December 7th, 2013 
CryptoBadger Now that the excellent Radeon 7950 video cards have essentially become unobtainable, I’m getting a lot of questions from people about how to get the most out of the new Radeon R9 280X in terms of mining. I managed to get my hands on a few Sapphire 280X cards this past week, and have done a little bit of experimenting.
I’ll get around to updating my guide for the 280X at some point in the next couple weeks, but if you’re looking for a good starting point as far as cgminer and undervolting settings go, read on.
Undervolting and Power Consumption
First, the 280X is essentially a re-branded 7970, which means they can be undervolted using VBE7 and the method I previously outlined for the 7950. My Sapphire cards run completely stable at 1137 mV, so if you’re following my undervolting guide, simply use 1137 in place of the 1081 that I recommended for the 7950. I’ll test lower voltages at some point soon, but the 7970/280X doesn’t have a stellar reputation as far as undervolting goes, so I wouldn’t expect too much more.
Running at 1137 mV, you can expect a 3 GPU rig to use between 850-900 watts at the wall, assuming you’re using the rest of the hardware that I recommend in my 7950 guide. You will, of course, need to upgrade the power supply to something a bit larger to accommodate the 280X’s larger power draw—I recommend a 1000+ watt gold or platinum rated PSU. The best choice for those of you without access to very cheap electricity is probably something like this platinum-rated Seasonic PSU. This gold-rated Corsair is a bit less expensive and also a good choice. If you’re looking to run four GPUs, this eVGA PSU has enough PCIe connectors without dealing with splitters.
Cgminer Settings and Performance
These are early test results, but they should get most of you up and running with acceptable performance. My tests were done on a platform with all of the hardware outlined in my guide, the only exception being that the PSU was substituted for a larger 1000 watt unit, and of course the 7950 GPUs were replaced with 280X cards.
For my tests, I was running Xubuntu 12.04 with the latest beta AMD Catalyst drivers installed, AMD SDK 2.7, and cgminer 3.7.2. I’m also using a 280X-optimized scrypt kernal in cgminer. I’ll be looking at Xubuntu 13.x and other driver versions (and Windows, too!) in the coming weeks.
The best results I saw were with the following cgminer settings:
-I 13 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192 --gpu-engine 1000 --gpu-memclock 1500 --temp-target 70 --auto-fan
This gets me roughly 710 Kh/sec per GPU, completely stable. Oddly enough, setting the core clock speed any higher than 1000 mhz results in lower cgminer performance.
I’ve read about some people getting speeds up to 750 Kh/sec with other brands and/or other configurations, but in my limited testing that was the best that I was able to achieve without compromising stability.
If you have another brand of 280X, you may want to try 1080 mhz or 1060 mhz for the core clock speed (--gpu-engine 1080), as many people report the best success in that range.
One interesting note is that the 280X seems to be pretty sensitive to heat, and will self-throttle its performance down whenever GPU temperatures get over 70C or so (at least, my Sapphires do). Setting cgminer’s auto-fan temp at 70C prevents this, although I run my rigs in a fairly cold room—some of you may need to point a box fan at your rigs to get the best performance out of your 280Xs.
I’ll be trying to get my hands on other brands of 280X cards, especially the ones by Gigabyte and MSI (although many are reporting good results with the Asus and PowerColor models, too)—although I’m sure most of you have noticed that they’re fairly difficult to come by right now.
If you’re currently running a 280X-based rig, please feel free to leave a comment below with your settings and results!
Posted in 
-g 2 -w 256 –lookup-gap 2 –thread-concurrency 8192
im not sure how to alter those setting. or what they are. i dont see a corresponding setting in cgminer 3.7.2.
is it only changeable with a script>?
my 280x asus hovers around 600kh/s with e-1000, m-1500, v-1.2, xubuntu.
with the above setting it tanked me to 550kh/s.
and –thread-concurrency 24000 broke it. wouldnt run. resetting to 21712.
Hi,
I can give you info on my 280x rig as soon as my PSU arrives (of course THAT’s the package that gets delayed -_- on a Saturday) and I got stuff running.
Rig consists of three XFX 280x Black Edition.
(They come with 1100 MHz core and 1650 MHz memory. We’ll see how that turns out.)
Greetings
Oliver
Getting stable ~707 with the cards.
Funny thing is, when throwing in an optimized kernel (https://litecointalk.org/index.php/topic,6058.0.html) the rate drops to ~540.
(yes, of course I adjusted the values to my configuration :D)
Config: http://nerdoholic.org/uploads/hiddencube/ltc/cgminer2.cfg
Performance: http://nerdoholic.org/uploads/hiddencube/ltc/rig_performance.png
Unfortunately I did already notice that throttling kicks in at around 77-81 °C. The core clock drops to 850 then, the KH/s to ~640-660 :/
Luckily that only lasts ~10 seconds and then everything is fine for a few minutes.
I hope that can be improved by the usb fan I was going to use for each rig…yet the perfectly abled engineer I am managed to cut BOTH wires (black plastic box, black wires…) on both rigs…so I’ll have to wait for the new ones 😀
Why I don’t have –gpu-memclock –gpu-engine –gpu-powertune options? What are the compilation options for them?
It’s basically what the parameter says:
–-gpu-memclock x
core clock of your GPU. Most r9 280x run optimal at 1020-1050
–-gpu-engine x
mem clock of your GPU. Most r9 280x run optimal at 1500
–-gpu-powertune [-20 – 20]
This allows your CPU to use 20% more power than the its TDP would usually allow…or 20% less 🙂
iam at stable 760kh/s per card on my sapphire r9 280x vapor-x.
im using this settings:
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer.exe –scrypt -o stratum+tcp://pool.org:3333 -u [user] -p [pass] –lookup-gap 2 –thread-concurrency 8192 -g 2 -I 13 -w 256 –auto-fan –gpu-fan 0-75 –temp-cutoff 90 –temp-overheat 85 –temp-target 70 –gpu-memclock 1500 –gpu-engine 1080 –gpu-powertune -20 –expiry 1 –scan-time 1 –queue 0 –no-submit-stale
not undervolted yet and running with the optimized kernel script
Thanks – I’m seeing a lot of Windows users report performance in the ~750 Kh/sec range with those settings on the Sapphire cards. Going to 1080 mhz lowered performance for me, so it’s possible that the linux drivers are to blame. I’ll try a few other versions.
Figured I’d post my info, as your blog helped get me started. working on putting my setup up on my own blog..I spent 2 and a half days getting this the way I wanted. I’ll let you know when that is done.
I’ve got my Sapphire R280X Vapor-X’s rock solid at 740kh/sec for 7 days now. (Have 3 more on the way for second rig)
I am running BAMT1.1 and then updated Catalyst drivers to 13.11 Beta
Using the VBE7 Rom tool. I’ve undervolted to 1.131 and boosted the fans to 100% at around 60C. I run two boards in a Haf XB case..hottest board is 65C. (Other one stays around 58C as it gets cool air from side of case. ambient room temp is 70F. I’ll be building a rack above the case and lifting the boards up so I can run 3 per rig.
I’m running ASROCK EXT4 970 mobo with the Athlon dual core CPU.
My boards peak at 275 watts each. Rig with all fans and two boards running all out is 610 watts.
cgminer is the BAMT included 3.1.0 (I tried 3.7.x but had no luck)
settings are as follows
“intensity” : “13,13”,
“vectors” : “1,1”,
“worksize” : “256,256”,
“kernel” : “scrypt,scrypt”,
“lookup-gap” : “0,0”,
“thread-concurrency” : “11200,11200”,
“shaders” : “0,0”,
“gpu-engine” : “1080-1080,1080-1080”,
“gpu-fan” : “0-100,0-100”,
“gpu-memclock” : “1500,1500”,
“gpu-memdiff” : “0,0”,
“gpu-powertune” : “20,20”,
“gpu-vddc” : “0.000,0.000”,
“temp-cutoff” : “95,95”,
“temp-overheat” : “85,85”,
“temp-target” : “75,75”,
“api-listen” : true,
“api-port” : “4028”,
“expiry” : “120”,
“failover-only” : true,
“gpu-dyninterval” : “7”,
“gpu-platform” : “0”,
“gpu-threads” : “2”,
“log” : “5”,
“no-pool-disable” : true,
“no-submit-stale” : true,
“queue” : “1”,
“scan-time” : “60”,
“scrypt” : true,
“temp-hysteresis” : “3”,
“shares” : “0”,
“kernel-path” : “/usr/local/bin”
That makes complete sense. I’m now running windows 7 i’m completely stable at 1125mv, @ 1070 GPU core, 1050 Mem. its getting ~725 per GPU. I can run at 1080 GPU core but because i’m still awaiting risers it approaches 78C and getting hot so i keep at 1070. I’m running Sapphire 280X Vapor-x.
Your right I was running linux xubuntu originally and also experienced the issue where increasing above 1020 GPU core would decrease performance drastically. under windows 7 its stable and allows increase in speed. The problem with this is the overhead of tight VNC drastically reduces performance so its difficult to tinker remotetely. unlike SSH. You must have a local monitor plugged in.
I’ll be trying the same settings as phreek, for example I’m not using target temp or powertune
Ok i running with similar settings to phreek with optimised scrypt bin file. Now getting 745Kh/s per GPU . 1080GPU core, 1125mV, ~75C but i’m not using risers I expect that temp to drastically drop (or noise reduce!) once the risers are in place. Currently fans spinning at 3,300 RPM
Forgot to mention get optimised script bin file here. scroll down to the 7970/280x files, download both bin files and past into your cgminer directory. It should replace the bin files created by your cgminer instance. If not your not running the matching thread concurrency and it will not start using the replaced bin. Just use Phreeks settings and it should work. ie, make sure you running thread concurrency ie, 8192
https://litecointalk.org/index.php?topic=6058.0
If you don’t mind me asking where did you pick up your cards from i can’t find them in stock anywhere
Build as per your guide but using 2 x R9-280 Sapphire Vapor X (have a 3rd but waiting for risers).
Xubuntu 13.10
Locally compiled cgminer 3.7.2
AMD catalyst 13.11-beta V9.4 Linux x86_64
AMD APP SDK v2.9 lnx64
ADL_SDK_6.0
export DISPLAY=:0
export GPU_MAX_ALLOC_PERCENT=100
export GPU_USE_SYNC_OBJECTS=1
/usr/local/bin/cgminer –scrypt -o stratum+tcp://pool:3333 -u username -p password -I 13 -g 2 -w 256 –lookup-gap 2 –thread-concurrency 8192 –gpu-engine 1080 –gpu-memclock 1500 –temp-target 70 –auto-fan
GPU 0: 77.0C 3379RPM | 744.4K/744.7Kh/s | A:17934 R: 0 HW:0 WU: 724.4/m I:13
GPU 1: 68.0C 2206RPM | 744.4K/744.8Kh/s | A:15296 R:225 HW:0 WU: 798.8/m I:13
No attempts to play with voltages yet and just got the crate in a room in the house (so not that cold). Garage is freezing by comparison so that will be next destination..
Yo Badger. Your guide is just simply awesome. I used it in combination with some other guides to get my rigs up an running. Thank you for all the work.
I reported my findings on Sapphire R9 280X pretty detailed over in the litecoin forums.
The summary is that There are various versions of Sapphire R9 280X out there that perform pretty different. I have 2 kinds running at the moment:
Sapphire Radeon R9 280X (11221-00-20G):
Gets up to 760Kh/s @ 1080/1500
This one is much easier to play around with: About every engine speed in the range of 1040-1100 MHz will yield >700Kh/s at he sweet spot memory speed. It is a power hog though. I am running it lower (1060/1490,730Kh/s) to “save” energy. (2.74 kH/W)
Sapphire Radeon R9 280X (11221-07-25G):
Best I can do is 725Kh/s @ 1028/1500
More finicky. 1500 memory speed is what this card likes. Anything else and it seems to tank. Engine sweet spots are between 1025-1040, depending on which card I look at. They all top out at 725 though. The good: More efficient than the card above. Getting 3.2 kH/W.
Thanks,
How do you determine the precise version, i.e what can I read off. Is it on the board or is there a command (i’m in linux).
Not sure which I have.
> More finicky. 1500 memory speed is what this card likes. Anything else and it seems to tank.
> Engine sweet spots are between 1025-1040, depending on which card I look at. They all top out at 725 though.
That is _exactly_ what has been driving me crazy the whole day.
Take a look:
http://nerdoholic.org/uploads/hiddencube/ltc/test_data.png
Oh, you changed the memclock up or down by 2-4 ? Well hello HW errors or -50 KH/s 😀
Doesn’t seem to be a ratio thing as far as I tried, no matter what core/mem combination at that (or a similar) ratio, the KH/s just drop significantly 🙁
dude, that is NICE.
awesome chart.
If you like I can upload it for you.
Did you change the value by keep editing the bat file or is there easier way to change value?
You can do that live from within cgminer.
Just press g, c, select your gpu and then e for core / m for memory MHz.
I am planning to mix a 280x and 2 7990s in a rig. Do anticipate any difficulty on optimization/configuration of this case?
Thanks crypto
I got a good deal on 3 HIS HD 7970 IceQ X² GHz Edition GDDR5 3072MB DVI/HDMI/2xMiniDP
http://www.amazon.co.uk/7970-Edition-GDDR5-3072MB-2xMiniDP/dp/B0093RJ08E/ref=sr_1_31?s=computers&ie=UTF8&qid=1386463868&sr=1-31&keywords=hd7970
Do you have any optimal settings for these please?
I am new to minging but this my experience with the HIS cards.
I have 2 HIS R9 280x cards in my first mining set up. They were the only ones I could find at the time. I followed Cryptobadger’s guide with the exception that I am using Xubuntu 13.10. I messed around with the mem and engine clocks for days trying to get past 600 kh/s. I actually had to bring the engine clock DOWN from 1000. If I increased the speed my Kh/s would take a nose dive, even down to 400! I eventually settled on a engine clock of 980 and a mem clock of 1500 giving me a steady 680 KH/s. That’s the best i could do. The HIS cards seem very finicky.
I purchased a third card, a MSI R9 280x. Getting 735 KH/s with 1050 engine, 1500 mem.
Hope that helps.
Hey thanks for the info mate. Do you have e-mail address or someway I can contact you so we can help each other out seeing as we have similar setup?
Cheers
Im in the UK and I cant get hold of any of the cards you recommend above. We always lag 6 months behind over here and get the crap stuff.
Any reason why you had to install the Beta and SDK? Can one not simply follow your Linux installation guide to setup 280x’s properly under Xubuntu?
I am running 2x MSI R9 280X. The best hashrate per GPU (697) I got was with these settings:
-I 13 -g 2 -w 256 –lookup-gap 2 –thread-concurrency 8192 –gpu-engine 1000 –gpu-memclock 1500 –temp-target 70 –auto-fan
Any attempt to overclocking caused decrease in hashrate. Strangely enough,
the pool is reporting combined hashrate for 2 gpus at 1500 kH/s: 750 kH/s pre GPU. Cgminer shows average hashrate at 1400 kH/s.
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer.exe –scrypt -o pool -u foo -p bar -d 0 -d 1 –no-submit-stale -I 13,13 –gpu-engine 1160,1160 –gpu-memclock 1800,1800 –thread-concurrency 8192,8192 -g 2 –auto-gpu –auto-fan –temp-target 75,75 –temp-overheat 85,85 –temp-cutoff 90,90
Both are ASUS R9280X-DC2T-3GD5 in a well ventilated open rig. Running Win 7 64bit, 13.11 Beta 9.5 drivers, 2.9 sdk, pulling 785 & 795KH @ 77C & 72C. Seems very stable, no voltage tweaking has been done yet, but will be working on them more soon.
First and foremost — Thank you sooo much! I know it takes time to output your thinking for others. Takes way longer you expect. So, again, thanks.
Question: Are you going to address the R9 290 separately? I didn’t want to confuse people by addressing the R9 290 here?
I built one of your 3x 7950 rigs to spec a few weeks back. Works great. I almost have all the components to build 4 3x R9 290 rigs and 3 4x R9 290s. Mostly Sapphire except for the 3 MSI GPUs. I would love yours or anyone’s guidance. Then I will contribute my results starting next weekend.
Silly question – what’s the definitive hashrate reading?
What number in cgminer et al should I be using? There are a few on the screen, and my pool worker dashboard shows a different number anyway.
It’s important we’re all reading the same number if we’re comparing 🙂
Hi CryptoBadger,
Could you please let me know if it would be worthwhile to use a R9 270X to mine Litecoin?
Thanks.
I’ve got 3 Gigabyte R9 280x and the 2 auxiliary cards have locked clock speeds and voltages, stuck at 300 and 150 mhz and .85V. What gives? I tried reinstalling drivers with no luck. I’m on Xubuntu.
Maybe I need to install dummy plugs? Anybody else experience this? Very frustrating.
Sam,
I’ve had 2 out or 3 Sapphire R9 280X OC cards exhibit the same clock speeds and voltages. Even after switching the “bad” cards into the primary Pcie slot alone with the monitor plugged into them, they still have clock speeds of 150 300 and voltage of .8 (or lower). Even after switching the bios switch to the other position, still no joy. The system would not even boot with one in the primary slot. This would indicate that dummy plugs are not the culprit.
I would really like some feed back on this issue as well. I just got another shippment of cards (exact same) so I’ll know more shortly.
I’ve had no luck even getting 2 R9 280X cards running at the same time.
I have the same issue. If I run one R9 280X card I get about a 620kh/sec rate. Added in the second card today and the second card is lockec at 0.85V and slows the whole system down to no more that about 550kh/second. I installed the system per the guide using Linux. Anyone have same problem but a successful solution?
Thanks!
I’m using 12.10 xubuntu, and for some reason I’m only able to get a mere ~560 on each of my 2 sapphire vapor 280x’s with -I 13 -g 2 -w 256 –lookup-gap 2 –thread-concurrency 8192 –gpu-engine 1080 –gpu-memclock 1500 –temp-target 70 –auto-fan
Wondering if anyone can help me out, cause I know there wouldn’t be that big of a drop from windows to xubuntu. I followed the linux setup on crypto’s guide, so perhaps the amd drivers I used from his guide are outdated? Thanks in advance if anyone can help me out.
Try running your engine at 1050 and not 1080. On my sapphire vapors, anything higher than 1050 causes a hash rate drop.
The sweetspot on my XFX 280’s is even lower: 1040/1499 (yes, not 1500. It gives like +10-15 KH/s :D)
Interesting… I went to AMD website to download the driver of R9 280X. When I tried to install, it’s telling me that I have wrong OS version. I did pick Win7 64 bit which I have. Now, I tried auto detect, and it’s feeding me back 7xxx driver instead of R9 series. What is this… -_-;;
These settings are netting me over 740Kh/s on the Sapphire 280x Vapor X’s using Xubuntu 13 and the catalyst drivers downloaded directly from the AMD site.
./cgminer –scrypt –lookup-gap 2 –thread-concurrency 8192 -g 2 -I 13 -w 256 –gpu-memclock 1500 –gpu-engine 1080 –gpu-powertune 0 –expiry 1 –scan-time 1 –queue 0 –auto-fan –temp-target 80 -o stratum+tcp://global.wemineltc.com:3334 -u user -p pass
I have two gigabyte r280x cards set up and one is both is peaking at 850kh/s but the second card seems to fluctuate between 600-850hkh/s and the second card is only showing about 70% activity. I’m running 1005 engine, 1500 memory, intensity 20. Can someone please let me know if the hash rates are meant to fluctuate around 100-200 and is the second card always less stable and run at a lower activity rate.
Would you post the bat file setting? I have same GPU and I am only getting 670s.
Do you all recommend going with Windows or Linux?
Got Sapphire R9 290X
I am running Win7.
I put intensity 20 and 1005/1500 and it’s feeding me 120kh/s…..
Anyone have good settings for Sapphire Radeon R9 290X 4GB GDDR5?
or Sapphire Radeon R9 290 4GB GDDR5 BF4 Edition Graphics Card ?
Also which is better to use Linux or Windows. It seems from most of the comments that Windows is better. I would have to buy some hard drives then.
Comments anyone?
I’m running dual Sapphire Vapor-X 280’s a 750 stable (some fluctuations now and then). When my tech bench arrives, I will be running 4, hopefully at 4x 750:)
Can anyone recommend what I can safely but aggressively run my heat settings and fan speed at? I have two Asus R9-280xs. Thanks Nicole!
I have two Gigabyte R9 290 with 880 khz each Card in an open test bench. They are very very noisy – rpm from the cooler are 4500 with 81C around.
I really wanted to test if undervolting is a way to have same Speed but not such a heat. Is it possbile now with the msi afterburner ?
Thanks in advance.
You may want to add a few fans. I think for most of people, temp aim is around 70s. Higher temp, higher fan speed.
running 4 sapphire r9 280x dual-x version
i’ve got stable undervolting of 1075 for 3 cards and 1081 for one card.
my sweet spot is 1035/1500 for 3 cards and 1024/1500 for one card.
Seems like I’ve got an odd ball in the bunch, but I’m getting 730kh/s x 3 and 720+kh/s on the oddball
stable and running for 4+days now, avging 2.9Mh/s. Lots of testing till I hit the optimal ranges. Just couldnt get over 730kh/s though
I’m pulling < 1100 watts at the wall with 3 120mm fans that are drawing .9A each (or about another 30 watts). Should replace those with more efficient ones, but they are keeping the cards sub 60 degrees Celsius.
Running windows 7, no dummy plugs, also using sapphire trixx to lock the fans at 70% speed on the dual-xs
chooface,
How do you get it to work without using dummy plugs ?
Do you have a specific version of the GPU driver ?
I have 3x GIGABYTE GV-R928XOC-3GD (rev. 2.0)
Thx
Grabbed the latest beta off amd site when I built it about a week ago. Dont know the driver # off the top of my head and I am at work, sorry.
in truth I’m not exactly sure why it is working without dummy plugs, but it does, so I figured I would leave well enough alone 🙂
Go to power settings and put “never” on both of “turn off display” and “put the computer to sleep”.
I have 3x XFX Double D R9-280X-TDFD running stable at 722K hash/sec.
Rig is open with a large circular house fan blowing on cards constantly.
Temps are about GPU0 ~ 70C, GPU1 ~ 72C, GPU2 ~ 74C
cgminer 3.7.2 with -I 13 -g 2 -w 256 –thread-concurrency 8192
I’m still playing around with the settings but it seems 725 K hash/sec is max. The cards run at 700Kh/s at GPU clock 1000Mhz, Memory clock 1500.
I get the 722Kh/s by overclocking GPU clock at 1035Mhz. I can get 725Kh/s with GPU clock at 1036-1037Mhz but Kh/s slightly less stable. Anything over 1037Mhz on GPU clock and speeds tank to 500’s. Same thing with any change to 1500Mhz memory clock either up or down. These cards are voltage unlocked. Nominal voltage is 1.250V. Catalyst 13.11.95 beta does not give any control for voltage so I downloaded MSI afterburner. That shows voltage slider control after enabling it in MSI AB settings. I actually moved voltage down to slightly under 1.0V and performance seemed OK. Not quite sure about how much advantage from undervolting so I moved voltage back to 1.25V. Anybody have a rough estimate of wattage saved per card by undervolting 0.25V ? They are 250-300W cards…Dont have a wattmeter.
I’m running Windows 7 x64, Asrock mb 990FX extreme 4, Athlon II X2 270 Regor 3.4 GHz 2×1 MB L2 Cache Socket AM3 65W Dual-Core and 4gb gskill ram. I use monitor on first card and dummy plugs (68 ohm resistors) on the other 2 cards. Windows would only detect 2 cards until I installed the resistors!
I’ve tried mining on 2 different large pools and my hashrate average is a little over 2Mh/s.
pic of rig…
https://www.facebook.com/photo.php?fbid=10201960852895145&set=pb.1599667230.-2207520000.1386798513.&type=3&theater
obtain a kill-a-watt.
test your wattage normally, then test again after undervolting.
record and report.
So here are some numbers I recorded with my setup (4x r9 280x)
4 cards 1025/1500 @ 1100mv = 1160 watts
4 cards @ 1090 (really 1087) = 1130 to 1140 watts
4 cards @ 1080 (really 1081) = 1120 to 1130 watts
So 13mv diff across 4 cards saved 20 to 30 watts
(Sanity check) 6 mv diff across 4 cards saved 10 to 20 watts
so my simple brain is telling me those numbers line up for the most part.
then we are left with about every 6 mv saves 2.5 to 5 watts per card
rough and crude observations made using my kill awatt…I think those numbers are within reason