PTS and NRS GPU Miner PtsGPUz v0.5b for Windows(GTX 750Ti~820cpm)

Started by abc123, January 09, 2014, 10:29:02 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.


lty616

Quote from: lty616 on February 13, 2014, 10:22:00 PM
Hi abc123, will be any performance improvement for Cuda? We all know AMD is much much faster. :-(
"AMD will is much much faster" is not true for SHA512, it is true for SHA256 and scrypt.
I don't think more significant performance increases can be made on GTX 660.
10~20% performance increases may be possible, but needs a lot of work.
Greate performance improvement for Tesla etc may be possible, or may not. But Tesla card is very expensive,
and whether the performance can be improved is uncertain, so I have not made decision to buy one.
[/quote]

My friend's R9 280X can have 3100/3500 (default/OC) cpm on clpts-v0.2.2, and my gtx690 has 2200 cpm in total. I believe this means AMD now is much faster in SHA 512, single chip beats my two. clpts also asks for 2 threads for each chip like your miner, apparently, it is much more efficient and optimized.
Check this post https://bitsharestalk.org/index.php?topic=2598.0

abc123

Quote from: d0mini on February 10, 2014, 07:24:02 PM
@abc123,
Any changes for SM 3.0 Nvidia cards like the 770? If not, do you think there will be any more significant performance increases for these cards?
There is no changes for Nvidia cards in version 0.4b and 0.4c.
I haven't tested my miner on GTX 770 till now. It is GTX 660 on my computer for develop, which is also SM 3.0.
I don't think more significant performance increases can be made on GTX 660, unless defects in PTS proof of work
algorithm dicovered. 10~20% performance increases may be possible, but needs a lot of work.

Quote from: Makezzila on February 11, 2014, 05:47:47 AM
Hello.
I using you new miner for my geat miner rig on AMD videocards.
6 GPUS: Radeon 7970+7870+5850+5850+5850+5850.
And i get from all videocards only 3200 collisions\minute.  :(
But When i using this miner: https://bitsharestalk.org/index.php?topic=2410.0
i  received from five videocard 4000 collisions  (miner on 6st the card didn't manage to start).
Please use version 0.4c released minutes ago. Version 0.4b is not optimized for AMD cards
other than Southern Islands (HD 7xxx) series with 2GB or more video memory.
    Version 0.4c is optimized and can be run on Evergreen (HD5xxx) series and later.

Quote from: krpetrov on February 11, 2014, 06:34:16 AM
How can I run this miner on Ubuntu ? ???
No. It is for Windows only till now.

Quote from: dongboy on February 13, 2014, 04:13:39 AM
hi abc123, 1gh?
Still not supported. 1gh use a different version of xpt protocol.

Quote from: lty616 on February 13, 2014, 10:22:00 PM
Hi abc123, will be any performance improvement for Cuda? We all know AMD is much much faster. :-(
"AMD will is much much faster" is not true for SHA512, it is true for SHA256 and scrypt.
I don't think more significant performance increases can be made on GTX 660.
10~20% performance increases may be possible, but needs a lot of work.
Greate performance improvement for Tesla etc may be possible, or may not. But Tesla card is very expensive,
and whether the performance can be improved is uncertain, so I have not made decision to buy one.

abc123

PtsGPUz v0.4c now released.
Main change log:
PtsGPUz v0.4c (2014-02-14):
MD5 of PtsGPUz0.4c.exe: e9a4a89c7348e2d7619c7bf39efe83d8
1. The performance of AMD cards other than Southern Islands (HD 7xxx) series
  with 2GB or more video memory increased greatly. In other words, the
  performance of AMD HD7xxx series cards with 1GB video memory and all HD6xxx
  series cards with 1GB or more video memory increased greatly.

Net performance list for AMD cards at default frequency, using driver "amd_catalyst_14.1_betav1.6.exe":
HD7850, 1024SP, 256bit, 2GB: 840 collisions/minute;
HD7850, 1024SP, 256bit, 1GB: 720 collisions/minute;
HD6670, 480SP,  128bit, 1GB: 380 collisions/minute.

lty616

Hi abc123, will be any performance improvement for Cuda? We all know AMD is much much faster. :-(

tecknogeek

PtsGPUz0.4b.exe -o ptspool.1gh.com:18120 -u PsT8AZjj2J7ivgxGyhYVZZC7WcJr2BNdi6 -p x

Just change for your own PTS address (Or feel free to leave mine there!! But you will make a bit less that way lolll)



Makezzila

Hello.
I using you new miner for my geat miner rig on AMD videocards.
6 GPUS: Radeon 7970+7870+5850+5850+5850+5850.

And i get from all videocards only 3200 collisions\minute.  :(
But When i using this miner: https://bitsharestalk.org/index.php?topic=2410.0
i  received from five videocard 4000 collisions  (miner on 6st the card didn't manage to start).



clever

Hello,

I receive this error, any ideas?

cudamalloc failed 4!try to use 256mb buffer memory

d0mini

@abc123,

Any changes for SM 3.0 Nvidia cards like the 770? If not, do you think there will be any more significant performance increases for these cards?

abc123

PtsGPUz v0.4b now released.
Main change log:
PtsGPUz v0.4b (2014-02-11):
MD5 of PtsGPUz0.4b.exe: 486f4d653aae492c8300dd5a06552753
1. Performance increased for AMD Southern Islands (HD 7xxx) Series display card.
Net performance for AMD card at default frequency:
HD7850, 1024SP, 256bit, 2GB: 770-810 collisions/minute.
HD6670, 480SP, 128bit, 1GB:  150-170 collisions/minute.
Note: For AMD Southern Islands Series display card with 1GB video memory,
the performance will be much poorer than 2GB's.

abc123

Quote from: tazbox on February 07, 2014, 04:08:04 PM
Why im connected on pts.rPool.net when i mining on ypool ?
Miner fee (develpoer fee) is 5%, it is send to ypool.net and one of pts.rpool.net or ptspool.com randomly.
User can change the address if the pool's port has been changed or pool has been shutdown etc,
but I can't as the pool's address is hard-coded in my miner...
So there are two connections for develpoer, together 5%, it is the initial design which is designed in hurry,
although now I think 3 minutes each hour for developer maybe more appropriate.

Quote from: lty616 on February 07, 2014, 05:41:30 PM
Hi, abc123
I am using GTX 690 for pts. I reported one issue here by my experience. The 0.3c and 0.4 version could not work properly for GTX690 which based on what I found, the 2nd GPU only has 700 Mb RAM occupied, while 0.3b version works properly, i.e., both GPUs take 1.2-1.3 Gb ram and give me 2100 cpm.
I am here just wondering if you have chance to optimize the performance like GTX 690 (I believe 590 should face same issue) which has 2 GPUs on one board. It will be appreciated.
Thanks for your report.

Quote from: sklye on February 09, 2014, 03:26:25 AM
hi abc123, i get one problem that  the miner runs on  Win8.1 64bit   physical machine with hyper-v service, and  it shows an error "sorry, this application cannot run under a virtual machine "   why?
Thanks for your report. This may due to the wrong setting when I pack the miner, I am sorry for this. Try PtsGPUz v0.3b before new version release.

sklye

hi abc123, i get one problem that  the miner runs on  Win8.1 64bit   physical machine with hyper-v service, and  it shows an error "sorry, this application cannot run under a virtual machine "
why?

lty616

Hi, abc123
I am using GTX 690 for pts. I reported one issue here by my experience. The 0.3c and 0.4 version could not work properly for GTX690 which based on what I found, the 2nd GPU only has 700 Mb RAM occupied, while 0.3b version works properly, i.e., both GPUs take 1.2-1.3 Gb ram and give me 2100 cpm.
I am here just wondering if you have chance to optimize the performance like GTX 690 (I believe 590 should face same issue) which has 2 GPUs on one board. It will be appreciated.