Radeon R9 280X GPU & mining: preliminary results

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!

You can leave a response, or trackback from your own site.

385 Responses to “Radeon R9 280X GPU & mining: preliminary results”

  1. Pete says:

    Hey there,
    I have a rig running with 3 Sapphire Vapor-X R9 280x, but I have some issues with the settings in cgminer. My set-up is Xubuntu 12.10, the latest amdcccle (Catalst control center) driver from ATI, 4GB Ram, Sempron 145. I’m using the following parameter with cgminer 3.7.2:

    –scrypt -I 13 -g 2 –thread-concurrency 8192 -w 256 –gpu-engine 1000 –gpu-memclock 1500 –temp-target 75 –auto-fan

    With these parameters all 3 GPUs are running at 540 MHash/s and I don’t now why, because most of you guys have 7xx MHash/s… Another mystery is that when I add –lookup-gap 2 cgminer crashes. Has anybody a clue what’s going on here. Apreciate really any help… Thx

    • Benton says:

      Pete…try adding in –shaders 2048 and see if that helps.

      • Wobbzz says:

        You should also change gpu-engine to 1080. You’re currently telling cgminer to run at lower clock speeds.

        • Pete says:

          Changing the –gpu-engine to 1080 brings me +20 to my hash rate. But it also pushes my power consumption +30 Watts up. So in the end there is no advantage, but thx for your idea.

      • Pete says:

        Hey Benton,

        thanks for your advise, but after adding –shaders 2048 (with and without –thread-concurrency 8192) didn’t change the hash rate.

        • Erebus says:

          Shaders should only be used for tuning. As such when a TC is determined the shaders like can and should be dropped as it does nothing. Unless you never set a TC and always let cgminer determine the highest possible TC attainable.. Which doesnt always guarrantee the best outputs.

          • Erebus says:

            Also ensure you are using proper syntax for your command lines. Intensity, work, threads can be abbreviated to a single preceding “-” but the others, as per cgminer’s readme.txt must have “–” as in “–thread-concurrency” and not “-thread-concurrency”. Devil is in the details. Don’t think it will make a diff but you never know.

          • Erebus says:

            Code wise, –lookup-gap 2 would be the only other line you could add. For linux, aren’t optimized mining drivers neccesary for max performance?

    • Remedin says:

      Don’t you need the AMD APP SDK on Linux?
      that brought me much performance on Windows.

    • Adam says:

      I find that if the filters on my case aren’t kept completely clean then my 280x will run slow, other wise I get 695kh with crypto’s setting on cgminer.

  2. Rob says:

    Try cgminr 3.2.2, was faster for me. Also try higher tc and 1 gpu thread. Mine’s 21-22000 ish for 760kh/s. Oddly,the “dropping your tc by 1” trick worked for me. I’d assumed it was non-tech/noob bs but it worked on 7/8 GPUs.

    • Pete says:

      I’m just about to get going with cgminer 3.2.2 and I’ll let you know, if it returns better results. Just by the way, what do you mean with “Dropping your tc by 1”-Trick ? Didn’t hear about that.

      • Rob says:

        When you get a threadcount you’re happy with, try it again minus 1 and see if it makes a difference. Against all odds it improved hashrate for me.
        So if you’re using 8192, try 8191.

        • Kay says:

          its not a -1 trick, + or – 1 would work, well actually it seems 0dd numbers tend to work better then even numbers. try 8193, i bet it will work just as good as 8191.
          for me, even numbers gave me different hash speeds by about 100khs, but odd thread numbers gave me the best hash rates. I can’t explain it, it just does.

          • Mies says:

            Good advise. I tried -1 and +1 for uneven tc’s and am sqeezing a bit more Khs out of my XFX 280x.

          • Brandon says:

            8191 is a prime. I’m completely speculating here, but might have something to do with it. With CPU’s, for example, using prime number array indices results in more efficient use of cache.

            Here are a few more if anyone wants to test that theory:
            8167 8171 8179 *8191* 8209 8219 8221

  3. Borut says:

    I’m running 3xGigabyte R9 280X (GV-R927XOC). Undervolted to 1.094, currently getting 715khash per card, underclocked to 1050MHz. Temperature of the cards is 75-80 degrees in an open rig without extra ventilation.

    • hab says:

      hi Borut,
      I am using same cards, but my best is 670 kh/s; what parameters are you using? I have undervolted to 1.094 v.
      thanks a lot

      • Borut says:

        These are my parameters:

        -I 13 -g 2 -w 256 –thread-concurrency 8192 –lookup-gap 2 –gpu-engine 1050 –gpu-memclock 1500 –gpu-powertune 10 –auto-fan –temp-target 80

        I’m running Cgminer 3.7.2 on Debian Linux 7.3 with AMD Catalyst 13.11 LINUX Beta V9.4 Driver (ADL SDK 6.0, AMD SDK 2.9). I think there is still some space for improvement, haven’t done any overclocking as there might be stability problems with undervolted cards. I’ll try dropping the voltage even a bit lower to the 7950 range – 1.081mV it might work.

        My power consumption is around 870W for 3 cards and Asus M5A97 motherboard, AMD Sempron 145 and 8GB Crucial Ram.

        • Ed says:

          I noticed that you’re using the Asus M5A97 motherboard. This is the one I chose to build a rig on, but I can’t determine 100% if all the PCIE slots can be used at the same time. I’ve read how some MBs will turn off some of the PCIE slots if a particular slot is occupied. Do you know if the M5A97 will allow me to run 4 GPUs? I see that you’re running 3 of them, so not sure if you’d even know this. The manual didn’t say anything about this, so I am looking for real world data.

  4. Borut says:

    auto-fan and temp-target require two dashes too:

    –auto-fan –temp-target 80

  5. wex83 says:

    hi Borut,

    I own this card, but I can not lower the voltage, I am interested in how the reduced voltage (more than any program)?

    thanks

  6. Hugo says:

    Hello.

    I read everything on this link. Thank you so far.

    Next monday I want to buy:
    3x GIGABYTE RADEON R9 280X OC WINDFORCE 3X 3GB GDDR5
    1x Asus M5A97 LE R2.0
    1x AMD Sempron 145 Box
    2x 2GB G.SKILL PC3-10600 1333MHZ 4GB CL9 DDR3V
    1x XFX Pro 1050W Black Edition (P1-1050-BEFX)

    I will spend total: ~~1185€
    I dont need to pay electricity because i will put in my office at work:))

    I want to help how can i calculate in how much time i have my investment paid, +/-.
    And an opinion to know if it worth it.

    Thank all,
    New Year

    • Patrik says:

      Hello all.My friend is building professional high-end water cooled computers for mining and gaming.He built like 300 pieces last year. From his rich experience i can say for all you guys:
      Do not buy Seasonic or Corsair and other psu’s, seasonics for example are well know for strange sounds when you go above 80% effiency, corsairs are burning out.
      It is based like on thousands feedbacks of his customers for all those years years.
      Best on the market are platinum editions of EVGA or SUPERFLOWER,they have best components and efficiency, very stable and no weird sounds.
      Here are really good reviews,check this :
      http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=361

      Mining needs best PSU,take care.

    • Erebus says:

      Coinwarz.com use their calculator for the coin you will mine. Say your system does arround 2000khs, depending on the coin u choose to mine at its current difficulty, could take u anywhere between 90 to 160 days (as an example). But difficulty will increase which means your returns will deminish over time.. So the timeframe given by the calculator cannot anticipate where things will be in months from now, or how much downtime your rig incurred or if the coin you chose to mine just doesnt perform well and its value crashes or on he flip side if it in in in reases dramatically in value. So could be 3 months to never in breaking even. There are no guarrantees. Setting realistic expectations are key here. Good luck wih that psu.. If it was me id get something with a little more capacity, 1050w with three 280x is cutting it close unless you’ll undervolt etc.

      • Phil says:

        1kw PSU is plenty for 3x 280x cards surely! They’re no more than 300w each?

        • Borut says:

          Before undervolting my system drew 1070W with 3 cards so with your psu undervolting is a must in my opinion.

        • Erebus says:

          Depends on the gpus or what you set them with (overclocked etc). Also depends a lot on your cooling solutions.. If u have a room thats just a couple degrees more then the gpu fans have to work harder so more power needed. But yeah normal settings say 1080/1500 with 3x 280x you’ll get anywhere from low 900s (absolute optimal conditions most 280x running in a open rig, well ventilated, ambient temps of say 21-23c) to upwards of 1200w (3x toxics, overclocked, fans running at 70-80%, ambient temps 23-26c).. I know from experience and the various things I’ve tried.

          Also idk about xfx psus but if he were to run it at 1000w unless this thing is rated gold or platinum it will push out the watts for as long as I can… How long that is is hard to say. Not even a 80 plus platinum certified psu can crank out 95% efficiency at almost 100% capacity and thats just the math, nevermind like others have said about exponent wear etc. these are consumer level decives.. Not enterprise server psus designed to work 24/7.

          • Erebus says:

            Meh.. Component not exponent. Anyways, based on specs I’m assuming you’re going with linux. All good, but if you’ve never touched linux, never worked with it, coupled with barely passable psu specs, this is a potential recipe for a lot of frustrations getting it actually up and running. It’s not plugging in a few setting and leaving for a week and expecting the rig to magically be stable and hash for 24/7. Takes work to get there. Takes access. Will you drive to work to do a manual reset if it hangs and the monitoring software isnt responsive? Love the enthusiasm, just know what you’re getting into is all. Basically, never cheap out on components, work within your knowledge base, if it’s lacking read more and set realistic goals. I knew windows like the back of my hand and yet still took a few weeks to dial things in when I first started building these rigs.

        • wbensky says:

          I have a rig running one gigabyte R9 280x and in total, it rarely ever uses more than 400w, but keep in mind that the entire rest of the rig (similar to cryptobadgers build) is included in there. So you would be cutting it a bit close with a 1000w psi.

  7. bryan says:

    I have two XFX 280X black editions. I can’t seem to find the bios switch? Can someone help me out?

  8. seb says:

    Hi,

    any idea of what psu wattage needed for 2x r9 280x from saphire @ 1080 / 1500 ?

    thanks for your help

  9. Erebus says:

    Thought some of you might find this interesting:

    A little while back did some tests with the sapphire 280x toxics and added my inital reuslts on here, by using the gpus overclocked at around 1165/1875, it brought up the hash rate to 815ish per gpu.. all well and good buuut is it sustainable? what are the accepted/rejects ratios? what’s the total wattage? well, decided to do a long term test to see if they woudl actually run fine and produce acceptable numbers of your burn electricity.. so here you go:

    Testing on 3 rigs:
    – Going on 4th day (one had to be restarted.. cgminer froze, the other two have been stable since I started them)
    – with 3 toxics each they get between 2.435 to 2.450 kh/s
    – The accepted/reject rates vary but are all within acceptable allowances of 2% to 3.5%.
    – I have the rigs auto-fan with target temps of 78c, seems to be holding fine, only a few gpus have run up intermittently to 3800 rpm.. the others vary. Now my cooling solutions are a work in progress, if I had a cold room, server room, heck a basement, things would be much simpler, but have to make do. Ambient temps in the room vary between 23-28c.. not optimal by any means so with some added cooling capacity I’ll be able to reduce some stress from the gpu fans having to work harder than they need to.
    -Finally which ties in to how much wattage these things use.. well toxics have 3 fans integrated into their cooling solution so they inherently use a bit more power.. that being said as these gpus are so highly over-clocked there is no room for undervolting etc (though I have not attempted this, so if there is an intrepid individual that would like to try I’d be interested to know how that went. That being said, being so highly overclocked pretty sure it needs all the voltage it can get to stabilize itself, but could be wrong). So, each rig pulls a whopping 1150 to 1250 watts. Yes it’s rather high, yes if one does the math one would say “Erebus, it’s not worth getting the extra 100 hashes per gpu..the electrity costs you crazy bastard”.. but think about this, if in 6mths valuations consistently go up for there various cryptocurrencies and difficulties make mining harder and harder.. every extra khs per gpu right now will net you a little more than you otherwise would.. and wouldn’t you prefer having as much in the wallet as possible? Personal choice naturally, just depends on your views, your own electricity costs where you live (mine is between 5 & 7 cents per wkh.. so a handful of canadian dollars per day, and this being the dead of winter, the rig heat the home which essentially eliminates or greatly reduces heating costs so almost a wash.. the summer will be another fun challenge to solve however lol).

    Hopee you guys find this interesting at the very least.

  10. Minorityof1 says:

    So I’ve been reading these pages for a couple months now, and my greatest thanks to Cryptobadger for so much help in setting this all up. I wanted to share my personal efforts to optimize my rig as a way to give back to the community.

    I’m currently running a modified rig based on Crypto’s design (I had been saving up for a while and right when I was ready to pull the trigger is when all this exploded in early December, so there were shortages!)

    CPU: AMD Athlon II X2 270 Regor 3.4 GHz
    MB: ASRock 990FX Extreme3
    Memory: G.SKILL 2 x 2GB DDR3 1333
    PSU: Thermaltake Toughpower Grand 1200W
    GPU: 2X ASUS TOP II 280X
    1X XFX DD 280X
    1X Gigabyte WINDFORCE Rev 2 280X (DIED!!!)

    Here is my current running script (CGMiner 3.7.2):

    -I 13 -g 2 –thread-concurrency 11200 –auto-gpu –gpu-engine 900-1060,900-1060,900-1040 –gpu-memclock 1500 –gpu-fan 20-67,20-60,20-85 –auto-fan –temp-target 72 –no-submit-stale

    Top Hash Rates: ASUS: 752 KHash/sec @ 1.118 V, ~71 C
    XFX: 730 KHash/sec @ 1.118 V, ~75 C

    Total Wattage: 886 W @ the wall.

    Reject Rate: ~1.5%

    Hardware Errors: 0 after ~3 days running.

    Random thoughts:

    1) My only complaint with the rig overall is the motherboard. I bought the 1200W PSU so I could run 4X cards, but the board kills the PCIe x1 slot when you have anything else plugged into the x16 slots, so that’s a bit of a bummer. ANY THOUGHTS WOULD BE WELCOME 🙂

    2) The ASUS have thus far outperformed both the Gigabyte and XFX, and they were only ~$20 more. I can easily push them to 1080/1500 and get 765 KHash/sec, but temps start running a little high. While they don’t have dual BIOS, I was able to undervolt to 1.118 and I honestly feel they could probably go even lower. Solid investment.

    3) The XFX is running nicely, but not as good as the ASUS. I can push it to 1050 MHz, but the hash rate gets finicky, 1040 is the farthest I could push it and keep it stable. However, it runs a lot hotter than the ASUS and the GPU is commonly throttled down so I’m usually averaging 700 KHash/sec @ 1000/1500. Runs stable at 1.118 V, but again I think I can push it further, maybe help keep it from running so hot.

    4) The Gigabyte was awful in my hands. Even with undervolting it (I got as low as 1.094 V, stable, which is impressive) and still it overheated. I was clocking it at ~700 KHash/sec at 1000/1500 MHz, but temps were running consistently into the 80’s and above. After a week it crapped out and I promptly returned it.

    Cheers to all the other newcomers to the world of cryptocurrency and my thanks again to Cryptobadger and fellow commentators for showing us the way!

    • PK_1 says:

      I would love to know if anyone has a solution to your issue with the PCIe x1 slots. I have just today ordered my rig and the motherboard I went for has 3 x16 slots and 2 x1 slots. I have bought 3 R9 290x gpus and when I have saved up a bit more money I was considering getting a couple more, and connecting them via some powered riser cables that are x1 to x16 adapters it would totally suck if the mobo just disables them.

      • Minority of One says:

        No word yet, though I did submit a technical inquiry with ASRock to see if they have any ideas.

        I think this is just a quirk of the Extreme3 M/B as I have heard of people putting 4, 5 and even 6 graphics cards in a single board. I can confirm that 16X to 1X risers work just fine as I have all my cards running through them, just only to the 16X slots.

        I’ll post if I hear back from ASRock. If anyone has any ideas or thoughts, let me know!

        • N8c says:

          Do you use molex risers ?

          • Minority of One says:

            Nope, using the ones without the MOLEX. They work just fine if I put them in the 16X slots despite the fact that they only cover the first little bit of the slot.

            Is there a difference how the cards are powered in the 1X vs. the 16x slots that could account for this?

        • Minority of One says:

          So I did some research following your inquiry and I think this explains it best:

          http://www.reddit.com/r/litecoinmining/comments/1dvit7/the_powered_riser_problem_is_there_a_hardware/

          It would appear the the failure is that the board cannot get enough wattage through the 1X slot (only 25W available v 75W at a 16X slot). I don’t have any powered riser cables handy to test this, but it would appear that if I use the powered risers, I should be able to use the 1X slot and maximize my M/B capacity. Now I just have to save up the money for a 4th card…

          Donations anyone? 😉

          • N8c says:

            That’s why I asked 😉

            They’re pretty cheap on ebay or stuff though (5-10€ when I ordered in Dec).

          • PK_1 says:

            Ok, I have ordered the ASRock Extreme 4, and I will be using the 3 x16 slots with powered risers for my pci-e x16 gpus, then I plan to use the 2 x1 slots with powered risers that are x1 to x16 when I save up a bit more money for two more gpus

          • N8c says:

            Please keep in mind that you shouldn’t take the overload from the motherboard and mindlessly put it onto the peripheral connectors.

            This would create an even worse situation since they are NOT designed to power like 4 GPUs over one cable connected to a single connector on the PSU.
            In the end, you will just permanently trigger your GPU’s overload protection or just burn it 🙂

            http://www.playtool.com/pages/psuconnectors/connectors.html#peripheral

  11. phil says:

    There you go…. 850w full load for 3 gpu’s! So what are you going on about lol.

    • Erebus says:

      Depends on setups.. overclocking, undervolting, ambient room temps etc. 850w with 3 280x is pretty great, well done. What are u getting as far as khs rates and are you overclocking/undervolting the cards?

  12. raycraps says:

    My Gigabyte 280X Rev 2 runs best on 1035mhz speed avg 730kh/s.

  13. Sam says:

    Hi all I’m new to all this and just setup my first rig, I have 3 R9 280x vapour x oc, can’t seem to get anything over 300ish on each card? Any ideas or possible issues I need to resolve? Thanks

  14. Bob says:

    Hi:

    I have a similar problem to Sam. I shelled out on 3 x XFX R9 290X cards. Loaded up on an ASUS Sabretooh Z87, 4GB ram and a Intel Pentium dual-core chip. (3000Ghz). Running on Ubuntu 12.04 and cgiminer 3.7.2 with the ATI drivers as per crypto badgers original guide.

    I can’t seem to get above 700kh with all 3 cards combined. However, independently run, each card can run about 500-600 Kh/s with fairly “standard” parameters.

    I tried all the usual parameters:

    -I 13 -g 1 -w 256 –thread-concurrency 8192 –lookup-gap 2 –gpu-engine 1000 –gpu-memclock 1500 –auto-fan –temp-target 70

    I tried playing with -g to 2, but this slows right down and same with thread concurrency. Changing the clocking speeds didn’t get any big difference.

    I split the scripts for each card and I can literally see the first card slow down to about from 500Kh down to 300Kh when I start the second card. When I start the 3rd card, everything slows down to about 200-250Kh.

    I looked at “top” and CPU doesn’t seem to be stressed and I have read from here that I thought CPU wasn’t important.

    Any suggestions?

    Thanks,

    Bob

    • Mies says:

      If you Run seperate cgminer instances they will interfere with eachother. Use the – – remove-disabled switch to work around that. With 3 seminar cards you should be ok with 1 config though.

      • Bob says:

        Success! I updated the drivers to the latest beta from ATI and the SDK. This solved my problem of running faster overall. Ran the config as per Crypto’s suggestions and I can get 705kh/card. The problem now is with cooling. Card 1 and 2 are running hot above 90C as I didn’t get the riser cables and they sit too close together.

        Now I know better. Will be installing riser cables and mounting the card somewhere else.

        • Juliano says:

          LOL!

          290x make 880-990khash easy! my two 7850 make 720khash (mixed) 8-.

          Use this config test:

          -I 20 -g 1 –thread-concurrency 32765

  15. New to the game says:

    Hey!
    I have a issue with my 3x sapphire 280X-rig.
    First 2 days were pure trial and error. I had to re-learn all about graphic-cards cus last time I cared about graphics was 8? years ago.
    The new cards are at a different other level and im a bit confused.

    I finally found a sweet setup for my 3 cards without risers to keep em at ~70C, minor adjustment sent them to 80, and the cooling became a issue with fans going up to 4000 rpm occasionally..
    Anyway; the .bat giving 715/card is fine for my settings, BUT
    The CG-miner stops working and I loose hours of mining before I notice and can restart it manually.

    My .bat and the error-message that appear can be seen at the pic \/

    Pic; http://s21.postimg.org/is680ygnr/ch_breakdown.jpg

    Now im curious about a few things; the biggest one is; WTF is going on? And how can I prevent it?
    I tried thesame .bat in 3.5.0 and 3.7.2 with thesame result. The server goes down randomly(?).
    Other .bat will work with lower hashrates and go down too.. It seems to be a matter of time(?).

    I’d love to get this thing sorted. If you have a guide explaining all I asked about Id be really grateful for a link, or if you have time to explain it to me Id be even more grateful!

    And while im here.. I might aswell ask about some other stuff that I not managed to understand while googling/trial-n’-error.

    Ive read http://rumorcity.com/2013/12/14/radeon-r9-280x-litecoin-mining-and-sweet-spot-for-700khashsec/
    and http://www.cryptobadger.com/2013/12/radeon-r9-280x-gpu-mining-results/ about 5 times each, and I think I figured out what each setting does, except for the undervoltage/stability-part.

    I go with 1020/1500 on the GPUs; How does the lower voltage effect the mining? I haven´t fully understood whats acctually happening with the card when I lower the volt (more then the consumed watts goes down) but what should I aim for, and how do I know whats the “best”?
    I find the spot where the GPU stops working and need a restart, but how do I know this is the best mvolt to provide at certain gpu/mem-clock?
    What define the miner as “unstable”?

    And whats the relation between TC and GPUmem/clock.
    I know its important to find the best TC, but I dont know WHY.

    Best regards from a newbie!

  16. Mike says:

    You all seem to know what youre talking about. My Sapphire 280x Vapor X will come in the mail in two weeks. This is my first and only mining card, so until i save up for a dedicated rig i have to make do with my current hardware. However, that means that i have to use my 375w power supply. Will i run into any problems running one vapor-x from 375 watts? Or will i have to buy a new power supply? Do i have to undervolt?

  17. Danny Mack says:

    So, I’m almost there. I built my first computer rig ever. I have four Sapphire r9 290x cards fired up and ready to go. Problem is – I installed windows and the screen is all scrambled. Even the bios screen was a little scrambled at first, but looked better w/the fourth start. Windows looks 8 bit, but worse and when you move the mouse, it just erases the screen.

    I’m sure it might be due to the drivers needing to be installed for the cards or even the drivers for the mobo…but I can’t load the drivers since I can’t see the screen.

    I’m doing it via the hdmi off of card/pcie #1. I tried using the dvi connection on the card w/a diff monitor, but there was no signal at all. I am forced to use an hdmi and a tv/monitor. I have enough power [1500w], a great cpu (amd3), 12gb ram (three 4gm cards], and a good mobo (gigabyte 990fx ud5). Everything is new and I wore a bracelet the entire time to ensure no static shocks.

    Any thoughts?

    • Kishkumen says:

      Having artifacts (scrambled as you call it) at your BIOS screen already indicates some hardware mailfunction. Drivers arent loaded untill Windows so a BIOS screen should always be error-free. What happens if you start off with 1-2-3 cards in stead of all 4?

  18. Kishkumen says:

    @Mike

    375w is tight, maybe too tight but it is could be possible. I tested my own setup with 1 card and it drew +/- 300w under full load. This is with a Asrock FM2A88X PRO+, AMD A6-5400K, 8 GB Ram, Vapor-X 280x and a SSD. No undervolting.

    Just remember that your really pushing it, could be done with proper undervolting.

    • Mike says:

      Ok, i was wrong, i only have 350 watts. However, i also found a 250W in the old pc in my closet. What if I put the 250 inside the case of my mining pc for the motherboard etc., then had the 350 sitting next to the case and used it to power my gpu? This way the card could consume up to 350 watts, and the pressure of the rest of the PC would be carried by the 250. This would, of course, be temporary until I organize and construct a dedicated rig, which would use something like a corsair 1000w or 1200w. Would this work for the time being?

      • Kishkumen says:

        In theory that would be possible. The only problem is that without a Mainboard connected (or a PSU tester) you cant turn your PSU on. So it wont give any current thus wont power your card(s). If you did have a PSU tester that would switch the PSU on it would work though.

        But to be honest, I wouldnt vote for the method 😛

        • Mike says:

          Alright, looks like im out some cash for a new power supply

          • Minority of One says:

            Probably for the best. According to the kill-a-watt, my CPU/MB/RAM/etc. pulls ~150 watts with each GPU pulling about 250 watts at full load, though I have a reasonably aggressive undervolting at 1.118V.

            Looking around, these seem to be pretty standard numbers so you can use them to estimate the power supply you’ll need given future expansion plans.

            Best of luck!

          • Mike says:

            I’ve heard of people undervolting to 1.093, and the rest of my system is about 100w. I think I can squeeze it in

  19. Kishkumen says:

    One more post, sorry for not doing it in 1 post… I have a question to the side:

    Sapphire R9 280X Vapor-X OC cards have a little switch with a Sapphire Logo that glows blue between the card PCB and the fan. Its on the same side as the power connector to the HDMI/DVI connector side.

    I heard this was a BIOS switch, but i cant find someone to confirm it or explain it. I was hoping someone could clarify the function of this small button. its about 0.8 cm x 0.8 cm large.

    • Mike says:

      I know the button youre talking about. I have heard it called the “Sapphire Button”, and I believe it is separate from the bios switch, but I too am unclear of its function. I will let you know when I get my card, hopefully between 1/25 and 2/1. The rest of my components are below-average power consumption, so with a respectable size undervolt I can make it work no problem?

    • Mike says:

      The Sapphire Button is the dual-bios. When it is pressed, the card is UEFI enabled. When its not pressed, the card is in legacy BIOS to support older motherboards and operating systems.

      http://www.sapphiretech.com/presentation/product/product_index.aspx?pid=2024&lid=1

      Scroll all the way to the bottom of that page to see Sapphire’s description of it.

  20. Anonymous says:

    I’ll share my mining rig adventure as well, might be helpfull to someone

    I finally got my rig running after numerous problems including:
    – Artifacts
    – BSOD when loading windows after reboot
    – Poor hashrates
    – Catalyst Control Center problems
    – Having a GPU not recognised
    – System not giving POST, BIOS hangs
    – CGminer not starting
    – CGwatcher raping my configs
    – …..and probally a few i forgot

    BUT

    It runs!!. End results so far (just have it running 30 mins so I’ll look again in the morning (0:23 at night atm) are as following:

    Hardware used:

    – Asrock FM2A88X Pro+ Mainboard (Socket FM2)
    – AMD A6-5400K (Sempron was not available)
    – 8GB Kingston ValueRam 1600 Mhz
    – 64GB SSD (had it laying around)
    – Corsair RM1000 PSU (1000w)
    – 3x Sapphire Vapor-X 280x (BIOS 015.041.000.000000 for all cards)
    – 1x –> 16x PCIe Riser Cables

    At the moment I got my settings at 1095 Core / 1500 Memory / 1.150 VDDC giving hashrates of 740/760/760 for the 3 cards. First card has the monitor on it so it performs a bit less.

    The rig draws about 820w of power according to my kill-a-watt with above settings. I have to say that 1.150 vs 1.200 VDDC results in a power consumption decrease of about 80w. Without it consumed 900w with peaks to 930.

    Side note: decreasing voltage also decreases temprature by about 5 degrees.

    Tempratures of the cards are 68/60/68 degrees (with mentioned settings), the middle card has the benefit of being almost blown out of the rig by fans, so thats why the temp difference is so huge. Fan speed is set at 50% wich is 2300 rpm.

    I’ll leave it at that for now, I have to let it run for a few days and then see if its stable. Then ill maybe tweak voltages or core speeds for a bit higher hashrate or lower power consumption and tempratures.

    Hope it was informative 🙂

    • Rad says:

      Hello Anonymous!

      I would like to ask you for some advice because I have almost the same mining rig as yours.

      – Asrock FM2A88X Extreme4+ (Latest BIOS 1.70)
      – AMD A4-6300 APU
      – 8GB Kingston 1600 Mhz
      – 500GB Toshiba 5400rpm HDD
      – Corsair RM750 PSU (750w)
      – 2x Sapphire Vapor-X 280x (BIOS 015.041.000.000000)
      – no PCIe Riser Cables
      – Windows 7 64-bit, SP1

      No matter what I’ve done so far, I couldn’t get the system to start windows with 2 GPUs installed and a working driver. With one gpu everything seems ok, 13.12 works and I was able to test mining for 30 minutes. Hashrate was around 740Kh/s. When I shut down and install the 2nd GPU, Windows 7 hangs with a black screen and a following reboot, just after the windows logo disappears from screen. I tried 13.6 and 13.10 as well – no luck. I quit on windows 7 and I am now trying under Windows 8.1, but I don’t think i will be successful. I tried to install 13.12 with both cards inserted and catalyst installer froze, windows rebooted and removed the drivers automatically. Trying one card at a time now.

      Any tips to get around the mess will be welcome.

      • Kishkumen says:

        Aka Anonymous (didnt enter name on post)

        Hello Rad,

        It’s strange you cant get the dual graphics to work. I am using riser cables but that shouldnt make any difference.

        A few questions pop into mind though:

        – Did you disable onboard grahpics in your bios?
        – Did you use crossfire bridges?
        – Does the system boot if you place just one card into either PCIe Slot 0 or 1? Did you test both slots seperatly?
        – Do you get a bluescreen? (if your not sure, disable auto-reboot on failure in Windows). Black screen –> reboot doesnt always mean you dont get a BSOD.
        – Did you tweak GPU voltages/clocks yet?
        – Did you try to boot again with 2 cards after the BSOD/reboot?
        – Are you sure you downloaded the right drivers (32-bit or 64-bit)?

        I’ll try and monitor for your reply 😉

        Good Luck

        Kishkumen

        • Rad says:

          Hello!

          Thanks for the quick reply. I’ve tried everything you suggested. No voltage tweaking yet. I tried with and without crossfire cables. It doesn’t matter if onboard graphichs is disabled or enabled either. Booting with one card from both pcie slots is fine, windows starts, drivers load. Whenever I added the second card – it crashes with a black screen. I tried disabling the “automatic restart on BSOD” from the windows startup menu, but it probably didn’t work and I never saw a BSOD, although I also suspect there should be one.

          I didn’t mention that I am building 2 rigs, both the same. So I do have 4 gpus at my disposal. I took out one of the other 2 available cards and inserted it into a freshly installed windows 8.1 on the same machine. It loaded without any problems, drivers installed and the rig is already mining for hours without any issues.

          Now, after I started building the 2nd rig, with one “new” card and one from the old machine – I ran into the same booting problem with 2 cards. The “new” card works fine alone, but the other one doesn’t even want to post when installed alone. With both installed I manage to get to windows 8 welcome screen, and once almost to desktop, but it eventually crashes and reboots.

          I think it’s a faulty card. It did post yesterday on the first machine (though never loaded windows) but now it seems a bit more dead – not even posting. It probably has nothing to do with windows 7 and bad drivers. I don’t want to tell you how pi$$ed I am after waiting for more than a month on queue to get 4 of these cards and now I will have to RMA.

          I will give it a few more tries and will bring it into service tomorrow if it doesn’t fix itself. I tried the UEFI and legacy BIOS button on the card without success. So unlucky…

          • Anonymous says:

            I’m sorry to hear that your having problems with a card. But based on your story about how you tested I suggest you try the following:

            Build the Rig with the faulty card only and install Windows on it, does it boot then? If not then I agree your card is faulty and needs to go through RMA (or DOA if your within the time period).

            The thing is, I had booting problems too with my Rig. It wouldnt load windows, wouldnt recognise the card etc. I installed Windows about 4 times before I got it working right and there were times I thought my card was faulty.

            My first installation was with 1 card, then adding the second card and third card later. That didnt work out so well. Second time I messed up my drivers by installing/reinstalling Catalyst versions and SDK versions. Then the card wasnt found in CGminer and finally to get it working I did the following:

            In my BIOS, set primary display output on PCIe, disable onboard VGA, started placing cards down from the first PCIe 16x slot (i have 1x –> 16x risers so i can connect a card to the 1x slot above the first PCIe). Because if I started from the 1x slot above the 16x
            I had BSOD’s after reboot in Windows or some weird other problem.

            TLDR; What finally did it for me was: Connect the system like it should be in the end, dont start with 1 card and add later. Fresh install Windows with everything connected. Install drivers, install Catalyst 13.12 as your last driver, Install CGminer, Install CGwatcher (if you want), Make config file and go.
            I can’t seem to remember that I installed SDK 2.7 or 2.9 and I can’t find anything on the Rig about it that I installed. So I think I only installed Catalyst.

            Sidenote: Why are you building 2 rigs with 2 cards per rig if you can just drop 4 cards on the board you have with riser cables?

            Good Luck

          • Kishkumen says:

            Anonymous again 😀

            P.s dont use crossfire bridges.
            p.p.s. And you dont need VGA dummy plugs anymore for R9 cards

  21. Steele Miner says:

    For anyone running Radeon R9 280x, 290 or 290x, you may want to use sgminer rather than cgminer.

    Sgminer is giving me same exact Kh/s and better WU at the same intensity. Sgminer also allows Intensity values up to 42.

    Also, the crash when you Quit is fixed in sgminer.

    See announcement here : https://litecointalk.org/index.php?topic=13190.0

    Enjoy.

  22. Howard says:

    Hi everyone,

    I just got my first rig start up running on asus r9 280x directcu top II. I’m currently mining with cgminer 3.7.2, catalyst 13.11 b9.5, sdk 2.9 hashing at 705 khs, 71 C, and undervolted to 1.1V. I’ve been trying to tunning the engine and memory around and it doesnt seems to help raising the hashrate. I know these cards can do a lot better than my current status. Any suggestion ?

    My current cgminer bat file:

    setx GPU_MAX_ALLOC_PERCENT 100
    setx GPU_USE_SYNC_OBJECTS 1
    cgminer –scrypt -o stratum+tcp://lite.wemineltc.com:3334 -u xxxx -p xxxx –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 72 –gpu-memclock 1500 –gpu-engine 1080 –gpu-powertune -20 –expiry 1 –scan-time 1 –queue 0 –no-submit-stale

    • Minority of One says:

      The only two things I can see that are different from my settings are the voltage and the scan/expiry times. I did notice that when I set my voltage to 1.1 (it’s at 1.118 now), the cards had a difficult time getting past ~700-710 KHash/sec, so maybe try bumping the voltage up a bit.

      I don’t know if scan/expiry time affect it at all, but with those settings, it seems like your computer might be looking for new work so often that it might affect the ability to hash the algorithm. Total shot in the dark, but work a try.

      Oh, and I see Powertune. It never did anything to my hash rate or power consumption when I adjusted it, but maybe at such low voltage it has a bigger effect?

      These are my random thoughts…

  23. Cryptosaurus says:

    Here is my config. Semprom 145, 4GB ram, ASUS M5A97 R2.0 AM3+ AMD 970 SATA 6Gb/s USB 3.0 ATX AMD Motherboard, 4 XFX R9 280x GPU. We used riser cables for each card. The cards are separated by ~2in. We are using BAMT for our operating system. Our cgminer config is: I 13, thread concurrency 8192, threads 2, lookup gap 2, workers 256, engine 990, memory 1720. The cards are sitting solidly 680 kH/s. I can’t get any more by tuning the engine or memory speeds. So it’s going to come from drivers, or kernel optimization. I have a small fan that blows over the rig to move the hot air out. Three cards are sitting at 70C and we have one at 74-75C. I plan to improve the airflow to try and get that card down. The rig runs right around 1280w. The computer use 135w under idle. That puts the cards at ~286w per card. That puts it at 2.377 kH/J. I can’t seem to access the catalyst drivers to try undervolting. Also I haven’t found any links for a scrypt optimized kernel. If anyone has any recommendations please let me know

  24. mattkn says:

    Hey Guys,

    I’ve managed to get a 3 GPU rig up and running with 3 XFX 280’s. All 3 are running smoothly at 550kh/s. I’ve overclocked. I’ve not undervolted as I don’t have too much faith in my ability not to brick them. I’m running the original Cgminer that is in the guide on Xubuntu 12.04. My question is this…

    Is it possible that if I get a newer version of Cgminer ie. 3.7.2 that would make a significant enough a difference to warrant changing it? I can provide more details if needs be. Cheers.

  25. maecky says:

    Hi,
    thanks for you great tutorial. I built a rig with 3 x R9 280x and used all of the components you descript except of the cpu (I took another one with TPD: 45W).

    I tried to undervolt my cards like you suggested and everything runs fine. BUT, my wattmeter shows, that i draw 1900W from the wall?!

    Is there something wrong with my wattmeter? I can hardly believe the values?

    Thanks,
    maecky

Leave a Reply