ATTN Litecoin GPU Miners - Scrypt support for cgminer

[Luke-Jr]

so if i am not mistaken the main difference between LTC and BTC was the fact that LTC was targeted at CPU mining only.

now that you can mine it on GPU, arent both currencies basically the same? making one redundant?

Litecoin was always redundant, and a scam. There is reason to believe the original designer of the scrypt algorithm had been GPU mining it from the first day. Also, it's impossible to make a CPU-only proof-of-work. Ironically, now that GPU mining is public, they're claiming it's ASIC-resistant which is even more absurd (nothing can be ASIC resistant).

[1714101078]

1714101078

[1714101078]

1714101078

[1714101078]

1714101078

[1714101078]

1714101078

[1714101078]

1714101078

[Tittiez]

There is already a windows build from ckolivas, why would someone download a windows version from somewhere else?

At the time of writing it there wasn't.

[smoothie]

so if i am not mistaken the main difference between LTC and BTC was the fact that LTC was targeted at CPU mining only.

now that you can mine it on GPU, arent both currencies basically the same? making one redundant?

Litecoin was always redundant, and a scam. There is reason to believe the original designer of the scrypt algorithm had been GPU mining it from the first day. Also, it's impossible to make a CPU-only proof-of-work. Ironically, now that GPU mining is public, they're claiming it's ASIC-resistant which is even more absurd (nothing can be ASIC resistant).

1. How is it a scam? Draw me a picture please.

2. Who lost money?

3. How is price fluctuations (like bitcoin) make it any more or less of a "scam"?

4. Wasn't ArtForz mining bitcoins with his GPU miner before any publicly released GPU miner existed? How is that any different (if it is true, which you have no proof of)?

5. "Nothing can be ASIC resistant". Well what you are doing is taking statements that are made by people who dont have a clue how litecoin works and using that to portray it as if every litecoin support believes that.

Do us all a set of favors:

1. Stop being a Douche.

2. Stop spreading bullshit and claiming that all LTC supporters  agree with certain statements.

3. Grow the fuck up. Most of the other early adopters/devs aren't as fucking childish as you are.

My rant has ended. Have a nice day!

 

[bitlane]

I seem to be experiencing a problem with the new CGMiner 2.6.1-official that was not affecting 2.5.0.Scrypt_beta_ee

CGMiner now seems to be seeing cards at times as sick and the program crashes. It has happened to each of my miners at least once since I switched from the 'EE' Beta 2.5.0 yesterday, to 2.6.1 Official.

The last beta ('EE') seemed to run fine for multiple days, but I have yet to experience a SOLID 24 hours from 2.6.1-Official.....without a crash.

The 'EE' version, for reference...

win32: http://www.mediafire.com/?0yxa1cf35pk5q2i

Latest Commits included!

[-ck]

so if i am not mistaken the main difference between LTC and BTC was the fact that LTC was targeted at CPU mining only.

now that you can mine it on GPU, arent both currencies basically the same? making one redundant?

Litecoin was always redundant, and a scam. There is reason to believe the original designer of the scrypt algorithm had been GPU mining it from the first day. Also, it's impossible to make a CPU-only proof-of-work. Ironically, now that GPU mining is public, they're claiming it's ASIC-resistant which is even more absurd (nothing can be ASIC resistant).

1. How is it a scam? Draw me a picture please.

2. Who lost money?

3. How is price fluctuations (like bitcoin) make it any more or less of a "scam"?

4. Wasn't ArtForz mining bitcoins with his GPU miner before any publicly released GPU miner existed? How is that any different (if it is true, which you have no proof of)?

5. "Nothing can be ASIC resistant". Well what you are doing is taking statements that are made by people who dont have a clue how litecoin works and using that to portray it as if every litecoin support believes that.

Do us all a set of favors:

1. Stop being a Douche.

2. Stop spreading bullshit and claiming that all LTC supporters  agree with certain statements.

3. Grow the fuck up. Most of the other early adopters/devs aren't as fucking childish as you are.

My rant has ended. Have a nice day!

 

Fatal mistake. Don't respond to trolls. The one thing that is MOST effective against them is ignoring them, or at most, talking about them in the 3rd person as though they're not present.

Luke-jr has failed to accept one of the universal truths: If you end up having a problem with the whole world, then perhaps it's not the whole world at fault.

[smoothie]

so if i am not mistaken the main difference between LTC and BTC was the fact that LTC was targeted at CPU mining only.

now that you can mine it on GPU, arent both currencies basically the same? making one redundant?

Litecoin was always redundant, and a scam. There is reason to believe the original designer of the scrypt algorithm had been GPU mining it from the first day. Also, it's impossible to make a CPU-only proof-of-work. Ironically, now that GPU mining is public, they're claiming it's ASIC-resistant which is even more absurd (nothing can be ASIC resistant).

1. How is it a scam? Draw me a picture please.

2. Who lost money?

3. How is price fluctuations (like bitcoin) make it any more or less of a "scam"?

4. Wasn't ArtForz mining bitcoins with his GPU miner before any publicly released GPU miner existed? How is that any different (if it is true, which you have no proof of)?

5. "Nothing can be ASIC resistant". Well what you are doing is taking statements that are made by people who dont have a clue how litecoin works and using that to portray it as if every litecoin support believes that.

Do us all a set of favors:

1. Stop being a Douche.

2. Stop spreading bullshit and claiming that all LTC supporters  agree with certain statements.

3. Grow the fuck up. Most of the other early adopters/devs aren't as fucking childish as you are.

My rant has ended. Have a nice day!

 

Fatal mistake. Don't respond to trolls. The one thing that is MOST effective against them is ignoring them, or at most, talking about them in the 3rd person as though they're not present.

Luke-jr has failed to accept one of the universal truths: If you end up having a problem with the whole world, then perhaps it's not the whole world at fault.

Words of wisdom.

Thanks.

[-ck]

I seem to be experiencing a problem with the new CGMiner 2.6.1-official that was not affecting 2.5.0.Scrypt_beta_ee

CGMiner now seems to be seeing cards at times as sick and the program crashes. It has happened to each of my miners at least once since I switched from the 'EE' Beta 2.5.0 yesterday, to 2.6.1 Official.

The last beta ('EE') seemed to run fine for multiple days, but I have yet to experience a SOLID 24 hours from 2.6.1-Official.....without a crash.

The 'EE' version, for reference...

win32: http://www.mediafire.com/?0yxa1cf35pk5q2i

Latest Commits included!

You know it's interesting, I'm in the middle of tracking down the cause of this exact issue for someone else. It turns out that the change to avoid flooding pools on longpoll means I limit outgoing connections - this means there is more likely to be a period a GPU is idle while waiting for work. Now what's interesting is that short periods of idle followed by huge bursts of activity again tend to make cards more unstable than running them flat out 100% of the time.  Some testing showed the cards needed to have their clocks decreased quite a lot to be able to tolerate this. So overclocking which was fine before suddenly becomes unstable. I'm experimenting with a feature I call no-idle which gives GPUs fake work to keep them busy to see if this improves stability, but it's still early stages. If you have more stability with the scrypt patched 2.5.0, then by all means go with that for now.

[bitlane]

I seem to be experiencing a problem with the new CGMiner 2.6.1-official that was not affecting 2.5.0.Scrypt_beta_ee

CGMiner now seems to be seeing cards at times as sick and the program crashes. It has happened to each of my miners at least once since I switched from the 'EE' Beta 2.5.0 yesterday, to 2.6.1 Official.

The last beta ('EE') seemed to run fine for multiple days, but I have yet to experience a SOLID 24 hours from 2.6.1-Official.....without a crash.

The 'EE' version, for reference...

win32: http://www.mediafire.com/?0yxa1cf35pk5q2i

Latest Commits included!

You know it's interesting, I'm in the middle of tracking down the cause of this exact issue for someone else. It turns out that the change to avoid flooding pools on longpoll means I limit outgoing connections - this means there is more likely to be a period a GPU is idle while waiting for work. Now what's interesting is that short periods of idle followed by huge bursts of activity again tend to make cards more unstable than running them flat out 100% of the time.  Some testing showed the cards needed to have their clocks decreased quite a lot to be able to tolerate this. So overclocking which was fine before suddenly becomes unstable. I'm experimenting with a feature I call no-idle which gives GPUs fake work to keep them busy to see if this improves stability, but it's still early stages. If you have more stability with the scrypt patched 2.5.0, then by all means go with that for now.

Could this problem simply be a bi-product of the switch to displaying MH/s while SCRYPT mining in 2.6.1 ?

In the 2.5.0 'EE' SCRYPT Beta, CGMiner displayed KH/s exclusively (GPU, Rig Summary/5s/AVG & API).
In 2.6.1, while SCRYPT mining, CGMiner seems to mix the 2 scales using KH/s for individual GPUs and then MH/s for Rig Summary/5s/AVG & API output.

Isn't this item the most significant difference that has changed from 2.5.0beta to 2.6.1official ? (for SCRYPT mining anyway).

Could the 'sickness' simply be a low value output in MH/s, causing CGMiner to panic ?

Would it be possible that when in --SCRYPT mode that the units displayed are 100% in KH/s ? (GPU, Rig Summary/5s/AVG & API output ?)
Or, perhaps have the option to specify in the CMD line using something like [ --units khash] or [--units mhash] ?

...Just a thought.
thanks again,
bitlane.

[BlackBison]

so has anyone tested cgminer with 5850s? is it faster than reaper and therefore worth me switching over, or is it roughly the the same hashrate?

[EskimoBob]

...
Would it be possible that when in --SCRYPT mode that the units displayed are 100% in KH/s ? (GPU, Rig Summary/5s/AVG & API output ?)
Or, perhaps have the option to specify in the CMD line using something like [ --units khash] or [--units mhash] ?

As you probably have noticed, there is no space for the Kh/s in the upper row when you are in 1000 range
Mine shows Kh/s  

[bitlane]

...
Would it be possible that when in --SCRYPT mode that the units displayed are 100% in KH/s ? (GPU, Rig Summary/5s/AVG & API output ?)
Or, perhaps have the option to specify in the CMD line using something like [ --units khash] or [--units mhash] ?

As you probably have noticed, there is no space for the Kh/s in the upper row when you are in 1000 range
Mine shows Kh/s  

You sure about THAT ?



<EDIT>

My API Web App displays whatever it's fed and does so automatically (kh rather than the MH as I showed above)...
(although values are dynamic, units are static in the app and need to be changed....not a big deal)

[burnside]

so if i am not mistaken the main difference between LTC and BTC was the fact that LTC was targeted at CPU mining only.

now that you can mine it on GPU, arent both currencies basically the same? making one redundant?

Nah, no FPGA's or ASIC's for LTC. (thus still accessible by the common man) LTC will have more coins in the wild.  Faster confirms.  LTC ftw.

edit... delayed reaction.  doh.

[-ck]

Could the 'sickness' simply be a low value output in MH/s, causing CGMiner to panic ?

No, but that bug WAS in 2.6.0 which was the main reason for 2.6.1

We have found that this problem you describe happens with driver 12.6 when it doesn't happen with the earlier drivers which were better (specifically the so-called 8.921 release which was the first beta driver supporting 7970s).

[VelvetLeaf]

That means the settings aren't at best. Play with it, try lower concurrencys and higher ones. The settings won't be the same as reaper. Try other worksizes.

2x5970, got 1022 kH/s with this parameter :
--scrypt --thread-concurrency 6144 --intensity 13 --lookup-gap 2 --gpu-threads 1 --gpu-memclock 1100 --gpu-engine 775 --auto-fan --temp-target 75 --worksize 256

Still less than 1.2 MH/s from reaper with the same configuration.
But temperature so far is good, better than reaper's temperature.

EDIT : Got 1.2 MH/s with intensity 15
I'm afraid to raise intensity because of the warning in readme : "Intensities over 13 start writing over the same ram and it is highly dependent on the GPU, but they can start actually DECREASING your hashrate, or even worse, start producing garbage with rejects skyrocketing."

EDIT 2 : Still fluctuating, now at 1.0 MH/s average.

[SAC]

That means the settings aren't at best. Play with it, try lower concurrencys and higher ones. The settings won't be the same as reaper. Try other worksizes.

2x5970, got 1022 kH/s with this parameter :
--scrypt --thread-concurrency 6144 --intensity 13 --lookup-gap 2 --gpu-threads 1 --gpu-memclock 1100 --gpu-engine 775 --auto-fan --temp-target 75 --worksize 256

Still less than 1.2 MH/s from reaper with the same configuration.
But temperature so far is good, better than reaper's temperature.

EDIT : Got 1.2 MH/s with intensity 15
I'm afraid to raise intensity because of the warning in readme : "Intensities over 13 start writing over the same ram and it is highly dependent on the GPU, but they can start actually DECREASING your hashrate, or even worse, start producing garbage with rejects skyrocketing."

EDIT 2 : Still fluctuating, now at 1.0 MH/s average.

2x5970, gets me 1388 kH/s, 347Kh/s per core with these parameters :
--scrypt --thread-concurrency 8000 --intensity 18 --lookup-gap 2 --gpu-threads 2 --gpu-memclock 1050 --gpu-engine 750 --auto-fan --auto-gpu --temp-target 81 --worksize 256

Stales at ~1-3% depending on how fast the network is running as most of them come on the new blocks.

[VelvetLeaf]

2x5970, gets me 1388 kH/s, 347Kh/s per core with these parameters :
--scrypt --thread-concurrency 8000 --intensity 18 --lookup-gap 2 --gpu-threads 2 --gpu-memclock 1050 --gpu-engine 750 --auto-fan --auto-gpu --temp-target 81 --worksize 256

Stales at ~1-3% depending on how fast the network is running as most of them come on the new blocks.

Got cgminer isn't responding. What's your driver version ?

[SAC]

2x5970, gets me 1388 kH/s, 347Kh/s per core with these parameters :
--scrypt --thread-concurrency 8000 --intensity 18 --lookup-gap 2 --gpu-threads 2 --gpu-memclock 1050 --gpu-engine 750 --auto-fan --auto-gpu --temp-target 81 --worksize 256

Stales at ~1-3% depending on how fast the network is running as most of them come on the new blocks.

Got cgminer isn't responding. What's your driver version ?

Ubuntu 11.04 12.6 display driver SDK 2.7.

[pekv2]

Is there a way to implement an option to add thread-concurrency configuration through cgminer.exe?

Everytime I do a [W], cgminer overwrites my config file, removing

Code:

"thread-concurrency" : "7200,7200",

from cgminer.conf.

Or is it already there and I am overlooking it?

[kano]

Is there a way to implement an option to add thread-concurrency configuration through cgminer.exe?

Everytime I do a [W], cgminer overwrites my config file, removing

Code:

"thread-concurrency" : "7200,7200",

from cgminer.conf.

Or is it already there and I am overlooking it?

https://github.com/ckolivas/cgminer/pull/279

[pekv2]

https://github.com/ckolivas/cgminer/pull/279

Oh, so it is a known problem. mk. Thanks.