Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 802. (Read 3426922 times)

Yacoin autotune still kills my Nvidia driver when I use -i 1:

cudaminer.exe -a scrypt-jane -i 1 -L 3 -o stratum+tcp://yac.coinmine.pl:9088 -O pato.3:password -H 2 -d 0 -C 2 -b 2048 -l f
....................
[2014-02-13 17:11:23] DEBUG: job_id='90a8' extranonce2=00000000 ntime=52fceeaa
[2014-02-13 17:11:23] Stratum detected new block
[2014-02-13 17:11:28]   8:0.062|0.122|0.182|0.239|0.294|0.346|0.392|0.435|
                          0.468|0.500|0.518|0.645|0.674|0.712|0.834|0.888|
                          0.913|0.966|0.978|1.031|     |     |     |     |
                               |     |     |     |     |     |     |     | kH/s

----------------------> around here display driver crashes and windows force closes cudaminer

[2014-02-13 17:11:51] GPU #0: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (C:/Users/Patrick/Desktop/cudaminer/cudami
ner/salsa_kernel.cu line 635)

[2014-02-13 17:11:51]   9: 69.4|137.5|204.5|269.2|330.6|389.5|441.3|489.4|
                          527.3|562.1|581.9|602.1|607.8|618.8|621.0|631.6|
                               |     |     |     |     |     |     |     |
                               |     |     |     |     |     |     |     |  H/s
[2014-02-13 17:11:51] GPU #0: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (C:/Users/Patrick/Desktop/cudaminer/cudami
ner/salsa_kernel.cu line 635)

[2014-02-13 17:11:51] GPU #0: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (C:/Users/Patrick/Desktop/cudaminer/cudami
ner/salsa_kernel.cu line 635)

The last line continues now a few hundred times more until the following stands at the end:

[2014-02-13 17:11:53] GPU #0: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaEventRecord(context_serialize[1][thr_id])' (C:/Users/Patrick/De
sktop/cudaminer/cudaminer/salsa_kernel.cu line 303)

[2014-02-13 17:11:53] GPU #0: cudaError 33 (invalid resource handle) calling 'cu
daStreamWaitEvent(context_streams[stream][thr_id], context_serialize[(stream+1)&
1][thr_id], 0)' (C:/Users/Patrick/Desktop/cudaminer/cudaminer/salsa_kernel.cu li
ne 931)

[2014-02-13 17:11:53] GPU #0: cudaError 33 (invalid resource handle) calling 'cu
daEventRecord(context_serialize[stream][thr_id], context_streams[stream][thr_id]
)' (C:/Users/Patrick/Desktop/cudaminer/cudaminer/salsa_kernel.cu line 937)

[2014-02-13 17:11:53] GPU #0: cudaError 33 (invalid resource handle) calling 'cu
daStreamQuery(context_streams[stream][thr_id])' (C:/Users/Patrick/Desktop/cudami
ner/cudaminer/salsa_kernel.cu line 943)

[2014-02-13 17:11:53] GPU #0: cudaError 33 (invalid resource handle) calling 'cu
daStreamWaitEvent(context_streams[stream][thr_id], context_serialize[(stream+1)&
1][thr_id], 0)' (C:/Users/Patrick/Desktop/cudaminer/cudaminer/salsa_kernel.cu li
ne 931)

[2014-02-13 17:11:53] GPU #0: cudaError 33 (invalid resource handle) calling 'cu
daEventRecord(context_serialize[stream][thr_id], context_streams[stream][thr_id]
)' (C:/Users/Patrick/Desktop/cudaminer/cudaminer/salsa_kernel.cu line 937)

[2014-02-13 17:11:53] GPU #0: cudaError 33 (invalid resource handle) calling 'cu
daStreamQuery(context_streams[stream][thr_id])' (C:/Users/Patrick/Desktop/cudami
ner/cudaminer/salsa_kernel.cu line 943)

[2014-02-13 17:11:53] GPU #0: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaMemcpyAsync(hash, context_hash[stream][thr_id], mem_size, cudaM
emcpyDeviceToHost, context_streams[stream][thr_id])' (C:/Users/Patrick/Desktop/c
udaminer/cudaminer/keccak.cu line 503)

Thank you Chris, I appreciate you taking the time to answer my question!

Just so you know, t in autotune, cudaminer crashes everytime it gets past 14x**

Cuda errors start to occur weirdly
So right now im doing t12x31 at 4.15khash/s

I now remember why I went to linux, and why I miss my old launch configs so bad that I lost when linux died

what's the output on the console before the crash? a screenshot would do.

I have a mainboard with three PCI express x16 slots spaced far apart and I do not yet want to experiment with risers. I do not really care about costs of initial investment ( I wouldn't be building nVidia farms if I did ). The GTX780Ti's can't get above 5 kHash/s each, but a single ROG MARS card might push close to 7 kHash/s.

All six GPUs will be mining Yacoin or some other scrypt-jane coins. Each GPU should do 3.3-3.4 kHash/s. I am going to run all 6 in one instance probably (under Linux).

Christian

Hm if i try to pass a launch configuration with a lower case k kernel cudaminer crashes with the generic error message "Cudaminer stopped working". The upper case works fine, also autotune sets a lower case, but if i do it manually cudaminer crashes. Tested the last 6 releases, doesn't work for me a single time. Any ideas on that?

I just upgraded the driver to 382.50 for my Tesla C2075's (I have two in a win8 64bit box)  and upgraded cudaminer from 12-18 to 2-9. The problem since DEcember is still there -- when I specify -d 0 I get a lot of errors (cuda error 30) and nvidia-smi shows GPU#1 consumes more power; when I specify -d 1 the cudaminer console shows GPU#0 working confirmed by nvidia-smi showing GPU#0 consumes more power.

Anyway for -d 1, GPU#0 gives 1.65kh/s with -i 0 -H 1 -C 2 -l f104x1 when mining YAC.

What kind of hash rate would you expect for LTC with the ASUS ROG MARS? Why did you choose these? they seem to cos US$650 on Newegg. I'm trying to learn before I upgrade my GTX 560ti.

Are you going to run all 3 in one instance for the same coin or you are going to have 3 instances running different coins in parallel. What's the best strategy at this point?

possibly, but lookup gap will help a bit. And the previous experience I've had with a 2 GB GTX 760 Model wasn't so bad...

For scrypt-jane yes
They would still be good for most other things. Although my 780 using over 2gb of memory is still only getting 2.6khash/s so im unsure how those 3 will get 20khash/s?

you should be getting about 4.5 kHash/s with -L 6 and some overclock. use the lower case t kernel and autotune.

Christian

Wouldn't the 2GB/GPU be a restriction then? Eventually?

Yes, I'm in a pool but I'm trying to learn the mechanics of mining and how fast I should expect a return at this point in the game. So if BTC and LTC aren't good for the average Joe due to demanding calculations which ones should I try just for fun? I'm hearing MicroCoin, NetCoin, etc... but I have only one machine and trying to learn as fast as I can so I can upgrade my GPU.

So far, it is costing me electricity and no coins!  

No I mixed up my yac balance and my microcoin balance lmao

So, YKMwM5jeTYp42YmiQ65dMCqJJavv6UpiHa is you?  So many people were wondering what the story was...

One million YAC - 27BTC, you're rich guy!

Oopse I mixed up yac and microcoin (dont ask how)
I have a few thousand yac lmao

2.66khash/s on my 780
Havent mined this in so long, is that about right now christian?

50 YAC (one block reward) should be worth 60 Euro cents. So excuse me doubting your balance. :-)

How is yac working out christian?
I have a few thousand but ive lost track of what their worth

3 ASUS ROG MARS cards are ordered. This should be close to 20 kHash/s YAC mining power - all on one mainboard.