Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 1390. (Read 6590215 times)

Ran v2 for 4-5 hours, rig 4 x R9-390 (stock settings), AMD 8-core CPU, 16GB RAM

Run command: "EthDcrMiner64.exe -epool asia1.ethermine.org:4444 -ewal 0x... -epsw x -dpool stratum+tcp://mine.zpool.ca:5744 -dwal 1... -dpsw c=BTC,decred"

POOL: ethermine.org
ETH - Total Speed: 117.066 Mh/s, Total Shares: 534, Rejected: 0, Time: 05:16
ETH: GPU0 29.388 Mh/s, GPU1 29.162 Mh/s, GPU2 29.239 Mh/s, GPU3 29.277 Mh/s

POOL: zpool.ca
DCR - Total Speed: 1755.988 Mh/s, Total Shares: 892, Rejected: 76
DCR: GPU0 440.823 Mh/s, GPU1 437.435 Mh/s, GPU2 438.581 Mh/s, GPU3 439.149 Mh/s

Summary:
Compared to previous config. running qtminer (ETH) and sgminer (DCR) in parallel.
1. ETH: averages 59-61Mhs total. DCR: 3.1 Gh/s
2. v2 delivers 2x more hashing so this is interesting and good.
3. However, in ethermine worker dashboard, hash info is consistent but no shares and invalid shares reported.
3b. Shares seems to be registered (unconfirmed) but the dashboard doesnt show it at all.
3c. Is my syntax used above correct? I have used plain vanilla settings without switches.
4. DCR: averages 1.756 Gh/s which is lower than sgminer's 3.1 Gh/s @ diff level 12
5. Reject rate is also quite high at 8.5% compared to sgminer 1.7%
6. During the run, 2 cpuminer instance were mining argon2 and yescrypt. They are using at least 4 threads each. I dont think this will impact v2 run since they dont touch the GPUs and vRAM.

After a few hours, I am at <1% DCR rejects on stratum.  No change to ethash output.  Big improvement from 20-30% rejects with http getwork.

I'm using this cmdline with the new version:


After 3 hours, my ETH shares decreased about 5% but my DCR efficiency is now at above 98%.

So, the results are promising and I'll give this new version some more hours to test it's stability.

Note: I'm using 6 270X, a 7870 and two 7950s, thus a mix of Pitcairn and Tahiti cards.

@Claymore: I'm looking forward to test your 270x optimized code!

same 4x r7 370
qtminer : 60mh/s
claymore : 55mh/s

ethminer--cointron-- 15mh\s

claymore--11mh\s  why? (R7 370)

After 1 hour on sprnova Decret stratum protocol i got from 2-3%. Very nice.

It is username.workername

EthDcrMiner64.exe -epool coinotron.com:3344 -ewal Redhex.rig1 -esm 2 -epsw x -dpool http://dcr.suprnova.cc:9111 -dwal Redhex.my -dpsw x -allpools 1

Redhex.rig1 is a wallet?

Let me implement most important features for pool version firstly, I think it will take 1-2 weeks.

I'm using r9 380, dont know if they are optimized. I updated the miner and now the links i gave are original qtminer vs Claymore eth-proxy method.

Thank you. I've testing v 2.0 now. Got above 10% Rej from zpool-stratum (DCR). Trying sprnova stratum (DCR), and it looks much better, about 3%. Need more time for better numbers. What about solo version, i am waiting ))

Check OP or readme.

Hello any example for coinotron?

I'm still working on it, also on improvements for Fury cards.

Nice work Claymore, thank you for frequent updates.

Any updates on 270/X, the "weak" cards?

Currently I use supronova though it does not support vardiff, it may cause problems if everyone starts using this pool. May be they will fix it.

What stratum DCR pool would you recommend to use with your miner?

 I'm at ethermine.org . Luck does make variance though the more time involved the less luck impacts anything. I ran for 24 hours. Since the miner hasnt yet existed for a week its difficult to give longer trials. Revisiting my first rig report found that over 2-days it was then gaining very slightly less than using just the first days data, but close enough to be giving a pretty accurate picture of whether this miner was giving more ethereum than ethminer.


 I agree your way is a greater scientific methodology, but I disagree that good comparisons can't be made any other ways. Given that the previous value of daily hashrate is well known and time proven, making just the one change of switching mining software does create a valid comparison between the two. While the new higher hashrate }could{ be a fluke result, todays addition 24 more hours of data, along with multiple others reporting similar results, suggest it is not a fluke at all.

 Perhaps you are using cards that Claymore doesnt give a benefit to? We dont know because nobody has compared the new improvement results across all gpu versions, other than we know 270s dont function well with it yet.

It stops immediately, you can see that miner sends a share, then gets new job, then pool sends "rejected" answer. The problem is on pool - I see sometimes it takes pool about 3-5 seconds to send "accepted" response for share sent. If pool gets new block within this time it sends "rejected".

For now (a short run) it has 145 total shares with 10 rejected. Rather bad statistics. Maybe the miner should stop working on old job immediately when the new job arrives?