cudaminer issues with dual nvidia GTX 590s

[npp1993]

Hello,


  I recently started mining on my rig with two Nvidia GTX 590s.  I use this command for cudaminer:

Code:

./cudaminer -H 1 -d 0,1 -i 0,0 -l F16x16,F16x16 -C 1,1 -m 1,1 -a scrypt -o stratum+tcp://ftc.give-me-coins.com:3336 -O user.1:1

Here is the output i get:

Code:

	   *** CudaMiner for nVidia GPUs by Christian Buchner ***
	             This is version 2013-12-01 (beta)
	based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
	       Cuda additions Copyright 2013 Christian Buchner
	   My donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm

[2013-12-04 19:46:42] 2 miner threads started, using 'scrypt' algorithm.
[2013-12-04 19:46:42] Starting Stratum on stratum+tcp://ftc.give-me-coins.com:3336
[2013-12-04 19:46:43] GPU #1: GeForce GTX 590 with compute capability 2.0
[2013-12-04 19:46:43] GPU #1: interactive: 0, tex-cache: 1D, single-alloc: 1
[2013-12-04 19:46:43] GPU #1: using launch configuration F16x16
[2013-12-04 19:46:43] GPU #1: GeForce GTX 590, 8192 hashes, 18.80 khash/s
[2013-12-04 19:46:44] GPU #0: GeForce GTX 590 with compute capability 2.0
[2013-12-04 19:46:44] GPU #0: interactive: 0, tex-cache: 1D, single-alloc: 1
[2013-12-04 19:46:44] GPU #0: using launch configuration F16x16
[2013-12-04 19:46:44] GPU #0: GeForce GTX 590, 8192 hashes, 9.32 khash/s
[2013-12-04 19:46:44] GPU #0: GeForce GTX 590, 565248 hashes, 968.70 khash/s
[2013-12-04 19:46:46] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:48] GPU #1: GeForce GTX 590, 1130496 hashes, 225.70 khash/s
[2013-12-04 19:46:49] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:50] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:51] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:52] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:55] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:55] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:56] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:56] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:57] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:57] GPU #1: GeForce GTX 590, 1949696 hashes, 228.38 khash/s
[2013-12-04 19:46:57] accepted: 1/1 (100.00%), 1197 khash/s (yay!!!)
[2013-12-04 19:46:57] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:58] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:46:59] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:00] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:00] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:01] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:02] Stratum detected new block
[2013-12-04 19:47:02] GPU #0: GeForce GTX 590, 17874944 hashes, 1018 khash/s
[2013-12-04 19:47:02] GPU #1: GeForce GTX 590, 1146880 hashes, 227.74 khash/s
[2013-12-04 19:47:05] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:05] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:06] GPU #0: GeForce GTX 590 result does not validate on CPU!
[2013-12-04 19:47:07] GPU #1: GeForce GTX 590, 1048576 hashes, 227.39 khash/s
[2013-12-04 19:47:07] accepted: 2/2 (100.00%), 1246 khash/s (yay!!!)
...

When I log into my pool account, i get a reported hashrate of around 220 KH/s.  Obviously, the 1000+ KH/s generated by GPU #0 are doing nothing (or at least that's what it seems like to me).

My questions for you guys:

Why the "result does not validate on CPU" message on GPU #0?  What does it mean? How can I fix it (if possible)?

Do you guys have any recommendations for settings I should use instead for my setup?  How can I get GPU #0 to actually mine?

I appreciate the help, this forum was instrumental in helping me set up my rig.

[Raptor2213]

As I understand it, that error means that it miscalculated the hash, rendering it's work invalid.  I messed with Cudaminer a bit myself and gave up because I could never get anything decent out of it - kept getting lots of "does not validate on CPU" errors too.  I kinda stopped caring once I got my ASIC BlockErupters.

[Gcr0n]

you might want to download an older version, latest does not support -H

[npp1993]

would the lack of support for the -H option cause this issue?

right now I've been playing around with passing different options to the -l flag.  If I leave it on auto tuned settings, then I don't have any issues. 

would anyone have any idea on what kind of KH/s rate I could expect out of this setup?

[Gcr0n]

i had sort of the same message when i used MSI afterburner..

do you use some kind of program to overclock your card?

[ruf_xu]

 

[npp1993]

i had sort of the same message when i used MSI afterburner..

do you use some kind of program to overclock your card?

nope, no overclocking.  i'm running on autotuned  settings now, getting about 220KH/s combined between both cards.  Would it be possible for me to get 220KH/s out of both cards at the same time?  Because the gain from running cudaminer with both cards is pretty marginal compared to running it with just one card (something close to 30KH/s).

[Kam800]

From the CUDAMiner readme

Code:

--texture-cache  [-C] list of flags (0 or 1 or 2) to enable use of the 
                 texture cache for reading from the scrypt scratchpad.
                 1 uses a 1D cache, whereas 2 uses a 2D texture layout.
                 Cached operation has proven to be slightly faster than
                 noncached operation on most GPUs.

--single-memory  [-m] list of flags (0 or 1) to make the devices
                 allocate their scrypt scratchpad in a single,
                 consecutive memory block. On Windows Vista, 7/8
                 this may lead to a smaller memory size being used.
                 When using the texture cache this option is implied.

Have you tried removing these flags? It seems '-m' implies '-C'; it's worth comparing results between removing these and just including '-m'

Are you using an SLI bridge?