2: Vega 64
3: Vega 64
4: Vega 56
5: Vega 56
6: Vega 56

and I'm running Cast XMR like this:

Code:
cast_xmr-vega.exe --remoteaccess -S pool:7777 -u address -p pass --opencl 2 -G 0,1,2,3,4

Once CAST XMR starts, I can see the list of cards:

Code:
GPU0: gfx901 | 64 Compute Units
GPU1: gfx901 | 64 Compute Units
GPU2: gfx901 | 64 Compute Units
GPU3: gfx901 | 64 Compute Units
GPU4: gfx901 | 64 Compute Units

but how can I tell exactly which GPUs are Vega 64s and which ones are 56s?
It is easy, just setup different rpm of fan for each gpu and you will see it.
member
Activity: 357
Merit: 26
Quick idiot question - how do you set failover pools? Thanks in advance!
newbie
Activity: 3
Merit: 0
Hello,

i got the following issue:

initially i got average around 1900h/s and after some time, it can be minutes or hours, i drops to 1500-1600h/s

Anyone having such issue and eventually a solution ?
newbie
Activity: 58
Merit: 0
Hi all!

I wonder how does Cast XMR enumerate GPUs? Does it have any relation with OverdriveNTool? For example, OverdriveNTool reports cards in my system in this way:

Code:
0: iGPU
1: Nvidia
2: Vega 64
3: Vega 64
4: Vega 56
5: Vega 56
6: Vega 56

and I'm running Cast XMR like this:

Code:
cast_xmr-vega.exe --remoteaccess -S pool:7777 -u address -p pass --opencl 2 -G 0,1,2,3,4

Once CAST XMR starts, I can see the list of cards:

Code:
GPU0: gfx901 | 64 Compute Units
GPU1: gfx901 | 64 Compute Units
GPU2: gfx901 | 64 Compute Units
GPU3: gfx901 | 64 Compute Units
GPU4: gfx901 | 64 Compute Units

but how can I tell exactly which GPUs are Vega 64s and which ones are 56s?
newbie
Activity: 10
Merit: 0
that is not that bad, which card is it again??
and what driver?
Blockchain driver, Sapphire Rx580 Nitro SE, they hold 1200/2250 with 850v on core and memory.

I've already seen similar questions about null hashes, but I did not find an answer
Code:
...
[05:59:50] GPU3 | 46°C | Fan 1285 RPM | 0.0 H/s
[05:59:50] GPU0 | 46°C | Fan 1326 RPM | 0.0 H/s
[05:59:50] GPU1 | 47°C | Fan 1350 RPM | 0.0 H/s
[05:59:50] GPU2 | 45°C | Fan 1313 RPM | 0.0 H/s
...

can I  do something with this problem? it can be with 1 card or 2 or 3. With the same settings, alternative miners work stably but give 7-9% less
I have it too, v0.7. It doesn't seem cause any problem apart that it spams into a console.

Yeah, it keeps spamming the console even though the statistics show that the GPU still running...

Code:
...
[Hash Rate Avg: 3147.9 H/s]
  717.4 H/s GPU0
  827.7 H/s GPU1
  805.1 H/s GPU2
  797.8 H/s GPU3

[Shares Found: 4 | Avg Search Time: 65.0 sec]
   4 (100%) Accepted
   0 (  0%) Rejected by pool
   0 (  0%) Invalid result computation failed
   0 (  0%) Could not be submit because of network error
   0 (  0%) Outdated because of job change

[14:44:04] GPU0 | 48°C | Fan 1639 RPM | 0.0 H/s
[14:44:04] GPU1 | 54°C | Fan 1879 RPM | 827.1 H/s
[14:44:04] GPU2 | 59°C | Fan 2070 RPM | 803.3 H/s
[14:44:04] GPU3 | 54°C | Fan 2025 RPM | 793.4 H/s
...
newbie
Activity: 10
Merit: 0
Error CL_MEM_OBJECT_ALLOCATION_FAILURE

How i solve this issue?
Make your pagefile huge

Make your virtual memory multiples of your GPU's VRAM, eg, 5 pcs RX570 4G will need 5*4 = 20GB virtual memory on hard disk.
newbie
Activity: 50
Merit: 0
that is not that bad, which card is it again??
and what driver?
Blockchain driver, Sapphire Rx580 Nitro SE, they hold 1200/2250 with 850v on core and memory.

I've already seen similar questions about null hashes, but I did not find an answer
Code:
...
[05:59:50] GPU3 | 46°C | Fan 1285 RPM | 0.0 H/s
[05:59:50] GPU0 | 46°C | Fan 1326 RPM | 0.0 H/s
[05:59:50] GPU1 | 47°C | Fan 1350 RPM | 0.0 H/s
[05:59:50] GPU2 | 45°C | Fan 1313 RPM | 0.0 H/s
...

can I  do something with this problem? it can be with 1 card or 2 or 3. With the same settings, alternative miners work stably but give 7-9% less
I have it too, v0.7. It doesn't seem cause any problem apart that it spams into a console.
sr. member
Activity: 449
Merit: 251
Error CL_MEM_OBJECT_ALLOCATION_FAILURE

How i solve this issue?
Make your pagefile huge
newbie
Activity: 2
Merit: 0
Error CL_MEM_OBJECT_ALLOCATION_FAILURE

How i solve this issue?
full member
Activity: 167
Merit: 100
try using this data in command line  -G 0,1,2,3 -O 1 , where -G num id cards , it helped me, had the same problem
Thanks, it worked. Apparently --opencl doesn't work in 0.7. Getting just 60h/s more then claymore's miner. Overall 920 h/s per card.

that is not that bad, which card is it again??

and what driver?
newbie
Activity: 154
Merit: 0
Can someone post complete list of configuration options, please  : (

I pull --help in bat file but nothing happens....Sadly

Really need readme.txt file Sad(

Thanks u guys.
newbie
Activity: 50
Merit: 0
try using this data in command line  -G 0,1,2,3 -O 1 , where -G num id cards , it helped me, had the same problem
Thanks, it worked. Apparently --opencl doesn't work in 0.7. Getting just 60h/s more then claymore's miner. Overall 920 h/s per card.
full member
Activity: 363
Merit: 100
try using this data in command line  -G 0,1,2,3 -O 1 , where -G num id cards , it helped me, had the same problem
newbie
Activity: 50
Merit: 0
Trying to run this miner on RX 580 rigs.
Getting "Error CL_DEVICE_NOT_FOUND calling clGetDeviceIDs for number of devices", tried different options like --opencl 0 /1. In device manager no integrated graphics, only rx 580s. Happens on all my rigs.
Any advise please?
full member
Activity: 363
Merit: 100
I've already seen similar questions about null hashes, but I did not find an answer

Code:
[05:59:47] GPU0 | 47°C | Fan 1327 RPM | 924.4 H/s
[05:59:47] GPU1 | 46°C | Fan 1348 RPM | 925.7 H/s
[05:59:47] GPU2 | 46°C | Fan 1318 RPM | 900.6 H/s
[05:59:47] GPU3 | 45°C | Fan 1288 RPM | 0.0 H/s
[05:59:49] GPU0 | 46°C | Fan 1326 RPM | 924.0 H/s
[05:59:50] GPU1 | 46°C | Fan 1347 RPM | 925.3 H/s
[05:59:50] GPU2 | 46°C | Fan 1314 RPM | 899.3 H/s
[05:59:50] GPU3 | 46°C | Fan 1285 RPM | 0.0 H/s
[05:59:50] GPU0 | 46°C | Fan 1326 RPM | 0.0 H/s
[05:59:50] GPU1 | 47°C | Fan 1350 RPM | 0.0 H/s
[05:59:50] GPU2 | 45°C | Fan 1313 RPM | 0.0 H/s
[05:59:50] GPU3 | 45°C | Fan 1285 RPM | 235.3 H/s
[05:59:53] GPU0 | 46°C | Fan 1327 RPM | 921.3 H/s
[05:59:53] GPU1 | 46°C | Fan 1344 RPM | 923.7 H/s
[05:59:53] GPU2 | 46°C | Fan 1317 RPM | 896.8 H/s
[05:59:53] GPU3 | 45°C | Fan 1286 RPM | 0.0 H/s
[05:59:55] New job received. Avg Job Time: 31.0 sec
[05:59:55] GPU0 | 47°C | Fan 1329 RPM | 921.7 H/s
[05:59:55] GPU1 | 46°C | Fan 1343 RPM | 921.7 H/s
[05:59:55] GPU2 | 46°C | Fan 1318 RPM | 895.9 H/s
[05:59:55] GPU3 | 46°C | Fan 1288 RPM | 0.0 H/s

can I  do something with this problem? it can be with 1 card or 2 or 3. With the same settings, alternative miners work stably but give 7-9% less
sr. member
Activity: 437
Merit: 250
my cast xmr fast (1950 h/s) but i see very slow on the pool site (1200-1600 h/s) Why?

This is a general thing in mining. The miner counts the hashes (the speed you are searching) while the pool only knows how many valid shares you've sent (so it counts the results).

The two are equivalent only when averaging on a very large time interval
plug it into nicehash. nicehash has nice graphs with average hashrate on longer periods. i woudnt worry about it too much. its not easy tho. we all would love to see highest hashrate on pools all the time
newbie
Activity: 13
Merit: 0
my cast xmr fast (1950 h/s) but i see very slow on the pool site (1200-1600 h/s) Why?

This is a general thing in mining. The miner counts the hashes (the speed you are searching) while the pool only knows how many valid shares you've sent (so it counts the results).

The two are equivalent only when avereaging on a very large time interval
member
Activity: 189
Merit: 10
my cast xmr fast (1950 h/s) but i see very slow on the pool site (1200-1600 h/s) Why?
newbie
Activity: 84
Merit: 0
how many do you have outdated shares because of job change ? I have 7% is this normal or I need try to change pool or something else?
Yes, exactly the same percentage, "job not found" on nicehash.
legendary
Activity: 1510
Merit: 1003
Quote
xmr-stak is open-source software, can you point me the place in source code where it is (fast/slow mode depending on drivers version)?

See above post.  Stak is slower (~1100h/s) with the 17.11.4 driver vs the blockchain driver.

I don't believe that's what they are talking about... Of course the blockchain drivers hash more than 17.11.4

All Red was saying is if you want to play around with the 17.11.4 drivers - XMR-Stak doesn't do a driver check vs Cast which may
Thanks for explanation. I wouldn't do better ))
Jump to: