CCminer(SP-MOD) Modded GPU kernels.

[myagui]

I dug around the Kopiemtu thread for the relevant bits, and seems this is the most important one:
In xorg.conf, just before the coolbits statement:

Code:

Option         "AllowEmptyInitialConfiguration" "True"

If I read right, this is how one sets up fake monitors on headless systems.

This will not work for every driver version. Will probably require this command to setup a fresh xorg.conf if a driver update is needed.

Code:

sudo nvidia-xconfig −−enable-all-gpus

I dunno, since I don't have a Linux system currently to test this on. I'd already been able to setup fake monitors (manually) before, so I know didn't use this at the time, but I did have 1 monitor attached to the 1st GPU, so mine was not a fully headless setup.

Edit: better yet, check the source. Hashbrown looks to be doing good stuff over there! 
https://litecointalk.org/index.php?topic=16800.msg266088#msg266088

Cheers!

[1714660672]

1714660672

[pallas]

Thanks everyone, I'll give all this a try asap.
Hopefully I'll then be able to test small kernel enhancements :-)

[scryptr]

Or maybe on windows it's different...

Yeps. On Windows it is extremely simple to set fixed clocks, fans, overclocking, so you can easily have a "benchmark platform".

With a headless Linux system I don't think there's any solution for fixed fans yet, others might know differently. I previously had fixed fans and clocks on Linux, but I perfectly recall that I specifically had to configure/attach a monitor in order to get that working at the time.

to set the fan on linux just test this

https://gist.github.com/squadbox/e5b5f7bcd86259d627ed

Thanks, but I fear it needs a monitor or the x session will not start... Or will it?

NOT ANY MORE--

If you have one of the latest driver packages, you can use nvidia-config to enable cards without monitors to be adjusted:

"sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration" (Source: https://bitcointalk.org/index.php?topic=826901.msg12279696#msg12279696)

The numbeer "28" is a bitsum that signifies clocks, fanspeed, and power controls.

This was in this thread, earlier. I think that "-a" and the longer "--allow-empty-initial-configuration" are equivalent. THEY ARE NOT EQUIVALENT!  Command "-a"  = "--enable-all-gpus"  

--scryptr

[Grim]

Final release of CUDA 7.5 toolkit (including some fixes)!

[t-nelson]

I remember getting the number of SMM/SMXs in the CryptoNight miner... can't remember how, though.

Yep. TSIV's Cryptonight ccminer complains whenever the blocks/threads (don't recall which one now) is not a multiple of the SMX/SMM on the card. IIRC, it does this also accurately for cards that did not exist when it was released, so the approach is not LUT based.

Right - I can't remember if I wrote that or not.

Nice, I'll have to dig into these APIs a bit deeper once I get all of the cleanup I have planned done.

[myagui]

@t-nelson: extending the invite that every cuda miner developer has probably received at some point:
Some of the usual suspects (both developers & community) hang out at #ccminer @freenode. It is not the most active channel in freenode, but it is a great place to exchange ideas with the other developers and the few of us there always try to help each other out (testing, etc).

In case that IRC is not your usual thing, maybe webchat will get you there: #ccminer webchat

[DragonSlayer]

TPB52 7 is best for mining. My gtx 980 make 11.2-11.6 Mh. 960 gtx make 6 Mh. Super!
For rig (2x980+1x960) hash is 28500!
PS. It is LYRA2v2! +100 git 1052 works!

Would you share your command line?  I am running a 1x980 and 1x960 in a box.  I have the hardest time to configuring both together.

Thanks

[Slava_K]

TPB52 7 is best for mining. My gtx 980 make 11.2-11.6 Mh. 960 gtx make 6 Mh. Super!
For rig (2x980+1x960) hash is 28500!
PS. It is LYRA2v2! +100 git 1052 works!

Would you share your command line?  I am running a 1x980 and 1x960 in a box.  I have the hardest time to configuring both together.

Thanks

All works by default, card OC'd to 1380 Mhz. Ccminer recompiled with TBP52=7 setted in lyra2REve.cu. Thanks to _SP!

[bathrobehero]

frames? ... i build them custom ...

the frame ( 1.5mm square aluminium tubing ) - edge connectors - screws and lugs - angle iron ( 90 degree edging ) ... all this comes to approx $70.00aud ...

but initially i had to buy the cutting and measuring equipment as well as the files and mallets ( yes plastic head mallets - not hammers ) ...

parts alone - not counting labour - if you are looking at adding teh countless hours to design the thing in the first place ...

prototype 1 - prototype 2 and now prototype 3 is probably the design we are going with ...

hope that helps bathrobehero ...

#crysx

Thanks, it does. I have similar prices but personally, I don't find that to be worth it and I'm much more comfortable spending that ~50 USD into better PSUs.

I must say that I didn't have any problem benchmarking amd cards: if the room was hot, I'd put the fans at high speed, run the miner and wait a couple minute to stabilize, and that's it. I could make 100 changes to a kernel in a day and check them all, accurately.

On nvidia I have throttling problems I can't easily fix (the cards reduce clock speed in a number of situations I just can't predict), overclocking/downclocking is more difficult as the cards tend to change clocks by themselves, and the hashrates fluctuates wildly, and even changes between ccminer runs.
The rig is headless so I only have nvidia-smi to work with, and it can't set the fan speed.
So when I make a little kernel speedup, I spend more time benchmarking it (to be sure it's indeed an improvement), than making the improvement itself :-/
Maybe there are some nvidia-smi settings to make it more stable?
Or maybe on windows it's different...
Finally I may need a workstation with a nvidia as main card, and work on it.

You should probably disable boost in the bios of the card. Or even set custom fan speed/clock speeds if you want.

[sp_]

TPB52 7 is best for mining. My gtx 980 make 11.2-11.6 Mh. 960 gtx make 6 Mh. Super!
For rig (2x980+1x960) hash is 28500!
PS. It is LYRA2v2! +100 git 1052 works!

Would you share your command line?  I am running a 1x980 and 1x960 in a box.  I have the hardest time to configuring both together.
Thanks

I have submitted a fix to autoconfig bether launch configs on the gtx 980 and the gtx 960. The 980ti and the 950 is probobly off. @latest github

[sp_]

It may be time for me to use multiple work items on one Lyra2 execution... but I'm worried the increased memory accesses for the matrix (which I've worked to eliminate) will eat me alive.

In my latests fixes I added more memory accesses and speed..

Perhaps this is the opposite of what you are supposed to do when doing a modded kernal, but I think it was fun..
Only latency wankers working for NVIDIA it seems...


Still slow though. You know the opensource is slow, because the market makes it not profitable to mine.

Your optimized kernals might end up here wolf0, And I think they can afford to pay you enough:

[hashbrown9000]

@myagui, yes you are on point about enabling the fake X screens on multiple cards, (which then allows you to set fan speeds, mem clocks and core clocks with Nvidia X Server settings program).  Details here: https://litecointalk.org/index.php?topic=16800.msg266088#msg266088.

I am running an older driver (346.59 with CUDA 6.5.19, LUbuntu 14.04), so the attribute names and procedure may be slightly different.

[GingerAle]

anybody got the standard and sp_ mod hashrate for Monero on the new 950?

and anyone know if there's a way in linux to get this card to overclock?

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121980&ignorebbr=1&cm_re=PPSSKQKDFPGTFN-_-14-121-980-_-Product

[scryptr]

anybody got the standard and sp_ mod hashrate for Monero on the new 950?

and anyone know if there's a way in linux to get this card to overclock?

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121980&ignorebbr=1&cm_re=PPSSKQKDFPGTFN-_-14-121-980-_-Product

LINUX OVERCLOCKING--

You will need to do some web research.  It depends on your Linux drivers.  HashBrown has written and posted on this topic in the Kopiemtu blog thread, his work is for slightly older drivers.  Only a few posts back in this thread, I mentioned how to use nvidia-xconfig to set CoolBits.  This is possible with the very latest nVidia drivers.

Frankly, I would write a tutorial but I haven't updated my drivers and actually done it myself.  My system is stable, my 970 cards are getting 16.5-17Mh/s on Quark as is.  I have been reading all over the web about Linux overclocking, though.  Useful search terms include CoolBits, nvidia-smi,  nvidia-xconfig.  The very latest nVidia drivers simplify things; some of the posts that are 2-3 years old are outdated and involve more complicated configuration editing.

Apparently, the "nvidia-xconfig" command will write out a config for your system, but only with the latest drivers.       --scryptr

[blue-yu]

Question from a friend: He have a couple of Nvidia GT 9600. Can he mine anything with them?

[scryptr]

Question from a friend: He have a couple of Nvidia GT 9600. Can he mine anything with them?

I have a 9800 GTX+ card, the top of the line for the series.  It mined script at 14kh/s using CudaMiner.  My 550ti got 90kh/s, and the 520 that came in the system got 21kh/s.        --scryptr

[blue-yu]

Question from a friend: He have a couple of Nvidia GT 9600. Can he mine anything with them?

I have a 9800 GTX+ card, the top of the line for the series.  It mined script at 14kh/s using CudaMiner.  My 550ti got 90kh/s, and the 520 that came in the system got 21kh/s.        --scryptr

Good old cudaminer I assume that ccminer don't work with old cards like that?

[scryptr]

Question from a friend: He have a couple of Nvidia GT 9600. Can he mine anything with them?

I have a 9800 GTX+ card, the top of the line for the series.  It mined script at 14kh/s using CudaMiner.  My 550ti got 90kh/s, and the 520 that came in the system got 21kh/s.        --scryptr

Good old cudaminer I assume that ccminer don't work with old cards like that?

COMPUTE 1.0 -

CCminer works with compute 2.0 and up, depending on the coder who wrote it.  A 550ti is "Compute 2.1".  SP_ codes for compute 5.0 (Maxwell, 750ti) and later. A 9800 is Compute 1.0.  So, no luck, for any practical mining with the 9XXX series.  You could play Tomb Raider with ease, though.         --scryptr

[GingerAle]

anybody got the standard and sp_ mod hashrate for Monero on the new 950?

and anyone know if there's a way in linux to get this card to overclock?

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121980&ignorebbr=1&cm_re=PPSSKQKDFPGTFN-_-14-121-980-_-Product

LINUX OVERCLOCKING--

You will need to do some web research.  It depends on your Linux drivers.  HashBrown has written and posted on this topic in the Kopiemtu blog thread, his work is for slightly older drivers.  Only a few posts back in this thread, I mentioned how to use nvidia-xconfig to set CoolBits.  This is possible with the very latest nVidia drivers.

Frankly, I would write a tutorial but I haven't updated my drivers and actually done it myself.  My system is stable, my 970 cards are getting 16.5-17Mh/s on Quark as is.  I have been reading all over the web about Linux overclocking, though.  Useful search terms include CoolBits, nvidia-smi,  nvidia-xconfig.  The very latest nVidia drivers simplify things; some of the posts that are 2-3 years old are outdated and involve more complicated configuration editing.

Apparently, the "nvidia-xconfig" command will write out a config for your system, but only with the latest drivers.       --scryptr

awesome, thanks for the leads. Hrm, forgot the 950 needs a 6 pin power line...

[frazier34567]

Noticed something interesting on my 6 card EVGA 750ti SC miner.  When I use -C I get about 39250 Kh/s (No OC).   If I do not use it I get about 39000-39190 Kh/s.  Now if I mine Axiom with the nicehash cpu launch miner with just one thread running (No -C 30% load on CPU).  https://github.com/nicehash/cpuminer-multi/releases  I get 39235-39240 Kh/s  I am running a Intel i3 4330 (dual Core 4 thread CPU)

This is with 1.5.66 commit. 1056  Cuda 6.5 Windows 8.1 Pro Driver 355.60

my command line is ccminer.exe -o stratum+tcp://stratum.nicehash.com:3345 -a quark -q -u bitcoinaddress.2 -p d=0.04 --cpu-priority 3