[BBR] Boolberry GPU Miner Discussion - AMD & Nvidia - OpenCL & CUDA - page 17.

sussex

hero member

Activity: 658

Merit: 500

Quote from: tygarbyte on August 07, 2014, 10:31:38 PM

4 card amd rig

boolbd.exe --start-mining=1K5BH.....rCLe9gxXF --mining-threads=4

boolbd.exe --start-mining=[address] --mining-threads=[n]

n - number of GPUs to use

Doesn't seem to be working though.

When I try 4 threads I get an error about CL_Name and platform index must be between 0 and some huge number.

Tried alter device index also, but still the same.

Only one gpu shown working in gpu-z

Do I need a certain version of driver maybe? I'm on 13.251 Catalyst 13.12

sussex

hero member

Activity: 658

Merit: 500

Quote from: Wolf0 on August 07, 2014, 10:28:17 PM

The shares are fake - just a check to make sure the card is working.

Thanks.

tygarbyte

sr. member

Activity: 300

Merit: 250

4mnth old Miner :-)

Quote from: sussex on August 07, 2014, 10:13:45 PM

OK, what am I doing wrong?

I have a 4 card AMD rig, but can only get one card mining using boolbd to solo.

My json:

"current_extra_message_index": 0,
  "device_index": 3,
  "difficulty": 16777215,
  "donation_decision": true,
  "donation_decision_made": false,
  "donation_descision": true,
  "donation_percent_opencl": "2.0",
  "kernel": "wild_keccak.cl",
  "kernel_type": 0,
  "platform_index": 0,
  "thread_delay": 100,
  "work_size": 262144

my start.bat:

boolbd.exe --start-mining=1K5BH.....rCLe9gxXF --mining-threads=1

Tried all sorts of combos but only ever get one card running.

Also, if I am soloing, why is it telling me I'm getting shares?

4 card amd rig

boolbd.exe --start-mining=1K5BH.....rCLe9gxXF --mining-threads=4

boolbd.exe --start-mining=[address] --mining-threads=[n]

n - number of GPUs to use

Wolf0

member

Activity: 81

Merit: 1002

It was only the wind.

Quote from: hahaerer01 on August 03, 2014, 04:30:52 PM

I have changed the cong,amd.cl is used. I got speed more fasted than old ,from 3.29 to 3.3 .

I set donate to 2%.I hope you release new version to get more fast speed !

Thank you!

The donation goes to mbk, but that's all right, as I probably would have been too lazy to write all the host code to run the OpenCL Tongue

sussex

hero member

Activity: 658

Merit: 500

OK, what am I doing wrong?

I have a 4 card AMD rig, but can only get one card mining using boolbd to solo.

My json:

"current_extra_message_index": 0,
  "device_index": 3,
  "difficulty": 16777215,
  "donation_decision": true,
  "donation_decision_made": false,
  "donation_descision": true,
  "donation_percent_opencl": "2.0",
  "kernel": "wild_keccak.cl",
  "kernel_type": 0,
  "platform_index": 0,
  "thread_delay": 100,
  "work_size": 262144

my start.bat:

boolbd.exe --start-mining=1K5BH.....rCLe9gxXF --mining-threads=1

Tried all sorts of combos but only ever get one card running.

Also, if I am soloing, why is it telling me I'm getting shares?

btc-mike

hero member

Activity: 938

Merit: 1001

Quote from: hahaerer01 on August 07, 2014, 09:27:40 PM

Please help me!

I can't solo .So I mine in pool. But the resault is:

2014-Aug-08 10:16:53.750857 Mining with 1 threads
2014-Aug-08 10:16:53.750857 [OCL 0/0] start with kernel wild_keccak.cl, type 0
2014-Aug-08 10:16:53.750857 [OCL 0/0] get Platform Info
2014-Aug-08 10:16:54.172760 [OCL 0/0] get Device Info
2014-Aug-08 10:16:54.172760 [OCL 0/0] create context
2014-Aug-08 10:16:54.172760 [OCL 0/0] create command queue
2014-Aug-08 10:16:54.219629 [OCL 0/0] init kernels
2014-Aug-08 10:16:54.610267 [OCL 0/0] allocate memory
2014-Aug-08 10:16:54.625892 [OCL 0/0] success
2014-Aug-08 10:16:54.625892 [OCL 0/0] using platform: AMD Accelerated Parallel P
rocessing
2014-Aug-08 10:16:54.625892 [OCL 0/0] using device:Pitcairn
2014-Aug-08 10:16:54.625892 Connecting bbr.cncoin.farm:1111....
2014-Aug-08 10:16:54.969669 Connected bbr.cncoin.farm:1111 OK
2014-Aug-08 10:16:55.329060 Getting scratchpad...
2014-Aug-08 10:19:00.365337 Failed to invoke http request to /json_rpc
2014-Aug-08 10:19:00.380950 Failed to get scratchpad. Disconnecting and retryin
g...

How can I do? thank you.

What OS?
What mining program?
Where did you get the mining program?
What GPU?
What command are you using to run it?

hahaerer01

newbie

Activity: 24

Merit: 0

Please help me!

I can't solo .So I mine in pool. But the resault is:

2014-Aug-08 10:16:53.750857 Mining with 1 threads
2014-Aug-08 10:16:53.750857 [OCL 0/0] start with kernel wild_keccak.cl, type 0
2014-Aug-08 10:16:53.750857 [OCL 0/0] get Platform Info
2014-Aug-08 10:16:54.172760 [OCL 0/0] get Device Info
2014-Aug-08 10:16:54.172760 [OCL 0/0] create context
2014-Aug-08 10:16:54.172760 [OCL 0/0] create command queue
2014-Aug-08 10:16:54.219629 [OCL 0/0] init kernels
2014-Aug-08 10:16:54.610267 [OCL 0/0] allocate memory
2014-Aug-08 10:16:54.625892 [OCL 0/0] success
2014-Aug-08 10:16:54.625892 [OCL 0/0] using platform: AMD Accelerated Parallel P
rocessing
2014-Aug-08 10:16:54.625892 [OCL 0/0] using device:Pitcairn
2014-Aug-08 10:16:54.625892 Connecting bbr.cncoin.farm:1111....
2014-Aug-08 10:16:54.969669 Connected bbr.cncoin.farm:1111 OK
2014-Aug-08 10:16:55.329060 Getting scratchpad...
2014-Aug-08 10:19:00.365337 Failed to invoke http request to /json_rpc
2014-Aug-08 10:19:00.380950 Failed to get scratchpad. Disconnecting and retryin
g...

How can I do? thank you.

maxcan

member

Activity: 117

Merit: 100

Quote from: Wolf0 on August 07, 2014, 05:53:11 PM

Okay, listen up!

Releasing my CUDA miner, with stratum support. I also cleaned up a LOT of the awful cpuminer code - now only supports stratum and Wild Keccak.

is there a way for me to get it working on windows? i'll test it with my 560 ti when i can.

whenever i get it working for a pool i'll mine to your address for a good bit

ltcnim

legendary

Activity: 914

Merit: 1001

yepp, I just used your example and changed the address. I'll install CUDA 6 and try again.

Wolf0

member

Activity: 81

Merit: 1002

It was only the wind.

Also, this does a little bit better on AMD cards. Well, the 7950 I tested on, at least.

https://ottrbutt.com/tmp/wild_keccak_amd.cl

ltcnim

legendary

Activity: 914

Merit: 1001

Quote from: Wolf0 on August 07, 2014, 05:53:11 PM

Okay, listen up!

Releasing my CUDA miner, with stratum support. I also cleaned up a LOT of the awful cpuminer code - now only supports stratum and Wild Keccak.

It's not polished, so READ THE GITHUB README. I also stripped out most of the garbage that came with cpuminer.

Best launch config I've found for 750Ti is 60x256.

Oh, and consider the -t option mandatory for now - set it to the number of GPUs you wish to mine on.

Have fun: https://github.com/wolf9466/bbr-cudaminer

Example launch (using sheepman's pool):

Code:

./cudaminerd -o stratum+tcp://38.93.234.101:40000 -u 1JYa1g52PN5aAq9gjTiDkwWc2Zu56ZkTUV49De8EuSFd9bQcSoRaGvdcJ3dut3XULBbAWqKassGp4Yr6PYodQpbsUKdGUtq -p x -k http://38.93.234.101/scratchpad.bin -t 1 -l60x256

I get ~350khash per 750Ti (5 card rig). I could not compile using maxwell (some error about compute_50), but using compute_35 it worked. does it need a newer CUDA toolkit than 5.5?

cbuchner1

hero member

Activity: 756

Merit: 502

Thanks for releasing this!

My own work on the stratum-enabled cpuminer-multi fork stalled when I got CPU/GPU validation errors and later fell sick for a couple of weeks.

Wolf0

member

Activity: 81

Merit: 1002

It was only the wind.

Quote from: PVmining on August 02, 2014, 01:17:02 PM

Quote from: Wolf0 on August 02, 2014, 10:14:37 AM

Here I go, doing some crazy shit.

See this? https://ottrbutt.com/tmp/bbropenclwolf2.png

It's around 60%+ better on 750Ti.

Now take it before I change my mind: https://ottrbutt.com/tmp/wild_keccak_nv.cl

wow. now I have to wear my fox suite Wink

thanks a lot - I will donate you something for this speed update.
is "@wolf" right for bbr?

Not sure if the alias works yet, I know bbr.farm said they would do it for me, though.

Address is 1JYa1g52PN5aAq9gjTiDkwWc2Zu56ZkTUV49De8EuSFd9bQcSoRaGvdcJ3dut3XULBbAWqKassGp4Yr 6PYodQpbsUKdGUtq

enerbyte

hero member

Activity: 556

Merit: 501

Quote from: Wolf0 on August 07, 2014, 05:53:11 PM

Okay, listen up!

Releasing my CUDA miner, with stratum support. I also cleaned up a LOT of the awful cpuminer code - now only supports stratum and Wild Keccak.

It's not polished, so READ THE GITHUB README. I also stripped out most of the garbage that came with cpuminer.

Best launch config I've found for 750Ti is 60x256.

Oh, and consider the -t option mandatory for now - set it to the number of GPUs you wish to mine on.

Have fun: https://github.com/wolf9466/bbr-cudaminer

Example launch (using sheepman's pool):

Code:

./cudaminerd -o stratum+tcp://38.93.234.101:40000 -u 1JYa1g52PN5aAq9gjTiDkwWc2Zu56ZkTUV49De8EuSFd9bQcSoRaGvdcJ3dut3XULBbAWqKassGp4Yr6PYodQpbsUKdGUtq -p x -k http://38.93.234.101/scratchpad.bin -t 1 -l60x256

Great job!

someone can compile this for windows?

neza777

newbie

Activity: 28

Merit: 0

i want to try this miner on my rig

tuwenxin

newbie

Activity: 4

Merit: 0

How online SOLO ah?

mbk

member

Activity: 106

Merit: 10

You mean ROUND? Or reordering "for" loops? Anyway it's hard to find any logic. The GPU spends 99% in MIX so other changes shouldn't make any difference but they do. In other miner I did some time ago I got boost accidentally switching on debug code... OpenCL compilers are still very far from perfect.

Quote from: Wolf0 on August 06, 2014, 01:57:27 PM

Quote from: mbk on August 06, 2014, 01:49:34 PM

What gives it boost on nvidia? That change in MIX macro? Keccak optimizations couldn't help that much cause all keccak code make just few percent of overall load.

Quote from: Wolf0 on August 06, 2014, 01:38:04 PM

Quote from: mbk on August 06, 2014, 01:36:23 PM

Merged all updates from the main branch. Added binary release too. Update if you use pool miner. Included wolf's nvidia optimized kernel. I cannot measure it's performance on nvidia but it gives about +2% on AMD cards too so it's the default kernel now.
https://github.com/mbkuperman/boolberry-opencl

Yeah, the small percentage bump on AMD is probably from the shortcuts.

Nah, it's what I did to the loop.

Wolf0

member

Activity: 81

Merit: 1002

It was only the wind.

Quote from: itod on August 02, 2014, 11:18:31 AM

Quote from: Wolf0 on August 02, 2014, 10:14:37 AM

Here I go, doing some crazy shit.

See this? https://ottrbutt.com/tmp/bbropenclwolf2.png

It's around 60%+ better on 750Ti.

Now take it before I change my mind: https://ottrbutt.com/tmp/wild_keccak_nv.cl

Can I be so lazy to ask for complete instruction how to compile this?

You don't. You download the file into the same dir as boolbd, rename it to wild_keccak.cl, and run the OpenCL miner like normal.

mbk

member

Activity: 106

Merit: 10

What gives it boost on nvidia? That change in MIX macro? Keccak optimizations couldn't help that much cause all keccak code make just few percent of overall load.

Quote from: Wolf0 on August 06, 2014, 01:38:04 PM

Quote from: mbk on August 06, 2014, 01:36:23 PM

Merged all updates from the main branch. Added binary release too. Update if you use pool miner. Included wolf's nvidia optimized kernel. I cannot measure it's performance on nvidia but it gives about +2% on AMD cards too so it's the default kernel now.
https://github.com/mbkuperman/boolberry-opencl

Yeah, the small percentage bump on AMD is probably from the shortcuts.

mbk

member

Activity: 106

Merit: 10

Merged all updates from the main branch. Added binary release too. Update if you use pool miner. Included wolf's nvidia optimized kernel. I cannot measure it's performance on nvidia but it gives about +2% on AMD cards too so it's the default kernel now.
https://github.com/mbkuperman/boolberry-opencl

Topic: [BBR] Boolberry GPU Miner Discussion - AMD & Nvidia - OpenCL & CUDA - page 17. (Read 82978 times)