[Cast XMR] high speed XMR/CryptoNight miner for RX Vega GPUs (2 KHash/s)

[huntingthesnark]

Quick idiot question - how do you set failover pools? Thanks in advance!

[Permonik]

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.

[errata]

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.

So, there's no other way except being close to the card and observing it?

[s3r9i0]

Have 7 RX580 GPUs, 8gb RAM

With 1 cards works good, but when us all 7 (-G 0,1,2,3,4,5,6) I see this error message:

CL_MEM_OBJECT_ALLOCATION_FAILURE calling clEnqueueNDRangeKernel for kernel 0

* increasing the Virtual Memory up to 41Gb makes no sense

[Peglar]

Have 7 RX580 GPUs, 8gb RAM

With 1 cards works good, but when us all 7 (-G 0,1,2,3,4,5,6) I see this error message:

CL_MEM_OBJECT_ALLOCATION_FAILURE calling clEnqueueNDRangeKernel for kernel 0

* increasing the Virtual Memory up to 41Gb makes no sense

yet increasing the virtual memory will fix it. i have it set at 60gb on my vega rigs

[th00ber]

Have 7 RX580 GPUs, 8gb RAM

With 1 cards works good, but when us all 7 (-G 0,1,2,3,4,5,6) I see this error message:

CL_MEM_OBJECT_ALLOCATION_FAILURE calling clEnqueueNDRangeKernel for kernel 0

* increasing the Virtual Memory up to 41Gb makes no sense

Who care about Gb lost on a rig ??
Mine is 60Gb and 5 Vega are running well.

[sikotic]

I've scanned this topic but couldn't find an answer for this.

I have 4 Vega 64s that were running NiceHash Cryptonight at the 1900 H/s rate, but since trying to switch the Cast XMR, I keep getting the error about not finding the Compute Drivers.

Yes they are installed.

Win10
16 GB Ram
120 GB SSD (40 GB Virtual Mem)

Any ideas or pointers would be very appreciated. Thanks!

[thilan29]

I've scanned this topic but couldn't find an answer for this.

I have 4 Vega 64s that were running NiceHash Cryptonight at the 1900 H/s rate, but since trying to switch the Cast XMR, I keep getting the error about not finding the Compute Drivers.

Yes they are installed.

Win10
16 GB Ram
120 GB SSD (40 GB Virtual Mem)

Any ideas or pointers would be very appreciated. Thanks!

Which driver are you actually running?
Which miner were you using before?

[nevermind41]

Why don't you put log option to miner? so that it is  easy to find the errors like  te gpu tat gives incorrect share and for many other errors.

[sikotic]

17.30.1029 BlockChain Drivers from Aug 23

The miner was a version of sgminer

[thilan29]

17.30.1029 BlockChain Drivers from Aug 23

The miner was a version of sgminer

Did you try a DDU and reinstall of the blockchain driver?

[sikotic]

No I haven't tried that yet as they have been the only driver on this build and they were just functioning until early this AM when NH went down. But I've seen stranger things work.

[wolverine5pl]

Is it possible to have information about Temp on HBM memory? as GPU temp doesn't tell us what we need to know. That's at least what I have heard

[Permonik]

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.

So, there's no other way except being close to the card and observing it?

You don´t need to be close to the cards, because within the miner you can see the fans rpms.

[entity279]

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 ?

I'm experiencing the same . No ideea how to fix it

[R0mi]

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 ?

I'm experiencing the same . No ideea how to fix it

Visit vegamining blog authored by @Circusdad.  This issue is discussed and a script for XMR-STAK to monitor hash rate and restart is provided; a few posts back in this thread, user StellarX88 explained how to modify that script to work with Cast.  Meaning, once you customize the script as laid out, and based on the number of cards that you have, the script will restart Cast if your hash rates drops below the hash limit that you define.  Not foolproof, but it has been working fairly well.

If your hash rate is dropping every few minutes, then your problem is with bad values for P7 and P3 (OverDriveNTool), or too aggressive settings in XMR-STAK config file.

[eldal]

Guys, who get result with zero hash (0.0 H/s) in last or first card in rig

https://pp.userapi.com/c834204/v834204065/42ef6/X1idS_wuzHo.jpg

Just random shuffle order -G parameters

E.g. "-G 0,1,2,3,4" to "-G 4,3,0,1,2" or similar

It works to me with RX580 (4 or 5 in rig) (Windows 10, 64bit, AMD Blockchain driver)

Thank you so much for this.  I used 1,0 and it started working for my 2 cards.

Address Question.  Are we to use the plain monero address or integrated address?  I put plain. Hope that is right?

[errata]

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.

So, there's no other way except being close to the card and observing it?

You don´t need to be close to the cards, because within the miner you can see the fans rpms.

Ah yes, of course :) me dumb ^^ thanks!

[nevermind41]

Gpu5 mines slow then the other cards in my 2 rigs. I couldn't find solution about this issue.Both of the rigs have 6 cards and they mine with too low hashes.
Edit: I foun the problem. Overdriventool loaded the profiles wrong with batch file.

[uDwcHYO]

who says that zeros in hashes are not a problem, so this is the problem because the video card does not work, judging by the drop in the load, the temperature and the total hash. There is an assumption that this depends on decrease voltage and core frequency change.