CCminer(SP-MOD) Modded GPU kernels.

[flipclip]

any hint how to use API? never even tried it...
just point my browser to api port ?

Add
--api-bind
to your command line of ccminer, then use a telnet program (putty if using windows) to telnet to 127.0.0.1 port 4068 (default setting). When it connects type:
threads
and hit enter.  A bunch of info will spit out, the bit you are interested in is "I=".  If you have multiple cards, you will have multiple I's (along with multiples of all the other information).  You can set different intensities per card with -i 21.8,22
If you cannot connect via telnet try:
--api-bind <IP Address of Computer>:4068
(example: --api-bind 192.168.2.2:4068) then try telneting to <IP Addres of Computer> port 4068

[Longsnowsm]

What kind of speeds are we seeing now with the latest build on the 750ti's?  I am finally about to do my nvidia rig build and would like to get the maximum perf from them possible.  Would love your input.  Thanks.

[restless]

10x!

But its not working :S

Win 8.1 - miner starts, binds to the port, when I telnet it just gices a blinking marker and on first letter i write disconnects and goes back to cmd prompt
Same behaviour on tpruvot 1.75 and sp-mod release 80 :/
Putty is just showing blank screen - no reaction :/

BTw, sp_mod release 80 build for Windows tells that its
SP-Mod 1.5.79
Compiled with Visual C++ 18 using Nvidia CUDA Toolkit 7.5

Binary is build on 5th of May so this is just a typo I guess

[Ayers]

What kind of speeds are we seeing now with the latest build on the 750ti's?  I am finally about to do my nvidia rig build and would like to get the maximum perf from them possible.  Would love your input.  Thanks.

are 750ti still a thing? nothing seems profitable with those little gpu lol, even quark is better on a 980ti

[crysx]

What kind of speeds are we seeing now with the latest build on the 750ti's?  I am finally about to do my nvidia rig build and would like to get the maximum perf from them possible.  Would love your input.  Thanks.

are 750ti still a thing? nothing seems profitable with those little gpu lol, even quark is better on a 980ti

they are VERY unique cards - that hash well and work well even in the heaviest pressure areas of mining ( except ethereum due to the dag ) ...

they are cheap in comparison - and on a price / watt basis - are still a very good card ...

though now the pascal chipset is out - we will probably see a decline in mining with these cards - and a flood up for sale ... which is one of the reasons i didnt sell all mine - just a few ...

#crysx

[restless]

Given the memory bus of 1080&1070, they won't be very good for ethereum - 256-bit bus GDDR5(x) at max

[bathrobehero]

What kind of speeds are we seeing now with the latest build on the 750ti's?  I am finally about to do my nvidia rig build and would like to get the maximum perf from them possible.  Would love your input.  Thanks.

are 750ti still a thing? nothing seems profitable with those little gpu lol, even quark is better on a 980ti

I don't have 980 Ti's but I'm pretty sure I have a much shorter time to reach ROI with 750 Ti's than it would be with 980 Ti's.

970's are the best for it though - at least for me.

Given the memory bus of 1080&1070, they won't be very good for ethereum - 256-bit bus GDDR5(x) at max

AFAIK bandwidth is not important, latency is.

Imagine a highway and bandwidth is basically how many lanes are on it while latency is how fast the cars are. We need fast cars in this case, not more lanes. Someone correct me if I'm wrong.

[restless]

GDDR5x is faster than GDDR5 - because can send more bytes per second (aka sdram-ddr-ddr2)
Looking Radeon speed in ethereum, it seems that wider bus gives most benefit (fury>290>280>270). Perhaps because bus is partitioned on 32/64-bit chinks with different memory controller/bus per chunk.
So more chunks - better latency.
1080 has 256-bit bus... so any benefit will come from faster controller or bigger caches or better management of caching (bigger memory pages, TLB etc)

[kahir]

will esper coin be implemented in the new version coming out soon ?

[rednoW]

What kind of speeds are we seeing now with the latest build on the 750ti's?  I am finally about to do my nvidia rig build and would like to get the maximum perf from them possible.  Would love your input.  Thanks.

are 750ti still a thing? nothing seems profitable with those little gpu lol, even quark is better on a 980ti

I don't have 980 Ti's but I'm pretty sure I have a much shorter time to reach ROI with 750 Ti's than it would be with 980 Ti's.

970's are the best for it though - at least for me.

Given the memory bus of 1080&1070, they won't be very good for ethereum - 256-bit bus GDDR5(x) at max

AFAIK bandwidth is not important, latency is.

Imagine a highway and bandwidth is basically how many lanes are on it while latency is how fast the cars are. We need fast cars in this case, not more lanes. Someone correct me if I'm wrong.

No, you are wrong. Car speed on highway is a memory frequency. But latency is a time needed for car to reach full speed if for some reason it was forced to stop before.

[Ayers]

What kind of speeds are we seeing now with the latest build on the 750ti's?  I am finally about to do my nvidia rig build and would like to get the maximum perf from them possible.  Would love your input.  Thanks.

are 750ti still a thing? nothing seems profitable with those little gpu lol, even quark is better on a 980ti

I don't have 980 Ti's but I'm pretty sure I have a much shorter time to reach ROI with 750 Ti's than it would be with 980 Ti's.

970's are the best for it though - at least for me.

Given the memory bus of 1080&1070, they won't be very good for ethereum - 256-bit bus GDDR5(x) at max

AFAIK bandwidth is not important, latency is.

Imagine a highway and bandwidth is basically how many lanes are on it while latency is how fast the cars are. We need fast cars in this case, not more lanes. Someone correct me if I'm wrong.

980ti was an example, 970 is much better tha 750ti currently, i think, and with 750ti you can not mine etheruem anymore

[bathrobehero]

No, you are wrong. Car speed on highway is a memory frequency. But latency is a time needed for car to reach full speed if for some reason it was forced to stop before.

I see. That makes sense.

980ti was an example, 970 is much better tha 750ti currently, i think, and with 750ti you can not mine etheruem anymore

Yes, currently I have about 140 days to ROI per 750 Ti and about 118 days per 970 after electricity.
I haven't mined Ethereum in weeks though.

For me a 980 Ti would cost exactly twice as much as a 970 but it's far from being twice as fast. I can't say for sure but the 980 Ti is probably the worst Maxwell card to buy for mining - unless you can get them very cheap.

The 1070 will likely be the winner for Pascal as well but we'll see.

[joblo]

No, you are wrong. Car speed on highway is a memory frequency. But latency is a time needed for car to reach full speed if for some reason it was forced to stop before.

I see. That makes sense.

980ti was an example, 970 is much better tha 750ti currently, i think, and with 750ti you can not mine etheruem anymore

Yes, currently I have about 140 days to ROI per 750 Ti and about 118 days per 970 after electricity.
I haven't mined Ethereum in weeks though.

For me a 980 Ti would cost exactly twice as much as a 970 but it's far from being twice as fast. I can't say for sure but the 980 Ti is probably the worst Maxwell card to buy for mining - unless you can get them very cheap.

The 1070 will likely be the winner for Pascal as well but we'll see.

I can't resist these types of discussions. Latency, using the highway analogy, is the time it takes to get from point A to
point B. Acceleration is ignored, you are assumed to be instantly at full speed. Throughput is how frequently the cars arrive
after the first one.

[Longsnowsm]

Well the weather is nasty outside today so I have been trying to get this Nvidia rig up and running.  I just cobbled together the basics, downloaded the Nvidia drivers, and the free copy of sp_ ccminer.  I only have one gpu connected to the mobo at the moment and wanted to see if it would at least run before I connect up the other 5 gpu's.  I start ccminer and it runs for about 10 seconds and then the nvidia driver crashes.  What is the preferred nvidia driver version to be running with these cards using ccminer?  This rig is Windows 7, i3 Haswell cpu, 16gb ram, 128gb ssd, and 6 ea 750ti's with 4gb ram. Any tips to get this rig up and running would be greatly appreciated. I am connecting to zpool.ca and attempting to mine neoscrypt.  Am I just missing some switches or configuration to make this work?  This is the first time trying to run ccminer. Thanks.

Added: I downloaded the 1.5.80, but now wonder if I should be running the 1.5.74 instead.  My nvidia driver version is 365.19.

Longsnowsm

[joblo]

Well the weather is nasty outside today so I have been trying to get this Nvidia rig up and running.  I just cobbled together the basics, downloaded the Nvidia drivers, and the free copy of sp_ ccminer.  I only have one gpu connected to the mobo at the moment and wanted to see if it would at least run before I connect up the other 5 gpu's.  I start ccminer and it runs for about 10 seconds and then the nvidia driver crashes.  What is the preferred nvidia driver version to be running with these cards using ccminer?  This rig is Windows 7, i3 Haswell cpu, 16gb ram, 128gb ssd, and 6 ea 750ti's with 4gb ram. Any tips to get this rig up and running would be greatly appreciated. I am connecting to zpool.ca and attempting to mine neoscrypt.  Am I just missing some switches or configuration to make this work?  This is the first time trying to run ccminer. Thanks.

Added: I downloaded the 1.5.80, but now wonder if I should be running the 1.5.74 instead.  My nvidia driver version is 365.19.

Longsnowsm

The only time I've seen a driver crash is when I OC too much. I presume you tried a different card.

[bathrobehero]

Well the weather is nasty outside today so I have been trying to get this Nvidia rig up and running.  I just cobbled together the basics, downloaded the Nvidia drivers, and the free copy of sp_ ccminer.  I only have one gpu connected to the mobo at the moment and wanted to see if it would at least run before I connect up the other 5 gpu's.  I start ccminer and it runs for about 10 seconds and then the nvidia driver crashes.  What is the preferred nvidia driver version to be running with these cards using ccminer?  This rig is Windows 7, i3 Haswell cpu, 16gb ram, 128gb ssd, and 6 ea 750ti's with 4gb ram. Any tips to get this rig up and running would be greatly appreciated. I am connecting to zpool.ca and attempting to mine neoscrypt.  Am I just missing some switches or configuration to make this work?  This is the first time trying to run ccminer. Thanks.

Added: I downloaded the 1.5.80, but now wonder if I should be running the 1.5.74 instead.  My nvidia driver version is 365.19.

Longsnowsm

You should increase the swapfile to something like 8GB for a 750 Ti rig or 16GB for higher end cards.

[bensam1231]

Well the weather is nasty outside today so I have been trying to get this Nvidia rig up and running.  I just cobbled together the basics, downloaded the Nvidia drivers, and the free copy of sp_ ccminer.  I only have one gpu connected to the mobo at the moment and wanted to see if it would at least run before I connect up the other 5 gpu's.  I start ccminer and it runs for about 10 seconds and then the nvidia driver crashes.  What is the preferred nvidia driver version to be running with these cards using ccminer?  This rig is Windows 7, i3 Haswell cpu, 16gb ram, 128gb ssd, and 6 ea 750ti's with 4gb ram. Any tips to get this rig up and running would be greatly appreciated. I am connecting to zpool.ca and attempting to mine neoscrypt.  Am I just missing some switches or configuration to make this work?  This is the first time trying to run ccminer. Thanks.

Added: I downloaded the 1.5.80, but now wonder if I should be running the 1.5.74 instead.  My nvidia driver version is 365.19.

Longsnowsm

The only time I've seen a driver crash is when I OC too much. I presume you tried a different card.

OCing too much, poor power delivery, crappy risers.

[Longsnowsm]

Thanks for the replies on this. Everything was on stock settings as I hadn't had a chance to try and tweak anything.  So I rebooted the box a couple more times and tried to watch MSI Afterburner to see if I could figure out what was happening.  Sure enough when I tried to start ccminer the power, and temp shot to the moon and then crashed the video driver.   Not sure why since it was still at the 1060 core speed and 1350 mem speed.  So I cranked it down to like 600mhz on the core and turned down the power to 80%.  I then started the miner and it would start and run at like 40khs.  I slowly put things back to the stock setting in MSI AB.  However the speeds did not increase from the card.  I noticed on the runs where it was crashing the speed was about 200khs before it crashed.  

So now when I try to adjust the speed up in MSI AB it still does not increase the mining speed.  So I don't know if I need to unlock these cards or if there is some trick to getting any speed from them.  I am still baffled why it would behave this way out of the box and act like it was being super OC'd and crash.  Now I can't seem to get it to pick the speed back up.  

Thougths? I checked the physical and virtual memory and both are at 16gb so it appears it should have enough swap.  Thanks again for the feedback.

Update: I just told MSI AB to reset, then I rebooted, and then tried to start ccminer and it appeared to overload the gpu again and just quit.  I did see the has rate increase to 140-160khs just before it hung.  So Windblows is updating right now, then I will try a different card to see if the same thing happens.

[bathrobehero]

Thanks for the replies on this. Everything was on stock settings as I hadn't had a chance to try and tweak anything.  So I rebooted the box a couple more times and tried to watch MSI Afterburner to see if I could figure out what was happening.  Sure enough when I tried to start ccminer the power, and temp shot to the moon and then crashed the video driver.   Not sure why since it was still at the 1060 core speed and 1350 mem speed.  So I cranked it down to like 600mhz on the core and turned down the power to 80%.  I then started the miner and it would start and run at like 40khs.  I slowly put things back to the stock setting in MSI AB.  However the speeds did not increase from the card.  I noticed on the runs where it was crashing the speed was about 200khs before it crashed.  

So now when I try to adjust the speed up in MSI AB it still does not increase the mining speed.  So I don't know if I need to unlock these cards or if there is some trick to getting any speed from them.  I am still baffled why it would behave this way out of the box and act like it was being super OC'd and crash.  Now I can't seem to get it to pick the speed back up.  

Thougths? I checked the physical and virtual memory and both are at 16gb so it appears it should have enough swap.  Thanks again for the feedback.

Update: I just told MSI AB to reset, then I rebooted, and then tried to start ccminer and it appeared to overload the gpu again and just quit.  I did see the has rate increase to 140-160khs just before it hung.  So Windblows is updating right now, then I will try a different card to see if the same thing happens.

How many GPUs do you have and what's your power supply model?

[joblo]

Thanks for the replies on this. Everything was on stock settings as I hadn't had a chance to try and tweak anything.  So I rebooted the box a couple more times and tried to watch MSI Afterburner to see if I could figure out what was happening.  Sure enough when I tried to start ccminer the power, and temp shot to the moon and then crashed the video driver.   Not sure why since it was still at the 1060 core speed and 1350 mem speed.  So I cranked it down to like 600mhz on the core and turned down the power to 80%.  I then started the miner and it would start and run at like 40khs.  I slowly put things back to the stock setting in MSI AB.  However the speeds did not increase from the card.  I noticed on the runs where it was crashing the speed was about 200khs before it crashed.  

So now when I try to adjust the speed up in MSI AB it still does not increase the mining speed.  So I don't know if I need to unlock these cards or if there is some trick to getting any speed from them.  I am still baffled why it would behave this way out of the box and act like it was being super OC'd and crash.  Now I can't seem to get it to pick the speed back up.  

Thougths? I checked the physical and virtual memory and both are at 16gb so it appears it should have enough swap.  Thanks again for the feedback.

Update: I just told MSI AB to reset, then I rebooted, and then tried to start ccminer and it appeared to overload the gpu again and just quit.  I did see the has rate increase to 140-160khs just before it hung.  So Windblows is updating right now, then I will try a different card to see if the same thing happens.

How many GPUs do you have and what's your power supply model?

I think he's trying to get one card working.

I think it's a HW problem, GPU or MB, maybe power but unlikely with only one midrange GPU.
Maybe a bad PCIe connection, are you using a riser? powered? Maybe a bad slot, try another slot.
Swapping GPUs should nail it.