Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 671. (Read 2347641 times)

That is very interesting crysx, and is matching a pattern I think that I am seeing here also.  I am still in testing at the moment and hopefully should have a better feel by tomorrow where I stand with these cards.

My mobo is a ASRock H97 Anniversary.  My GPU's are Gigabyte Windforce 750ti's. 

I can honestly say that you are wrong bensam1231.

no - you dont seem to be doing anything wrong at all mate ...

if you read back quite a few pages - i had the same problem with 6 x gigabyte 980ti g1 / extreme cards ...

after almost two weeks of research and stuffing around - the fix? ...

run 5 cards ...

it seems that the nvidia driver craps itself when 6 cards are placed altogether ... i can assure you that i am no novice to mining - and you will find the same thing on your end ... i have yet to find a 'real' fix for this issue ...

6 x gigabyte 750ti oc lp cards work fine - but the 980ti g1 / extreme cards refuse to and stop functioning at 6 cards ...

i run fedora 23 x64 - with the latest drivers from the repo ( not the manually downloaded and installed nvidia drivers from their site ) and it is the same issue - time and time again ...

when thefarm is in its new home - ill be in a better position to test on different motherboards also - as i honestly believe it is a BIOS allocation issue ... i feel VERY confident that if i placed the cards on a supermicro server board that is built for hpc ( larger BIOS and designed to handle VGA allocation in BIOS ) - it will run fine with these big cards ...

what is the make and model of your motherboard? ...

#crysx

Please stop quoting Bensam.........it makes the ignore function kinda pointless!!!

I don't want to start another conversation with you, Bensam, because I know it leads nowhere.
Let me just say you are, again, recalling what you want and forgetting who made the miners in the first place.
There is Wolf0's name even in the old opencl code for scrypt, but you surely don't know or don't want to remember.
Now show me YOUR contributions :-D

Why don't you rehash the 'whole miners history'? I guess I don't have that down pat.

Are we going to talk about the part where he releases his kernels after they get leaked and people try to sell them, so it looks like a act of 'good faith'... or the part where he graciously makes miners (usually getting paid by the developers) for coins that no one uses or cares about?

He knows where the money is and he knows what miners need. He refuses to sell to small or even medium size miners... even large miners with the coin if he doesn't like them... I lost a good amount of money to that fucktwat due to his schenanagins and he talks about me running my mouth?

Lets look at the real world for a second: https://www.ftc.gov/enforcement/anticompetitive-practices

Specifically, this guy: https://www.ftc.gov/tips-advice/competition-guidance/guide-antitrust-laws/single-firm-conduct/exclusive-supply-or

...but, you know... it doesn't matter cause this is all online, right?

You just forgot the whole miners history. Or you are remembering only what you want to.

So where are your miners you openly sell to the public? Oh yeah...

Say hello to the guy that fucked over everyone except the giant farms on AMD's side with his private kernels. Don't worry, I didn't forget.

How is your dual miner coming along or optimized kernel for DCR?

It's great and everything when devs see kernel development as a hobby and they put time into whatever they feel like, whenever they feel like it... but mining is a job for some of us and we need developers that see it as a job. I am very much willing to pay for it... and have, but you can't pay money for something that either does not exist or someone wont sell you.

Good plan, just make sure you rotate the 6th GPU into a working slot and also try with a working GPU
in the 6th slot but with another slot empty so there are fewer than 6 cards installed. That should isolate
the problem of one of:

- the GPU
- the slot
- the number of GPUs installed

Thanks everyone for the advice on how to get this Nvidia rig to run with ccminer.  Last night I tore the rig all down and started from scratch, replaced a bunch of powered risers with versions that I think are better quality.  I checked the mobo PCI-E slots with a good GPU to validate that the rig boots and works normally in each of the PCIE slots.  Then attempted to plug in each of the GPU's as a single GPU to make sure they booted and worked.  Next I tried to plug in each GPU one at a time adding to the previous one to see if there were any issues.  The rig seems fine until you plug in that last(6th) GPU then it acts strange...

So I decided the only way to figure out if I have a bad GPU in the bunch is to plug in one GPU and launch the miner and let it run overnight if it will stay up and not crash... So I did that last night, connected a single GPU and ran the miner.  No intensity settings are set.  It mined fine, stayed up all night.  So this morning first thing I connected a second GPU and let it run all day with just 2 GPU's.  It ran all day, no issues.  This evening I connected a 3rd GPU and will let it run until the morning.  I will keep adding GPU's until I start to see the "validate on CPU" errors and crashing. 

I thought maybe I was having the issue where the number of GPU's isn't supported, but after digging around today that issue appeared to be related to AMD GPU's, and I didn't see anything about Nvidia gpu's having that issue.  But I am starting to wonder what might be going on if I get down to the 6th GPU and the errors return... I don't want to get the cart before the horse here, but it seems kind of strange. 

I will have to order some more powered risers.  I used the last of my spares last night replacing ones that I suspected might be an issue.   Hopefully things will stay stable with ccminer.  I am using the affinity 0, cpu priority 0, -C switches to startup the miner, but I left out the intensity and it seems to be working so far.  Any other thoughts or inputs would be great.  Thanks again...

I am sorry to hear that SP_ is talking about retiring.  It must be a great feeling to be ready to retire, but we really do need his talent here... I hope he will reconsider for a little while longer... :-)

yes sure :p please stop saying shit please

What opensource code competes with what SP puts out? The only reason any of the other developers feel a reason to contribute is some convoluted form of rivalry or superiority they need to try and flaunt over him when he puts out a new version.

Part of the reason I skipped off of AMD was because of the work the developers put into Nvidia mining software, most of that was being perpetuated by SP. If SP skips out, Nvidia mining will more then likely suffer greatly or just end up in the same boat as AMD, where only a few people are allowed to pay a arm and a leg for private kernels and everyone else is very much SOL.

Simply because there's enough competition, an opensource one with similar (if not comparable with intensity) speed.

Imagine the satisfaction of whoever donated, due this immediate retirement. Easy money, no obligations.

You and everyone elses mom. The real test is going to be see what sort of performance the 1xxx series gets with Ethereum. If it's no better then current cards there is really no reason to upgrade.

Ethereum is still the only GPU crypto keeping everything afloat. If it goes PoS everyone is completely fucked.

why he want to retire, i think he was making money with his private miner, if he can support pascal and offer some other private miner, he can rise some other btc, i really need someone that can do the pascal miner, i'm, going to invest big in them

O, that sweet word "retirement"! There is a problem, the amount of money to retire is comparable with cost of creating new mining IC. If I'll ever be in such situation the temptation will be too strong

Double-check all connectors, those @PSU and all others. USB risers must not be pressed etc. Powered risers must use several lanes/cables...

Sorry about that scryptr.  I have been dialing back intensity and I currently am at 12.  It still is dying.  I am running 1.5.80 on Win 7.  I have tried coupling the affinity and priority in the startup and the -C and still have this thing dying.

The rig details:
EVGA 1000w P2 PSU
ASRock H97 Anniversary Mobo
6 Windforce 750ti's
i3 Haswell proc
16gb ram
128gb SSD
Powered USB risers

It has configured 16gb of swap space.  Task manager shows cpu generally below 16-17%, memory stays around 2.5gb used. 

Miner is 1.5.80, mining Neoscrypt, on Win 7.  Sorry for the lack of details... Still delirious from smacking my head on the keyboard I guess! ROFL

yes it may take a while but it will happen eventually, none here is saying that it will happen in 7 weesk, in the same date of the halving