[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 1083.

aa

hero member

Activity: 544

Merit: 500

Litecoin is right coin

Hey everyone, I want to update this chart: https://litecoin.info/Mining_hardware_comparison#NVIDIA Especially people with GPUs that aren't listed, newer GPUs, or if you have better hashrates on your model than what's listed. You can add the info yourself if you register and get your account approved with the CryptoPayment system (it completely stops spammers, coin goes to Litecoin Association to fund the server).

I can add the info for you: ~~PM me the info that's in the chart, please (referenced below).~~ Add the info to the URL below!

Model, kH/s, Watts (if you know it), Core (MHz), Memory (MHz), Miner (I guess, only if it's different from the current version in OP), after that I'd like the flags used, the hardware specs for the rest of the system, OS, drivers, etc.

I'd appreciate it greatly, and I'm sure all NVIDIA Litecoin miners would appreciate having an updated chart to refer to.

Oh, along with the info, a link to a screenshot of cudaMiner running would be a nice bonus.

Editing to say that I made this to easily add your information: http://www.lurkmore.com/hardware-compare/

muto

sr. member

Activity: 457

Merit: 250

Did anybody have this problem? Have had it tonight. For some reason the hashrate seemed to drop a little bit and the cudaminer freezed. It didn't react to CTRL+C this morning.

Eli0t

sr. member

Activity: 252

Merit: 250

ah the 64 extra cuda cores would explain it, iv got a 384 version of the 560ti. still managed to get a few more kh up to 144 now

and almost 0 cpu usage 90% of the time, other version was going from 2 to 4%. thanks for sharing

fruittool

newbie

Activity: 19

Merit: 0

Quote from: Eli0t on September 24, 2013, 08:27:43 AM

Quote from: fruittool on September 24, 2013, 05:12:01 AM

I have compiled the new repo for 64 bit. Meant i had to compile a 64 bit libcurl but now im getting up to 228khash/s but at the same time getting some nasty cpu usage.
This is a 2.4GHz Core2Quad overclocked to 3GHz.

228kh is amazing for a 560ti, can you share this build?

I can but, don't be surprised if it craps out on anything other than a 560ti 448.

http://www.filedropper.com/cudaminerx64

Eli0t

sr. member

Activity: 252

Merit: 250

Quote from: fruittool on September 24, 2013, 05:12:01 AM

I have compiled the new repo for 64 bit. Meant i had to compile a 64 bit libcurl but now im getting up to 228khash/s but at the same time getting some nasty cpu usage.
This is a 2.4GHz Core2Quad overclocked to 3GHz.

Adding an extra sleep before MyStreamSynchronize returns seems to keep it down without impacting performance.

228kh is amazing for a 560ti, can you share this build?

Eli0t

sr. member

Activity: 252

Merit: 250

try -i 0 -C 2 -m 1 -l 14x8

EDIT oops missed a page

muto

sr. member

Activity: 457

Merit: 250

Quote from: fruittool on September 24, 2013, 04:41:30 AM

I can make two suggestions. Make sure the -C flag is optimal for your card. I get best performance using -C 1. You could try -i 0 but this might make desktop lag unbearable, it depends on the power of your card.

Thanks for that, the -i 0 made the difference. GPU load is now constantly at 99% and not between 85% and 98%. -C 2 is still the best value for my card.

Undervolting to 925 mV is stable for me. Further undervolting got the cudaminer to stop mining.

I'm mining now with constant 66°C at 53% fan speed

fruittool

newbie

Activity: 19

Merit: 0

I have compiled the new repo for 64 bit. Meant i had to compile a 64 bit libcurl but now im getting up to 228khash/s but at the same time getting some nasty cpu usage.
This is a 2.4GHz Core2Quad overclocked to 3GHz.

http://s24.postimg.org/yfxb7fa4l/Capture.jpg

Adding an extra sleep before MyStreamSynchronize returns seems to keep it down without impacting performance.

http://s18.postimg.org/mr5rvzk49/Captureb.jpg

fruittool

newbie

Activity: 19

Merit: 0

Quote from: muto on September 24, 2013, 01:37:34 AM

I think in point of overclocking this is near te optimum. I'm wondering if soemeone could suggest some flags for the miner, which could improve the hashing power?

I can make two suggestions. Make sure the -C flag is optimal for your card. I get best performance using -C 1. You could try -i 0 but this might make desktop lag unbearable, it depends on the power of your card.

muto

sr. member

Activity: 457

Merit: 250

At stock my card got about 90 kH/S. But since the 1GB Version does have 256bit memoryinterface (instead of 192bit) it seems legit, that the card is faster at stock settings.

Mine is now between 110 and 118 kH/S and ran undervolted to 937mV stable for more the 12 hours now. Will try further undervolting to save energy.

I think in point of overclocking this is near te optimum. I'm wondering if soemeone could suggest some flags for the miner, which could improve the hashing power?

Stoneysilence

member

Activity: 104

Merit: 10

Quote from: muto on September 23, 2013, 05:54:24 AM

Does anybody mine with a GTX460 768mb RAM? I realy wondered how good it's performance is. Getting about 110-115KH/s after about 10 minutes of really basic performance tweaking. Since the GPU is at about 85% load on average, I think there is room for more.

Currently my card is overclocked to 830/2000 and undervolted to 962 mV. I'm using the following parameters: -C 2 -l 28x4

Does anybody have an idea about further improving the hashing power?

Edit: further overclocking to 835/2000 got the Card to perform 3-4 KH/ better. Further undervolting to 950 mV seems to be stable.

I have a 1GB GTX460 780/1800 and I only get about 90-110KH/s (average 105). Mine is stock speeds though. My card gets real unstable when I OC.

muto

sr. member

Activity: 457

Merit: 250

Does anybody mine with a GTX460 768mb RAM? I realy wondered how good it's performance is. Getting about 110-115KH/s after about 10 minutes of really basic performance tweaking. Since the GPU is at about 85% load on average, I think there is room for more.

Currently my card is overclocked to 830/2000 and undervolted to 962 mV. I'm using the following parameters: -C 2 -l 28x4

Does anybody have an idea about further improving the hashing power?

Edit: further overclocking to 835/2000 got the Card to perform 3-4 KH/ better. Further undervolting to 950 mV seems to be stable.

CrowJuice

newbie

Activity: 2

Merit: 0

I am having a problem with Cudaminer. If I use any other port then 8337 then it automatically tries to connect to use a Stratum connection. I am trying to connect to a LongPool connection with the following line:
cudaminer.exe --url=http://mining.eu.hypernova.pw:9332 --user=user.1 --pass=pass -q -d 0

fruittool

newbie

Activity: 19

Merit: 0

Quote from: cbuchner1 on September 13, 2013, 04:38:06 AM

Try passing the flag -H1 for a bit more speed. I get 225 kHash/s on 560Ti 448 core. The default is -H 0 which uses a non-parallel approach on the CPU to do the SHA256 hashing - while it reduces CPU load by about a third it also takes some percent off my kHash values on the fastest cards.

I am actually using -C 1 -i 0 -l F14x16 -H 1. What are your clock speeds? Here i have stock 732/1900.

I'll try downloading your updated repo and a fresh compile.

*edit* No, still the same.

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: fruittool on September 13, 2013, 04:26:45 AM

There is some increase in reported speed. Before, i was using 28x8 with a GTX560Ti 448 core and peaking around 210khash/s. Now using F14x16 im getting up to 218khash/s.

Not much more but still more. Great work, thanks.

First, congrats for being able to set up compilation on Windows. This certainly is no piece of cake.

That windows compilation issue has been fixed meanwhile, also the x16 warps issue you noted.

Try passing the flag -H1 for a bit more speed. I get 225 kHash/s on 560Ti 448 core. The default is -H 0 which uses a non-parallel approach on the CPU to do the SHA256 hashing - while it reduces CPU load by about a third it also takes some percent off my kHash values on the fastest cards.

fruittool

newbie

Activity: 19

Merit: 0

Hi Christian. I've tried your new repo under windows. Just needed to #undef HAVE_ALLOCA_H and HAVE_SYSLOG_H in cpuminer-config.h to get it to compile under vc2010 and then changed max_warps_per_block() to return 16 in fermi_kernel.h to make use of the extra WARPS_PER_BLOCK.

There is some increase in reported speed. Before, i was using 28x8 with a GTX560Ti 448 core and peaking around 210khash/s. Now using F14x16 im getting up to 218khash/s.

Not much more but still more. Great work, thanks.

mhps

hero member

Activity: 516

Merit: 500

CAT.EX Exchange

Quote from: cbuchner1 on September 09, 2013, 06:30:45 AM

Quote from: mhps on September 09, 2013, 06:25:56 AM

Is there a windows binary coming out soon?

after some more days of testing and tinkering (a timeframe of about a week).

Thanks. I have a Tesla c2075.

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: mhps on September 09, 2013, 06:25:56 AM

Is there a windows binary coming out soon?

after some more days of testing and tinkering (a timeframe of about a week).

mhps

hero member

Activity: 516

Merit: 500

CAT.EX Exchange

Quote from: cbuchner1 on September 09, 2013, 05:18:43 AM

The brave among you might want to try out this code repo under Linux (it's a straight fork from pooler's cpuminer with my CUDA additions).

Is there a windows binary coming out soon?

cbuchner1

hero member

Activity: 756

Merit: 502

The brave among you might want to try out this code repo under Linux (it's a straight fork from pooler's cpuminer with my CUDA additions).

https://github.com/cbuchner1/cpuminer

There are 4 kernels, accessible with the monikers
L - Legacy (for compute 1.x devices)
F - Fermi (for Fermi class devices, compute 2.x)
K or S - Kepler kernel (using spinlocks to guard shared memory, compute 3.0)
T - Tesla (compute 3.5)

to autotune for a specific kernel, you can just pass the letter representing the specific kernel to the -l option, otherwise it just picks the kernel that matches your architecture.

some of the kernels (Fermi and Kepler) have been sped up a bit using optimizations I've received from a nice guy named Alex from Greece. 5-15% speed up can be obtained. Note that sometimes a kernel for an older architecture may run at same or better speed than the kernel for your hardware architecture. Wink

Currently getting 52 kHash/s on GT 640M (compute 3.5) and GT 750M (compute 3.0)

Open issues:
-the Fermi kernel currently doesn't run for me on Linux 64bit on a GT750M with compute 3.0
-still not enough error checking is done in the CUDA code
-that Stratum parse error and subsequent protocol freeze

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 1083. (Read 3426918 times)