Author

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 776. (Read 3426922 times)

hero member
Activity: 756
Merit: 502
I'm mining both yac and pts... Cheesy

Nice results, looking forward to higher clocks. Why no h81 btc pro with risers and 6 750tis?

So far we have no experience with risers, and I've seen reports that nVidia cards have been crashing on risers.
No experiments. Wink  We wanted to get mining fast.

An extension to 10 GPUs is a possibility, there are currenty 2 unused PCI-x x16 slots on each mainboard.
This will require some risers.

Christian
hero member
Activity: 812
Merit: 1000
I'm mining both yac and pts... Cheesy

Nice results, looking forward to higher clocks. Why no h81 btc pro with risers and 6 750tis?
hero member
Activity: 756
Merit: 502
I present a friend's first nVidia-only rig. Six 750Ti from Gigabyte (WindForce with 6 Pin power). Yes, we've built it into a suitcase.
Call it a "travel rig" if you want.



Currently this runs Linux and we are getting 3kHash/s Yacoin performance per card with the factory overclock.
We may have to switch to Windows for better overclocking possibilies.

The cards run cool at 56 deg C, even the ones that have almost no air breathing room.


Yacoin could be interesting in the future. Today was a price spike to triple the usual price.
https://bter.com/trade/yac_btc

Also Yacoin is back on cryptsy.com now.

I am supposed to be remote administering this rig. But at the moment we do not know how  (he's behind one router and one WLAN repeater, making this tricky). Maybe that's the final straw that will make me put a remote administration/monitoring API into cudaminer Wink

Christian
legendary
Activity: 1400
Merit: 1050
Do anyone know a good cudaminer config for a Gigabyte 780 Ti OC?
What do you want to mine exactly ? (configuration depends of the algo which is used)
newbie
Activity: 18
Merit: 0
I see some people complaining about performance decrease with the new nvidia driver, but I have found the opposite.  I have a pair of Gigabyte 770 OC 4GB and I am getting 40-60 KH/s more per card with the update. 
newbie
Activity: 30
Merit: 0
Do anyone know a good cudaminer config for a Gigabyte 780 Ti OC?
hero member
Activity: 812
Merit: 1000
What clocks are you running? Cheesy

I bet you can do better with some more time and of course, after Mr Buchner improves cudaminer for Maxwell cards. (looking forward to that!)

Well stock on my card is 1189 base / 1268 boost and upped that by 36.
Memory stock is 5400 and I upped that by 516... And this morning I had it up to 660 or so...

Nice, looking forward to getting one.
newbie
Activity: 52
Merit: 0
What clocks are you running? Cheesy

I bet you can do better with some more time and of course, after Mr Buchner improves cudaminer for Maxwell cards. (looking forward to that!)

Well stock on my card is 1189 base / 1268 boost and upped that by 36.
Memory stock is 5400 and I upped that by 516... And this morning I had it up to 660 or so...
member
Activity: 76
Merit: 10
Ah, makes sense. So the performance boost on the 750ti due to this new instruction is already accounted for by using the T kernel Smiley
hero member
Activity: 756
Merit: 502

Quote
The barrel shifter returns in Maxwell, and Nvidia claims the GM107 can mine digital currencies quite nicely, especially given its focus on power efficiency.

Would really like to hear Christians thoughts on this.

the funnel shifter is already present in GT 780, 780 Ti, Geforce Titan, Geforce GT 630 (GK208), Geforce GT 635 OEM (GK208), Geforce GT 640 (GK208). It helps with power efficiency and mining speed, but its instruction throughput is very limited compared to e.g. floating point instructions. When you've been mining with the T kernel, you've already used the funnel shifter.

Christian

ils
newbie
Activity: 5
Merit: 0
On your Config take a alook at reducing the "SMX" number:

ex. -l K7x32

You can try dropping the 7 to something lower. This can sometimes help. Some games just can't really deal well with sharing gfx. BF4 or BF3 are a great example... my Titan has more then enough power to play them but not with CUDAMiner running at even super low settings. While League of Legends I can just use my normal config and switch it to  -i 0 and boom frame rate loss but well within playable.

Thanks! Smiley
but not working Sad
still got 10-20 fps using K1x32 (i use K6x32 as default)
reducing it to K1x16 makes no difference in fps too
hero member
Activity: 756
Merit: 502
I wanna see Christian DO A BARREL SHIFT!!! LOL  Grin

nVidia calls it a funnel shifter.

Christian
member
Activity: 70
Merit: 10
hello, newbie here
i'm using cudaminer to mine LTC using my GTX 760 and so far no problem

however i just bought Bioshock Infinite and want to play it
i'm using -i 1 as option, but the game only ran @ 10-20 fps

is there any other option i can use to increase my fps?
or telling cudaminer to "only use remaining GPU power"? because Bioshock only use 50% GPU power to maintain 60 fps
i'm hoping to use remaining 50% to mine coins Smiley

Thank you Smiley

On your Config take a alook at reducing the "SMX" number:

ex. -l K7x32

You can try dropping the 7 to something lower. This can sometimes help. Some games just can't really deal well with sharing gfx. BF4 or BF3 are a great example... my Titan has more then enough power to play them but not with CUDAMiner running at even super low settings. While League of Legends I can just use my normal config and switch it to  -i 0 and boom frame rate loss but well within playable.
ils
newbie
Activity: 5
Merit: 0
hello, newbie here
i'm using cudaminer to mine LTC using my GTX 760 and so far no problem

however i just bought Bioshock Infinite and want to play it
i'm using -i 1 as option, but the game only ran @ 10-20 fps

is there any other option i can use to increase my fps?
or telling cudaminer to "only use remaining GPU power"? because Bioshock only use 50% GPU power to maintain 60 fps
i'm hoping to use remaining 50% to mine coins Smiley

Thank you Smiley
sr. member
Activity: 840
Merit: 255
SportsIcon - Connect With Your Sports Heroes
The Barrel shifter is not present on 3.0 cards like the GTX760, but is already on 3.5 Titans. You can see that on Wikipedia.

I may be wrong since I don't program CUDA neither I have a 750Ti in hand, but my impression is that NVidia catched up and fixed it's own problems, not surpassed (by much) the already efficient ATI design. Turning the tables would depend in impressive engineering feat by NVidia but also ATI sloppiness in it's current line, which nobody managed to point out yet.

Then there's always the physical limit of using the same TSMC silicon and 28nm technology.

I'd be happy if both are on an even playing field , that would likely normalize prices, unless even Nvidia and AMD combined can't meet demand from miners (which I doubt).
Indeed. That's exactly what I hope for.
member
Activity: 70
Merit: 10
A pretty good review of the actual Maxwell architecture changes can be found here:

http://techreport.com/review/26050/nvidia-geforce-gtx-750-ti-maxwell-graphics-processor
hero member
Activity: 840
Merit: 1000
The Barrel shifter is not present on 3.0 cards like the GTX760, but is already on 3.5 Titans. You can see that on Wikipedia.

I may be wrong since I don't program CUDA neither I have a 750Ti in hand, but my impression is that NVidia catched up and fixed it's own problems, not surpassed (by much) the already efficient ATI design. Turning the tables would depend in impressive engineering feat by NVidia but also ATI sloppiness in it's current line, which nobody managed to point out yet.

Then there's always the physical limit of using the same TSMC silicon and 28nm technology.

I'd be happy if both are on an even playing field , that would likely normalize prices, unless even Nvidia and AMD combined can't meet demand from miners (which I doubt).
sr. member
Activity: 840
Merit: 255
SportsIcon - Connect With Your Sports Heroes
The Barrel shifter is not present on 3.0 cards like the GTX760, but is already on 3.5 Titans. You can see that on Wikipedia.

I may be wrong since I don't program CUDA neither I have a 750Ti in hand, but my impression is that NVidia catched up and fixed it's own problems, not surpassed (by much) the already efficient ATI design. Turning the tables would depend in impressive engineering feat by NVidia but also ATI sloppiness in it's current line, which nobody managed to point out yet. Perhaps high latency on memory accesses is the only thing holding ATI back on mining.

Then there's always the physical limit of using the same TSMC silicon and 28nm technology.
member
Activity: 76
Merit: 10

Quote
One change in the new SM will be especially consequential for certain customers—and possibly for the entire GPU market. Maxwell restores a key execution resource that was left out of Kepler: the barrel shifter. The absence of this hardware doesn't seem to have negative consequences for graphics, but it means Kepler isn't well-suited to the make-work algorithms used by Litecoin and other digital currencies. AMD's GCN architecture handles this work quite well, and Radeons are currently quite scarce in North America since coin miners have bought up all of the graphics cards. The barrel shifter returns in Maxwell, and Nvidia claims the GM107 can mine digital currencies quite nicely, especially given its focus on power efficiency.

Would really like to hear Christians thoughts on this.
full member
Activity: 173
Merit: 100
What clocks are you running? Cheesy

I bet you can do better with some more time and of course, after Mr Buchner improves cudaminer for Maxwell cards. (looking forward to that!)

I cannot really promise that this produces actual speed benefits. There's no specific new features that a compilation for sm_50 would provide and which sm_35 (Titan kernel) doesn't already offer.

One can only hope that the compiler does some better code generation when targeting sm_50.

Christian


What about the Barrel Shifter that AMD crows so much about that Kepler was missing? This was added to Maxwell Smiley

I wanna see Christian DO A BARREL SHIFT!!! LOL  Grin
Jump to: