[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX]

[Eli0t]

now poolers cpu miner (which cudaminer is based on) has stratum support https://bitcointalk.org/index.php?topic=55038.msg2451455#msg2451455
would it be easier to implement stratum in cudaminer?

[BlIiTtZz]

Thanks for this Christian; LTC coming your way as soon as I actually have some.

After careful configuration, driver upgrade (620.18 works sooo much better for me, despite others' troubles), and memory overclock, I'm pulling:
238kH/s from my dual GeForce 680Ms.

Not bad for a laptop. And it's chilling at 70degC.

I might try to hack stratum support into this if I get time. noo promises.

Nice1 {=

[420]

if someone can get me over 180mhash for stock gtx 680 khash you'll get a finder's fee. PM me

[K1773R]

if someone can get me over 180mhash for stock gtx 680 khash you'll get a finder's fee. PM me

this is scrypt not sha256d

[Schleicher]

Okay it was my AV, i've disabled it for now. As for the batch file I created it doesn't seem to be working. it keeps going to default settings.
maybe I typed it wrong?  "mining_proxy.exe -pa scrypt -o pool.de1.bigltc.stratum.cc  -p 8336"

Should be working.
My batch file looks like this:

Code:

@echo off
D:
cd \stratumproxy-ltc
mining_proxy.exe -pa scrypt -nm --host coinotron.com --port 3334
pause

And cudaMiner:

Code:

cudaminer -O name:password -o http://localhost:8332 -C 2 -i 0 --no-autotune -l 14x8

[cbuchner1]

There are some new affordable cards on the market, based on the GK208 chip. They are called GT640, clocked at 902 MHz and they have the same hardware features as the GeForce Titan. That is compute capability 3.5.

I expect that they obtain a formidable scrypt hashing power for their 45 Watts peak power consumption.

I ordered one, so I can finally optimize and debug this Titan kernel without relying on third parties for test runs

Christian

P.S. Watch out that you don't get the older GT640 models based on the GK107 chip, if you buy one.

[brioche]

There are some new affordable cards on the market, based on the GK208 chip. They are called GT640, clocked at 902 MHz and they have the same hardware features as the GeForce Titan. That is compute capability 3.5.

I expect that they obtain a formidable scrypt hashing power for their 45 Watts peak power consumption.

I ordered one, so I can finally optimize and debug this Titan kernel without relying on third parties for test runs

Christian

P.S. Watch out that you don't get the older GT640 models based on the GK107 chip, if you buy one.

Oh where did buy it from? Oh okay never mind I see one on eBay now, cool.

[BlIiTtZz]

Hey Christian,

Updated Google sheet with my results after flashing the vBIOS of my dual 680Ms (SLI) (letting me to mess with the GPU Clock), and pushing the GPU Clock to +135MHz and the Memory Clock to +350MHz.

Now pulling 280kH/s from just under 2V which is GRAVY! AND even though I'm not in interactive mode, it's having cray little slow-down effect on my laptop.


One thing that I noticed is that with the OC settings in place, auto-tune acted all kinds of batshit when I was trying to tune both cards at once. Took upwards of an hour and debug output was out of sync and held no reality to the actual speeds I could pull.

So I just tuned one card at a time, that worked out fine.


Time to get desktop on it meow. Let us know how you go with that GT640 an if I buy one I'll totefully hit you with some LTC 4 the ref.

I dont know enough about mining to compare this for me, but how do you think our nvidia mining is stacking up against amd mining now? Are we at like 60%, or higher?


Be perfect!! =!= all my bests

[GoldBit89]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

[BrewCrewFan]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

That is not something new. I get that from time to time... not sure what causes this. It seems to happen way more on my 660 than the 450.

[BrewCrewFan]

Hey Christian,

Updated Google sheet with my results after flashing the vBIOS of my dual 680Ms (SLI) (letting me to mess with the GPU Clock), and pushing the GPU Clock to +135MHz and the Memory Clock to +350MHz.

Now pulling 280kH/s from just under 2V which is GRAVY! AND even though I'm not in interactive mode, it's having cray little slow-down effect on my laptop.


One thing that I noticed is that with the OC settings in place, auto-tune acted all kinds of batshit when I was trying to tune both cards at once. Took upwards of an hour and debug output was out of sync and held no reality to the actual speeds I could pull.

So I just tuned one card at a time, that worked out fine.


Time to get desktop on it meow. Let us know how you go with that GT640 an if I buy one I'll totefully hit you with some LTC 4 the ref.

I dont know enough about mining to compare this for me, but how do you think our nvidia mining is stacking up against amd mining now? Are we at like 60%, or higher?


Be perfect!! =!= all my bests

Price per card vs return is very poor when using Nvidia. I guess if you currently have em then its much better... but I would not go out to buy one.

[mhps]

"result does not validate on CPU!"

Do a search and find out more.

[Lacan82]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

It means you're using an invalid configuration for your card. If you were using an old version and upgrade. You must Autotune with the new version. The old configuration may not be valid anymore

[BrewCrewFan]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

It means you're using an invalid configuration for your card. If you were using an old version and upgrade. You must Autotune with the new version. The old configuration may not be valid anymore

Thats Bullshit. All my cards auto tune every time and still get the error.

[Charles999]

i am watching this aswell, hope it works

[K1773R]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

It means you're using an invalid configuration for your card. If you were using an old version and upgrade. You must Autotune with the new version. The old configuration may not be valid anymore

Thats Bullshit. All my cards auto tune every time and still get the error.

yes its bullshit.

result does not validate happen and has nothing to do with the card config.

your mining for pool X, connection gets lost and in meantime pool X finds a new block and cudaminer dosnt get the LP, connection restores gets new work but the work thats being put on the GPU is still the old one, which obviously dosnt match the actual work. after this no LP/reconnct fixes it, you have to restart it.
i created a simple bash + expect script to catch this and automatically restart cudaminer

PS: if you look at his posts you see a high % of his post are total bullshit or just FUD -> https://bitcointalk.org/index.php?action=profile;u=95137;sa=showPosts

[cbuchner1]

the "does not validate on CPU" message really indicates a computation error. Try running without texture cache to see if it makes a difference. Also maybe lower any aggressive overclocking you may be running.

[Lacan82]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

It means you're using an invalid configuration for your card. If you were using an old version and upgrade. You must Autotune with the new version. The old configuration may not be valid anymore

Thats Bullshit. All my cards auto tune every time and still get the error.

yes its bullshit.

result does not validate happen and has nothing to do with the card config.

your mining for pool X, connection gets lost and in meantime pool X finds a new block and cudaminer dosnt get the LP, connection restores gets new work but the work thats being put on the GPU is still the old one, which obviously dosnt match the actual work. after this no LP/reconnct fixes it, you have to restart it.
i created a simple bash + expect script to catch this and automatically restart cudaminer

PS: if you look at his posts you see a high % of his post are total bullshit or just FUD -> https://bitcointalk.org/index.php?action=profile;u=95137;sa=showPosts

my posts are bullshit? did you read any of these thread? CPU validation was put in to make sure your card was doing work. When you get that area it means your card isn't actually doing any work at all.

[UNOE]

What kind of yield does a Titan get

[BrewCrewFan]

i have discovered a weird error: i have been getting shares ok and been reporting to pools ok and having them match and since i am new to this still virtually especially with mining, this is a first of this error i have seen on any of the forums.
It appears that when i have found and solved a block, i get this error and i am not credited with finding it but the pool gets credit for solving it tho jsut with a "unknown" finder. error is this:

"result does not validate on CPU!"

this is the latest version and i love the program, as it has boosted my hashes greatly but i could not figure out this odd bug.

anyone have any ideas?

It means you're using an invalid configuration for your card. If you were using an old version and upgrade. You must Autotune with the new version. The old configuration may not be valid anymore

Thats Bullshit. All my cards auto tune every time and still get the error.

yes its bullshit.

result does not validate happen and has nothing to do with the card config.

your mining for pool X, connection gets lost and in meantime pool X finds a new block and cudaminer dosnt get the LP, connection restores gets new work but the work thats being put on the GPU is still the old one, which obviously dosnt match the actual work. after this no LP/reconnct fixes it, you have to restart it.
i created a simple bash + expect script to catch this and automatically restart cudaminer

PS: if you look at his posts you see a high % of his post are total bullshit or just FUD -> https://bitcointalk.org/index.php?action=profile;u=95137;sa=showPosts

my posts are bullshit? did you read any of these thread? CPU validation was put in to make sure your card was doing work. When you get that area it means your card isn't actually doing any work at all.

So getting that every so often means the card is not doing the work?

Anyways I think I came off wrong  in what I said before as being forceful or what not. That was not my intentions ...just the general way of talking with people sometimes.

Back to the issue.

-I have starting paying more attention to when the errors do occur. Its seems to be only on the 1 slot GPU 660.
-Only happens on pools. If I solo I can go all day long and not see this issue.
-I cant seem to figure out if its the other flags like -m "x", or the -C "x"... I always auto tune for the most part because getting that extra 5khash really is not worth the headach.

When I have more time I can sit and switch out cards or whatnot and play around with it..

Bottom line is that I should say I am happy this program IS out there because otherwise it would be totally worthless to mine at all on nvidia.