Bump... there is little info out there. Which algorthim would one use on Nicehash for primecoin mining?
Nicehash don't support primecoin mining now. Sapphire R9 390 Fury HBM 4Gb. - 6.8 cpd
Sapphire R9 390 Nitro+ GDDR5 8Gb - 6.5 cpd
Sapphire R9 390X Nitro+ GDDR5 8Gb - 6.8 cpd
Win7x64, driver 16.3.2
I don't know why so slow.
16.3.2 not suitable for XPM mining, try old Catalyst 14.6 driver. Are there any plans to write a fresh version of the miner? After new Radeon RX Vega comes. |
|
|
|
eXtremal, Why not go pool payments (xpmforall)  xpmforall (coinsforall too) XPM payments work properly last 2 years.. unlike zec pool on startup  Sometimes pool can delay payment up to 4-5 hours, not more. |
|
|
|
Anyway, I have two questions. What are the preferred GPUs for Primecoin mining now, and is there any reason to use coinsforall over xpmforall?
Preferred GPUs is Hawaii - R9 290(X)/390(X). Other GPUs also can be profitable on XPM rather than ETH and others, need use calculator to check them. If you use coinsforall you can set payment threshold, do manual payments.. and no any lags on website. |
|
|
|
hi
how much CPD can i get with r9 390x and r9 nano?
R9 390x more than 15CPD, r9 nano/fury x don't know What powertune options are valid? 0 - 20?
If ADL is working with your drivers, try 20. Any idea to optimize the miner for rx480? I only get ~10 cpd.
Polaris too slow in XPM mining.. using GCN assembly can give about +5-10% on this GPUs, but it's not easy. |
|
|
|
This patch gives +5-6% on NVidia GTX10xx cards: --- a/Core/kernel/equihash.cl
+++ b/Core/kernel/equihash.cl
@@ -102,7 +102,11 @@ typedef __global slot_t *global_pointer_to_slot_t;
__global char *get_slot_ptr(__global char *ht, uint round, uint row, uint slot)
{
- return ht + (row * _NR_SLOTS(round) + slot) * _SLOT_LEN(round);
+ // Split row into several sub-rows with 2^RowFragmentLog slots, it gives more L2 cache hits
+ const uint RowFragmentLog = 4;
+ const uint SlotsInRow = 1 << RowFragmentLog;
+ const uint SlotMask = (1 << RowFragmentLog) - 1;
+ return ht + ((slot >> RowFragmentLog)*_NR_ROWS(round)*_SLOT_LEN(round)*SlotsInRow) + (row*_SLOT_LEN(round)*SlotsInRow) + (slot & SlotMask)*_SLOT_LEN(round);
}
Can't test on AMD now, may be it gives same performance advantage. Unfortunatelly, this miner useless for NV cards.. only 280sols/s on GTX1070. |
|
|
|
|
Did anybody implement NR_ROWS less than 16 support already ?
|
|
|
|
Did you make any progress with AMD?
Last release don't working on AMD, but if I found a reason, it will be same +10-15% on AMD cards. For more speedup, see my previous post. |
|
|
|
Speedup +10-15% for NVIdia only: http://coinsforall.io/distr/nvidia/input.clhttp://coinsforall.io/distr/nvidia/param.hSorry, but can't work more than 1 hour a day on miner now. For other developers, you need: - Decrease NR_ROWS_LOG to 13 or 12. ht_store function works much faster with low NR_ROWS values and when you decrease NR_ROWS, you also decrease total slots amount, because you can use less values for OVERHEAD constant. - Optimize equihash round for big NR_SLOTS values. I begin do it in last NVidia release, but need much more work.. |
|
|
|
Because it's not native and we are discussing open source project here.If you don't want to share your kernel - that's fine. At least give a little hint to mrb,nerdralph and eXtremal for the new version.
Thank you.
I know what need to do, but have a time problem. I'll make an update at this week and instruction for other developers how to get other miners performance. |
|
|
|
|
Wait next +10% in 12h, then 2 days pause before major release (~+50% ).
|
|
|
|
Hi eXtremal, I have a problem with your pool and can't contact you using admin@coinsforall.io : Delivery to the following recipient failed permanently:
admin@coinsforall.io
Technical details of permanent failure:
read error: generic::failed_precondition: read error (0): error I'm waiting for a payment to a zaddr (requested days ago) but nothing happens. Thank you I changed email on website, use it plz. |
|
|
|
"Capitalism vs communism mining" shows that with proper $ fee incentive, capitalism beat communism due to human (greedy) nature. Now we have minner sw wars where one side gets 2.5% and thrive with speed well motivated by earnings, while other side gets %0 or free will people donation. We see how it ends. Both author and its users of people version earn less then author and users of paid version. What an mind fk
Sad but mostly true... Game not finished, wait a several days plz. Just say the price and we'll gather donations for you(At least I am willing to donate hashpower for you) I don't want talk about donations before new SA version. |
|
|
|
|
For NVidia: processing 2 nonces per iteration gives +5-6% performance, AMD not affected.
|
|
|
|
Genoil's package not working? r9 270 : 70-72 sol
r9 380 : 90-95 sol
I know what need to do for get same performance, but it's not trivial patch, it's full equihash round code rewritting, need a time. |
|
|
|
Hey there,
Wondering what the connection details are for XPM pool?
cheers
See config.txt file from native miner archive https://www.dropbox.com/s/2976f8tnhc6p4pj/xpmclient_v9.4.1.tar.gz?dl=0 : server = "xpmforall.org";
port = "6666";
Use "xpmforall.org" as server address for mining on old pool or "coinsforall.io" for new XPM pool. New pool does not pay for immature coins, you need wait 2,5 days for payments or continue to use xpmforall.org. |
|
|
|
Yeah, I think it's totally worth it to put up a serious open source competition against Sir. Claymore  We shall be back. I'll try get Claymore's v5/optiminer speed in 2-3 days. p.s. he have a more motivation because his model (miner with fee) works perfectly - opposite to my (opensource miner and own pool), miners don't want support coinsforall.io  |
|
|
|
|
Show Posts
