Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 705. (Read 3426922 times)

yeh i checked all of that, the power saving is off, and system is on high performance
seems strange that its convinced it only runs at x2 speed

There is in power management an option in the "advanced" tab to manage pcie power usage.

what about qurak?
 
that's a good idea!

yep, everything is on high performance

EDIT:
ok then some interesting information.
So i just flashed my bios to the newest one (unsure why i hadnt already) and my system is running much faster. But here are some GPU related specs ive found

GPU-Z before bio update: Bus Interface: PCI-E 3.0 x16 @ x2 2.0
GPU-Z after bio update: Bus Interface: PCI-E 3.0 x16 @ x2 3.0

now my system is full PCI-E x16 and my GPU is the only device plugged in, when idle it drops to x2 1.1 which seems weird that it only ever runs at x2 speed

has anyone else noticed this? perhaps GPU-Z is messing up?

Did you check the power management options ?

Thinking about it more my situation seems to be a bios hard lock because of the pci-e bandwidth being used. As if the system had locked the reset button would still cycle the system to reboot.

Going to try checking for bios updates and flash it again

This might be related to the system slow downs I've experienced too. I thought they'd gone away but I was wrong. If I leave Cudamanager running after a certain length of time (yet unsure) the rig becomes VERY sluggish and unresponsive. This is with multiple cards and risers. Haven't tried with multiple cards directly on the board (though I know 1 directly on the board is fine), or without cudamanager and just using cudaminer. Won't be able to do more testing until next week now.

Mine Gigabyte GTX 750 Ti (N75TOC-2GI) arrived yesterday.
It's installed inside a regular closed case desktop directly at the PCI-E x16 slot.

I spent the whole night testing it with Cudaminer and EVGA Precision X.


1) Chrome Trick

First, I tested the "Chrome trick" with several other software.
No OC e no parameters on Cudaminer.
Below the results:



As seen, the best average rates that you will achieve is 243KHs for cudaminer 32 bits version and 258KHs for the 64 bits version.

As Chrome proved to be the best option (Google Earth is a "heavy" program to be running in the background), I decided to use only it for further tests.

The following tests were performed with Chrome running (doesn't need to be maximized).


2) Cudaminer Tunning

The minimum set of parameters for Cudaminer that give some real improvement is this: -l T5x24  -i 0  -C 1
With it I got 264KHs in both 32 and 64 bits version.

So, I tried almost all possible combinations of the most usual parameters. Here's my conclusions:
- Changing T5x24 to T10x24 makes no difference.
- Changing C1 to C2 gives you around 1KHs more.
- Adding -m1 gives you around 1KHs more.

Thus, the following set is recommended for a average rate of 266KHs: -l T5x24  -i 0  -C 2  -m 1

You can gain another 10KHs adding the -H1 parameter. I achieved around 275KHs with it (-l T5x24  -i 0  -C 2  -m 1  -H 1).
However, it will make use of your CPU together with the GPU, which will cause greater temperature and, specially, noise.
Using -H1 my CPU remained constantly above 60ºC and the fan was very loud.
I don't have a Kill-A-Watt to evaluate precisely, but, I can imagine that the KHs gained will not pay the extra KW.
Using -H2 makes no sense because its the default value and brings no gain.


3) Overclock

I wasn't so thorough in this part as previous ones (I'm without sleep and tired), but, I managed to perform some tests.
Assuming the recommended Cudaminer parameter configuration (-l T5x24 -i 0 -C 2 -m 1), I could achieve around 300KHs with the following EVGA Precision X profile:
- GPU +135
- MEM +700
- Power Target 100%

GPU temperature remained at 58ºC and fan speed at 70%. Very stable and quiet OC.
I think much more than that cannot be "power effective".

Some more aggressive values crashed the GPU.

Will research more OC options next days.

I did experience this multiple times when overclocking my card. I was surprised because I'm using the onboard GPU while mining, but my system kept on freezing completely when I added another 200mHz core clock to my gtx760, 180 seems fine.
It first starts off with some lag on mouse movement, 10 seconds later the system can only be rebooted with a hard reset. If I'm just in time to close the process before the freeze, I'm able to save it from happening =)

Yeh a driver crash will cause a BSOD. It's definitely a full lock up, even the power buttons can't shut it down and the reboot button does nothing. Which should reboot either way

Yes, only when mining.

First thing I checked on the first lockup was num lock, doesn't work so interrupts aren't even working. True hard lock.

Unrecoverable driver crash should cause a BSOD, I believe.

Sounds for me like a driver crash which can't anymore be recovered by Windows. I guess your machine won't anymore respond to Ping once it's crashed? (Please note that the Windows firewall might block it always)
Does the num-lock button still work on the keyboard? (Light turns off and on) If yes, then the system isn't dead, if no, system is dead

Does yours only happen when mining like mine?

FYI: I don't have any wallets on the machine that locks up. There's no guarantee our lockups are due to the same issue though...

Internet doesn't cut off. I have something checking that for me. And it's running straight off an SSD,  I'm going to see what it's like with the wallets on a virtual for a few days. At least if the virtual dies it's not a big issue

The bit bugging me is that it only happens at night while I'm asleep. It will happily mine all day, system temps of my god and cup never hit the 50°C mark so I know it's not temps

Hmm the wallets cause a crash, perhaps some kind of internet/harddrive issue then?

I noticed the same. But I have disabled sleep for the monitor and still have it. Although I do turn the monitor off. But it's off all day as well :-s

Weirdly all the new wallets seem to cause it. Mining to pools doesn't seem as bad. So I'm going through the process of moving all my wallets to virtual machines to see if it helps

Same thing happens to me. Seems to happen more frequently with scrypt-n than scrypt for me, but I have no data to back that up. I *think* it may have to do with waking the monitor, it never happens on my machine that has monitor sleeping disabled, but the machine that does put the monitor to sleep locks up occasionally. Once I checked the pool it was running against and it had been submitting work right up until the moment I tried to use the machine (thus waking the monitor). It's a weak hypothesis, but who knows...

No, Asrock fatal1ty Proffessional zip
It's very strange. I never had a single issue mining on Linux till the OS died

Hmm I have a hang too at random times, with very nasty sound effects...
I have been looking at temperatures and both my CPU and GPU were acceptable, possibly some motherboard failure? I don't mind about myself just trying to help you.
You got a MSI motherboard by chance?