Open Source XPM (Primecoin) GPU Miner & Pool xpmforall.org

[abrgeza]

CPD pool dropped by 25% per day.

Because some miners uses target=10 others target=11, pool can't show real CPD now.

both of your pools count shares sended by clients so real CPD does not matter?

[eXtremal]

@all: read this F.A.Q. please. If you have more questions, I will add it.
https://bitcointalk.org/index.php?topic=831708
briefly: set target="10", then wait and install xpmclient 10.2.1 - after this release no more changes in target switching procedure.
Sorry for xpmclient 10.2 users, if more miners update their clients, you will not need change settings.

both of your pools count shares sended by clients so real CPD does not matter?

Pool reject shares with wrong target (less than current difficulty). Without it, you can set target="6" and generate 81 times more shares without blocks. Why pool must pay for this shares?

[buny]

...and install xpmclient 10.2.1

it was already 10.2.1?! Where? saw only 10.2 ...

[eXtremal]

...and install xpmclient 10.2.1

it was already 10.2.1?! Where? saw only 10.2 ...

https://bitcointalk.org/index.php?topic=831708
Most recommended miner is xpmclient 10.2.1 (will be available in 24 hours)

how relevant is the text of "Primecoin mining calculator (1 Scrypt Mh = 15.6 CPD)"?

Ohh, about 4 years ago Litecoin GPU mining was actual, overclocked Radeon R9 290X did 1000KHash/s in scrypt algo.. and 15.6 CPD on XPM. Now it's useless information

[buny]

how relevant is the text of "Primecoin mining calculator (1 Scrypt Mh = 15.6 CPD)"?

Ohh, about 4 years ago Litecoin GPU mining was actual, overclocked Radeon R9 290X did 1000KHash/s in scrypt algo.. and 15.6 CPD on XPM. Now it's useless information

I repaired the RX560 ...
3xRX560 = 0.46
but it's still much smaller than it was ...

[abrgeza]

both of your pools count shares sended by clients so real CPD does not matter?

Pool reject shares with wrong target (less than current difficulty). Without it, you can set target="6" and generate 81 times more shares without blocks. Why pool must pay for this shares?

i was talking about valid shares, not interest in cheating.
what i saw is that xpmclient 10.2 with target=10 generates more shares than same client with target=11 for the same difficulty.
i didn't direct compare it with xpmclient9.4.1 but it seems that is better to force old xpmclient until EOL.

[eXtremal]

both of your pools count shares sended by clients so real CPD does not matter?

Pool reject shares with wrong target (less than current difficulty). Without it, you can set target="6" and generate 81 times more shares without blocks. Why pool must pay for this shares?

i was talking about valid shares, not interest in cheating.
what i saw is that xpmclient 10.2 with target=10 generates more shares than same client with target=11 for the same difficulty.
i didn't direct compare it with xpmclient9.4.1 but it seems that is better to force old xpmclient until EOL.

See F.A.Q. https://bitcointalk.org/index.php?topic=831708.0

What happens if I will use invalid target settings?

You will get less profit:
* difficulty < 10.994 and target="11": miner will generate less amount if shares.
* difficulty > 10.994 and target="10": big amount of shares will be marked as invalid.

When pool switches to "11ch" mode, it begin reject shares from xpmclient with target="10", because it's cheating same as target="9","8" etc. Now pool again in 10ch mode until 10.994 difficulty, best option now is using target="10" and wait xpmclient 10.2.1 release.
I try release xpmclient 10.2 and switch pool to "11ch" at 10.991 difficulty, but less than 10% miners install xpmclient 10.2 or put target="11" to config.txt, all others continue using old settings. I decide that miners don't ready and make rollback.

[farfui]

what is the best setting for 1080ti

[abrgeza]

what is the best setting for 1080ti

try this settings in config.txt:

onCrash = "1";
platform = "nvidia";
cpuload = "1";
target = "10";
sieveSize = "420";
weaveDepth = "65536";
width = "20";
windowSize = "12284";
devices = ["1", "1", "1", "1", "1", "1"];
sieveprimorial = ["13", "13", "13", "13", "13", "13"];
sievePerRound = ["5", "5", "5", "5", "5", "5"];

1. run DDU in safe mode and uninstall all nvidia driver
2. reboot and install driver 372.54
3. put this into .bat file:
    @echo off
    color 02
    setx GPU_FORCE_64BIT_PTR 1
    setx GPU_MAX_ALLOC_PERCENT 100
    setx GPU_USE_SYNC_OBJECTS 1
    setx GPU_MAX_HEAP_SIZE 100
    echo XPMclient NVidia
    ping -n 10 127.0.0.1>nul
    echo XPMclient NVidia
    start xpmclient
    :1
    ping -n 10 127.0.0.1>nul
    tasklist | find /i "xpmclient.exe"  || echo %date% %time% restart >> log.txt && start xpmclient.exe
    goto 1
4. run .bat file and compile kernel
5. close xpmclient and .bat file
6. run DDU in safe mode and uninstall all nvidia driver
7. reboot and install driver 388.71
8. run .bat file and inform us what is your cpd

works fine on win10

[abrgeza]

something is wrong on coinsforall.io?
cpd is raised and droped like crazy.

[buny]

I'll soon be a millionaire ...

[abrgeza]

I'll soon be a millionaire ...

Only if other gpu are like that one. 

[eXtremal]

xpmclient 10.2.1 is out http://coinsforall.io/distr/xpmclient-10.2.1-beta.tar.gz
Difference from 10.2 is only target switch point, 10.994 now and it will not be changed.

something is wrong on coinsforall.io?
cpd is raised and droped like crazy.

coinsforall.io pool display only sent CPD from clients. On xpmforall.org you can see true CPD in "24h Share Count". Sum of 10ch and 11ch is real CPD

[abrgeza]

xpmclient 10.2.1 is out http://coinsforall.io/distr/xpmclient-10.2.1-beta.tar.gz
Difference from 10.2 is only target switch point, 10.994 now and it will not be changed.

thanks for the info.
what is the reason that you forcing target=11 before difficulty reach 11.00?

[eXtremal]

xpmclient 10.2.1 is out http://coinsforall.io/distr/xpmclient-10.2.1-beta.tar.gz
Difference from 10.2 is only target switch point, 10.994 now and it will not be changed.

thanks for the info.
what is the reason that you forcing target=11 before difficulty reach 11.00?

Difficulty also contains fractional part, it's a shares/blocks ratio. For example, with difficulty 10.0 all 10ch shares are blocks, with 10.5 - only 50% 10ch shares are blocks, with 10.994 - 0.6% 10ch shares are blocks.
When you switch miner target to 11, it generates 60% less 10ch shares, but 10-20% more 11ch shares (depend on type of GPU). So, miner with target="11" will generate more blocks, than miner with target="10" at diffuculty >= 10.99x.

[abrgeza]

Difficulty also contains fractional part, it's a shares/blocks ratio. For example, with difficulty 10.0 all 10ch shares are blocks, with 10.5 - only 50% 10ch shares are blocks, with 10.994 - 0.6% 10ch shares are blocks.
When you switch miner target to 11, it generates 60% less 10ch shares, but 10-20% more 11ch shares (depend on type of GPU). So, miner with target="11" will generate more blocks, than miner with target="10" at diffuculty >= 10.99x.

really appreciate for that info and thanks for your great effort.
in a next couple of days we will reach 10.994 and miners should switch to target=11. what if they stay at target=10? your pools will accept those shares or reject those shares?
i am asking that because most of the network hashrate is generated by tonga, tahiti, hawaii cards and those cards performs very well because you made great 9.4.1 client with superb kernel for those cards. any newly compiled kernel give as 20-35% less hashrate. can you make new superb kernel with target=11 for those cards?
thanks in advance.

[ser_ator]

270 card 820/1250:
  new miner   < 3.2 cpd
 9.4.1 < 4.8 cpd

7850 820/1250:
 new miner <2.5 cpd
9.4.1 <3.7

difference 50%

for 9.4.1 driver 15.7.1 with ocl 14.6
for  the new miner work only witout ocl driver in miner (15.7.1)

[abrgeza]

270 card 820/1250:
  new miner   < 3.2 cpd
 9.4.1 < 4.8 cpd

7850 820/1250:
 new miner <2.5 cpd
9.4.1 <3.7

difference 50%

it is not so bad, it is worse. 

[HuanHuan1997]

Hi, eXtermal, where can I find the source code for version 10.X? I want to compile it my own. The provided pre-compiled binary version hangs my PC which when I quit the miner (for playing my game). The github version dates back to 2014 and no source code for the new version. the old version no longer builts on modern ubuntu.

New version 9.4 with next speedup: https://www.dropbox.com/s/ny027u62vbeb2u5/xpmclient_v9.4.tar.gz?dl=0

Radeon R9 290X @ 1000/1250 (stock) - 14.9 CPD
  R9 290X @ 1040/1500 - 15.35CPD
Radeon R9 290 @ 947/1250 (stock) - 13.2 CPD
Radeon R9 280X @ 1000/1250 (stock) - 9.35 CPD
  R9 280X @ 1100/1500 - 10.25CPD

[eXtremal]

Hi, eXtermal, where can I find the source code for version 10.X? I want to compile it my own. The provided pre-compiled binary version hangs my PC which when I quit the miner (for playing my game). The github version dates back to 2014 and no source code for the new version. the old version no longer builts on modern ubuntu.

New version 9.4 with next speedup: https://www.dropbox.com/s/ny027u62vbeb2u5/xpmclient_v9.4.tar.gz?dl=0

Radeon R9 290X @ 1000/1250 (stock) - 14.9 CPD
  R9 290X @ 1040/1500 - 15.35CPD
Radeon R9 290 @ 947/1250 (stock) - 13.2 CPD
Radeon R9 280X @ 1000/1250 (stock) - 9.35 CPD
  R9 280X @ 1100/1500 - 10.25CPD

https://github.com/eXtremal-ik7/xpmclient/tree/version/zcash

270 card 820/1250:
  new miner   < 3.2 cpd
 9.4.1 < 4.8 cpd

7850 820/1250:
 new miner <2.5 cpd
9.4.1 <3.7

difference 50%

for 9.4.1 driver 15.7.1 with ocl 14.6
for  the new miner work only witout ocl driver in miner (15.7.1)

Miners 9.4.1 and 10.2.1 shows almost identically CPD with same drivers version. What's problem with xpmclient 10.2 on system with catalyst 15.7.1 and ocl 14.6 ?

what if they stay at target=10? your pools will accept those shares or reject those shares?

Pool will reject around 60% shares from miners with target=10, but continue accept blocks from it.

can you make new superb kernel with target=11 for those cards?

I can share this kernel, but everyone can do this.
Best option for Tahiti/Hawaii (R9 280/290) is Catalyst 14.6, for R9 390 - Catalyst 15.12 with ocl from 14.6.