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

[buny]

New version xpmclient 10.3 available.

All is well. Continue on.   

[1714077928]

1714077928

[1714077928]

1714077928

[suona]

@eXtremal.ik71

Is there any way to completely disable any interaction of xpmclient with AMD WattMan? As we see, even if value of "powertune" parameter is set to "-1", it still proceeds interaction with WattMan setting powerlimit value to -1%, instead of changing nothing (as mentioned in comment, value "-1 means don't change", that is correct for else parameters of this section, but not for "powertune" as we have seen).

Previously, with older AMD drivers, this was not a problem.
But now the fact is that recent AMD drivers have a lot of new problems, on some PCs interaction of 3-rd party software with WattMan causes improper functioning of the driver that eventually ends up with driver crash, and in some cases - OS crash.

If PC is used for running xpmclient solely, then no any issues occur (though the problem is there, and we can see this if to run AMD control panel and open WattMan tab - driver will get crashed, and in about 3 of 5 cases OS will also get crashed several seconds later).
But in case if PC is also being used for another tasks, this becomes a problem. Attempt to launch any another software that utilizes GPU, especially simultaneously with xpmclient being running, often ends up with driver or OS crash.

On those PCs where we down-graded AMD drivers to versions of 2018 this problem disappeared. But older drivers have lack of some options and have another common issues, so users prefer to use recent drivers versions.

There is also another interesting fact - that problem with driver remains until OS restart even if xpmclient application has been already closed. This indicates that interaction of xpmclient with WattMan causes some sort of corruption to driver that remains until OS restart and causes improper functioning of the driver itself and of another software that utilizes GPU.

If xpmclient was not launched after OS restart then no any issues with driver or with any software occur.

So i'd suggest for the new xpmclient build to add an option that would allow to control whether interaction with WattMan is enabled or is completely disabled.


These issues were confirmed to occur with driver version v19.5.1.
I'm gonna check this with v19.5.2 as well next days, and will add one more comment for that.

[suona]

Checked xpmclient with AMD driver version 19.5.2 - no such problems with WattMan, everything works stable.

[wadimkilll]

what is this shit ..... 2882 confirmations


Pending ([163 of 2880)   XPM   90.52027785

2019-06-08 08:53:19

Address: AMePdnePJaCDMjZYvBwm1GL8NF3rH
Txid: dbecf6fe93bcfa2090ab41b4610fb58ed29401cd21dcc79fc3865770b4ad
https://c.radikal.ru/c07/1906/0d/36145e9c4246.png][url=https://radikal.ru]https://c.radikal.ru/c07/1906/0d/36145e9c4246.png[/url]

[buny]

@eXtremal.ik71, let's have another scam? As with bitcoin prime. And that is boring already. 

[buny]

it is agony ... a new client from march 2018 .... hurry to see! 

[buny]

really. how many years ago ppc on btc-e sent the fuck ....

[suona]

Apparently he develops this project alone. It is quite difficult for a single developer to maintain such a complicated freeware and opensource project. Especially if person has the job and the family when free time for any hobby like that becomes a very rare and precious resource, and furthermore when 3-rd parties add their own problems to the project - like the recent unstable AMD drivers. Not to mention of attacks against the pools - reliable protection costs more than those pools ever earn.
So this is normal he releases updates so rarely.
Unlike many other projects that exist due to money income, this one seems to still exist due to enthusiasm of one person. This rather deserves thankfullness even if result isn't that good as was expected.
Finally nobody is forced to use it, there are many else alt-coins. Also it is opensource - anyone could take part in development, but no one else seems to offer anything better for xpm.

I think writing constructive messages about existing problems may help developer to fix them. But simply negative feedbacks might rather suppress his motivation to proceed working on it, and barely may help to fix the problems.

[tar1s]

On coinsforall problems?
Some "Requested" go to "Paid", but they are not in "Transaction".
Already the third day this crap.
Or is it with me

[a7269118]

It is now 2880 times to confirm the transaction to the transaction. Does anyone know why?
It takes almost three days, very long

[Odyysee]

Share information how many grant RX 474 и RX 478 on xpmclient 10.2 and 10.3

[a7269118]

Share information how many grant RX 474 и RX 478 on xpmclient 10.2 and 10.3

If only RX4 series and 5 series are used
You can get a higher CPD on xpmclient 10.2.2 The impression of using xpmclient 10.3 is that the CPD is lower, and the higher overclocking parameters are not acceptable.


If you insist on a higher overclocking parameter, xpmclient 10.3 will continue to display the reset display card parameters, and the display card stops responding.

In addition, digging xpmclient is not related to the memory capacity of the display card. The RX480 will have a higher CPD than the RX470. In addition, the label is required. The ASUS RX470 is always lower than other CPDs such as PowerColor or MSI.


Look at the core/memory frequency, overclocking parameters of 1250/1750
RX470's CPD is about 0.34~0.42
Look at the compatible version of the BIOS is good

In the xpmclient 10.2.2 version, 25 RX4 series and RX5 series can get 9.425 CPD. If it is in xpmclient 10.3 version, then 9.0 CPD is left, and the amount dug in one day is directly worse than 3 XPM.

[Odyysee]

Share information how many grant RX 474 и RX 478 on xpmclient 10.2 and 10.3

Look at the core/memory frequency, overclocking parameters of 1250/1750
RX470's CPD is about 0.34~0.42
Look at the compatible version of the BIOS is good

In the xpmclient 10.2.2 version, 25 RX4 series and RX5 series can get 9.425 CPD. If it is in xpmclient 10.3 version, then 9.0 CPD is left, and the amount dug in one day is directly worse than 3 XPM.

Please let me know the settings in the Config to get the maximum CPD (xpmclient 10.2.2) for RX 588.
Somewhere seen what happens with one RX 588 power 0.5 CPD.
thank you
 

[buny]

for the year fell 15 times. what predictions? 

[tar1s]

for the year fell 15 times. what predictions? 

Soon, soon to be x40.
I already reserve the island

[a7269118]

Share information how many grant RX 474 и RX 478 on xpmclient 10.2 and 10.3

Look at the core/memory frequency, overclocking parameters of 1250/1750
RX470's CPD is about 0.34~0.42
Look at the compatible version of the BIOS is good

In the xpmclient 10.2.2 version, 25 RX4 series and RX5 series can get 9.425 CPD. If it is in xpmclient 10.3 version, then 9.0 CPD is left, and the amount dug in one day is directly worse than 3 XPM.

Please let me know the settings in the Config to get the maximum CPD (xpmclient 10.2.2) for RX 588.
Somewhere seen what happens with one RX 588 power 0.5 CPD.
thank you
 

Core 1300
Memory 1750
Xpmclient only looks at the core frequency, the higher the better, but also the power and temperature

[klaidez]

Whats up with the difficulty? Its stuck at 10.995592
It looks like a bug

[eXtremal.ik71]

New beta version of xpmclient for AMD cards available:
http://coinsforall.io/distr/xpmclient-opencl-10.5-beta1-win64.zip
http://coinsforall.io/distr/xpmclient-opencl-10.5-beta1-linux.tar.gz

SHA256:
c6a1d06eb4711c252883bdcd4ddd17033957a197467bfa631a5855d5dc35ce8f  xpmclient-opencl-10.5-beta1-linux.tar.gz
12f1abec56cfc39e7e4ad10db06989e799fffdaccf44dff2154a49cd0da5cb71 xpmclient-opencl-10.5-beta1-win64.zip

Main feature of this version - new experimental kernel partially writed using GCN assembly language (https://github.com/CLRX/CLRX-mirror). For enable this, change parameter in config.txt:

# Kernel type
#   generic - OpenCL kernel
#   asm - experimental kernel created using GCN assembly, no any warranty!!!
#     this kernel doesn't support Tahiti (HD7970/R9 280X) and Navi (Radeon RX5700)
#     other GPUs can work
kernelType = "asm";

"asm" kernel is in not final release, it can works and show big performance advantage (such as 10-12% for Vega and up to 30% for Radeon 4xx/5xx GPUs on Linux.. so big speed up is possible because Linux amdgpu-pro drivers not so good for XPM mining), or fails to launch.
Also, interesting to see result of different GPUs in benchmark mode (xpmclient -b) with generic and asm kernels.

[tkachev]

"asm" kernel not work on R9 290 and 390, only generic works but slower than the older version of miner.
Got this error on win 8.1, 10 and diffent drivers:

2019-07-22 18:38:40.945 (   0.608s) INFO| compiling Hawaii_sha256.bin ...
2019-07-22 18:38:42.698 (   2.360s) INFO| compiling Hawaii_sieve.bin ...
2019-07-22 18:38:46.786 (   6.449s) INFO| compiling Hawaii_sieve_utils.bin ...
2019-07-22 18:38:47.104 (   6.767s) INFO| assembling Hawaii_Fermat.bin ...
2019-07-22 18:38:47.175 (   6.838s) INFO| compiling Hawaii_Fermat_utils.bin ...
2019-07-22 18:38:48.101 (   7.764s)  ERR| Existing OpenCL kernel (Hawaii_Fermat.bin) incompatible with configuration
2019-07-22 18:38:48.106 (   7.769s)  ERR| Please remove all *.bin files and restart miner

I tried different parameters in config, but still got error. Fury and Nano works fine. Some 480-s hangs at start up, mostly in first gpu's slot.
The big trouble is also log file. Need to have swich it off

[eXtremal.ik71]

"asm" kernel not work on R9 290 and 390, only generic works but slower than the older version of miner.

asm kernel can work on R9 290/390, may be next version will support it.

I tried different parameters in config, but still got error. Fury and Nano works fine. Some 480-s hangs at start up, mostly in first gpu's slot.

Can you post results of Fury and RX4xx with old miner version and 10.5 with asm kernel for compare ?
It seems, normal CPD for stock RX470 is 0.4-0.41CPD, for RX580 - 0.5 CPD now.

The big trouble is also log file. Need to have swich it off

Will be in 10.5 beta2.