Trivia Challenge:
What does dsm34 stand for? ...a typo
Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 206. (Read 3426932 times)
July 30, 2014, 02:03:17 AM
What does dsm34 stand for? ...a typo
July 30, 2014, 01:39:14 AM
Hey all, just started trying my hand at monero and I'm getting an odd issue. Using NVMiner 1.2D6 my GTX 750 tis are hashing in the 280s or so, but my GTX titan is only getting around 220 or so. Am I missing something?
Yes, 750Ti is the bomb.
yes and no. In all other coins, my Titan is 2-3x a 750 ti. So yes, it rock, but it's not that big of a bomb if you got a titan. Seems like the code is not set for this card. Much like the early ccminer was for x11.
Any tips?
What's the watt consumption of your Titan? On a non-OCed 750 Ti it's around 60 Watt. Can your Titan beat the Ti's hash/watt ratio?
Edit: after looking at the GTX Titan specs, I see 250W consumption... Clear cut case here!
Win8 x64, G1620, Asrock H61 PRO BTC, 4Gb, nvminer-v1.2U-D5, 337.88-whql
6x750Ti - x11 - 410W rig
[2014-07-29 00:43:44] Stratum detected new block
[2014-07-29 00:43:44] GPU #5: GeForce GTX 750 Ti, 2653 khash/s
[2014-07-29 00:43:44] GPU #4: GeForce GTX 750 Ti, 2659 khash/s
[2014-07-29 00:43:44] GPU #1: GeForce GTX 750 Ti, 2653 khash/s
[2014-07-29 00:43:44] GPU #3: GeForce GTX 750 Ti, 2669 khash/s
[2014-07-29 00:43:44] GPU #2: GeForce GTX 750 Ti, 2642 khash/s
[2014-07-29 00:43:44] GPU #0: GeForce GTX 750 Ti, 2659 khash/s
[2014-07-29 00:43:46] accepted: 5/5 (100.00%), 15913 kh/s - Yay
6x750Ti - CryptoNight - 316W rig
[2014-07-28 15:52:38] Stratum detected new block
[2014-07-28 15:52:39] GPU #3: GeForce GTX 750 Ti, 283.68 H/s
[2014-07-28 15:52:39] GPU #5: GeForce GTX 750 Ti, 283.17 H/s
[2014-07-28 15:52:39] GPU #4: GeForce GTX 750 Ti, 283.37 H/s
[2014-07-28 15:52:39] GPU #1: GeForce GTX 750 Ti, 283.25 H/s
[2014-07-28 15:52:40] GPU #0: GeForce GTX 750 Ti, 283.72 H/s
[2014-07-28 15:52:40] GPU #2: GeForce GTX 750 Ti, 283.14 H/s
[2014-07-28 15:53:11] accepted: 201/201 (100.00%), 1700.33 H/s - Yay
July 30, 2014, 01:15:49 AM
Trivia Challenge:
What does dsm34 stand for? ...a typo
July 30, 2014, 12:01:25 AM
Trivia Challenge:
What does tsiv stand for?
What does dsm34 stand for?
what does cayars stand for <-- super easy
what does CB stand for <-- even easier
4 people are disqualified from answering
Bonus,
What is the first name of each (2 are super easy). How well do you know thy "neighbors"?
What does tsiv stand for?
What does dsm34 stand for?
what does cayars stand for <-- super easy
what does CB stand for <-- even easier
4 people are disqualified from answering
Bonus,
What is the first name of each (2 are super easy). How well do you know thy "neighbors"?
What does tsiv stand for? Two-Sample Instrumental Variables
What does dsm34 stand for? Digital sex machine 34
what does cayars stand for ? Car yard
what does CB stand for ? Cyber Bus
Do I win ?
July 29, 2014, 11:52:25 PM
Hey all, just started trying my hand at monero and I'm getting an odd issue. Using NVMiner 1.2D6 my GTX 750 tis are hashing in the 280s or so, but my GTX titan is only getting around 220 or so. Am I missing something?
Yes, 750Ti is the bomb.
yes and no. In all other coins, my Titan is 2-3x a 750 ti. So yes, it rock, but it's not that big of a bomb if you got a titan. Seems like the code is not set for this card. Much like the early ccminer was for x11.
Any tips?
What's the watt consumption of your Titan? On a non-OCed 750 Ti it's around 60 Watt. Can your Titan beat the Ti's hash/watt ratio?
Edit: after looking at the GTX Titan specs, I see 250W consumption... Clear cut case here!
July 29, 2014, 11:25:00 PM
Hey all, just started trying my hand at monero and I'm getting an odd issue. Using NVMiner 1.2D6 my GTX 750 tis are hashing in the 280s or so, but my GTX titan is only getting around 220 or so. Am I missing something?
Yes, 750Ti is the bomb.
yes and no. In all other coins, my Titan is 2-3x a 750 ti. So yes, it rock, but it's not that big of a bomb if you got a titan. Seems like the code is not set for this card. Much like the early ccminer was for x11.
Any tips?
July 29, 2014, 09:19:44 PM
Carlo,
Are you able to connect?
Are you able to connect?
No I sent you email. VPN is no connect
July 29, 2014, 09:19:15 PM
6.5 atm.
6.0 was a dud, but was using 5.5 up until recently.
6.0 was a dud, but was using 5.5 up until recently.
6.5 cool
that's what we are moving to so NICE!
July 29, 2014, 09:14:32 PM
Carlo,
Are you able to connect?
Are you able to connect?
July 29, 2014, 09:06:37 PM
6.5 atm.
6.0 was a dud, but was using 5.5 up until recently.
6.0 was a dud, but was using 5.5 up until recently.
July 29, 2014, 09:04:06 PM
Whoops sorry...one last update.
Forgot a change to floating point precision (fast) and enabled /mp (multi-processor compilation) compiles at least twice as fast. Can do with it what you want, just something to fiddle with.
Anyway seems a tiny / negligable hash increase, but compiles way faster...a lot faster.
http://d-h.st/AWY
Forgot a change to floating point precision (fast) and enabled /mp (multi-processor compilation) compiles at least twice as fast. Can do with it what you want, just something to fiddle with.
Anyway seems a tiny / negligable hash increase, but compiles way faster...a lot faster.
http://d-h.st/AWY
Nice, will have to check it out. BTW, what version of CUDA toolkit do you have installed. (haven't looked at VS yet).
Carlo
July 29, 2014, 08:45:35 PM
Whoops sorry...one last update.
Forgot a change to floating point precision (fast) and enabled /mp (multi-processor compilation) compiles at least twice as fast. Can do with it what you want, just something to fiddle with.
Anyway seems a tiny / negligable hash increase, but compiles way faster...a lot faster.
http://d-h.st/AWY
Forgot a change to floating point precision (fast) and enabled /mp (multi-processor compilation) compiles at least twice as fast. Can do with it what you want, just something to fiddle with.
Anyway seems a tiny / negligable hash increase, but compiles way faster...a lot faster.
http://d-h.st/AWY
July 29, 2014, 08:45:12 PM
Won't let me download nvminer on http://cudamining.cc/url/releases
That's cause we don't like you.
Kidding aside, what type of error message are you getting?
It's working now, the server kept saying internal server error.
July 29, 2014, 08:37:07 PM
Trivia Challenge:
What does tsiv stand for?
What does dsm34 stand for?
what does cayars stand for <-- super easy
what does CB stand for <-- even easier
4 people are disqualified from answering
Bonus,
What is the first name of each (2 are super easy). How well do you know thy "neighbors"?
What does tsiv stand for?
What does dsm34 stand for?
what does cayars stand for <-- super easy
what does CB stand for <-- even easier
4 people are disqualified from answering
Bonus,
What is the first name of each (2 are super easy). How well do you know thy "neighbors"?
July 29, 2014, 07:44:57 PM
Won't let me download nvminer on http://cudamining.cc/url/releases
That's cause we don't like you.
Kidding aside, what type of error message are you getting?
July 29, 2014, 07:35:01 PM
How about posting your code and we can compare to see what you did differently?
Me just wants to get my grubby hands on it.
Me just wants to get my grubby hands on it.
Right, I cloned djm's latest and only modified doom.cu and created doom_luffa512.cu. And obviously you'll need to modify Makefile.am and the VC project files to include doom_luffa512.cu in the build. Built it and voila, ~80 MH/s per 750 Ti. Let me know how it turns out.
Source for both files here: http://pastebin.com/3kUMumEm
this :
Code:
#pragma unroll 8
for (int i = 7; i >= 0; i--) {
if (hash[i] > d_target[i]) {
if(position < i) {
position = i;
rc = false;
}
}
if (hash[i] < d_target[i]) {
if(position < i) {
position = i;
rc = true;
}
}
}
give a gain of performance while this doesn't:
Code:
#pragma unroll 8
for (int i = 7; i >= 0; i--) {
if (Hash[i] > pTarget[i]) {
rc = false;
break;
}
if (Hash[i] < pTarget[i]) {
rc = true;
break;
}
}
Now if someone has the explanations, I will be happy to hear it... (I would tend to believe that the second is faster....) but no...
It's all them capital letters, yo. Can't fit through them pipelines.
You've got the Hash array defined in the kernel and not still a pointer to gmem?
may-be the compiler doesn't like the break instruction which might get translated into 100 line of asm
actually it is the ptx file generated during compilation we should look for an explanation...
July 29, 2014, 07:30:42 PM
Won't let me download nvminer on http://cudamining.cc/url/releases
July 29, 2014, 07:07:46 PM
Dinner is waiting on me! lol =)
So I just re-zipped my VS project folder (ccminer-master)...
A few notes...I've been building with release - x64 - CUDA 6.5.
Changed the project file to load from CUDA 6.5 not 5.5.
If you have a processor that does not have AVX instructions you'll need to change DEBUG/ccminer properties/'C/C++'/Code Generation/ to whatever it does support or back to default.
I do not have a page file, so if you have very little ram / rely on a page file you'll need remove the two '0's under DEBUG/ccminer properties/Linker/System for good measure.
Look through DEBUG/ccminer properties under the maintool bar in visual studio to see the settings I've changed. I don't claim this to increase hashrate any significant amount, but every little bit helps and it definitely is more stable.
Also as I said before this is with CUDA 6.5 and x64 architecture, but can easily be modified for any CUDA toolkit and architecture.
http://d-h.st/bBe
So I just re-zipped my VS project folder (ccminer-master)...
A few notes...I've been building with release - x64 - CUDA 6.5.
Changed the project file to load from CUDA 6.5 not 5.5.
If you have a processor that does not have AVX instructions you'll need to change DEBUG/ccminer properties/'C/C++'/Code Generation/ to whatever it does support or back to default.
I do not have a page file, so if you have very little ram / rely on a page file you'll need remove the two '0's under DEBUG/ccminer properties/Linker/System for good measure.
Look through DEBUG/ccminer properties under the maintool bar in visual studio to see the settings I've changed. I don't claim this to increase hashrate any significant amount, but every little bit helps and it definitely is more stable.
Also as I said before this is with CUDA 6.5 and x64 architecture, but can easily be modified for any CUDA toolkit and architecture.
http://d-h.st/bBe
Cool, thank you for posting.
We have been getting some very good collective knowledge, algos and support this last month from many members.
Your contribution is most welcome.
July 29, 2014, 07:05:38 PM
djm34 when you compiled doom does it matter what graphics drivers you have in the outcome? because i cant mine it with 9.18.13.3788 drivers. I can with the lastest but the new drivers crash my old bat files. thz
July 29, 2014, 07:04:29 PM
Dinner is waiting on me! lol =)
So I just re-zipped my VS project folder (ccminer-master)...
A few notes...I've been building with release - x64 - CUDA 6.5.
Changed the project file to load from CUDA 6.5 not 5.5.
If you have a processor that does not have AVX instructions you'll need to change DEBUG/ccminer properties/'C/C++'/Code Generation/ to whatever it does support or back to default.
I do not have a page file, so if you have very little ram / rely on a page file you'll need remove the two '0's under DEBUG/ccminer properties/Linker/System for good measure.
Look through DEBUG/ccminer properties under the main tool bar in visual studio to see the settings I've changed. I don't claim this to increase hashrate any significant amount, but every little bit helps and it definitely is more stable.
Also as I said before this is with CUDA 6.5 and x64 architecture, but can easily be modified for any CUDA toolkit and architecture.
Also removed the ccminer.rc file that appeared in the project file within the last few commits. It shouldn't be there and breaks building.
http://d-h.st/bBe
So I just re-zipped my VS project folder (ccminer-master)...
A few notes...I've been building with release - x64 - CUDA 6.5.
Changed the project file to load from CUDA 6.5 not 5.5.
If you have a processor that does not have AVX instructions you'll need to change DEBUG/ccminer properties/'C/C++'/Code Generation/ to whatever it does support or back to default.
I do not have a page file, so if you have very little ram / rely on a page file you'll need remove the two '0's under DEBUG/ccminer properties/Linker/System for good measure.
Look through DEBUG/ccminer properties under the main tool bar in visual studio to see the settings I've changed. I don't claim this to increase hashrate any significant amount, but every little bit helps and it definitely is more stable.
Also as I said before this is with CUDA 6.5 and x64 architecture, but can easily be modified for any CUDA toolkit and architecture.
Also removed the ccminer.rc file that appeared in the project file within the last few commits. It shouldn't be there and breaks building.
http://d-h.st/bBe
Jump to: