Sapphire dual-x 7970 issue 15LTC for solution. - page 2.

JoeMattie

full member

Activity: 220

Merit: 100

Quote from: FiiNALiZE on July 15, 2013, 07:29:28 PM

No.

7970's cannot obtain stable settings at intensities above 13.

I don't know why, but I haven't heard of one case where anyone was able to attain a stable configuration with an -I 13+

Set -g to 2, -I to 13, and --thread-concurrency to 8192

Anyone who says intensities above 13 are stable for 7970s don't know what they are talking about.

btw, did anyone receive your 15LTC reward yet?

Incorrect.

7970s run rock solid stable at -I 20 as long as the thread-concurrency is above 16384 and gpu-threads = 1.
They're just slightly faster with -I13 / 8192 / gpu-threads:2 [680kH vs 710 or so]

whitedragon

member

Activity: 84

Merit: 10

Quote from: FiiNALiZE on July 15, 2013, 07:29:28 PM

No.

7970's cannot obtain stable settings at intensities above 13.

I don't know why, but I haven't heard of one case where anyone was able to attain a stable configuration with an -I 13+

Set -g to 2, -I to 13, and --thread-concurrency to 8192

Anyone who says intensities above 13 are stable for 7970s don't know what they are talking about.

btw, did anyone receive your 15LTC reward yet?

^^^ What he said.

FiiNALiZE

hero member

Activity: 868

Merit: 500

CryptoTalk.Org - Get Paid for every Post!

No.

7970's cannot obtain stable settings at intensities above 13.

I don't know why, but I haven't heard of one case where anyone was able to attain a stable configuration with an -I 13+

Set -g to 2, -I to 13, and --thread-concurrency to 8192

Anyone who says intensities above 13 are stable for 7970s don't know what they are talking about.

btw, did anyone receive your 15LTC reward yet?

pyromaniac

hero member

Activity: 639

Merit: 500

Ideal parameters for me is:
first, exec this from Administrator:

Quote

setx GPU_MAX_ALLOC_PERCENT 100

then add to your bath file this:

Quote

bfgminer.exe --scrypt --gpu-engine 1075 --gpu-memclock 1485 --thread-concurrency 8192 -I 13 -g 2 -w 256 --lookup-gap 2 --no-submit-stale --auto-fan

It's should give you around 750 k/hash per card.
LXAhq2uLqXqV8EZet62bvunZuqUA8i7mdX

ReCat

sr. member

Activity: 406

Merit: 250

Nonsense. I ran my 7950 at intensity 19 all the time. A 7970 should handle higher intensities just fine.

sikman

hero member

Activity: 572

Merit: 500

This is what I use to achieve 710kh/s rock solid

cgminer --scrypt -o stratum+tcp://coinotron.com:3334 -u XXXX -p XXXX --auto-fan --temp-overheat 85 --temp-target 77 --temp-cutoff 92 --intensity 13 --worksize 256 -g 2 --lookup-gap 2 --thread-concurrency 8192 --gpu-engine 1044 --gpu-memclock 1500

Im using ATI drivers 13.3.beta 3 on Win 7 64bit

I used to get heaps of nounce errors, hw errors etc until I put more MEMORY in the PC, I spend hours playing with different settings and soon as I put in 4GB of ram up from 2GB all errors went away and now happily mining away.

geronimo77

sr. member

Activity: 476

Merit: 250

for 7970 It's not recommended to have intensity higher than 13. Anything above I:13 will give you HW errors. Try play with the numbers (concurrency and clocks)

try this settings:

--gpu-engine 1050 --gpu-memclock 1750 --shaders 2048 --thread-concurrency 8192 -I 13 -g 2 -w 256 --lookup-gap 2

illpoet

sr. member

Activity: 341

Merit: 250

try just using 1 gpu thread

cdog

hero member

Activity: 1036

Merit: 500

I dont have a 7970 but I would try adjusting the voltage on both core and mem, and see what happens.

Intensity 17-18 is usually the sweet spot for my 7950s, with a 1:1.5 core to mem speed ratio.

If this helps you, PM me I can give you my LTC address :-)

ReCat

sr. member

Activity: 406

Merit: 250

Leave the current GPU drivers. They are fine.

Maybe CGMiner isn't able to allocate enough GPU memory for scrypt?

What happens if you run

setx GPU_MAX_ALLOC_PERCENT 100

In the command prompt before starting CGminer?

Also, try running CGMiner with the following tuning flags.

--worksize 256 -v 1 -I 13 -g 2 --thread-concurrency 8192 --lookup-gap 2 --shaders 2048

If you'd like, We can communicate via skype (Mine is therecreate) or IRC or w/e, and I can resolve the problem for you directly via teamviewer immediately. Wink

If this resolves your problem, my LTC address is LeT4SyiWWstdvB42thgekYsBLrub9UWvS4

Beans

hero member

Activity: 518

Merit: 500

Quote from: slarty on July 15, 2013, 11:58:53 AM

Quote from: Beans on July 15, 2013, 07:18:42 AM

BFGminer seems to be more informative as to what's going on. This is after a short time period but it keeps doing the same thing.

bfgminer --scrypt --thread-concurrency 10752 -g 2 -I 20 -w 256

I tested one of these cards on a different pc and I get the same issue. There must be something in my config or drivers. I just installed 13.4 ccc on the new pc I tested. They were running 11.11 previously.

"auto-gpu" : true,
"auto-fan" : true,
"gpu-engine" : "900-1040,900-1040",
"gpu-fan" : "65-85,65-85",
"gpu-memclock" : "1450,1450",
"gpu-vddc" : "1.060,1.060",
"temp-target" : "70,70",
"temp-cutoff" : "80,80",
"temp-overheat" : "75,75",

"algo" : "c",
"api-port" : "4028",
"expiry" : "120",
"gpu-threads" : "4",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "60",
"temp-hysteresis" : "3",

"donation" : "0.00",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}

I'm not familiar with BFGMINER settings, but looks more or less the same than cgminer conf.
If I get it right with the config from above EACH GPU will use 4 threads. Shouldn't it be "gpu-threads" : "2,2" for each GPU to use 2 threads?
What does algo: c mean?
you can also try to use only one thread, in this case you can double the thread-concurrency.

RGDS

I've been messing with the settings all day, I forgot to switch that back. I normally just have 1 or 2. Don't think it needs to be 2,2

slarty

newbie

Activity: 12

Merit: 0

Quote from: Beans on July 15, 2013, 07:18:42 AM

BFGminer seems to be more informative as to what's going on. This is after a short time period but it keeps doing the same thing.

bfgminer --scrypt --thread-concurrency 10752 -g 2 -I 20 -w 256

http://i40.tinypic.com/2e5o5qr.png

I tested one of these cards on a different pc and I get the same issue. There must be something in my config or drivers. I just installed 13.4 ccc on the new pc I tested. They were running 11.11 previously.

"auto-gpu" : true,
"auto-fan" : true,
"gpu-engine" : "900-1040,900-1040",
"gpu-fan" : "65-85,65-85",
"gpu-memclock" : "1450,1450",
"gpu-vddc" : "1.060,1.060",
"temp-target" : "70,70",
"temp-cutoff" : "80,80",
"temp-overheat" : "75,75",

"algo" : "c",
"api-port" : "4028",
"expiry" : "120",
"gpu-threads" : "4",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "60",
"temp-hysteresis" : "3",

"donation" : "0.00",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}

I'm not familiar with BFGMINER settings, but looks more or less the same than cgminer conf.
If I get it right with the config from above EACH GPU will use 4 threads. Shouldn't it be "gpu-threads" : "2,2" for each GPU to use 2 threads?
What does algo: c mean?
you can also try to use only one thread, in this case you can double the thread-concurrency.

RGDS

Damnsammit

sr. member

Activity: 406

Merit: 250

Everyone already touched on it, but just wanted to reiterate that intensity 20 is rarely the best setting.

Usually you want to play with that number and find the best one for each particular card.

For example, my 6850 runs great at intensity 19 (if I am only using the computer for mining, I drop it down to 14 if I want to use the computer for something else)

My 5850 will get nothing by HW errors at i20, and it gets a lot at i19. Similarly, at i18 it will run perfectly for a few hours, then start getting HW errors, so I turn it on i17 and it runs flawlessly without really suffering a lower hashing rate.

You just gotta play with it a little bit and find the best for each card

Edit: Also, what PSU are you using?

JoeMattie

full member

Activity: 220

Merit: 100

Remove all AMD drivers with the AMD driver removal tool from here: http://forums.guru3d.com/showthread.php?t=379505
reboot
Install 11.12 7900 series drivers from here: http://www2.ati.com/drivers/hotfix/radeon_7900_series/amd_radeon_hd7900_win7_64.exe

Use:
thread-concurrency: 8192
gpu-threads: 2
intensity: 13

If you want to mine at -I 20 you have to use tc=22400 or tc=40960 with gpu-threads=1

jdebunt

legendary

Activity: 1596

Merit: 1010

-I 20 is most likely the issue there, i run this config for 750 khash/sec (you can add more variables etc) :

(keep in mind i use cgminer, not bfgminer) :

cgminer -o server:port -u user.1 -p x --scrypt --gpu-engine 1085 --gpu-memclock 1500 --thread-concurrency 8192 -I 13 -g 2 -w 256 --lookup-gap 2 --no-submit-stale

anything with intensity over 13 will give hardware errors

Dum3

member

Activity: 75

Merit: 10

Try 13.1 whql.Intensity 13 worked excellent for me too,but drivers matter...

i had this parameters and yielded good mh/s

--shaders 2048 --thread-concurrency 8192 --intensity 13 --worksize 256 -g 2

Beans

hero member

Activity: 518

Merit: 500

Quote from: slarty on July 15, 2013, 02:32:36 AM

Quote from: Beans on July 14, 2013, 10:18:58 AM

I set up a new rig today, sapphire dual-x 7970's with cgminer on windows. I got it to 715k per card, but only around 10% of the work gets credited to the pool. The shares are not being submitted in cgminer either.The problem seems to be the intensity, I've tried a couple pools, and different settings. It's works fine at intensity 13 and starts going down with anything above that. This issue seems strange to me but I only switched to ltc recently. Anyone seen this before?

could you provide some stting you are working with? OS?
Do you see HW Errors in the cgminer?

RGDS

BFGminer seems to be more informative as to what's going on. This is after a short time period but it keeps doing the same thing.

bfgminer --scrypt --thread-concurrency 10752 -g 2 -I 20 -w 256

I tested one of these cards on a different pc and I get the same issue. There must be something in my config or drivers. I just installed 13.4 ccc on the new pc I tested. They were running 11.11 previously.

"auto-gpu" : true,
"auto-fan" : true,
"gpu-engine" : "900-1040,900-1040",
"gpu-fan" : "65-85,65-85",
"gpu-memclock" : "1450,1450",
"gpu-vddc" : "1.060,1.060",
"temp-target" : "70,70",
"temp-cutoff" : "80,80",
"temp-overheat" : "75,75",

"algo" : "c",
"api-port" : "4028",
"expiry" : "120",
"gpu-threads" : "4",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "60",
"temp-hysteresis" : "3",

"donation" : "0.00",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}

slarty

newbie

Activity: 12

Merit: 0

Quote from: Beans on July 14, 2013, 10:18:58 AM

I set up a new rig today, sapphire dual-x 7970's with cgminer on windows. I got it to 715k per card, but only around 10% of the work gets credited to the pool. The shares are not being submitted in cgminer either.The problem seems to be the intensity, I've tried a couple pools, and different settings. It's works fine at intensity 13 and starts going down with anything above that. This issue seems strange to me but I only switched to ltc recently. Anyone seen this before?

could you provide some stting you are working with? OS?
Do you see HW Errors in the cgminer?

RGDS

Beans

hero member

Activity: 518

Merit: 500

I set up a new rig today, sapphire dual-x 7970's with cgminer on windows. I got it to 715k per card, but only around 10% of the work gets credited to the pool. The shares are not being submitted in cgminer either.The problem seems to be the intensity, I've tried a couple pools, and different settings. It's works fine at intensity 13 and starts going down with anything above that. This issue seems strange to me but I only switched to ltc recently. Anyone seen this before?

Topic: Sapphire dual-x 7970 issue 15LTC for solution. - page 2. (Read 2416 times)