Top Posts

sapphire r9 280x vapor x tri x 6 pin pcie power required paper clip trick power supply 6 pin power connector for graphics card pci to pci express adapter card how to install windows inside ubuntu what is a kernel power critical error build a bitcoin mining rig windows 7 error recovery mode asus p8z77 v lx drivers molex to sata power adaptor which pcie slot for graphics card sapphire vapor-x r9 280x oc virtual memory for 8gb ram how to open a terminal window in ubuntu 6 pin graphics card power adapter atikmdag sys blue screen windows 7 how to connect 4 pin molex to psu display driver amd driver stopped responding and has recovered how to open terminal window in ubuntu critical error kernel power 41 pci to pci express x16 adapter sata to 6 pin adapter how to check gpu bios version usb 3.0 maximum cable length power supply jump start connector amd driver stopped responding and has successfully recovered 128 bit vs 256 bit video card difference between pcie and pci express p vs np for dummies display stopped responding and has recovered pci e 16x graphics card run batch file at startup atx 24 pin 12v power supply pinout

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

(only) 50mh/s with two r9 290

VabVab Member Posts: 34
Hello,
I've read that an r9 290 have a hashrate of 29mh/s. However, I can only have 25mh/s with my sapphire r9 290 (SAPPHIRE AMD ATI Radeon R9 290 4G GDDR5 PCI-E DUAL DVI-D / HDMI / DP TRI-X OC VERSION (UEFI)).
I'm on ubuntu, how can I maximize my hashrate to have the expected one?

Thank you
«1

Comments

  • philakonephilakone Member Posts: 163 ✭✭
    won't happen with a sapphire card. try overclocking?
  • ImAMiner?ImAMiner? Member Posts: 208 ✭✭
    @Vab This looks about right for stock speeds as of this date. 29mh/s could have been captured when the dag was smaller and/or overclocking as @philakone said

  • happytreefriendshappytreefriends Member Posts: 533 ✭✭
    Be happy with 25MH per card..
  • philakonephilakone Member Posts: 163 ✭✭
    The 29 mh/s could've been an inner mean as well, which is very inaccurate since graphs of mining looks like a tooth saw with ups and downs that can be +/- 50% of your inner mean bench mark.

    Trust me my friend 25 is REALLY REALLY GOOD!! The card you have is one of the best performance to power to cost to hashing ratio.

    Keep up the good work!
  • VabVab Member Posts: 34
    edited March 2016
    Ooh ok make sense.
    i've searched for solution during this time and saw that I can use "--cl-local-work" and "--cl-global-work".
    When I enter ""--cl-local-work 256" and "--cl-global-work 49920" my hashrate can go up to 60 (sometimes 70). But sometimes it drop to 46 so it's less stable. It's difficult to take an average because of this but most of the time i'm between 53 to 59.
    Is it still better to put these?
  • ethfanethfan Member Posts: 458 ✭✭✭
    I acquired 2 x 290 to test. They will be returned soon as they are getting less than 2 x Tahiti! AND using more Watts in the process.
  • happytreefriendshappytreefriends Member Posts: 533 ✭✭
    edited March 2016
    --cl-local-worksize 256" and "--cl-global-worksize 49920 is way too high.
    I'm surprised your 'range is not 0-110MH with that crazy setting.

    Besides, wouldn't it be --cl-local-work ...?
  • VabVab Member Posts: 34
    edited March 2016

    --cl-local-worksize 256" and "--cl-global-worksize 49920 is way too high.
    I'm surprised your 'range is not 0-110MH with that crazy setting.

    --cl-local-worksize 256" and "--cl-global-worksize 49920 is way too high.
    I'm surprised your 'range is not 0-110MH with that crazy setting.

    First time I'm mining in my life, I don't know what does these relly mean. Is this bad to have these high values?
    And yes sorry it is work not worksize
  • workwork Member Posts: 2,084 ✭✭✭✭
    edited March 2016
    I experimented quite a bit, and the most optimal settings for my Tri-X OC card are:
    --cl-global-work 16384 --cl-local-work 256

    I'm getting (right now, with current DAG size) a bit over 29MH/s (1040mhz) at moderate overclock, and 31.5MH/s with extreme overclock (1120mhz). I also downclock the memory from 1300mhz to 1250mhz, as it has no effect on hashrate.

    My regular Sapphire R9 290s (non-tri-x) put out 27.5MH/s at 1000mhz OC and those same cl work values.

    All hashrates confirmed by share output.
  • happytreefriendshappytreefriends Member Posts: 533 ✭✭
    edited March 2016
    Dunno man. I'm having a real hard time seeing an R9-290 pulling 31.5MH even with an overclock at 1120/1300Mh speeds.


    How exactly do you 'confirm' your hash rate by shares?
  • workwork Member Posts: 2,084 ✭✭✭✭
    edited March 2016

    Dunno man. I'm having a real hard time seeing an R9-290 pulling 31.5MH even with an overclock at 1120/1300Mh speeds.


    How exactly do you 'confirm' your hash rate by shares?

    By confirmed with shares, I mean I used a low share target and confirmed share output matched reported hashrate over a period of 24hr. Have been mining on ethpool for a week at this hashrate, and pool reported average consistently agrees with mining software hashrate.

    31.5MH/s is no problem with a significant overclock, and with a moderate overclock 27-29MH/s is easy.



    Edit: note, you can't get quite these speeds with eth-proxy/ethminer. When I tested with eth-proxy, I was averaging 1-2MH/s lower.
    Post edited by work on
  • VabVab Member Posts: 34
    Now, my hashrate on the shell is between 55 et 63 but on coinotron, it says 24mh/s. Idea?(with the same settings "--cl-global-work 49920 and local 256)
  • happytreefriendshappytreefriends Member Posts: 533 ✭✭
    edited March 2016
    Still calling BS, unless you're using a custom miner which is misreading your hashes or something, or you're using a diff card. B)

    You'e the only person in the net getting that speed from a 290..
  • workwork Member Posts: 2,084 ✭✭✭✭
    edited March 2016

    Still calling BS, unless you're using a custom miner which is misreading your hashes or something, or you're using a diff card. B)

    You'e the only person in the net getting that speed from a 290..

    Dunno what to tell you. I don't have anything to gain by bullshitting, and this qtminer build definitely isn't over reporting hashrate as evidenced by pool shares over a long period. (Including testing against a local "pool" with very low share target).

    I've had a bunch of R9 290 cards for years (since before the doge mining boom), fired them back up to mine Eth, and got these hashrates with minimal tweaking. 27.5MH/s on the reference cards took no effort at all, and 31.5MH/s on the Tri-X card wasn't hard.
  • philakonephilakone Member Posts: 163 ✭✭
    If you haven't mined for a continuous 24 hours to get a large sample, your data is inaccurate if you look at 2-3 minutes of it hashing. Try getting a better sample over a 24 hour period.
  • VabVab Member Posts: 34
    No it's not BS
  • VabVab Member Posts: 34
    edited March 2016
    Vab said:


    My GPUs are not overclocked, I just used arguments I written here before. As you can see, it's less stable than without arguments but I have better peaks. The question is, is it better to have 53mh/s continually or to have this?
  • workwork Member Posts: 2,084 ✭✭✭✭
    Vab said:

    My GPUs are not overclocked, I just used arguments I written here before. As you can see, it's less stable than without arguments but I have better peaks. The question is, is it better to have 53mh/s continually or to have this?

    Try --cl-global-work 16384 --cl-local-work 256

    I experimented a lot, and these options produced optimal results for my rigs. Your cl-global-work is definitely too high, and will only serve to increase variance in reporting without increasing hash output. At stock clocks (1000MHz on the Tri-X OC), I get 27.2MH/s; at 1060mhz I get ~29.3MH/s.
    philakone said:

    If you haven't mined for a continuous 24 hours to get a large sample, your data is inaccurate if you look at 2-3 minutes of it hashing. Try getting a better sample over a 24 hour period.

    If you're talking to me, then re-read my posts. I tested for 24hr against a local "pool" to confirm hashrate equated to share output and have been mining on ethpool for over a week. Hashrate reported by qtminer has been confirmed by share count.
  • eddy4eddy4 Member Posts: 30
    @work Which qtminer version and OS do you use for your miner?
  • workwork Member Posts: 2,084 ✭✭✭✭
    eddy4 said:

    @work Which qtminer version and OS do you use for your miner?

    Win 10 x64 Ent LTSB N; I believe there is only one version of qtminer from ~5 months ago - https://github.com/etherchain-org/qtminer
  • eddy4eddy4 Member Posts: 30
    edited March 2016
    @work Ok, great! Do you think I could have a chance to reproduce the hash rate with Ubuntu and how do you protect your rig from failovers and switch to a different pool?
  • workwork Member Posts: 2,084 ✭✭✭✭
    @eddy4 I'm not using anything for fail-over; hasn't been a problem thanks to ethpool's stability so far. I do monitor the rigs quite closely tho, and can remote in to all the rigs from my phone if needed. So if the pool goes down I'll probably only lose a few minutes of mining.

    About ubuntu, I have no idea. I don't use binary linux distros, and I gave up after 2 days of trying to compile cpp-ethereum/qtminer on gentoo (god... what dev uses v8 but doesn't include the specific v8 version used? the api changes constantly and you aren't ever supposed to rely on a shared copy of v8 /rant) and just installed win10 on all my rigs since it was working well for mining on my gaming computer.
  • VabVab Member Posts: 34
    Thank you for your answers guys. I've asked coinotron about difference between my estimated hashrate and my ethminer hashrate.
    He responded "You are using getwork protocol.
    Could you please switch to stratum proxy or ethminer-0.9.41-genoil in stratum mode ?"
    I'm on Ubuntu but I don't have a clue about how switch to stratum proxy. Everyting on Internet is for Bitcoin and not nyoc. Do you have any idea please? (I'm using Linux for 2days)
  • workwork Member Posts: 2,084 ✭✭✭✭
    Try qtminer with the params I suggested? Qtminer supports stratum natively and doesn't need the lame eth-proxy layer in-between a getwork miner.
  • BiodomBiodom Member Posts: 693 ✭✭✭
    edited March 2016
    @Vab
    read here re proxy from the Linux guru himself (@dlehenky)
    /discussion/2695/frontier-mining-setup-notes-ubuntu-15-04-geth-v1-0-amd-ethminer/p5
    welcome to the club of Linux novices (I am at it for three weeks or so)
  • CaptForrestCaptForrest Member Posts: 28
    qtminer is a lot slower for me. Is this common?
  • eddy4eddy4 Member Posts: 30
    @CaptForrest I get with qtminer about 10% less compared with Genoil's ethminer. What means a lot?
  • workwork Member Posts: 2,084 ✭✭✭✭

    qtminer is a lot slower for me. Is this common?

    eddy4 said:

    @CaptForrest I get with qtminer about 10% less compared with Genoil's ethminer. What means a lot?

    With stock params?... Stock params are trash.
  • eddy4eddy4 Member Posts: 30
    @work Yes. What are your params?
or to comment.