Trial 1... 0 - starting to GPU mine with ethminer on Ubuntu

jeffmakesjeffmakes Member Posts: 4
Hi all,

Finally joining the ethereum party with a spot of mining!

I have a pair of R280X cards, and Ubuntu 14.04.2.

Following Stephan's mining FAQ and the Frontier guide I have set up the fglrx drivers and everything looks encouraging:
[email protected]:~/ethereum/repos/cpp-ethereum/build$ fglrxinfo
display: :0 screen: 0
OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: AMD Radeon HD 7900 Series
OpenGL version string: 4.4.13374 Compatibility Profile Context 15.20.1013

display: :0 screen: 1
OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: AMD Radeon HD 7900 Series
OpenGL version string: 4.4.13374 Compatibility Profile Context 15.20.1013

[email protected]:~/ethereum/repos/cpp-ethereum/build$ aticonfig --list-adapters
* 0. 01:00.0 AMD Radeon HD 7900 Series
1. 04:00.0 AMD Radeon HD 7900 Series
I've built go-ethereum from source (a906a84950f84a3cae178b2aca49fa6bba54a4c9 cloned today), built it, and geth syncs up to the network ok. It downloads the blocks in chunks of 256, then follows the chain.

I've built cpp-ethereum from source (01a2289c857ca7beb1576e5fe9c8638d655b43fd cloned today), built it with
cmake .. -DGUI=0 -DETHASHCL=1

But when I run the ethminer benchmark I get:
[email protected]:~/ethereum/repos/cpp-ethereum/build$ ./ethminer/ethminer -M -G --opencl-device 0 --opencl-platform 0
Benchmarking on platform: { "platform": "AMD Accelerated Parallel Processing", "device": "Tahiti", "version": "OpenCL 1.2 AMD-APP (1729.3)" }
Preparing DAG...
Warming up...
ℹ 19:14:13|gpuminer0 workLoop 0 #00000000… #00000000…
ℹ 19:14:13|gpuminer0 Initialising miner...
Using platform: AMD Accelerated Parallel Processing
Using device: Tahiti(OpenCL 1.2 AMD-APP (1729.3))
Trial 1... 0
Trial 2... 0
Trial 3... 0
Trial 4... 0
Trial 5... 0
⚡ 19:21:26|ethminer Stop worker 27.6873 s
min/mean/max: 0/0/0 H/s
inner mean: 0 H/s
Phoning home to find world ranking...
Error phoning home. ET is sad.
eth produces the same results.

Various combinations of --opencl-device and --opencl-platform produce the same results. I've also tested the output of the print-devices tool and it looks good
[email protected]:~/ethereum/tools/tools-master$ ./print-devices 
platform 0: vendor 'Advanced Micro Devices, Inc.'
device 0: 'Tahiti'
device 1: 'Tahiti'
device 2: 'AMD Sempron(tm) 145 Processor'
I've also tried the commit of cpp-ethereum that Stephan used in his guide (b821f0e7dc66bcc1ed35ecbc46c84fe83562b192) but it produces the same results.

I'm all out of ideas. Can anyone help?

Comments

  • gcolbourngcolbourn Member Posts: 48
    I've also got the same problem. I couldn't get the 64-bit Ubuntu install from usb stick to usb stick so ended up using the 32-bit version. Don't know if that would have anything to do with it?
  • KLmoneyKLmoney Member Posts: 27
    @gcolbourn, are you using a Gigabyte motherboard by any chance?
  • gcolbourngcolbourn Member Posts: 48
    edited June 2015
    @KLmoney Yes - Gigabyte GA-990FXA-UD3 990FX Socket AM3+ 8 Channel Audio ATX Motherboard
  • jeffmakesjeffmakes Member Posts: 4
    I have the same motherboard but didn't have any problems installing Ubuntu 64 bit. Is that what you mean? Or do you mean that with 64-bit Ubuntu on the Gigabyte motherboard you get the "trial 1... 0" behavior?
  • KLmoneyKLmoney Member Posts: 27
    Sorry @jeffmakes, yes I meant the former. The post I referenced is only a solution for installing 64 bit Ubuntu on certain Gigabyte motherboards, like the problem @gcolbourn seems to be having.
  • KLmoneyKLmoney Member Posts: 27
    edited June 2015
    I am no expert, but based on much discussion I've seen on this particular MB, it seems to have issues when 64 bit Ubuntu is installed on it without the suggested workaround.
  • jeffmakesjeffmakes Member Posts: 4
    OK thanks well ill try it anyway because i think my usb3 ports aren't working. But anyway, returning to the topic of the thread @gcolbourn you say you have the same problem; have you seen this "Trial 1... 0" problem and did it disappear with 32 bit Ubuntu?

    Before I submit a bug against ethminer i would like to know if it's user error on my part.
  • gcolbourngcolbourn Member Posts: 48
    I couldn't get Ubuntu 64bit to install, so I'm using Ubuntu 32bit, and have the same "Trial 1... 0" problem with it. I've not yet got up to date with the blockchain on geth, is that an issue? I would've thought the benchmarking would work anyway.
  • gcolbourngcolbourn Member Posts: 48
    Regarding the USB3.0/2.0 issue with the motherboard, all the ports were working, it was just a case of the Ubuntu-64bit installer crashing with errors, and when I checked the install drive for errors (from the boot screen), it said it had 1 error (but didn't say what it was). I thought it might be something to do with my crappy USB 2.0 flash drive overheating, but it eventually worked with Ubuntu 32-bit (which might've just been a coincidence - i.e. it didn't overheat and could've done that with the 64-bit if I kept trying).
  • jeffmakesjeffmakes Member Posts: 4
    Yes, I've synced up to the blockchain with geth, but I don't think that effects benchmarking anyway. I can CPU mine in geth, but GPU and CPU benchmarking in ethminer both produce the "Trial 1... 0" problem, and mining doesn't produce anything either (hash rate of 0).

    Anyone else got any ideas?
  • keokeo Member Posts: 21
    Same here. ethminer does 0 H/s trials.

    tried fgrlx and fgrlx-updates. The drivers seems to be installed properly.

    Gigabyte brand R9 280x, ASRock 970 Ex4 mobo, Ubunt 14.04.2 64bit.
    [email protected]:~/cpp-ethereum$ aticonfig --list-adapters
    * 0. 01:00.0 AMD Radeon HD 7900 Series
    
    * - Default adapter
    [email protected]:~/cpp-ethereum$ fglrxinfo
    display: :0  screen: 0
    OpenGL vendor string: Advanced Micro Devices, Inc.
    OpenGL renderer string: AMD Radeon HD 7900 Series
    OpenGL version string: 4.4.13374 Compatibility Profile Context 15.20.1013
    
    geth fully synced, DAG got created, ethminer was hand compiled based on Stephan's recommendation (b821f0).

    I'll try compiling the develop branch now.
  • gcolbourngcolbourn Member Posts: 48
    Anyone had any luck resolving this? I keep sudo apt-get upgrading hoping that the issue would be fixed, but I'm still getting the same: Trial 1... 0
  • keokeo Member Posts: 21
    Develop branch didn't work for me... Will try with Ubuntu 15.04 again next week and share results.
  • keokeo Member Posts: 21
    ok, finally could get 15.04 to work (my thumb drive died and to install it to another one).

    I've set it up based on Stephan's mining FAQ. Trials now show 12.5 MH/S.
  • gcolbourngcolbourn Member Posts: 48
    @keo That's great :-) Unfortunately I couldn't get wifi to work with 15.04(!) so I've tried 14.10 and get:

    [email protected]:~$ ethminer -M -G
    [OPENCL]:Found suitable OpenCL device [Tahiti] with 2946498560 bytes of GPU memory
    Benchmarking on platform: { "platform": "AMD Accelerated Parallel Processing", "device": "Tahiti", "version": "OpenCL 1.2 AMD-APP (1526.3)" }
    Preparing DAG...
    Warming up...
    ℹ 08:42:16|gpuminer0 workLoop 0 #00000000… #00000000…
    ℹ 08:42:16|gpuminer0 Initialising miner...
    [OPENCL]:Using platform: AMD Accelerated Parallel Processing
    [OPENCL]:Using device: Tahiti(OpenCL 1.2 AMD-APP (1526.3))
    [OPENCL]:Using 1 big chunk. Max OpenCL allocateable memory is 2102919168
    [OPENCL]:Printing program log
    [OPENCL]:LOOP UNROLL: pragma unroll (line 300)
    Unrolled as requested!
    LOOP UNROLL: pragma unroll (line 266)
    Unrolled as requested!
    LOOP UNROLL: pragma unroll (line 361)
    Unrolled as requested!
    LOOP UNROLL: pragma unroll (line 361)
    Unrolled as requested!

    [OPENCL]:Loading single big chunk kernels
    [OPENCL]:Creating one big buffer
    [OPENCL]:Creating buffer for header.
    [OPENCL]:Mapping one big chunk.
    Trial 1... 0
    Trial 2... 0
    Trial 3... [OPENCL]:Creating mining buffer 0
    [OPENCL]:Creating mining buffer 1
    2097152
    Trial 4... 20272469
    Trial 5... 20272469
    min/mean/max: 0/8528418/20272469 H/s
    inner mean: 7456540 H/s
    Phoning home to find world ranking...
    Error phoning home. ET is sad.

    Not sure what the LOOP UNROLL is about, and it seems to create a buffer, but often too late (mostly not by Trial 5, above is the best I've got). Still, it seems to be doing *something* at least. A little disappointed that I'm only getting 20MH/s with my R9 280X when everyone else seems to be getting 25, but I can't complain as I only paid £66 for that card.

    Still not tried mining proper, as it's taking forever to sync the blockchain. On my previous attempt using 14.04 I spent 2 weeks(!) trying to sync and only got up to 540000 blocks. Have to start again now..
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    @gcolbourn if your card is a slow starter (which it seems), try adding --benchmark-warmup 10 to the launch params. Then it will only start benchmarking after 10 seconds. Default is 3.
  • gcolbourngcolbourn Member Posts: 48
    @Genoil Thanks - It works with --benchmark-warmup 20:

    [OPENCL]:Creating mining buffer 0
    [OPENCL]:Creating mining buffer 1
    Trial 1... 20447232
    Trial 2... 20272469
    Trial 3... 20359850
    Trial 4... 20272469
    Trial 5... 20447232
    min/mean/max: 20272469/20359850/20447232 H/s
    inner mean: 13602360 H/s
  • TStewartTStewart Member Posts: 2
    Genoil said:

    @gcolbourn if your card is a slow starter (which it seems), try adding --benchmark-warmup 10 to the launch params. Then it will only start benchmarking after 10 seconds. Default is 3.

    Thank you very much for the help!

    "--benchmark-warmup 10" did it for me as well!
  • leon_proleon_pro Member Posts: 1
    edited June 2016
    There's a bug in the code. To be short: benchmark is started before mining thread is actually initialized, so it shows zeroes because there were no calculations done yet. Mention text starting with [OPENCL]:
    gcolbourn said:

    @keo T
    Trial 1... 0
    Trial 2... 0
    Trial 3... [OPENCL]:Creating mining buffer 0
    [OPENCL]:Creating mining buffer 1
    2097152
    Trial 4... 20272469
    Trial 5... 20272469
    min/mean/max: 0/8528418/20272469 H/s
    inner mean: 7456540 H/s

    On reasonably fast system you'll probably won't hit this bug. You can use --benchmark-warmup N as workaround, where N depends on how fast your system is.
    I have a patch for ethminer and genoil-ethminer, which waits while hashrate is zero and starts benchmark only after non-zero result returned.
    I'll submit the patches later, but any C++ programmer can also implement similar fix.

Sign In or Register to comment.