Pages:
Author

Topic: [Cast XMR] high speed XMR/CryptoNight miner for RX Vega GPUs (2 KHash/s) - page 38. (Read 206424 times)

newbie
Activity: 89
Merit: 0
....MEM alloc error, says computer has no enough memory, pagefile is 60 gb??? What's with this error?
12 gpu's rig

if all your card is 8 GB then you need to set virtual memory 12 * 8 = 96 GB to be safe you also can added 8 GB so it would 104 GB.

Jeez, that much?? Thank you
jr. member
Activity: 269
Merit: 4
I understand it's different algorithm and used more memory, but just don't understand how come RX 580 8 GB can get almost the same speed or sometimes even better but for RX 580 4 GB and RX Vega the speed decrease drastically. maybe someone can explain about this?

My 8gb RX cards get around 200+ hashses more than my 4gb cards do.  Somebody earlier said it's because the algo chokes with only 4gb vram.  I've never owned a Vega, so I can't really speak to it, but I'd suspect it has to do with the mining software you're using and the Devs are still figuring out the Vega cards.

full member
Activity: 325
Merit: 110
Hi, I'm using Cast XMR Version 1.0.0 (2018/4/24) on Vega 56 for Nicehash.


cast_xmr-vega -S cryptonightv7.usa.nicehash.com:3363 -u WALLET_ADDRESS -I 9 --maxmem


My AMD driver is Radeon Adrenalin Edition 18.4.1.

The hashrate is ~1900H, which is OK. But the rejection percentage is too high, almost 40% of the shares are rejected (alert: Share above target).

Is there any solution?

Thank you!



Adding "--algo=1"  helped me in my case

Thanks! this resolved for me too. I've probably lost 30% shares since the fork... lol
only noticed today.
jr. member
Activity: 176
Merit: 2
....MEM alloc error, says computer has no enough memory, pagefile is 60 gb??? What's with this error?
12 gpu's rig

if all your card is 8 GB then you need to set virtual memory 12 * 8 = 96 GB to be safe you also can added 8 GB so it would 104 GB.
jr. member
Activity: 176
Merit: 2
BECAUSE

IT
 
IS

A

DIFFERENT

ALGO

Its called HEAVY because it uses a bigger block of memory(4Mb versus 2Mb) to calculate the hash.

Baz

I understand it's different algorithm and used more memory, but just don't understand how come RX 580 8 GB can get almost the same speed or sometimes even better but for RX 580 4 GB and RX Vega the speed decrease drastically. maybe someone can explain about this?
jr. member
Activity: 75
Merit: 1
BECAUSE

IT
 
IS

A

DIFFERENT

ALGO

Its called HEAVY because it uses a bigger block of memory(4Mb versus 2Mb) to calculate the hash.

Baz
newbie
Activity: 27
Merit: 0
Why do the Vega 56s lose hash power when mining the Cryptonight Heavy Algo?
newbie
Activity: 37
Merit: 0
....MEM alloc error, says computer has no enough memory, pagefile is 60 gb??? What's with this error?
12 gpu's rig

you need to increase your Virtual memory, i have set to 90GB for 6 vega 56, ofcourse its windows
newbie
Activity: 89
Merit: 0
....MEM alloc error, says computer has no enough memory, pagefile is 60 gb??? What's with this error?
12 gpu's rig
newbie
Activity: 23
Merit: 3
hi gandalph, can u add that miner restarts if you get 3 concurrent errors from pool: "unathenticated"? Ratewatchdog doesnt help...

thanks
newbie
Activity: 89
Merit: 0
win10, bitcoin crimson drivers, it only detects 1 gpu, dont see any settings in bat to specify more.

says multi gpu up to more, search on bitcointalk is garbage so i cant find comments fast enough at all. what do i do? i gotta use latest drivers?

and cryptonight7 runs alot better now? new benchmarks for vega 56 and 64 now too maybe?


Use -G 0,1,2,3......as many GPU's you got
newbie
Activity: 2
Merit: 0
Hi, I'm using Cast XMR Version 1.0.0 (2018/4/24) on Vega 56 for Nicehash.


cast_xmr-vega -S cryptonightv7.usa.nicehash.com:3363 -u WALLET_ADDRESS -I 9 --maxmem


My AMD driver is Radeon Adrenalin Edition 18.4.1.

The hashrate is ~1900H, which is OK. But the rejection percentage is too high, almost 40% of the shares are rejected (alert: Share above target).

Is there any solution?

Thank you!



Adding "--algo=1"  helped me in my case

Thank you!!! It helped!
newbie
Activity: 266
Merit: 0
win10, bitcoin crimson drivers, it only detects 1 gpu, dont see any settings in bat to specify more.

says multi gpu up to more, search on bitcointalk is garbage so i cant find comments fast enough at all. what do i do? i gotta use latest drivers?

and cryptonight7 runs alot better now? new benchmarks for vega 56 and 64 now too maybe?
newbie
Activity: 42
Merit: 0
MIner tool supported Cryptonight arto HuhHuh?
jr. member
Activity: 113
Merit: 1
@Gandalph3000
are you busy?
Is there a keepalive settings for castxmr? i cant seem to use --help option for it . Miner is almost perfect just need keepalive, hashrate average on logs, and failover pool settings.
newbie
Activity: 84
Merit: 0
Well, I know at least for one miner where hashrate displayed in the miner is net hashrate, i.e. what you see is what you get, miner switches to dev pool in the background. There are 2 connections all the time, there is not switching to dev pool and back like on most of the miners. As result there is no hashrate drops when miner switches to dev pool and back, what you see in the miner window is what you get on the pool side (average result).
So what I'm asking is: if miner shows that my hashrate is 9650 H/s, does that mean that this is my net hashrate or is it total hashrate from with will be subtracted 1.5% for dev fee so my net hashrate will be 9505 H/s (average)?
full member
Activity: 322
Merit: 101
One quick question: Hashrate displayed in this miner is gross hashrate i.e. total hashrate (including what is subtracted for devfee) or net hashrate i.e. what miner gets?
Dev didn't mentioned this info.

Pretty sure it is net hashrate, as when you think about it when it switches to mine at the devs pool it is not mining for you.
newbie
Activity: 84
Merit: 0
One quick question: Hashrate displayed in this miner is gross hashrate i.e. total hashrate (including what is subtracted for devfee) or net hashrate i.e. what miner gets?
Dev didn't mentioned this info.
newbie
Activity: 33
Merit: 0
Just an update on this....the only algo I am having trouble with is Heavy, and specifically Sumokoin.  This is universal, across all my machines running Cast.  Have tested both with EasyHash and HashVault with same result.  Anyone have a solution to this? 

Geez, really don’t know what to tell you.  I’ve never mined Sumo, but been mining Haven with little problem, and it also uses Heavy.  Been running Cast v 1.0.


I am also mining haven but with 9.6.  I tried the latest version (v 1.0) but was having all sorts of stability problems.  Even upgrading my video drivers to 18.3.4 I still had issues.

I don't want to sound like I am bashing Cast because a) it is "strange" that the Sumokoin pools are generating an error suggesting the you use XMRig or XMR-STAK (but no message of Cast), b) Cast 1.0.0.0 Cryptonightv7 with the --nomem flag is hashing for me at around 2000 KH/s with the 18.3.4 driver, and c) I have, for some time, found Cast to be a very solid program that just runs (and I think the devfee is reasonable).

The above being said, I would like to hear from the Cast development team on the Sumo issue....wouldn't be such a big deal but Sumo is one of the more profitable cryptos right now.

Sumo is cryptonight heavy.  Use the proper flag in your batch file.

I have.  It still doesn't work.  "support for CryptoNight Heavy set --algo=2 to mine" is not working.  --algo=1 does work as to other variants.  --algo=2 does not work.

Getting exactly the same issue myself, sumokoin will not work at all for some reason, algo flag is set, but it gets rejected by the pool stating wrong algo, please use xmrig/xmr-stak instead.

Haven works just fine though with the exact same setup, looks like it is more of an issue with hashvault rather than the miner itself.
newbie
Activity: 2
Merit: 0
Has some strange errors to my rigs too, mostly hang miner (stop mining but not crash), hashdrop, reset v..v.. in ver 1.0. Mining IPBC so i have no choice, may back to 0.9.5 or .6



Same issue here! Using vega rigs and 18.3.4 driver.

It's a shame considering the hashrate improvement.
Pages:
Jump to: