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

[amanda89]

this new ver cast xmr really not stable for cn heavy . I'm using blockchain driver + vega 56 , it crashed alot when I'm using --ratewatchdog or --fastjobswitch mode maybe (idk which one but I'm pretty sure that two command really caused alot problem to my rigs)   after a few hours later, hash drop half & sometimes my rigs getting freeze . 

[dragonmike]

I'm getting just under 800h/s per RX570 clocked at 1220/1900 with the following command line using 0.9.6:

cast_xmr-vega -I 0 -S europe.cryptonight-hub.miningpoolhub.com:17024 -u xxx.yyy -p x -R --nonicehash --fastjobswitch -G 0,1,2,3,4,5,6

Anything I could do to improve? Setting --forcecompute just cuts the hashrate in half...

Hi,

which driver version are you using?

Unless you have the blockchain driver installed:

 --forcecompute is the way to go to get the extra hashes. Are you sure that all cards are switched to "Compute" mode in the Radeon settings?


Regards,
glph3k

I'm using 18.3.4.
If I toggle --forcecompute in the miner, hashrates gets cut in half... I have not installed Radeon Settings (as it just clogs up the miner and tends to crash more than anything else), so I overclock using Overdriventool. I'll try forcing compute mode in registry. Treanski shared the procedure in another thread. I did not realise I would have to do it again after I updated the drivers... Let me do it and I'll report in here.

Ok, so I created the new DWORD in registry to be able to enable compute mode... as per Treanski's instructions:

add this DWORD to the below registry path: KMD_EnableInternalLargePage, set value to 2 to enable compute mode

"HKLM\SYSTEM\ControlSet001\Control\Class\{4d36e968-e325-11ce-bfc1-08002be10318}\XXXX"
where "XXXX" is the card number.

...rebooted, and added --forcecompute in the miner and wham! Instant hashrate boost. I now average 900h/s per RX 570. Fantastic!

[AlexanderDumas]

Hi anyone having issues like this with a 4x Vega 56 Build    :

10:38:12] GPU1 | 56°C | Fan 4429 RPM | 1886.6 H/s
[10:38:12] GPU0 | 53°C | Fan 4319 RPM | 62.5 H/s
[10:38:17] GPU1 | 56°C | Fan 4437 RPM | 1884.0 H/s
[10:38:23] GPU1 | 56°C | Fan 4439 RPM | 1886.0 H/s

[Pool: 'cryptonightv7.usa.nicehash.com:3363' | Connected: 2018-04-19 06:14:00]
  19:26:55 (100%) Online
  0:00:40 (  0%) Offline (2 Reconnects)

[Job: #168 | Difficulty: 100001 | Running: 194.0 sec | Avg Job Time: 418.1 sec]

[Hash Rate Avg: 4268.2 H/s]
 816.2 H/s GPU0
  1874.2 H/s GPU1
  790.3 H/s GPU2
  787.6 H/s GPU3

[Shares Found: 1576 | Avg Search Time: 44.4 sec]
1576 (100%) Accepted
0 (  0%) Rejected by pool
0 (  0%) Invalid result computation failed
0 (  0%) Could not be submitted because of network error
0 (  0%) Outdated because of job change

[10:38:29] GPU1 | 56°C | Fan 4428 RPM | 1885.0 H/s
[10:38:34] GPU1 | 56°C | Fan 4445 RPM | 1885.7 H/s


After some time, the hash rate just drops drastically

Thank you

It may seems counterintuitive, but removing the ratewatchdog command resolved the problem for me.
I also had those low 60hs rate per cards showing up, but removing the "watchdog" eliminated the problem for good.

[rpinola]

There's no API to monitor this miner, right?

I would prefer this miner, but need the monitoring to work (checking if it's dead or not). How would you do it?

[dragonmike]

There's no API to monitor this miner, right?

I would prefer this miner, but need the monitoring to work (checking if it's dead or not). How would you do it?

Add "-R" to your command line and access the API by the ip address of the rig and port 7777, eg. 192.168.1.10:7777

[duyquang06]

We love ur tool and don't want to switch any other tool. Need update for IPBC and TRTL algo

[RAJNIFICENT]

It may seems counterintuitive, but removing the ratewatchdog command resolved the problem for me.
I also had those low 60hs rate per cards showing up, but removing the "watchdog" eliminated the problem for good.

When I remove it, my hash rate drops after 1-2 hours from 18XX to 15 or 16 XX 

Ratewatch is crucial for me 

[AlexanderDumas]

It may seems counterintuitive, but removing the ratewatchdog command resolved the problem for me.
I also had those low 60hs rate per cards showing up, but removing the "watchdog" eliminated the problem for good.

When I remove it, my hash rate drops after 1-2 hours from 18XX to 15 or 16 XX 

Ratewatch is crucial for me 

Try this, unless you already did it:
Check if in your MotherBoard BIOS setting for the PCIe slots are set on GEN1 or GEN2, instead of auto.
GEN1 or GEN2 are the appropriate settings for mining.

[Poseidon_]

Still not working here at sumokoin, criptonight heavy. I tried many times mines with so many ways... and ever and ever crashed and got lots of erros

One way work here and just with the rx 580 but it gets only 443h/s... but toke just 443h/s is the same as not working

Btw I hope that the new update will be good

[polofitted007]

Hi,

try the option "--ratewatchdog" to monitor the hash rate and automatic restart of a dropping GPU.

Or use the cast-xmr-ui for monitoring:

https://github.com/anadventureisu/cast-xmr-ui

Btw. are you using the Blockchain Beta or 18.3.4 driver?

Regards,
glph3k

I'm using --ratewatchdog and Blockchain Beta From August

It seems like after sometime if watchdog kicks in, the hash rate drops lower.  It doesn't appear as if my graphics driver crashed, because normally it will reset all of my OC settings.  Which also I tried to adjust OC settings however I'm not having much luck.

In the above sample I posted, each card will start off by doing roughly 1870Hs, but after a few kernel resets I guess over the course of the day, some of the cards just become obsolete.  Maybe ratewatchdog is not fully bringing the vega to it's potential.  When the hashrate drops, I also notice that particular card no longer reports hashes that often.  As you can tell by that small insert, GPU 1 started reporting hashes but 0 2 3 slowly fell off.  So it's interesting!

OC is 0 Core 905 Mem and -20% power

Thank you for any advice you have.

UPDATE:  I was able to watch this happen today and I wanted to share the log

[19:04:10] GPU2 | 48°C | Fan 4229 RPM | 1896.3 H/s
[19:04:11] GPU0 | 56°C | Fan 4634 RPM | 1863.8 H/s
[19:04:11] GPU3 Kernel Resetted.
[19:04:11] GPU3 | 54°C | Fan 4328 RPM | 86.0 H/s
[19:04:14] GPU1 | 55°C | Fan 4342 RPM | 1880.0 H/s
[19:04:16] GPU2 | 48°C | Fan 4219 RPM | 1892.3 H/s
[19:04:16] GPU0 | 56°C | Fan 4636 RPM | 1861.8 H/s
[19:04:20] GPU1 | 55°C | Fan 4352 RPM | 1883.0 H/s
[19:04:21] GPU2 | 51°C | Fan 4180 RPM | 1894.6 H/s
[19:04:22] GPU0 | 56°C | Fan 4636 RPM | 1862.8 H/s
[19:04:26] GPU1 | 55°C | Fan 4352 RPM | 1879.7 H/s
[19:04:27] GPU2 | 51°C | Fan 4178 RPM | 1894.3 H/s
[19:04:28] GPU0 | 56°C | Fan 4650 RPM | 1864.7 H/s
[19:04:31] GPU1 | 55°C | Fan 4363 RPM | 1880.7 H/s
[19:04:33] GPU2 | 51°C | Fan 4178 RPM | 1895.0 H/s
[19:04:34] GPU0 | 56°C | Fan 4632 RPM | 1865.0 H/s
[19:04:37] GPU1 | 55°C | Fan 4362 RPM | 1881.0 H/s
[19:04:39] GPU2 | 52°C | Fan 4169 RPM | 1895.3 H/s
[19:04:39] GPU0 | 56°C | Fan 4629 RPM | 1864.7 H/s
[19:04:43] GPU1 | 55°C | Fan 4373 RPM | 1881.4 H/s
[19:04:44] GPU2 | 52°C | Fan 4144 RPM | 1894.6 H/s
[19:04:45] GPU0 | 56°C | Fan 4620 RPM | 1865.0 H/s
[19:04:48] GPU1 | 55°C | Fan 4368 RPM | 1883.0 H/s
[19:04:50] GPU2 | 49°C | Fan 4274 RPM | 1893.0 H/s
[19:04:51] GPU0 | 57°C | Fan 4631 RPM | 1865.4 H/s
[19:04:54] GPU1 | 55°C | Fan 4356 RPM | 1883.0 H/s
[19:04:56] GPU2 | 49°C | Fan 4273 RPM | 1893.0 H/s
[19:04:57] GPU0 | 56°C | Fan 4630 RPM | 1865.0 H/s
[19:05:00] GPU1 | 55°C | Fan 4362 RPM | 1883.7 H/s
[19:05:01] GPU2 | 52°C | Fan 4279 RPM | 1884.7 H/s
[19:05:03] GPU0 | 56°C | Fan 4636 RPM | 1860.9 H/s
[19:05:06] GPU1 | 55°C | Fan 4363 RPM | 1871.2 H/s
[19:05:07] GPU2 | 52°C | Fan 4278 RPM | 1886.0 H/s
[19:05:08] GPU0 | 57°C | Fan 4669 RPM | 1866.7 H/s
[19:05:11] GPU1 | 55°C | Fan 4358 RPM | 1881.7 H/s
[19:05:11] GPU3 | 54°C | Fan 4301 RPM | 59.7 H/s
[19:05:13] GPU2 | 49°C | Fan 4249 RPM | 1900.7 H/s
[19:05:14] GPU0 | 56°C | Fan 4636 RPM | 1862.1 H/s

[Pool: 'cryptonightv7.usa.nicehash.com:3363' | Connected: 2018-04-19 12:02:04]
  7:03:11 (100%) Online
  0:00:00 (  0%) Offline

[Job: #446 | Difficulty: 400015 | Running: 95.0 sec | Avg Job Time: 56.8 sec]

[Hash Rate Avg: 7443.7 H/s]
  1852.4 H/s GPU0
  1867.1 H/s GPU1
  1879.9 H/s GPU2
  1844.3 H/s GPU3

[Shares Found: 621 | Avg Search Time: 40.8 sec]
620 (100%) Accepted
1 (  0%) Rejected by pool
0 (  0%) Invalid result computation failed
0 (  0%) Could not be submitted because of network error
0 (  0%) Outdated because of job change

[19:05:17] GPU1 | 55°C | Fan 4336 RPM | 1882.0 H/s
[19:05:18] GPU2 | 50°C | Fan 4244 RPM | 1896.6 H/s
[19:05:20] GPU0 | 57°C | Fan 4623 RPM | 1873.5 H/s
[19:05:23] GPU1 | 55°C | Fan 4341 RPM | 1881.7 H/s
[19:05:24] GPU2 | 50°C | Fan 4216 RPM | 1894.6 H/s
[19:05:26] GPU0 | 57°C | Fan 4618 RPM | 1864.4 H/s
[19:05:28] GPU1 | 55°C | Fan 4329 RPM | 1881.0 H/s
[19:05:30] GPU2 | 52°C | Fan 4185 RPM | 1889.6 H/s
[19:05:31] GPU0 | 56°C | Fan 4605 RPM | 1866.7 H/s
[19:05:34] GPU1 | 55°C | Fan 4322 RPM | 1878.1 H/s
[19:05:35] GPU2 | 52°C | Fan 4171 RPM | 1890.6 H/s
[19:05:37] GPU0 | 55°C | Fan 4615 RPM | 1858.9 H/s
[19:05:40] GPU1 | 55°C | Fan 4317 RPM | 1881.4 H/s
[19:05:41] GPU2 | 49°C | Fan 4172 RPM | 1893.3 H/s

Now after GPU 3 was reset > it dropped to 56 hashes and it's no longer showing up like GPU 0 1 2  (3 the one which was just reset is now MIA)

Any update here, same thing just started happening to me.  4x Vega 56s.  two running at 17* and the other two running normal.  Any help?

[citronick]

glph3k,

Any recommendations on which is the preferred driver for CAST XMR miner for Vega64s ?

Latest 2018 Blockchain driver or 18.3.4 driver ?

Testing..
Rig 1: 4 x Vega64
Rig 2: 8 x Vega64
Rig 3: 2 x Vega56

Thanks

[P00P135]

glph3k,

Any recommendations on which is the preferred driver for CAST XMR miner for Vega64s ?

Latest 2018 Blockchain driver or 18.3.4 driver ?

Testing..
Rig 1: 4 x Vega64
Rig 2: 8 x Vega64
Rig 3: 2 x Vega56

Thanks

I just made the switch from blockchain to new adrenaline.  Only issue i had was finding a soft power play table for one of my AIB 56's.  Everything running the same but saving around 10w a GPU over blockchain drivers.

[Rechit]

[13:42:42] Difficulty changed. Now: 150000.
[13:42:46] GPU1 | 27°C | Fan 2373 RPM | 1778.2 H/s
[13:42:47] GPU2 | 27°C | Fan 2324 RPM | 1714.4 H/s
[13:42:48] GPU0 | 25°C | Fan 2460 RPM | 1868.9 H/s
[13:42:52] GPU1 | 31°C | Fan 2557 RPM | 1919.7 H/s
[13:42:52] GPU2 | 34°C | Fan 2510 RPM | 1868.0 H/s
[13:42:54] GPU0 | 28°C | Fan 2630 RPM | 1975.7 H/s
[13:42:58] GPU1 | 35°C | Fan 2736 RPM | 1918.6 H/s
[13:42:58] GPU2 | 38°C | Fan 2684 RPM | 1893.3 H/s
[13:42:59] GPU0 | 31°C | Fan 2795 RPM | 1975.0 H/s
[13:43:03] GPU1 | 37°C | Fan 2907 RPM | 1916.6 H/s
[13:43:04] GPU2 | 43°C | Fan 2861 RPM | 1882.0 H/s
[13:43:04] GPU0 | 33°C | Fan 2959 RPM | 1974.7 H/s
[13:43:09] GPU1 | 40°C | Fan 3069 RPM | 1916.6 H/s
[13:43:09] GPU2 | 46°C | Fan 3027 RPM | 1896.0 H/s
[13:43:10] GPU0 | 34°C | Fan 3130 RPM | 1973.9 H/s
[13:43:14] GPU1 | 42°C | Fan 3223 RPM | 1915.6 H/s
[13:43:15] GPU2 | 49°C | Fan 3196 RPM | 1888.3 H/s
[13:43:15] GPU0 | 35°C | Fan 3279 RPM | 1974.3 H/s
[13:43:19] GPU0 Found Nonce, submitting...
[13:43:20] !Rejected by pool! Reason: Wrong algo, use monero7 miner
[13:43:20] Shares: 0 Accepted, 1 Errors | Hash Rate Avg: 5718.8 H/s
[13:43:20] GPU1 | 42°C | Fan 3391 RPM | 1909.8 H/s
[13:43:21] GPU0 | 37°C | Fan 3423 RPM | 1974.3 H/s
[13:43:21] GPU2 | 52°C | Fan 3363 RPM | 1807.7 H/s


cast_xmr-vega-win64_096
driver blockchain
Hashing Algorithm: Cryptonight-heavy with Haven modification

Immediately after starting the error, why?

[optii]

i got vega 56 and vega 64. When i upgrade to 9.6 from 9.5 , mining on vega 56 is fine, but on vega 64, HS drops from 2000 to 1600. mining monero, with watchdog, power tables etc. and i have latest drivers, hbcc disabled

[coke15]

hi

something goes wrong with 0.95 and 0.96.
BC drivers,soft table reg

i don't know why,cast stop hashing ,maybe 20min maybe 10hours.

cast not crashing,just stop hashing :/

everything fine with 0.92...for weeks....

[RAJNIFICENT]

Any update here, same thing just started happening to me.  4x Vega 56s.  two running at 17* and the other two running normal.  Any help?

I solved my problem by using an auto mouse and keyboard macro recorder.  I set it to Q(quit) out of Cast XMR every 6 hours give it a 1 minute breather and restart it.  Every 6 hours something happens to my Cast so rather than figure it out, I'm just restarting it right before it starts getting flaky.

To be fair, this is nothing against Cast.  I have a 13 card rig and I'm mixing Nvidia and AMD together.  Although on older versions I never had issues WITHOUT the --ratewatch, I'm looking forward to further iterations of that command.

Still love the project, hope to see more updates!

[P00P135]

glph3k,

Any recommendations on which is the preferred driver for CAST XMR miner for Vega64s ?

Latest 2018 Blockchain driver or 18.3.4 driver ?

Testing..
Rig 1: 4 x Vega64
Rig 2: 8 x Vega64
Rig 3: 2 x Vega56

Thanks

I just made the switch from blockchain to new adrenaline.  Only issue i had was finding a soft power play table for one of my AIB 56's.  Everything running the same but saving around 10w a GPU over blockchain drivers.

Can you report back if, in the future, you get a crash and Vegas start disappearing of the PCIe bus?  I never experienced this with the old August blockchain drivers, but I have seen this with the new Adrenaline drivers.  I am wondering whether it is a motherboard/bios issue or a driver issue.

Thanks.

I havent had this happen since switching, but I did have that happen once on blockchain drivers. It was a strange windows hang that required a power cycle of my server PSU's.  I think its related to not shutting down Cast-xmr properly with the Q command.  I'll let you know if i run into it again on adrenaline drivers.

[almost30]

Rejected by pool 50%     why  

Using Nicehash  Cryptonightv7 algo

[Radman07]

Just tried out your program today....so far the best performance I have seen with any mining software I have used.  I do have some questions....
1)  Can't get the help function to work.  I have tried --help command and i can't get it to work.  Can you put out a separate message with all of the line commands for this version?
2)  Is there a way to have the temperature and fan speeds remain on one part of the screen?  It's pretty annoying to see scrolling lines of temperature and fan speeds.
3)  From your latest message I gather there is no performance advantage to update from the blockchain drivers (17.30.1029)?

Thanks for your time...... 

[Radman07]

Just tried out your program today....so far the best performance I have seen with any mining software I have used.  I do have some questions....
1)  Can't get the help function to work.  I have tried --help command and i can't get it to work.  Can you put out a separate message with all of the line commands for this version?
2)  Is there a way to have the temperature and fan speeds remain on one part of the screen?  It's pretty annoying to see scrolling lines of temperature and fan speeds.
3)  From your latest message I gather there is no performance advantage to update from the blockchain drivers (17.30.1029)?

Thanks for your time...... 

Just for completeness I'm running seven RX470 and one RX570 GPUs.  Getting about 6.4 kH/s on ~7A (~800W measured at wall).  The RX470's are getting about 820 kH/s and the RX570 is getting about 700 KH/s....