Bitcoin Forum

Hi Guys,

I have recently bought a sapphire radeon r9 280x for litecoin mining.
I have tried several different cgminer settings and so far this is giving me the best results so far ( 700 kh/s )

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer --scrypt --intensity 13 -g 2 -w 256 --lookup-gap 2 --temp-target 70 --auto-fan --temp-cutoff 90 --thread-concurrency 8192 --gpu-engine 1000 --gpu-memclock 1500 -o stratum+tcp://world.wemineltc.com:3333 -u workername -p password


Ive read people get 750 kh/s with this card, so if anyone can help me out with that?
I tried playing with thread concurrency and set it to 11200, also set memclock lower but none of those help.

Also i heard about undervolting the card, but i am not really confident doing that.

Picked up one of these today & having a hard time with it so far, its also voltage locked which is a bummer.

Currently I get 550 ish without tweaking, confident I should get it up to the 700's so will post back my config when I do

You should be able to get at least 735-740. Try changing TC to 8191 and gpu engine to 1045 or 1050.

Just mess around with the values of gpu engine and TC. You'll find a sweet spot.

You could try increasing intensity to 16 or 17. I've seen some go as high as 19 and even 20.

I have my intensity set to 20 on a r9 290, non x.

I'm getting 735K on my 280x Toxic with this settings:

-I 13 -g 2 -w 256 --gpu-memclock 1500 --gpu-engine 1080 --gpu-powertune 20 --thread-concurrency 8192 --auto-fan --gpu-fan 30-75 --temp-cutoff 90 --temp-overheat 85 --temp-target 70

Hope it helps :D

Hi Guys, i get four Sapphire R9 280X Vapor-X OC, the hasrate is very poor, only 550 mhs per card. I have tested the cards in two different Mobos with the same result. With two MSI R9 280x i get 720MHs per card.  Have anybody a hint for me.

Try : cgminer.exe --scrypt -I 20 -g 1 -w 256 --lookup-gap 2 --temp-target 75 --auto-fan --temp-cutoff 90 --thread-concurrency 24000 --gpu-engine 1050 --gpu-memclock 1500 (if You see HW Errors set -I 18-19)

I'm running these settings:

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer --scrypt -o URL -u USERNAME -p PASSWORD --lookup-gap 2 --thread-concurrency 8193 -g 2 -I 13 -w 256 --auto-fan --gpu-fan 30-85 --temp-cutoff 90 --temp-overheat 85 --temp-target 70 --gpu-memclock 1500 --gpu-engine 1052 --gpu-powertune -20

Don't remember where I got them exactly...I know they are a bit quirky, but I was able to consistently get ~730 Kh/s with these settings.  Good luck.

These are the core/memory settings that have worked for me for the various 280x s ive purchased. 

1085/1500

1035/1500

1050/1500

My cards' performance have varied.  The older cards I got could hit 750 kh/s while all the newer ones can only get around 725kh/s.

:o its so strange, i install trixx and activate disable ULPS on the Win7 system and play with the other settings, i see three cards runs with 715khs and the card with monitor con runs with 560khs. When all cards run with full power they will crash after a view minutes, so i reduced the GPU freq. and VDDV  a litte bit with the result they hash now stable with 705khs.

I just ordered a 280X.  My first card actually capable of some high hashrate.  So excited.  Want to buy about 10 more but not at $300.  I got 'lucky' i suppose in this market and have one on order for $310.  Wishing I had a pulse around black friday ><

anyways... this website is a really good reference for some ideas on settings.. but also on configuration:  https://litecoin.info/Mining_hardware_comparison

Give you some idea for miner configs and also how you setup your system... for example, method of cooling, operating system and such

here's to falling video card prices in early 2014

Thats nice.. I am running mine at 1049/1500 pushing 732 and another one is stable only at 1020 and pushing only 702 khs.. but as long as I even work on that PC, its very good, because hash rates dont fell much and PC is perfectly controlable.. way better than 7950 on I200 then you cant do anything on such PC

I would be interested to know what your temps are at when doing these various overclocks.  The hotter it is the more noise and errors.  Cooling is critical to OC.  Whenyou look at the comparison charts, it's the folks with great cooling that are hitting really good kh/s.  But I'm not sure how much other things like cpu mfr and OS may play in as well.

people were getting around 800 khs with 1150/1600 overclock on a 280x

Check that site https://litecoin.info/Mining_hardware_comparison and "Notes (OS/TEMPS/DRIVERS/HARDWARE/FIRMWARE)" about your card version - what conf to do for make xx kh/s.

I owe two 280x cards, one xfx one from gigabyte. And i would recommend to overclock them too heavy. I have them stable at 700 KHs. I know it is possible to have them running higher but with normal settings they will life longer. My settings are just --thread-concurrency 8192 -I 13 -g 2 -w 256 --gpu-fan 85 --lookup-gap 2 --gpu-engine 1030 --gpu-memclock 1500 --expiry 1 --scan-time 1 --queue 0 --no-submit-stale

If you have video BIOS Version: 015.041.000.000.000000, you are out of luck as it has an interal temperature limit that throttles the card if it exceeds 72c. Typically you'll have a problem in multi gpu configs. My card gets 700 khsec in a single config but hashes poorly in a multigpu config (less than and no better than 550khsec).


Flashing the card to an earlier bios say 015.039 does not help as it results in the card not posting. Cards with this bios version don't have this problem. Maybe dumping the existing 015.041 and editing it and reflashing may help? This is just a guess. Also pressing the sapphire button on the card so it operates the 7970 bios may help, I haven't tried yet (will give this a shot tho).

After investigations its due to the vrms running really hot (over 100c). Sapphire has been contacted but won't fix it and says the temps are acceptable :|

To fix it yourself, you may swap it for a toxic edition which has heatsinks on the vrms vs just heatpads that the vaporx uses.

I'm going to try increase the TDP e.g powertune to 30 to see if this helps with hashing performance.

Sorry to be the bearer of bad news if you aren't aware of this.

I am mining litecoins with 5 Sapphire R9 280X Vapor-X cards and the following bat file:

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer --scrypt -o URL -u USERNAME -p PASSWORD --lookup-gap 2 --thread-concurrency 8192 -g 2 -I 13 -w 256 --auto-fan --gpu-fan 30-85 --temp-cutoff 95 --temp-overheat 85 --temp-target 70 --gpu-memclock 1500 --gpu-engine 1045 --gpu-powertune -20

The rig will run file for 2 or 3 hours and then crash.

Any suggestions for finding and fixing the problem?

What happens when it crashes?  Does the whole computer shutdown?  is there a warning before it shuts down?  What are your temperatures when it crashes?  Is the card directly connected to the mobo or are you using risers?

As @pengoau suggested with the newer spec Sapphire R9 280x Vapor-X models have a newer bios version; you can try undervolting the cards and see how your performance goes. I'm testing a 4x Sapphire R9 280x Vapor-X rig and it's hashrate is 710-20KHs that are undervolted to 1.093V with default settings besides core clock @ 1030 and thread concurrency @ 8192.

Also, I've been in touch with Sapphire Support trying to find a solution. For anyone who owns Sapphire products or would like to support, please sign an Online Petition to have Sapphire (Althon Micro Inc) address the situation with a bios update.

Google Form Sapphire R9 280x Performance Issue Online Petition: http://goo.gl/F6kSWj (http://goo.gl/F6kSWj)

Thanks in advance.

@dougtpx in addition to @grendel25's questions.. Let us know about your PSU (and configuration).. Try and isolate if it is a supply of power issue by either trying with 2 GPUs; 3 GPUs; 4 GPUs; Differing PSU PCIe power plug assignment (T configuration).

https://litecointalk.org/index.php?topic=10313.45

Have a read, tell me if my findings help you.

I have been getting about 745kh/s per GPU with three of these cards in a rig.
My settings are similar to most people apart from the gpu engine, which i've clocked to 1100:

-I 13 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192 --gpu-engine 1100 --gpu-memclock 1500 --temp-target 70 --auto-fan

My rig's been extremely stable since I installed it about two weeks ago, however that was with two cards, I've only recently installed the third and am planning to add a 4th GPU (a 7950 this time). Temp's have been around 70c so far, an zero HW errors.

I'm just wondering about my gpu engine clock setting of 1100, as I don't really see anyone else posting such a high OC value. Would it be considered too high a risk to clock it that far? (factory setting is 1000)

Try reading the thread I linked to, contains a lot of good info (including settings, clocks etc to try).

on the newer cards with the new bios has the hardware changed ie new vrms or do you think saphire were getting to many failures returned due to blown vrms so they made the new restricted bios?

Problems seems to be with the vram, if you use Elpida you may or not have a problem as its cheaper (and not as good) vram than using Hynix (which is better).

Looks like they cut corners on the vram and possibly the vrms vs other oems to make a cheaper product.

I just bought a R9 280X tri-x today, I had an Nvidia before. Sorry I'm really new at mining, every time I open Cgminer and run the the bat file, it closes automatically. I just downloaded the 3.9 version, which version should I use and which settings would work best?

Thanks!!

3.7.2 is last to support gpus.

Settings, you gotta read this thread and see what works best for you. No one setting works well for all.

GL.

Use at your own risk: http://k-dev.net/cgminer/

Thanks to yours settings I'm getting 700-705 kH/s!!! Any other setting fail to break 700...
Can I know what bios version u have on those cards? Any problem with temperature? And have you checked VRM temp? One more (:P), what kH/s you obtain with that config?

Sorry, I have not 280X, my card it's MSI R7970 Lightning, but that's the same what all card's R9 280X. With this settings I have above 750 kH/s and temp 74-75, without any problems.

I got a few Sapphire R9 280X Dual X
Hashrate is generally between 700 and 730 KH/s

my settings
-I 13 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 11200 --gpu-engine 1035 --gpu-memclock 1500

I never managed to get more than 650 KH/s with only 1 gpu-thread  (-g 1) even whith higher intensities (-I 20)

I'm using Trixx to handle the fans (custom graph with 100% around 70°C)  and to set the VVDC to 1.100
The low voltage gives no noticeable hashrate drops but lowers significantly the temperature.

I am running 5 Sapphire Vapor-X R9 280X cards with a Lepa G1600 PSU. When I try to run them fast I use -I 13,13,13,12,12 to ensure I do not pull over 1400 watts. I already tried switching out G1600 PSUs.

The cards will run for days at about 575 kh/s using the following settings:

 -I 13 -w 256 --auto-fan --gpu-memclock 1500 --gpu-engine 1080 --thread-concurrency 8192 --expiry 1 --scan-time 1 --queue 0 --no-submit-stale

The cards with -I 13 will run at about 720 kh/s with the following settings:

 --thread-concurrency 8192 -I 13,13,13,12,12 -g 2 --lookup-gap 2 -w 256 --auto-fan --auto-gpu --gpu-memclock 1500 --gpu-engine 1070

... but either the machine will simply reboot in a couple minutes, or I will start seeing messages stating that the display driver as crashed and recovered.

I have tried running each of the 5 cards at a time with the same result.

It appears that without " -g 2 --lookup-gap 2" the cards will be stable but slow and when I add ' -g 2 --lookup-gap 2" the cards will briefly run fast and then crash.

Any suggestions?  Sapphire tech support has been of no help.

I only have one vapor-x, in that rig there are 2 dual-x as well. This are my settings:

color 02
setx GPU_MAX_ALLOC_PERCENT 100
cgminer.exe --scrypt -u *.1 -p x --worksize 384,384,384 -I 13,13,13 -g 2 -v 1 --temp-target 75 --temp-cutoff 95 --gpu-engine 1090,1020,1070 --gpu-memclock 1500,1500,1500 --gpu-threads 2 --thread-concurrency 8192,8192,8192 --shaders 2048,2048,2048 --lookup-gap 2,2,2 -o stratum+tcp://pool1.us.multipool.us:7777 --failover-only -o stratum+tcp://pool1.eu.multipool.us:7777 -u *.1 -p x

I don't experience crashes with this settings. The vapor-x is the third one in the values. It runs at 739 k/h at 80c (though 35º outside temp...). The oced dual-x runs at 752 k/h 73º and the other at 690 94º. I need to get some risers for that rig so I can separate them and get better temps. But if I can cool them I don't see why all of them wouldn't run at 750+...

Notice the extra options and that I'm not using auto-fan or auto-gpu. Perhaps that is what is causing the crashing? I think also that the fan speed works automatically as well when you define your temp-target.


Feel free to tip  8)

Im trying this setup in the OP on a saphire 280x and it doesn't show my GPU temp, but if i just use a basic script the temp shows is there a command in the OP's config that stop the temp from showing?

EDIT:   I think i figured it out...im mining on linux and ssh'ing in so i had to change the SETX to EXPORT this is my config below im getting about 708 kh/s on a Sapphire R280x B4 Edition, does this seem about right or can i bump it up a little?

export DISPLAY=:0
export GPU_MAX_ALLOC_PERCENT=100
export GPU_USE_SYNC_OBJECTS=1
cd /home/matt/cgminer-3.7.2-x86_64-built
./cgminer --scrypt --intensity 13 -g 2 -w 256 --lookup-gap 2 --temp-target 70 --auto-fan --temp-cutoff 90 --thread-concurrency 8192 --gpu-engine 1000 --gpu-memclock 1500 -o tcp://pool.dogechain.info:3333  -u user.user -p password

I'm accepting 0.05 BTC reward per GPU (280X only) for a setup producing at least 700KH/s and also undervolting a card. Team viewer or VNC is required. PM me if interested with details about your system.

After days of problems with one of the cards overheating the following settings work best for me.
Especially since the cards are not cheap i think everyone would like to get the maximum from their card without exhausting it so it will last more then a few months

--scrypt --gpu-dyninterval 7 --shaders 2048 --vectors 1 --lookup-gap 2 --thread-concurrency 11200 -g 2 -I 13 -w 256 --auto-fan --gpu-fan 30-75 --temp-cutoff 90 --temp-overheat 85 --temp-target 75 --gpu-memclock 1500 --gpu-engine 1080 --gpu-powertune -20 --expiry 1 --scan-time 1 --queue 0 --no-submit-stale

*Some coins seem to work better with thread concurrency 8192.
*Powertune works better on -20 then it does on +20.

This info is free as it should be but donations/tips are always welcome  ;D


--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
bitcoin@1Du9DFKtMFaVz4hFW1FGpRamcdYzkGTAFW
litecoin@LdNz7mvNn7Swyym8Uj53mMr87oJTpcmgtK
noblecoin@9URzRiG8eX7x9X1uZWXmDzmPqJS61Wfckp

how many kh/s ???

With the R9's do NOT use cgminer/bfgminer to adjust the voltage, fan speed, core or memory clocks.
Use a manual overclocking tool like Sapphire TRIXX(will let you set individual cards separately too), and set the fan at a high enough speed that it won't overheat at load.
CGMiner, when it's adjusting settings on the fly will go lower than it should, as it doesn't realize that the r9s can operate in the higher temp ranges.  I keep my 290 at 80 degrees with the fan set at a static 4400rpm.

Now use the scrypt mining tweak guide in your cgminer directory to tune it, and you should be good.

Also note, depending on what pool and what coin you are mining, it can sometimes make a 100kh difference, so choose a pool with good connection speeds.

That is all.

Hello,

I'm getting around 768-771 kH/s average

The best way to do this is to use conf file. You can find the conf file, called example.conf inside the cgminer dir. Edit it to look like this

{
"pools" : [
   {
      "url" : "YourPool",
      "user" : "YourWorker",
      "pass" : "YourPassword"
   }
],

"intensity" : "13",
"gpu-engine" : "1100",
"gpu-memclock" : "1770",
"gpu-fan" : "65",
"gpu-powertune" : "5",
"thread-concurrency" : "8913" ,
"shaders" : "2048" ,
"gpu-threads" : "2" ,
"auto-fan" : true,
"expiry" : "120",
"failover-only" : true,
"gpu-threads" : "2",
"log" : "5",
"queue" : "1",
"scan-time" : "30",
"temp-hysteresis" : "3",
"scrypt" : true,
"kernel-path" : "/usr/local/bin"
}


After this I will save it and name it YourConf.conf and start it over cmd. CD into the cgminer folder and type:
cgminer --config YourConf.conf

* If the temperature goes higher than 66C then change the "gpu-fan" option to 70 or more

I hope this helps a lot of you guys

Hi everyone. I have a R9 280x sapphire and i mine with 560 hr/s. Can you tell me if this is my best hash rate and if not how can i increase it? Thanks in advance!

At first check your BIOS version via GPU-Z if it contains string similar to 15.41.XXXXXXXXXXV42, you're not the lucky one and bought one of the "throttling" series. On this card the VRM temperature is to high and you need to lower the voltage. Try voltage change to 1,089V (use Sapphire Trixx software not cgminer argument) and provide good cooling (temps 73'C or less). With this card you're able to do approx 740kH/s.

If you have other BIOS version, you can do up to 780kH/s with increasing GPU engine to 1130MHz (1150mV / 63'C in  my case).

The best params I've found are generally based on 8193 thread concurrency.

CGMiner settings:

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
del *.bin
cgminer --scrypt --intensity 13 -g 2 -w 256 --lookup-gap 2 --temp-target 68 --auto-fan --temp-cutoff 90 --thread-concurrency 8193 --gpu-engine 1050 --gpu-memclock 1500 -o stratum+tcp://middlecoin.com:3333 -u 1ETYyETSWhkVxyj7HrBVEr5iAxAx2uSfTe -p x

 ;)

If you're desperate, try PM me and I can try to solve it for you via remote desktop or similar SW.

Alright, I just got my mining rig started today with two of these cards and I can only get 690kh/s after a bunch of messing around. Should I undervault this card and can anyone pm me and help me out with my rig?

Hi ,

I bought today a R9 280X sapphire !!!

My KH/s si actually of 580

with

Intel core Duo2 E8400 @ 3Ghz
4 Giga Memory at DDr2
ModtherBoard : PQ5

Windows 7 x64

My script connexion :

Frist file :
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1

Second file :
cgminer --scrypt -o stratum+tcp://usa.wemineltc.com:3335 -u XXXX -p XXXX--intensity 13 --auto-fan --temp-cutoff 90 --thread-concurrency 8192 --gpu-engine 1100 --gpu-memclock 1500 --temp-target 70

Actually, my temperature is 80°C not more...

I can't put over -I 13 , otherwise i got some HW...

Any idea to get better score ?

Try to use -g 2 (gpu-threads) and reduce gpu-engine. In your case, the command line should be:
cgminer --scrypt -o stratum+tcp://usa.wemineltc.com:3335 -u XXXX -p XXXX--intensity 13 -g 2 --auto-fan --temp-cutoff 90 --thread-concurrency 8192 --gpu-engine 1050 --gpu-memclock 1500 --temp-target 70