[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 740.

heajin5921

member

Activity: 90

Merit: 10

I wiil try my 780ti for this.
thanks

CodyF86

full member

Activity: 161

Merit: 100

First post so hi everyone!!!

Just wanted to say thank you for this awesome miner!

Zotac Nvidia Geforce 670 @ 1267 Core & 3416 Memory - 375 kh/s!

cudaminer.exe -H 1 -d 0 -i 0 -m 1 -l auto -o stratum+tcp://etc.etc:etc -O pass

Takashi267

newbie

Activity: 6

Merit: 0

Quote from: cvax on March 01, 2014, 06:52:35 PM

Quote from: bigjme on March 01, 2014, 04:02:55 PM

Quote from: cvax on March 01, 2014, 02:24:30 PM

Quote from: bigjme on March 01, 2014, 01:50:41 PM

I can check my 780 on pci 3 x16 if you guys want those results

That would help tremendously. What kind of risers do you have? Can you try a 16x to 16x riser or even a 1x to 16x in that PCI-E 3.0 versus the card being put into the slot directly? Thanks.

I missed the riser part sorry. Would test it but it means ripping my water cooler apart :-(

Ah, no worries. Would still be curious if 3.0 slots yield any benefit versus 2.0 slots for cudaminer if there was any way you had a means of doing that without messing up the water cooling.

While I dont have any risers laying around my motherboard has PCI-E 2.0 x16 slot running at x2 and a heavily overclocked GTX 780.
Here is my config for my GTX 780: cudaminer.exe -q -i 0 -m 0 -H 2 -l T12x24
PCI-E 3.0 x16: ~715Khash/s with a bus usage of ~13%
PCI-E 2.0 x16(x2): ~685Khash/s with a bus usage of ~40-50% (I forgot the exact number)

If you have any more questions feel free to ask.

cdogster

newbie

Activity: 47

Merit: 0

Quote from: jstabb on March 02, 2014, 03:45:07 AM

Specs:
EVGA GTX 750Ti FTW (has 6 pin power connector)
Intel G3220 on ASRock Fatil1ty Z87 Killer

Conservative overclock: +50 GPU, +500 Memory

And the results (scrypt):
Baseline (plugged straight into x16 slot): 290-299 kh/s

Non-powered x1-x16 in x1 slot: 270-280 kh/s
Powered x1-x16 in x1 slot: 272-282 kh/s
USB powered x1-x16 in x1 slot: 270-277 kh/s
USB powered x1-x16 in x16 slot (gen 1): 266-269 kh/s
USB powered x1-x16 in x16 slot (gen 2): 288-295 kh/s
USB powered x1-x16 in x16 slot (gen 3): 288-297 kh/s
Non-powered x16-x16 riser: 292-297 kh/s

Thanks for this info. I was expecting more from these cards. Did you try any higher overclocking? And how many cards did you try at one time?

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: ollyweg on March 02, 2014, 07:50:28 AM

Quote from: cbuchner1 on March 02, 2014, 07:22:41 AM

swap the upper case K against a lower case k.

It's probably best to auto-tune the lower case k kernel with various lookup gap settings and pick the best one.

Christian

Okay I didn´t know that makes a difference. Trying it out right now. Already looks better.
Thanks!

EDIT: For some reason my graphics driver crashes when autotuning with higher lookup-gaps. Not immediately, but when it´s almost done.

the crashing is due to a time-out... haven't found a way to solve it yet. Use the -D option to at least get some readings before the crash.

Christian

ollyweg

newbie

Activity: 19

Merit: 0

Quote from: cbuchner1 on March 02, 2014, 07:22:41 AM

swap the upper case K against a lower case k.

It's probably best to auto-tune the lower case k kernel with various lookup gap settings and pick the best one.

Christian

Okay I didn´t know that makes a difference. Trying it out right now. Already looks better.
Thanks!

EDIT: For some reason my graphics driver crashes when autotuning with higher lookup-gaps. Not immediately, but when it´s almost done.

djm34

legendary

Activity: 1400

Merit: 1050

I looked a bit into bus usage (with AB), not sure if it is very helpful here what I found:

On scrypt; scrypt:2048 with H2: The bus usage is around 14%
On scrypt-jane, it is 2%
The strange thing is when using autotune (on scrypt-jane), this one is all over the place going up to 100%

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: ollyweg on March 02, 2014, 06:21:52 AM

Ok I tried the settings, but K27x3 only gets me "result does not validate".

swap the upper case K against a lower case k.

It's probably best to auto-tune the lower case k kernel with various lookup gap settings and pick the best one.

Christian

ollyweg

newbie

Activity: 19

Merit: 0

Quote from: cbuchner1 on March 01, 2014, 08:30:02 PM

Quote from: djm34 on March 01, 2014, 07:17:08 PM

Quote from: ollyweg on March 01, 2014, 07:11:39 PM

Earlier I used --interactive=0 --hash-parallel=1 --launch-config=K7x23 --texture-cache=0 --single-memory=0 --lookup-gap=3
I´ve tried several combinations of the other options, but nothing helps. (also batchsize doesn´t seem to change anything for me.)

BTW I want to report a small bug: --batchsize=x isn´t recognized correctly. Only -b x works.

You should try to increase even more the look-up gap value

--launch-config=K7x23 should be k27x3

and it's probably --batch-size=...

Christian

Ok I tried the settings, but K27x3 only gets me "result does not validate". I have to increase lookup-gap to 9 in order for it to run, but then it only gets me around 1kHash. I´ve already tried other launch-configs (autotune) with higher lookup-gap, but the higher I go the lower my hashrate.

When autotuning on some settings, it crashes my driver. And somehow any configuration never uses a lot of ram.
For example autotune found K14x2, it works but only uses 1295MB of vram on lookup-gap 3.

EDIT: seems to be the same with the older version I was using before. (2014-02-09).
I last used scrypt-jane when the feature was in beta. (i think 2014-01-22)
With that version and my old config I now only get 0.5kHash.

Also I found out that -batchsize=x is recognized as a command (only one -). However this prevents cudaminer from starting to mine. It stops at:

Code:

                     This is version 2014-02-28 (beta)
        based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
            Cuda additions Copyright 2013,2014 Christian Buchner
          LTC donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm
          BTC donation address: 16hJF5mceSojnTD3ZTUDqdRhDyPJzoRakM
          YAC donation address: Y87sptDEcpLkLeAuex6qZioDbvy1qXZEj4
[2014-03-02 12:24:05] 1 miner threads started, using 'scrypt-jane' algorithm.
[2014-03-02 12:24:05] Nfactor is 14 (N=32768)!
[2014-03-02 12:24:05] GPU #0: GeForce GTX 660 Ti with compute capability 3.0
[2014-03-02 12:24:05] GPU #0: interactive: 0, tex-cache: 0 , single-alloc: 0
[2014-03-02 12:24:05] GPU #0: 32 hashes / 14.2 MB per warp.
[2014-03-02 12:24:05] GPU #0: using launch configuration K27x3

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: haigent on March 01, 2014, 05:35:42 AM

Just donated 1 LTC to you, keep your great work on cudaMiner.

Thank you very much! The same goes to the anonymous donor with the 2.0795371 LTC donation Wink

jstabb

newbie

Activity: 7

Merit: 0

Hi, I'm new to the forums and crypto mining in general. But I thought I'd share the performance results of some riser testing I did.

Specs:
EVGA GTX 750Ti FTW (has 6 pin power connector)
Intel G3220 on ASRock Fatil1ty Z87 Killer

Conservative overclock: +50 GPU, +500 Memory

And the results (scrypt):
Baseline (plugged straight into x16 slot): 290-299 kh/s

Non-powered x1-x16 in x1 slot: 270-280 kh/s
Powered x1-x16 in x1 slot: 272-282 kh/s
USB powered x1-x16 in x1 slot: 270-277 kh/s
USB powered x1-x16 in x16 slot (gen 1): 266-269 kh/s
USB powered x1-x16 in x16 slot (gen 2): 288-295 kh/s
USB powered x1-x16 in x16 slot (gen 3): 288-297 kh/s
Non-powered x16-x16 riser: 292-297 kh/s

As you can see, there's not much penalty for the riser as long as you're able to plug into a gen2+ x16 slot. If you're plugging into an x1 or gen1 x16 slot, there's a 10-15% performance penalty.

I hope this information proves useful.

Edit: Command line was: cudaminer.exe -H 2 -i 0 -m 1 -l T5x24 --benchmark

guzzzi

newbie

Activity: 53

Merit: 0

So 350/360kh/s each gtx780ti is max for this Card? All other settings i done cudaminer gets me error or not enough memory msg. or the nvidia drivers breaks.

cvax

member

Activity: 84

Merit: 10

Quote from: gazzamc on March 01, 2014, 10:25:10 PM

@anyone

Hey guys I registered just to ask this... I've got 2 gigabyte 750ti's and one is kinda acting strange.. i have one OC'd @ +100 core & +500 Memory which yields me 295 kh/s roughly... But my second card keeps on crashing whether it's OC'd or not..

Should I RMA that card? as I even tried Underclocking it and it still crashed.

The error has something to do with memory, Is it VRam or Ram? As i could always add more.. If it helps i'm on Win7 x64..

I hope someone could help as this is kind of irritating me..

Appreciate any replies, Thanks

Sounds like a busted card to me, but try it as a single card only first.

whitesand77

full member

Activity: 125

Merit: 100

Quote from: gazzamc on March 01, 2014, 10:25:10 PM

@anyone

Hey guys I registered just to ask this... I've got 2 gigabyte 750ti's and one is kinda acting strange.. i have one OC'd @ +100 core & +500 Memory which yields me 295 kh/s roughly... But my second card keeps on crashing whether it's OC'd or not..

Should I RMA that card? as I even tried Underclocking it and it still crashed.

The error has something to do with memory, Is it VRam or Ram? As i could always add more.. If it helps i'm on Win7 x64..

I hope someone could help as this is kind of irritating me..

Appreciate any replies, Thanks

Swap the cards in the PCI-E slots and see if the same card gives you issues.

gazzamc

full member

Activity: 189

Merit: 100

@anyone

Hey guys I registered just to ask this... I've got 2 gigabyte 750ti's and one is kinda acting strange.. i have one OC'd @ +100 core & +500 Memory which yields me 295 kh/s roughly... But my second card keeps on crashing whether it's OC'd or not..

Should I RMA that card? as I even tried Underclocking it and it still crashed.

The error has something to do with memory, Is it VRam or Ram? As i could always add more.. If it helps i'm on Win7 x64..

I hope someone could help as this is kind of irritating me..

Appreciate any replies, Thanks

ivanlabrie

hero member

Activity: 812

Merit: 1000

My board seems ideal for this (or any X79 with 4 slots for that matter)...

Not sure if this still applies though: http://nvidia.custhelp.com/app/answers/detail/a_id/3135/session/L3RpbWUvMTM0MDIyMzU2OC9zaWQvaDEzbE45X2s=

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: djm34 on March 01, 2014, 07:17:08 PM

Quote from: ollyweg on March 01, 2014, 07:11:39 PM

Earlier I used --interactive=0 --hash-parallel=1 --launch-config=K7x23 --texture-cache=0 --single-memory=0 --lookup-gap=3
I´ve tried several combinations of the other options, but nothing helps. (also batchsize doesn´t seem to change anything for me.)

BTW I want to report a small bug: --batchsize=x isn´t recognized correctly. Only -b x works.

You should try to increase even more the look-up gap value

--launch-config=K7x23 should be k27x3

and it's probably --batch-size=...

Christian

RustyShackleford

newbie

Activity: 43

Merit: 0

Nevermind. Had libcurl5 installed, needed 4

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: ollyweg on March 01, 2014, 07:11:39 PM

I wanted to see how much the new version improves Yacoin performance on my 660Ti.
But I can´t seem to get it to mine at an acceptable rate.
With the older versions I was getting about 3.2kHash. (my card is highly oc´d and used only for mining)
But now i can´t seem to get near that.

Earlier I used --interactive=0 --hash-parallel=1 --launch-config=K7x23 --texture-cache=0 --single-memory=0 --lookup-gap=3
Now that same launch config doesn´t start and via autotune I only get around 1.7kHash.
I´ve tried several combinations of the other options, but nothing helps. (also batchsize doesn´t seem to change anything for me.)

BTW I want to report a small bug: --batchsize=x isn´t recognized correctly. Only -b x works.

You should try to increase even more the look-up gap value

ollyweg

newbie

Activity: 19

Merit: 0

I wanted to see how much the new version improves Yacoin performance on my 660Ti.
But I can´t seem to get it to mine at an acceptable rate.
With the older versions I was getting about 3.2kHash. (my card is highly oc´d and used only for mining)
But now i can´t seem to get near that.

Earlier I used --interactive=0 --hash-parallel=1 --launch-config=K7x23 --texture-cache=0 --single-memory=0 --lookup-gap=3
Now that same launch config doesn´t start and via autotune I only get around 1.7kHash.
I´ve tried several combinations of the other options, but nothing helps. (also batchsize doesn´t seem to change anything for me.)

BTW I want to report a small bug: --batchsize=x isn´t recognized correctly. Only -b x works.

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 740. (Read 3426976 times)