italeffect
|
|
October 31, 2014, 09:33:30 PM |
|
I have a few 750ti x6 rigs running Windows 8.
I've always used EVGA Precision X to overclock and set fans. I recently discovered that if I open Precision to set the clocks and fans, launch ccminer, and then quit Precision - the hash rate goes up. Of course I loose the fan control but the overclocks stay.
Any theories on why this is? I first though maybe CPU, but it's never using higher then about 35% with Precision running and about 25% just ccminer running. Ram usage is about 30% of total.
On x11 it's the difference between 17.2 MH/s or 17.5 MH/s for 6 cards.
|
Dash: Xdopotr3eAHpsSCMkUyU2YWP3WQWb5X3t8
|
|
|
myagui
Legendary
Offline
Activity: 1154
Merit: 1001
|
|
October 31, 2014, 11:17:31 PM |
|
I have a few 750ti x6 rigs running Windows 8.
I've always used EVGA Precision X to overclock and set fans. I recently discovered that if I open Precision to set the clocks and fans, launch ccminer, and then quit Precision - the hash rate goes up. Of course I loose the fan control but the overclocks stay.
Any theories on why this is? I first though maybe CPU, but it's never using higher then about 35% with Precision running and about 25% just ccminer running. Ram usage is about 30% of total.
On x11 it's the difference between 17.2 MH/s or 17.5 MH/s for 6 cards.
The reason is that Precision X adds a few hardware locks when doing the hardware monitoring & particularly the software fan controls. You can tweak the settings in Precision X to make it less noticeable, but for the absolute maximum hashrate, you'll need to close Precision X...
|
|
|
|
|
jpouza
Legendary
Offline
Activity: 2828
Merit: 1119
|
|
November 01, 2014, 08:22:26 AM |
|
|
|
|
|
bathrobehero
Legendary
Offline
Activity: 2002
Merit: 1051
ICO? Not even once.
|
|
November 01, 2014, 10:27:38 AM |
|
I have a few 750ti x6 rigs running Windows 8.
I've always used EVGA Precision X to overclock and set fans. I recently discovered that if I open Precision to set the clocks and fans, launch ccminer, and then quit Precision - the hash rate goes up. Of course I loose the fan control but the overclocks stay.
Any theories on why this is? I first though maybe CPU, but it's never using higher then about 35% with Precision running and about 25% just ccminer running. Ram usage is about 30% of total.
On x11 it's the difference between 17.2 MH/s or 17.5 MH/s for 6 cards.
Decrese the polling rate to like 5000 ms.
|
Not your keys, not your coins!
|
|
|
Odeyin
Member
Offline
Activity: 86
Merit: 10
|
|
November 01, 2014, 05:02:51 PM |
|
unfortunately this time seems to be over... Actually...Gapcoin may be the next big opportunity to do this same thing right now..at least for an early GPU miner..and since it is dealing with primes..and Integral approach might blaze...
|
|
|
|
djm34
Legendary
Offline
Activity: 1400
Merit: 1050
|
|
November 01, 2014, 06:00:41 PM |
|
unfortunately this time seems to be over... Actually...Gapcoin may be the next big opportunity to do this same thing right now..at least for an early GPU miner..and since it is dealing with primes..and Integral approach might blaze... Is it on an exchange ?
|
djm34 facebook pageBTC: 1NENYmxwZGHsKFmyjTc5WferTn5VTFb7Ze Pledge for neoscrypt ccminer to that address: 16UoC4DmTz2pvhFvcfTQrzkPTrXkWijzXw
|
|
|
Odeyin
Member
Offline
Activity: 86
Merit: 10
|
|
November 01, 2014, 06:05:59 PM |
|
|
|
|
|
SRBOOTH
|
|
November 01, 2014, 07:12:02 PM |
|
Also here.... https://coin-swap.net/market/GAP/BTCBTW Gap supporters ARE interested in a GPU miner...trying to reach a mathematical world record, very interesting stuff at least. Fairly profitable at the moment as well, I can only speculate where it might go if that world record is achieved. And if it doesn't go anywhere at least it is an interesting endeavor as opposed to a useless one.
|
|
|
|
antonio8
Legendary
Offline
Activity: 1400
Merit: 1000
|
|
November 02, 2014, 04:18:11 AM |
|
Not sure if all you miners from Europe have seen this but here you go: http://www.ecoinmining.eu/news/bitcoin-wallet-platform-coinbase-opens-europe/Quote from inside article "The 13 countries Coinbase is adding include Italy, Spain, France, Belgium, the Netherlands, Austria, Cyprus, Finland, Greece, Latvia, Malta, Portugal, and Slovakia."
|
If you are going to leave your BTC on an exchange please send it to this address instead 1GH3ub3UUHbU5qDJW5u3E9jZ96ZEmzaXtG, I will at least use the money better than someone who steals it from the exchange. Thanks
|
|
|
italeffect
|
|
November 02, 2014, 06:52:02 AM |
|
I have a few 750ti x6 rigs running Windows 8.
I've always used EVGA Precision X to overclock and set fans. I recently discovered that if I open Precision to set the clocks and fans, launch ccminer, and then quit Precision - the hash rate goes up. Of course I loose the fan control but the overclocks stay.
Any theories on why this is? I first though maybe CPU, but it's never using higher then about 35% with Precision running and about 25% just ccminer running. Ram usage is about 30% of total.
On x11 it's the difference between 17.2 MH/s or 17.5 MH/s for 6 cards.
The reason is that Precision X adds a few hardware locks when doing the hardware monitoring & particularly the software fan controls. You can tweak the settings in Precision X to make it less noticeable, but for the absolute maximum hashrate, you'll need to close Precision X... Thank you (and bathrobehero). Such an obvious reason.
|
Dash: Xdopotr3eAHpsSCMkUyU2YWP3WQWb5X3t8
|
|
|
djm34
Legendary
Offline
Activity: 1400
Merit: 1050
|
|
November 02, 2014, 11:12:10 AM |
|
say I want to multiply by two some big integer (256bit ie 8 uint32_t) What would be the fastest ?
* multiplying by 2 with mad/madc ? * just adding the numbers to itself ? * shifting with shfl/shl ?
in terms of number of lines of code it seems to be more or less the same...
|
djm34 facebook pageBTC: 1NENYmxwZGHsKFmyjTc5WferTn5VTFb7Ze Pledge for neoscrypt ccminer to that address: 16UoC4DmTz2pvhFvcfTQrzkPTrXkWijzXw
|
|
|
bigjme
|
|
November 02, 2014, 01:41:39 PM |
|
Sooooo, what have i missed? Been a while since i checked in here, I don't get notifications any more for some reason but yehhhhhhh. How's everyone doing?
|
Owner of: cudamining.co.uk
|
|
|
Namsbreh
|
|
November 02, 2014, 02:53:13 PM |
|
Sooooo, what have i missed? Been a while since i checked in here, I don't get notifications any more for some reason but yehhhhhhh. How's everyone doing? Excellent thanks. Wondered where u had got too. We have had GANJA, DANK, FTC switching to Neoscrypt (with great profit on day 1). Devs have been working hard too, a couple of new miners have been released. Shame BTC keeps tanking.
|
|
|
|
bigjme
|
|
November 02, 2014, 03:12:42 PM |
|
Yeh the bitcoin price drop is a killer
|
Owner of: cudamining.co.uk
|
|
|
georgidimitrov
Member
Offline
Activity: 101
Merit: 10
|
|
November 02, 2014, 03:14:53 PM |
|
Is there any working neoscrypt miner with cuda support?
|
|
|
|
antonio8
Legendary
Offline
Activity: 1400
Merit: 1000
|
|
November 02, 2014, 03:47:37 PM |
|
Is there any working neoscrypt miner with cuda support?
Here you go: http://cryptomining-blog.com/category/mining-software/Second article. It will work for you.
|
If you are going to leave your BTC on an exchange please send it to this address instead 1GH3ub3UUHbU5qDJW5u3E9jZ96ZEmzaXtG, I will at least use the money better than someone who steals it from the exchange. Thanks
|
|
|
georgidimitrov
Member
Offline
Activity: 101
Merit: 10
|
|
November 02, 2014, 03:57:15 PM |
|
Thank you very much for the fast reply. Here comes my next question, is anyone working on neoscrypt mining with cudaminer?
|
|
|
|
Schleicher
|
|
November 02, 2014, 04:20:06 PM Last edit: November 02, 2014, 04:37:42 PM by Schleicher |
|
say I want to multiply by two some big integer (256bit ie 8 uint32_t) What would be the fastest ?
* multiplying by 2 with mad/madc ? * just adding the numbers to itself ? * shifting with shfl/shl ?
in terms of number of lines of code it seems to be more or less the same...
From the Programming Guide: 32bit multiply: multiple instructions 32bit add: 128 operations / clock cycle 32bit shift: 64 operations / clock cycle But I think in reality it's more complicated than that.
|
|
|
|
Odeyin
Member
Offline
Activity: 86
Merit: 10
|
|
November 02, 2014, 05:12:33 PM |
|
say I want to multiply by two some big integer (256bit ie 8 uint32_t) What would be the fastest ?
* multiplying by 2 with mad/madc ? * just adding the numbers to itself ? * shifting with shfl/shl ?
in terms of number of lines of code it seems to be more or less the same...
I think they are the same. But If we use the divide and conquer algorithm proposed by Dr. Fürer described here: http://wwwmath.uni-muenster.de/u/cl/WS20078/mult.pdfModular Multiplication. Maybe?
|
|
|
|
|