[Cast XMR] high speed XMR/CryptoNight miner for RX Vega GPUs (2 KHash/s) - page 11.

Yerba

member

Activity: 146

Merit: 10

Quote from: glph3k on October 18, 2018, 09:08:45 AM

Quote from: ingyenfrag on October 18, 2018, 08:51:46 AM

Quote from: glph3k on October 18, 2018, 08:49:34 AM

Quote from: huntingthesnark on October 18, 2018, 08:44:17 AM

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

This is a known problem with the blockchain drivers. Update to 18.5.1, they are more stable and don't need the hardware reset trick and will fix performance with V8. HBCC should be turned off. Vega64 can go with intensity 9 if used headless (no desktop running on it),

18.5.1 exactly or any newer one?

Can recommend 18.5.1 .

Later ones (including 18.10.1) only run with lower intensity settings if a desktop is running on the card. But try for yourself.

Any real information, that with 18.5.1 hashrate drops only 10%, not from 2000 to 600?

ingyenfrag

newbie

Activity: 84

Merit: 0

Quote from: glph3k on October 18, 2018, 09:08:45 AM

Quote from: ingyenfrag on October 18, 2018, 08:51:46 AM

Quote from: glph3k on October 18, 2018, 08:49:34 AM

Quote from: huntingthesnark on October 18, 2018, 08:44:17 AM

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

This is a known problem with the blockchain drivers. Update to 18.5.1, they are more stable and don't need the hardware reset trick and will fix performance with V8. HBCC should be turned off. Vega64 can go with intensity 9 if used headless (no desktop running on it),

18.5.1 exactly or any newer one?

Can recommend 18.5.1 .

Later ones (including 18.10.1) only run with lower intensity settings if a desktop is running on the card. But try for yourself.

I believe you.

Until today I used the blockchain. A few month ago I've tried the 18.7.1 but it was a little bit slower but at least the consumption was a little bit better.

glph3k

member

Activity: 91

Merit: 29

Quote from: ingyenfrag on October 18, 2018, 08:51:46 AM

Quote from: glph3k on October 18, 2018, 08:49:34 AM

Quote from: huntingthesnark on October 18, 2018, 08:44:17 AM

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

This is a known problem with the blockchain drivers. Update to 18.5.1, they are more stable and don't need the hardware reset trick and will fix performance with V8. HBCC should be turned off. Vega64 can go with intensity 9 if used headless (no desktop running on it),

18.5.1 exactly or any newer one?

Can recommend 18.5.1 .

Later ones (including 18.10.1) only run with lower intensity settings if a desktop is running on the card. But try for yourself.

ingyenfrag

newbie

Activity: 84

Merit: 0

Quote from: glph3k on October 18, 2018, 08:49:34 AM

Quote from: huntingthesnark on October 18, 2018, 08:44:17 AM

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

This is a known problem with the blockchain drivers. Update to 18.5.1, they are more stable and don't need the hardware reset trick and will fix performance with V8. HBCC should be turned off. Vega64 can go with intensity 9 if used headless (no desktop running on it),

18.5.1 exactly or any newer one?

ingyenfrag

newbie

Activity: 84

Merit: 0

Quote from: huntingthesnark on October 18, 2018, 08:44:17 AM

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

Even with a new driver?

With blockchain driver my 5 card vega 56 rig dropped from 9750 H/s to ~3800 H/s.

glph3k

member

Activity: 91

Merit: 29

Quote from: huntingthesnark on October 18, 2018, 08:44:17 AM

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

This is a known problem with the blockchain drivers. Update to 18.5.1, they are more stable and don't need the hardware reset trick and will fix performance with V8. HBCC should be turned off. Vega64 can go with intensity 9 if used headless (no desktop running on it),

huntingthesnark

member

Activity: 357

Merit: 26

Hashrate on cnv8 has cratered. Vega 64s/blockchain drivers/HBCC enabled. intensity 7. hashing at under 1k...

ciki022

newbie

Activity: 26

Merit: 0

Hi,

I am trying to connect to nanopool, but I am receiving this :

Blockchain Compute Driver detected.
GPU0: Radeon (TM) RX 470 Graphics (Ellesmere) | 32 Compute Units | Memory (MB): 4096 | Intensity: 8 / 10
GPU1: Radeon (TM) RX 470 Graphics (Ellesmere) | 32 Compute Units | Memory (MB): 4096 | Intensity: 8 / 10
GPU2: Radeon (TM) RX 470 Graphics (Ellesmere) | 32 Compute Units | Memory (MB): 4096 | Intensity: 8 / 10
GPU3: Radeon (TM) RX 470 Graphics (Ellesmere) | 32 Compute Units | Memory (MB): 4096 | Intensity: 8 / 10
GPU4: Radeon (TM) RX 470 Graphics (Ellesmere) | 32 Compute Units | Memory (MB): 4096 | Intensity: 8 / 10
[12:16:49] Connecting to xmr-eu1.nanopool.org:14433 ...
[12:16:49] Connected to pool.
[12:16:49] Socket Error - Recv Error: socket closed
[12:16:49] Connecting to Pool failed. Retrying in 20 secs ...
[12:17:09] Connecting to xmr-eu1.nanopool.org:14433 ...
[12:17:09] Connected to pool.
[12:17:09] Socket Error - Recv Error: socket closed
[12:17:09] Connecting to Pool failed. Retrying in 20 secs ...

What causing this ?

mattstyle

newbie

Activity: 31

Merit: 0

Quote from: rainbowfcc on October 18, 2018, 01:31:08 AM

which one is more profitable for vega 56 cards now? cryptonight v7 or v8? Seems like v7 is still more profitable now to me..

BR,

umm wait 5-6 hrs.. v8 fork happens at block height 1685555

rainbowfcc

newbie

Activity: 10

Merit: 0

which one is more profitable for vega 56 cards now? cryptonight v7 or v8? Seems like v7 is still more profitable now to me..

BR,

mattstyle

newbie

Activity: 31

Merit: 0

do you have to restart the drivers still? running 18.5.1

getting 9.1k with 6 vega 56 -- 1.55-1.6k each.. v8
intensity 10 and 8

over 10k on v7
intensity 8

1375mhz/900 and 925 mhx mem / 900v

9.2k on the other rig.. 5 airboost 56 and one amd reference 64..
pool shows 10k on both rigs..

glph3k

member

Activity: 91

Merit: 29

Quote from: alucard20724 on October 17, 2018, 09:47:26 AM

cnv8 available on nicehash.

for 1.5.0, i'm getting for by vega56 @1.9k hash on cnv7, while only getting around @1k hash for cnv8. win10, hbcc enabled, 18.10.1 drivers.

Try lowering the intensity to 7 or 6. HBCC should be turned off it saves memory. Would recommend 18.5.1 drivers for mining.1.8k cnv8 hashes/s are doable on a Vega 56.

alucard20724

sr. member

Activity: 703

Merit: 272

cnv8 available on nicehash.

for 1.5.0, i'm getting for by vega56 @1.9k hash on cnv7, while only getting around @1k hash for cnv8. win10, hbcc enabled, 18.10.1 drivers.

ntshb

newbie

Activity: 95

Merit: 0

after the fork, would anyone still mine cnv7? we should all switch to cnv8 right?

NinaJk221

newbie

Activity: 4

Merit: 0

want to know about the software and the OS version.

carlosmonaco

newbie

Activity: 105

Merit: 0

Quote from: glph3k on October 16, 2018, 08:04:18 AM

Quote from: carlosmonaco on October 15, 2018, 04:56:50 AM

Quote from: carlosmonaco on September 10, 2018, 10:27:24 AM

Quote from: unsivilaudio on September 08, 2018, 11:12:59 AM

Don't know if its an issues on my end, but I noticed GPU numbers are being incorrectly assigned. I currently have 5 vega 56s, HWInfo shows them 0,1,2,3,4 --> 0,1,4,3,2 in Cast, I was hoping I could reorder the gpu numbers by reordering the -G numbers, but this is not the case.

Also I noticed RX580s are having issues being erroneously detected as not in compute mode and also having a low hashrate because of it, RX570s appear to be fine.

Adrenaline 18.5.1 driver.

Same issues with RX580 Nitro + , SWitching mode fail , but even after manually switch ON , detected as not in compute mode and low hashrate.
--> back to 1.0 with --forcecompute for now

I still have this issue, I tried latest driver but same issue, still being erroneously detected as not in compute mode. Only Cast 1.0 working with forcecompute.
Any solution?

Use the --intensity option to override the automatic detection. One can set the intensity for each card by separating it by comma in the order the GPUs are specified Like --intensity=7,8,10,7

Try --intensity=7 it should have the same effect as the legacy --forcecompute option.

Thanks, I just tried --intensity=7 with cast 1.5, I still get "Compute Mode Not Enabled" error but hashrate is better. It's still a little slower than CAST but only few h/s

glph3k

member

Activity: 91

Merit: 29

Quote from: carlosmonaco on October 15, 2018, 04:56:50 AM

Quote from: carlosmonaco on September 10, 2018, 10:27:24 AM

Quote from: unsivilaudio on September 08, 2018, 11:12:59 AM

Don't know if its an issues on my end, but I noticed GPU numbers are being incorrectly assigned. I currently have 5 vega 56s, HWInfo shows them 0,1,2,3,4 --> 0,1,4,3,2 in Cast, I was hoping I could reorder the gpu numbers by reordering the -G numbers, but this is not the case.

Also I noticed RX580s are having issues being erroneously detected as not in compute mode and also having a low hashrate because of it, RX570s appear to be fine.

Adrenaline 18.5.1 driver.

Same issues with RX580 Nitro + , SWitching mode fail , but even after manually switch ON , detected as not in compute mode and low hashrate.
--> back to 1.0 with --forcecompute for now

I still have this issue, I tried latest driver but same issue, still being erroneously detected as not in compute mode. Only Cast 1.0 working with forcecompute.
Any solution?

Use the --intensity option to override the automatic detection. One can set the intensity for each card by separating it by comma in the order the GPUs are specified Like --intensity=7,8,10,7

Try --intensity=7 it should have the same effect as the legacy --forcecompute option.

carlosmonaco

newbie

Activity: 105

Merit: 0

Quote from: carlosmonaco on September 10, 2018, 10:27:24 AM

Quote from: unsivilaudio on September 08, 2018, 11:12:59 AM

Don't know if its an issues on my end, but I noticed GPU numbers are being incorrectly assigned. I currently have 5 vega 56s, HWInfo shows them 0,1,2,3,4 --> 0,1,4,3,2 in Cast, I was hoping I could reorder the gpu numbers by reordering the -G numbers, but this is not the case.

Also I noticed RX580s are having issues being erroneously detected as not in compute mode and also having a low hashrate because of it, RX570s appear to be fine.

Adrenaline 18.5.1 driver.

Same issues with RX580 Nitro + , SWitching mode fail , but even after manually switch ON , detected as not in compute mode and low hashrate.
--> back to 1.0 with --forcecompute for now

I still have this issue, I tried latest driver but same issue, still being erroneously detected as not in compute mode. Only Cast 1.0 working with forcecompute.
Any solution?

rainbowfcc

newbie

Activity: 10

Merit: 0

Quote from: glph3k on October 14, 2018, 05:35:56 PM

Quote from: rainbowfcc on October 14, 2018, 10:54:06 AM

Quote from: glph3k on October 14, 2018, 10:29:13 AM

Quote from: rainbowfcc on October 14, 2018, 08:45:20 AM

I have been using all cast xmr versions for 1 year.. and i did not have any issue since v8 came up.. now i have installed cast xmr 1.5 .. when i get v7 requests , my rig does it without any problem.. but when i get sometimes v8, my rig is running 500-600 hash rate per card.. i have 5 vega 56.. whats wrong with v8? do you think i should add -1 switch?

On official site of cast xmr saying that there will be a %10 decrease but this is more than %70 decrease..

BR,

Hi,

this is a known issue (related to some mainboard chipsets?). The cure seems to install the AMD drivers version 18.5.1 (if that does not help version 18.6.1.).

Independent form the hash rate drop. Please use the "--algo=-1" switch or leave the algo setting out completely to make the automatic switch from CryptoNighV7 to V8.

Regards,
glph3k

I have just installed 18.5.1 adrenaline driver. Now my hashrate fixed to 1800 for v8 . Blockchain driver was causing the decrease of hashrate! . But now i am receiving for all v8 shares rejected . Reason: Share above target. I dont have any problem with v7. Is this a driver related issue? I can install Adrenalin Edition 18.10.1 Optional if you think it helps to get rid off "share above target" issue..

Thank you very much for your quick response..

BR,

Note: Soon i will install Adrenalin Edition 18.10.1 Optional and will inform you...

Note v2: seemslike it is a beta driver.. i had other stability issues.. and than back to 18.5.1.. now v7 is ok .. same issue for v8.. share above target..

I would also recommend 18.5.1. Later drivers are not that stable or have a lower performance.

Did you test V8 with a testnet pool like http://killallasics.moneroworld.com/ ? Submitting V8 shares to the current chain will give "share above target" error.

Yes .. you are right.. Now i am using 18.5.1.. and it is stable.. i fixed my rig with +1 parameter to v7... now i am not receiving any v8 share.. and its going very smooth.. tonight i will test using the test pool..

glph3k

member

Activity: 91

Merit: 29

Quote from: rainbowfcc on October 14, 2018, 10:54:06 AM

Quote from: glph3k on October 14, 2018, 10:29:13 AM

Quote from: rainbowfcc on October 14, 2018, 08:45:20 AM

I have been using all cast xmr versions for 1 year.. and i did not have any issue since v8 came up.. now i have installed cast xmr 1.5 .. when i get v7 requests , my rig does it without any problem.. but when i get sometimes v8, my rig is running 500-600 hash rate per card.. i have 5 vega 56.. whats wrong with v8? do you think i should add -1 switch?

On official site of cast xmr saying that there will be a %10 decrease but this is more than %70 decrease..

BR,

Hi,

this is a known issue (related to some mainboard chipsets?). The cure seems to install the AMD drivers version 18.5.1 (if that does not help version 18.6.1.).

Independent form the hash rate drop. Please use the "--algo=-1" switch or leave the algo setting out completely to make the automatic switch from CryptoNighV7 to V8.

Regards,
glph3k

I have just installed 18.5.1 adrenaline driver. Now my hashrate fixed to 1800 for v8 . Blockchain driver was causing the decrease of hashrate! . But now i am receiving for all v8 shares rejected . Reason: Share above target. I dont have any problem with v7. Is this a driver related issue? I can install Adrenalin Edition 18.10.1 Optional if you think it helps to get rid off "share above target" issue..

Thank you very much for your quick response..

BR,

Note: Soon i will install Adrenalin Edition 18.10.1 Optional and will inform you...

Note v2: seemslike it is a beta driver.. i had other stability issues.. and than back to 18.5.1.. now v7 is ok .. same issue for v8.. share above target..

I would also recommend 18.5.1. Later drivers are not that stable or have a lower performance.

Did you test V8 with a testnet pool like http://killallasics.moneroworld.com/ ? Submitting V8 shares to the current chain will give "share above target" error.

Topic: [Cast XMR] high speed XMR/CryptoNight miner for RX Vega GPUs (2 KHash/s) - page 11. (Read 206352 times)