Topic: The big VTC Vertcoin settings thread - page 7. (Read 264278 times)

Thanks for your reply. I will look in to it this weekend. I will come back with the results.


Yes i tried from 1000-1090. I will try 950-1000 with increments of 10 this weekend. I will post the results here after.

i really dont know know why 1600.. i try 1000 1500 but no luck ... get worse hash rate actually 270khs...

the 1085 seems neceesary if i go lower than that i can worse hash rates but ill try changing the memclock maybe i better hash rate

Why 1600? Try 1500, or 1250. You might also want to play with gpu clock, going down from 1085 may increase the hashrate.

alright you guys i need some help configuring this miner.. i am running linux xubutu os and i need to figure out why i cannot get past 300 khs.. i think my cards can go higher

brand: sapphire 280x dual-x

here are my settings

{
"pools" : [
        {
                "url" :"",
                "user" : "",
                "pass" : ""
        }
],
"failover-only" : true,
"intensity" : "13",
"gpu-threads" : "1",
"worksize" : "256",
"auto-fan" : true,
"temp-cutoff" : "85",
"temp-overheat" : "75",
"temp-target" : "70",
"temp-hysteresis" : "3",
"no-submit-stale" : true,
"shaders" : "2048",
"gpu-engine" : "1085",
"lookup-gap" : "2",
"gpu-memclock" : "1600"
}

someone wanna help me get past 300? please let me know thanks !

I agree do not run with regular HW


"thread-concurrency" : "22400",
"shaders" : "1792",

Cant do both.  TC > 
Cut TC and see if runs better, shader alone gives me 300 but tc could maybe give 330 but really be happy at 300 or near.  TC is too high evidential in any case

http://i.imgur.com/b4aDTyr.png
Gpu 0 = 280x, Gpu 1 = 270x, Gpu 2 = 6950

anyone has a setting conf file for powercolor pcs 270x. I need a setting for it. the best I can do is 175 kh/s. I need to be at least 200 kh/s.

ty

m50b25:  You mentioned going down to 1030 on core speed, but have you tried going lower? If you haven't, try setting one card to 950 in vertminer, and move it up in increments of 10.

you shouldn't be getting any errors at all if you are change your settings and fix the problem.
6% HW errors is pretty nasty.. don't keep running it with that.

oh and my pool always reports faster than what i am reporting in my client by a good amount
and pools calculate your hash rate based on what your sending share wise..
pools have no way of knowing what your client is actually hashing so that is a good indicator if you working good..

m50b25: If you're connecting to a p2pool node, try specifying the difficulty by multiplying your hashrate by 0.00000116, in your case it should be something like  -u Vd8EUUAq...+0.001624 -p x. I remember having low bitrate for some of my 7970s with some algo on p2pool before and setting the diff manually helped, not sure whether it was scrypt-n though or some other algo, but worth a try. It's your worker name (VTC address in your case) + 0.001624 without any spaces. Another suggesting is to get rid of the config file temporarily, you've got a lot of stuff in there and not everything might work with vertminer, when troubleshooting like this it's better to just start with default simple settings, no config, just bare minimum command line.

About my 177 KH/s with errors...

It looks like my pool is reporting only an 81 KH/s hashrate. Does that have anything to do with my ~6% error rate? Or is that normal?

Thanks in advance.

No luck at all. I was able to use -g 2 with 8gb ram but i dont get any better hashrate. It just gets worse. I have tried just about every setting i can find. different tc, core speed, memory, intensity, lookup-gap. Yeah you name it.

Here is someone who posted in this thread at page 41. He got 4x powercolor 280x turboduo just like me. I tried his config. Stil not getting any better hashrate.
https://bitcointalksearch.org/topic/m.5430886

Here are two pictures of cgminer and config with -g2 and -g 1
The rig was running about 20 minutes with each config.

-g 2

http://i.imgur.com/Uj8v3cv.png

-g 1

http://i.imgur.com/rA9u8P1.png

I am having 3 asus r9 280x cards and with the following settings one card is getting 16K/hash can you please let me know how to get proper hash on all 3 cards

vertminer.exe --scrypt-vert -o stratum+tcp://gns.pool.mn:3333 -u xxx -p xxxx -w 256,256,256 -I 13,13,13 -g 2 --lookup-gap 2,2,2 --gpu-powertune -20,-20,-20 --temp-overheat 85,85,85 --temp-cutoff 90,90,90 --gpu-memclock 1500,1500,1500 --gpu-engine 1000,1000,1000 --thread-concurrency 8191,8191,8191

Thanks

I've got a 6x R9 280x BAMT distro mining currently... What steps do I have to take to get it to mine Vertcoins? Is it even doable on BAMT? Or any Linux distro out there? (Windows won't recognise 6 GPU's)+1

I am using a Powercolor Radeon 7950 to mine VTC (with an nvidia gpu and ocasional other hardware)
anyway i have used GPU-Z to max out the gpu memory usage to the limit.

i use the following in my conf.. cgwatcher reports 260 on average (card is dedicated for mining on Win7 x64)

"intensity" : "18",
"vectors" : "1",
"worksize" : "256",
"lookup-gap" : "2",
"thread-concurrency" : "22400",
"shaders" : "1792",
"gpu-engine" : "1000",
"gpu-fan" : "60",
"gpu-memclock" : "1400",
"gpu-memdiff" : "0",
"gpu-powertune" : "0",
"gpu-vddc" : "0.000",
"temp-cutoff" : "95",
"temp-overheat" : "85",
"temp-target" : "75",
"gpu-threads" : "1",


what i did was i multiplied my stream processor count.. 1792 x 12 = 21504 thread-concurrency then i added a bit more with an equal amount to max it out speed wise.
i report no hardware errors with all this.

but i get errors using far less concurrency and far less intensity.. the key issue is finding the sweet spot.
since i am using vertminer-0.5.2 work size is stuck on 256 ..won't matter if you change it.. same with 0.5.3 (i get same hash rate fiddling around with that one)
various 7950's by different companies and even the same model can act very differently so listening to 50 pages of people telling you to use their settings is dumb.
at best use it as a guideline to find what works best for you..
And what i said earlier was i experimented using GPU-Z to maximize the amount of ram on the gfx card that is used.. how much are you using ?
read the cgminer doc's files for info on scrypt tuning too maybe.. that is where i got the idea to do what i did
hint you *should be using the correct multiples.. and not random odd numbers either.. use even numbers (@guy above me talking)

I could just borrow 4 gb from one of my other rigs just to try that. Thanks



I know. I should be able to use -g 2. Yes i already tried using thred concurrency 8191 and 8193. I usually use 8193 when mining other scrypt coins. I will try lowering the core speed but i seriously doubt it will do anything because i already tried lowering to 1030 mhz.
With that config i have and using -g 2 i get 745 kh/s when mining ltc. And the cards are stable for weeks.
I will post later today what kind of error i get in vertminer when using -g 2

7x MSI Radeon HD 7990 (Rig 1 = 8GPUs, Rig 2 = 6GPUs)

My YouTube Video


Image

Software:

• Microsoft Windows 8.1 Update 1 (64-bit)
• AMD Catalyst 14.3 Beta v1.0
• AMD APP SDK 2.9
• CGMiner 3.7.2, sgminer 4.1.153, vertminer 0.5.2

Settings:

• GPU Voltage: 1.050V
• GPU Clock: 950MHz
• Memory Clock: 1500MHz

Batch File:

ONLY IDIOTS USE: setx GPU_MAX_ALLOC_PERCENT 200  


cgminer 3.7.2 - Scrypt (AuroraCoin): Rig 1 = 5240KH/s (100%)



sgminer 4.1.153 - Scrypt-N (SpainCoin): Rig 1 = 2580KH/s (49%)

Image




vertminer 0.5.2 - Scrypt-N (SpainCoin): Rig 1 = 2550KH/s (48.7%)

Image



vertminer 0.5.3 - Scrypt-N (SpainCoin): Rig 1 = 2415KH/s (46%) w/ HW Errors

Image


vertminer 0.5.2 vs 0.5.3 - THE SAME CONFIG BUT HW ERRORS!!!

Okay in that case it's acting like a bad card, but that could also be PCI-e slot or riser or some BIOS setting. Both cards behave identically? I would try to confirm the bad behavior as a hardware level thing by downloading a USB-bootable Linux miner like BAMT, and failing that, either switch to Linux or try different Catalyst driver versions in Windows. @wacko's suggestion to try more system RAM might help, and I've heard dual channel (two sticks) vs single can make a difference.

But before all of that, make sure -g2 is having the same problems. For -g2 you start by dropping TC all the way to 8192 and intensity to 13. I'm actually surprised the miner started at all with 2 threads and those other settings as high as you said.

Don't know how high can I go with TC? Tried a random 35400 but it ended with a BSOD. With TC=29400 hashing speed decreased to about 200 kh/s per card and there are still as many HWs. I was getting many HWs even with the video cards on stock settings and with a blank config file.

with 2 gpu threads, intensity 18, and TC 21000 (I can't set it lower because vertminer says I need to decrease TC or increase LG) I'm having 220 kh/s per card and a little less HWs.

No matter what I do, I'm still having HWs and not more than 280 kh/s, help

Does going to a higher thread concurrency make the HW errors go away? How much higher does it need to be, and does that cost you any hash rate? Then you may have some headroom with your memory clock, don't know if you've tested that to the point of glitches/failures. Higher memory speed at a good error-free TC could help you get more out of the high engine clock. But you'll probably want to reoptimize engine too at that point. Also, have you tried at all with 2 for gpu-threads (at low TC and intensity)? Hope this helps.