Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1202. (Read 2347597 times)

Hi SP. I have a little problem, i have 2GTX 970s (ASUS) in SLI. Mining works fine, but whenever i close the miner my computer hangs ( i have to hard reboot it)
any ideas?

SP, would you kindly add a tag in git when you release a beta?  It's hard to know if i'm on the release commit or building from an in between version.

here is the latest stats on a single low profile gigabyte 750ti low profile - latest build under linux a few moments ago - x11 ... in case you wanted to know ...

[2014-12-27 16:07:00] accepted: 3/6 (50.00%), 2586 khash/s yay!!!
[2014-12-27 16:07:07] GPU #0: GeForce GTX 750 Ti, 2591 kH/s
[2014-12-27 16:07:10] Stratum difficulty set to 0.03125
[2014-12-27 16:07:10] accepted: 4/7 (57.14%), 2586 khash/s yay!!!
[2014-12-27 16:07:11] GPU #0: GeForce GTX 750 Ti, 2591 kH/s
[2014-12-27 16:07:23] GPU #0: GeForce GTX 750 Ti, 2593 kH/s
[2014-12-27 16:07:26] accepted: 5/8 (62.50%), 2586 khash/s yay!!!
[2014-12-27 16:07:31] Stratum difficulty set to 0.0625
[2014-12-27 16:07:55] Stratum difficulty set to 0.03125
[2014-12-27 16:08:24] GPU #0: GeForce GTX 750 Ti, 2592 kH/s
[2014-12-27 16:08:27] Stratum difficulty set to 0.015625
[2014-12-27 16:08:48] Stratum difficulty set to 0.03125
[2014-12-27 16:08:59] GPU #0: GeForce GTX 750 Ti, 2593 kH/s
[2014-12-27 16:09:19] 192.168.0.205:8332/ x11 block 7569
[2014-12-27 16:09:19] GPU #0: GeForce GTX 750 Ti, 2593 kH/s
[2014-12-27 16:09:21] GPU #0: GeForce GTX 750 Ti, 2591 kH/s
[2014-12-27 16:09:22] accepted: 6/9 (66.67%), 2587 khash/s yay!!!

a little bit less hashrate - but so much less space and im testing the power when i get back into the office ...

#crysx

not if i dont want to pop my power circuit on a regular level ...

the 750ti oc is powerless now ( low profile edition ) AND only around the $170AUD mark which uses one 875W PSU ... as opposed to the GTX970 which requires two power connectors ( 6 pin and 8 pin ) which means you need 2 x 875W PSU running in tandem ( or a large 1200-1500W - which escalates the price dramatically )  if you want more than 4 cards in a single machine and it costs around the $540AUD mark ... which means more cost per machine and a higher $/hash ratio ...

no doubt that the GTX 970 is one hell of a gutsy card - but the cost of the machines to build becomes almost ridiculous ...

either way - i have tested the non-powered ( no power connector to the card ) low profile 750ti and am getting some nice hashrates ( approx 2700-2750) at approx 60W per card ...

this power consumption and the price that they are at the moment - make them the better value card for a stock build ... if you OC and play with the BIOS ( if you have the windforce edition as sp_ was mentioning ) it gets even better value ...

#crysx

Hello, I have a drop performance in lyra2re in the last build from near 680 to near 460 with GTX 750 Ti from previous build.  

Yeah, let's ditch xtranonce support, the api and the nice colors <.< I don't think it's useful to waste your time on merging your code into his, might be better to try and locate that bug.

I have reproduced this on one of my rigs. I will try to merge the kernals into the djm34 fork. I think this version is more stable. But after xmax. Now I have some days of.

When I run with the -i at best it gives me the same hash rate. But sometimes does a hard crash.
Edit: On the 750ti

sp_ Why do you use an intensity level ? Why not just over-clock? Or under-clock?

I wish I could download that all in PDF format. 

What is it?

in god we trast!

Looks like one of the kernel calls failed.
Since we don't do any error checks at all, we only see the error the next time x11_simd512_cpu_init() is called.

I don't have any clue what files those errors are referring to. they were just the error ccminer.exe spit out while mining, posted the screenshots of them 2 pages ago. The typo could have been me, I didn't copy/paste them I just didn't want to re post the same screenshots.

edit, yes the typo was me. the screenshot shows simd but I typed smid

first one looks like a typo...
may-be if you show wjat there is at line 462 and 308 that would help...

So nobody has an answer or idea about these two errors?

"Cuda error in func 'x11_smid512_cpu_init' at line 462 : unspecified launch failure."
"Cuda error in func 'quark_blake512_cpu_setBlock_80' at line 308 : unspecified launch failure."

happens regardless of how low I set the intensity, using stock clocks on my MSI GeForce GTX 750 Ti OC TwinFrozr IV's (7 card rig running windows 8.1 with latest nvidia driver)

A real cudaminer has read it all!

to be fair that thread is over 1100 pages long

haha good point , I think Ive missed it , shame ... thanks for the hint..

get involve in reading  :
I committed some optimization to m7 at the same time I released lyra2re (last try...   ), that was announced on the cudaminer thread.