sgminer 4.1 won't obey my commands... 5x 280x Vapor-X

yidakee

yidakee

Well-known member
Foundation Member
Hi all, new around these parts...

So I started mining DRK yesterdays and I'm loving it. Got my rig to run 10,35mH/s @ 675w yeee-ha! Not bad. 2,07mh/s per card.
But I wanted to play around with some setting to see if I could squeeze a little more/less juice depending on results.

Thing is, some commands work, but I find intensity is locked to 13, power tune does not budge, shaders, gpu-engine... nothing.
Once in a while when I restart the miner, it lowers the hash rate slightly, yet consumes less 50w and runs cooler.

Using CGwatcher to monitor sgminer. Any ideas?
 
I have no clue to be honest. Don't know what that is but would highly appreciate some tutoring!
For reference, was mining scrypt with cgminer 3.7.2 and everything works fine. Managed to get 750kh/s each card under scrypt

EDIT: ADL SDK - Display Library? 2.2 mb? this it? What is the purpose of this? Sorry, general geek, but I don't want to crash several months of rig troubleshooting!! aaarg!! :p

http://developer.amd.com/tools-and-sdks/graphics-development/display-library-adl-sdk/
 
But why do I need this? cgminer doesn't complain... Is it just a matter of installing or do I need extra arguments in the conf?

here is my current config, as you can see, doesn't really make sense (gnu-engine 1000 instead of 1080, shaders 1792 instead of 2048)

Code: 
-k darkcoin -o stratum+tcp://stratum-eu.suchpool.pw:3335 -u cryptog.1 -p x --auto-fan --gpu-fan 40-85 --temp-cutoff 92 --temp-overheat 88 --temp-target 74 -I 13 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192 --gpu-powertune 20 --gpu-engine 1000 --shaders 1792 --gpu-memclock 1500
 
well, sgminer is reporting temps and fan speed fine... but I'll give it a shot... 121mb downloading...
 
Nope, didn't work, but I found out something. Before installing SDK I fired up cgminer in scrypt mode to check power consumption, and 1260W, with engine 1065. Installed SDK and rebooted. Same deal. sgminer won't obey, but the GPU's are now running at 1065, and not 1070 like before. I just gained -30W (total 635w at wall), -5ºC each card, and rig still at 10.35 mh/s... not bad at all.

Seems CGWatcher is buggy and locking cgminer parameters... Whatever. I like where this is going! shame I have to run cgminer then switch back to sgminar... ug!

These are the setting so far (except pointing to srcypt pool...)
Code: 
--scrypt -o stratum+tcp://eu.wafflepool.com:3333 -u xyz123 -p x --auto-fan --gpu-fan 50-85,50-85,50-85,95,50-85 --temp-cutoff 92 --temp-overheat 88 --temp-target 74 -I 13 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192 --gpu-powertune 20 --gpu-engine 1065 --gpu-memclock 1500

let the tweaking begin... :D

EDIT: 10.5mH @ 650w. engine 1080, tried intensity 12, but its locked at 13. Oddly enough, temperature is even lower now?? weird stuff.. I put my hands above the GPUs and they're really nice and cool at 62ºC - another cool thing is I had headaches with GPU 4 registering +8ºC than the others, now its GPU 5 thats doing it...

EDIT2: engine 1050 mem-clock 1000 = 575w @ 10 mh/s = best so far at 0.056 w/mh and cards at 55ºC !! yet !! Its a damn shame I can't mess around with intensity :(
 
Last edited by a moderator:
fra55 said:
Did you delete the .bin files in the sgminer folder?
Click to expand...

Yes, same deal unfortunately...

EDIT: Got it !! deleted the sgminar.conf file from the folder. For some reason CGWatcher tried to overwrite it but can't. Some options get modified, others don't. Strange because CGWatcher does not write a new .conf file, its stores the profiles with the arguments with in. Stupid bug.

Yippee.... :rolleyes: :D:p
 
Last edited by a moderator:
Ok... lets start seeing the proper figures, the ones before don't count due to CGwatcher bug.. I let them run about 5 minutes until final reading

1- 10.35mH/s @ 675w = 0.065 w/mh - gpu-engine 1070, memclock 1500, i-13
2 - 9.8mh/s @ 575w = 0.0586 w/mh - gpu-engine 1050, memclock 1000, i-13 - 59ºC
3 - 9.95mh/s @ 588w = 0.005909 w/mh - gpu-engine 1070, memclock 1000, i-13 - uneven temps 58-64ºC - (weird, higher engine = lower mh)
4 - 1010mh/s @ 605w = 0.599 w/mh - gpu-engine 1070, memclock 1161, i-13 - 62ºC ... nice even temperatures across all GPU's, actually numbers almost exactly the same!
5 - 9.22 mh/s @ 630w = 0.067 w/mh - gpu-engine 1070, memclock 1500, i-12 - 64ºC

I'll try n.2 with i-14...

6 - 9.5mh/s @ 577w = 0.0607 w/mh - gpu-engine 1050, memclock 1000, i-14 ... 58-64ºC
7 - 9.5mh/s @ 565w = 0.059 w/mh - gpu-engine 1050, memclock 1000, i-20 ... 57ºC
8 - 0 mh/s @ 000w = 0.000 w/mh - gpu-engine 1050, memclock 1000, i-5 ... sgminer won't start
9 - 2.5mh/s @ 394w = 0.1739 w/mh - gpu-engine 1050, memclock 1000, i-10 ... 46ºC
10 - 2.5mh/s @ 452w = 0.1844 w/mh - gpu-engine 1070, memclock 1500, i-10 ... 48ºC

mixing up n.1 and n.7 for low temps and high hash rate ...

11- 10.02mH/s @ 634w = 0.06327 w/mh - gpu-engine 1070, memclock 1500, i-20 - 62ºC
12- 10.07mH/s @ 635w = 0.06325 w/mh - gpu-engine 1060, memclock 1500, i-20 - 62ºC

Well, that was fun :p Back to n.2 (checking...)
2 - 9.95mh/s @ 575w = 0.0577 w/mh, 57ºC - fans @ 1100rpm - silent - bingo!

Love Darkcoin!

Code: 
-k darkcoin -o stratum+tcp://stratum-eu.suchpool.pw:3335 -u xxx.1 -p x --auto-fan --gpu-fan 20-85 --temp-cutoff 92 --temp-overheat 88 --temp-target 64 -I 13 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192 --gpu-powertune 20 --gpu-engine 1060 --gpu-memclock 1000

Anyone want to throw in any ideas with some other args?
 
You should be able to get more hash out of a 5x rig, ideally you're aiming for 2.2-2.35 per card. Here's what I use, lower the core if you have stability problems (1150 works on half my rigs, the others need 1100 to stay stable):

-k darkcoin -I 18 -g 4 -w 128 --gpu-engine 1150 --gpu-memclock 1500

Of course add the basics such as powertune and fan speed if you care to have them in there.
 
I'll give shot. Like this I managed to crape 100w at the expense of 350-400kh and saved 0.26 w/mh, but can only check again wattage tomorrow. My test n.1 give 10.35mh/s
... 1150 crashed the drivers instantly. was needing a reboot anyways... Nope, just keeps crashing the drivers. 1100 too... 1050 seems stable but only gives me 10.32mh/s (2.06 per card).

Best is 1100, gave me 10.8mh (2.16 per card), but crashes after a few mins, and rejects 2%.
Going with 1050, with -i 13 -g 4 -w 256 seems stable at 9.95mh.

maybe -w 64... or mem clock 1075.

EDIT: This is the best so far. 10.8mh (2.16 per card). 1085 was the one with -w 128,
2am 55ºC. Silent. I'm in heaven

Code: 
-k darkcoin -o stratum+tcp://stratum-eu.suchpool.pw:3335 -u xxxx.1 -p x --auto-fan --gpu-fan 40-85 --temp-cutoff 92 --temp-overheat 88 --temp-target 64 -I 18 -g 4 -w 128 --lookup-gap 2 --thread-concurrency 8192 --gpu-powertune 20 --gpu-engine 1085 --gpu-memclock 1500

Thank you jakecrow!!
 
Last edited by a moderator:
Yeah you always have to play around with the core speeds, brands are a bit different in their tolerances. Love my Asus cards that can run 1150 24/7 and not give me any problems, some of my MSI cards get very grumpy above 1100. Still I haven't had any cards that wouldn't run at 1100, that seems a bit odd but who knows.

2.16 isn't bad, you PROBABLY could squeeze more though. You are running at mem 1500?
 
I got better results with my Asus with -g 1 and intensity 17. Moreover, I don't use TC but --shaders 2048. You could also try out xintensity instead of standard intensity

I get 2.1 M/hash with Vcore @ 1.075 mv and memory clock @1.500/6.000 mhz instead of 1.600/6.500. GPU clock is the standard one, 1070 mhz. Temps are usually at 50°, maximum 52°
 
7h, avg hashrate 10.63... will try -g 1 and i-17 fra55

EDIT: 10.75mh @ 625w = 0.0581 - highest mh with lowest watts !! Yess !!... let me try i-18 & 16 with -g 1
-i 16 = 10.6
-i 18 = 10.8 mh/s ... going to increase core again now that using only 1 thread

got it !! core 1150 crashes, 1100 seems stable. 11 mh/s! @ 625w @ 53ºC = 0.05727.. lets see how it runs today
thank you all !!
Code: 
-k darkcoin -o stratum+tcp://stratum-eu.suchpool.pw:3335 -u xxxx.1 -p x --auto-fan --gpu-fan 40-85 --temp-cutoff 92 --temp-overheat 88 --temp-target 64 -I 18 -g 1 -w 128 --lookup-gap 2 --thread-concurrency 8192 --gpu-powertune 20 --gpu-engine 1100 --gpu-memclock 1500
 
Last edited by a moderator:
WU= 0.152/m ... how does it look? 0.25% rejects so far. SuchPool is reading between 11 to 12mh/s.
 
tried -i 20, but got lightly worse, went with -i 19 and WU=0.160 @ 11.02mh/s, 0% rejects - I can interpret scrypt WU but not yet X11. Also increased -w to 128 because 64 was crashing with -i 20

Code: 
-k darkcoin -o stratum+tcp://stratum-eu.suchpool.pw:3335 -u xxxxx.1 -p x --auto-fan --gpu-fan 40-85 --temp-cutoff 92 --temp-overheat 88 --temp-target 64 -I 19 -g 1 -w 128 --lookup-gap 2 --thread-concurrency 8192 --gpu-powertune 20 --gpu-engine 1100 --gpu-memclock 1500

All card have scary identical values. Like they're all perfect in sync! Except temperatures, but around 56ºC, who cares :p
Ok, rollin with this one for the rest of the day and report back
 
Last edited by a moderator:
You must log in or register to reply here.
Back
Top