Author

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 734. (Read 2347659 times)

legendary
Activity: 1764
Merit: 1024
Anyone else having issues with decred not showing correct speeds on pool with #3 and #4?

Coinmine, suprnova, and mnpool all have the same problem. Missing about 25% of my speed, 100% accepts.
legendary
Activity: 1400
Merit: 1000
Anyone know of a solution that allows changing clock speeds from the command line in windows? Nvidia-smi -ac doesn't seem to work on the core clock, it only really changes the mem speed.
I tried changing the clock speeds to as low as 8xx Mhz but it's still hashing away at 14xx Mhz. And nvidia-smi also doesn't allow disabling GPU Boost.

With decred mod 4 I get 1725 Mh/s per 970 and 590 Mh/s per 750 Ti with medium OC (970 WFOC: 1465-1478 Mhz, 750 Ti: 1288 Mhz).
I could get more with higher OC but that would not be stable for some other algos and I'm switching between coins a lot so it would be nice if I could switch between different OC presets from within the .bat files.

Ps.: that 1900 Mh/s is just due to oscillation (just like the 989 Mh/s before that) and it's not realistic (yet  Grin).

Try using Nvidia Inspector.

Click on "show overclock".

Change to P0 in the drop down menu. Top right you might have to click "Unlock Max". Use the Base Clock Offset and then when done click on the "Apply Clocks & Voltage".

I believe P2 is used for the memory.

At least this works for me.
legendary
Activity: 3164
Merit: 1003
Get it?
Got it now.  thx

EDIT: The first one 86x is about the same as 64x. Mod 4 I have --cpu-priority 3....mod 4 cpu  no --cpu-priority 3.
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
legendary
Activity: 3164
Merit: 1003
The 64 bit build has been sendt to the donators.

I am now working on the vanillacoin sp-mod #1. You can test the beta version in the 64bit build (the hash is broken, but the speed will be similar or faster)


Edit:

I get  +7% +8.5% on the gtx 960mobile. And fixed the broken hash.
I didn't receive mine sp...... I still owe you .0175 BTC ...will send tomorrow. Smiley

Thx
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Tnx for testing. Nice results.
legendary
Activity: 1176
Merit: 1015
Some quick decred sp-mod #4 tests:

Windows 7 gtx 970 +22%

Windows 10 gtx 980 g1 +18%

OC'd gtx 980 g1 did 2330 2330 on windows 10.

-edit-

Ok I didn't know you can overclock this much. Insane!

Need to test that 980 g1 again.

-edit2-

No more hash for poor gtx 980. Gtx 970 win7 1.9GH vs gtx 980 win10 2.3GH.
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Vanilla coin sp-mod #1 is now available for a 0.2BTC donation It includes the decred sp-mod #4. The decred donators will get a discount of 0.1BTC. There wil be no more updates on the Decred algo. You need to upgrade to the Decred+Vanilla miner to recieve decred #5.

Vanillacoin 8.5% faster:


sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
The 64 bit build has been sendt to the donators.

I am now working on the vanillacoin sp-mod #1. You can test the beta version in the 64bit build (the hash is broken, but the speed will be similar or faster)


Edit:

I get   +7% +8.5% on the gtx 960mobile. And fixed the broken hash.
legendary
Activity: 3164
Merit: 1003
Something came up. I will make another build in a few hours
ok  Smiley

1.7.4 64x does higher hashrate then 1.7.4  86x
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Something came up. I will make another build in a few hours
legendary
Activity: 3164
Merit: 1003
x86 cuda 7.5
64x gives higher hashrate  that's how come my lower hashrate.

I am building x64 version now, +a version without the blocking sync. (more cpu usage)
More cpu....ok I'll try.


EDIT: then i'll go to yiimp pool.
Suprnova is having problems.

sp what should I put for the -p options
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
x86 cuda 7.5
64x gives higher hashrate  that's how come my lower hashrate.

I am building x64 version now, +a version without the blocking sync. (more cpu usage)
legendary
Activity: 3164
Merit: 1003
x86 cuda 7.5
64x gives higher hashrate  that's how come my lower hashrate.
Can you compile it in 64x plz sp   see if that works.
thx
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
10min. test, 970@1586 core reaches 1919mh/s.Need any other proof? Smiley)) First card Msi can fluctates while Remote connect,don't know why..
8178.52 Mh/s total with 5 cards equals 1635.7 Mh/s per card. Those 19xx figures are being offset by the sub 1000 numbers. Both the MSI and EVGA cards of yours fluctuate like that for some reason. I don't have that:

He said that when he remote login to the rig, one of the cards (the msi card dropped to 900MHASH) This is a known issue when remoting into a mining rig. Its bether to user teamviewer or another tool to remote to the rig,
legendary
Activity: 3164
Merit: 1003
x86 cuda 7.5
64x gives higher hashrate  that's how come my lower hashrate.
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
x86 cuda 7.5
legendary
Activity: 3164
Merit: 1003
Wonder if I need to add anything to my bat file like   -s 10 or difficulty setting?

Just got up to 5.1 gh/s now dropping.

Mine at yiimp.ccminer.org or zpool.ca
There's no profit there sp. Undecided

Mine for 24 hours and you will see profit. The decred blocks are slow to confirn. It takes 24 hours
I know.
but did you compile DECRED r 4 in 86x  or 64x  ?
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Wonder if I need to add anything to my bat file like   -s 10 or difficulty setting?

Just got up to 5.1 gh/s now dropping.

Mine at yiimp.ccminer.org or zpool.ca
There's no profit there sp. Undecided

Mine for 24 hours and you will see profit. The decred blocks are slow to confirn. It takes 24 hours
legendary
Activity: 3164
Merit: 1003
Wonder if I need to add anything to my bat file like   -s 10 or difficulty setting?

Just got up to 5.1 gh/s now dropping.

Mine at yiimp.ccminer.org or zpool.ca
There's no profit there sp. Undecided

EDIT: did you compile DECRED r 4 in 86x  or 64x  ?
Jump to: