whats wrong with my sapphire r9 290 settings?

[rodneycrypto]

i have 2 sapphire r9 290s using 4 gigs of ram, a sempron 145, and a asrock 990fx extreme 4 mobo

using cgminer 3.72 with these settings

-I 13 -g 2 --gpu-engine 1000 --gpu-memclock 1250 --thread-concurrency 8192 --gpu-fan 85

right now those settings only are getting me 513kh per card. i know these get 800kh out the box so its gotta be a settings issue.
also if i close cgminer windows crashes with a bluescreen and restarts the computer, but as long as i dont close it it is up and mining.


if i increase the intensity i get all HW errors.

any help is greatly appreciated 

[gilad215]

youre --thread-concurrency is wrong.

set it to atleast 24550 or 32765. do stock clocks, and intensity to 19 and youre good to go. 4gb ram should be enough

[rodneycrypto]

i get error -4 Enqueing kernel onto command queue

when using -I 19 -g 2 --thread-concurrency 24550 --gpu-fan 85

[atp1916]

i get error -4 Enqueing kernel onto command queue

when using -I 19 -g 2 --thread-concurrency 24550 --gpu-fan 85

4GB ram is not enough to run even 24k TC.  My 290s all give that exact error with anything over 22-23k in any of my systems that have 4GB ram. You also may have performance issues with the memory at only 1250. None of my 290s wanted to come close to even 800khs until i started making the ram run @ 1500.  Also, not all 290s like the --gpu thread count @ 2.  Most like -g 1.  

To avoid that blue screen, close out by hitting the 'x'.  Don't attempt to close "nicely".  I am also assuming you are running the official 13.12's.  I do not recommend any of the 14.x series drivers.

Change TC to 20481.
Change -g to 1
Add -w 256

Your config should look like this:

Code:

-I 19 -g 1 -w 256 --thread-concurrency 20481 --gpu-engine 1000 --gpu-memclock 1250 --gpu-fan 85

[rodneycrypto]

i get error -4 Enqueing kernel onto command queue

when using -I 19 -g 2 --thread-concurrency 24550 --gpu-fan 85

4GB ram is not enough to run even 24k TC.  My 290s all give that exact error with anything over 22-23k in any of my systems that have 4GB ram. You also may have performance issues with the memory at only 1250. None of my 290s wanted to come close to even 800khs until i started making the ram run @ 1500.  Also, not all 290s like the --gpu thread count @ 2.  Most like -g 1.  

To avoid that blue screen, close out by hitting the 'x'.  Don't attempt to close "nicely".  I am also assuming you are running the official 13.12's.  I do not recommend any of the 14.x series drivers.

Change TC to 20481.
Change -g to 1
Add -w 256

Your config should look like this:

Code:

-I 19 -g 1 -w 256 --thread-concurrency 20481 --gpu-engine 1000 --gpu-memclock 1250 --gpu-fan 85

OK FANTASTIC! that is some great info and will post results.  yes i am running the 13.12 from the disc that came with the cards. will it hurt to bump the ram up to 8gig? its a windows 7 pro machine

[rodneycrypto]

damn i get the same results using -I 19 -g 1 -w 256 --thread-concurrency 20481 --gpu-engine 1000 --gpu-memclock 1250 --gpu-fan 85

error -4


i would have to think its the ram, i am going to add 4 more gigs and see if that helps out.

[atp1916]

Well, the fact that it runs with the TC at 8192 tells me that it shoudn't be a driver issue or miner issue.   Did you try running with higher clocked ram with that TC val? Say 1500?

4GB ram is enough to run 4 290s @ 20481 np....

That being said, -4 Enqueing kernel onto command queue = not enough system ram.

[sgi02]

Fellow R9 290 miner here. Ive been tweaking my cards for a couple months in order get the best possible performance, and ive found that most / if not all advice you will find is purely speculative and varies significantly per setup. With that being said, here is what has yielded me the best results...

1.) Use cgminer 3.7.3 kalroth edition, this has provided me with the most stable results as compared to cgminer 3.7.2 and sgminer. These are the settings I use personally use in my bat file...

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer --thread-concurrency 24550 --xintensity 48 -g 1 --scrypt -o stratum+tcp://yourpool.com -u r290.gpu -p gpu

2.) BIOS Update - Upgrade to a BIOS from The Stilt, this will provide you 30-60 k/hs higher overall hashrate with less power usage. Elpida vs Hynix memory makes a significant difference. If you have Hynix memory overclocking will help your overall performance, with Elpida ive found that overclocking yeilds slight increases at the cost of reliability and power usage, IMO not worth it.

3.) Odds and ends - Ive personally found 8GB memory works more reliably than 4GB. YMMV and you'll surely get a million different opinions on this. If you have access to additional ram / its cheap enough then just do it. A final note is that your hashrate is meaningless in the grand scheme of things, your WU efficiency is what counts the most. Ive cranked my 290 to 875 k/hs and found that I had less accepted shares and a lot of rejects. With the config above I get 830 k/hs solid, no rejects on the vast majority of pools and a 92%+ WU.

I hope some of this has helped you out, good luck!

[rodneycrypto]

wow now what you said makes a lot of sense and will tryu when i get home tonight. becuase of my ram uissues io just put each 290 in a specerate miner becuase when i tried to take the 2 sticks of ram out of my seond miner and added it the to first it woruldnt boot, so ok incompatible ram. so i am running two miners with one 290 each, not the best for power effecency but hey who cares, btc will be at 10,000 in a few years after total adoption and all these articles about how gpus are dead is bs. asics can only do scrypt and no other algos so gpus are so safe, new coins switch algos and gpus will make money forever in my mind. anyway, now i put my second 290 in my 2nd miner , which consits of 4 gig ram, msi millitary class mobo, sempron, cgminer 3.72, ati stock driver cd that came with 290, and windows 7 pro


but using the exact same setting as other miner, which has been hashing away at 860kh for over 24 hours straigh (big ups and thanks to atp1916 for that gangster shit) , it doesnt even run. so yeah even a different mobo or ram brand makes a difference.

ok not sure how i should explain, if i use
-I 19 -g 1 -w 256 --thread-concurrency 20481 --gpu-engine 1000 --gpu-memclock 1250 --gpu-fan 85

i get
error -61: clCreateBuffer <padbuffer8>, decrease TC or increase LG

what the hell is LG?

but if i use
my original settings that i posted on the OP

-I 13 -g 2 --gpu-engine 1000 --gpu-memclock 1250 --thread-concurrency 8192 --gpu-fan 85

then i do infact get the 500+kh that i first talked about.

i can give eact msi model if that is needed and ram specs.

please help me obi wan your my only hope

(is anyone else as scared as me that disney is gonna fuck up our star wars franchise?) lol. off topic. :}

[rupayacoin]

i have 2 sapphire r9 290s using 4 gigs of ram, a sempron 145, and a asrock 990fx extreme 4 mobo

using cgminer 3.72 with these settings

-I 13 -g 2 --gpu-engine 1000 --gpu-memclock 1250 --thread-concurrency 8192 --gpu-fan 85

right now those settings only are getting me 513kh per card. i know these get 800kh out the box so its gotta be a settings issue.
also if i close cgminer windows crashes with a bluescreen and restarts the computer, but as long as i dont close it it is up and mining.


if i increase the intensity i get all HW errors.

any help is greatly appreciated  

I suggest you use kalroth version of cgminer.

make a cgminer.conf and paste the config below:

If you have hynix memory use this change pools obviously

Code:

{
"pools" : [
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos",
                "pass" : "123"
        },
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos1",
                "pass" : "123"
        },
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos2",
                "pass" : "123"
        },
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos3",
                "pass" : "123"
        }
],
"api-listen" : true,
"intensity" : "20",
"vectors" : "1",
"worksize" : "512",
"kernel" : "scrypt",
"auto-fan" : true,
"temp-cutoff" : "85",
"temp-overheat" : "75",
"temp-target" : "70",
"expiry" : "30",
"gpu-dyninterval" : "7",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "30",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"shaders" : "2560",
"thread-concurrency" : "25599",
"gpu-thread" : "1",
"gpu-engine" : "1045",
"sharethreads" : "32",
"lookup-gap" : "2",
"gpu-powertune" : "20",
"gpu-memclock" : "1250"
}

Elpida Ram use :

Code:

{
"pools" : [
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos",
                "pass" : "123"
        },
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos1",
                "pass" : "123"
        },
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos2",
                "pass" : "123"
        },
        {
                "url" : "stratum+tcp://eu.wemineltc.com:3333",
                "user" : "tabakerica.smos3",
                "pass" : "123"
        }
],
"api-listen" : true,
"intensity" : "20",
"vectors" : "1",
"worksize" : "256",
"kernel" : "scrypt",
"gpu-fan" : "75",
"temp-cutoff" : "98",
"temp-overheat" : "96",
"temp-target" : "90",
"expiry" : "30",
"gpu-dyninterval" : "7",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "30",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"shaders" : "1792",
"thread-concurrency" : "24500",
"gpu-thread" : "1",
"gpu-engine" : "1000",
"sharethreads" : "32",
"lookup-gap" : "2",
"gpu-powertune" : "20",
"gpu-memclock" : "1500"
}