GPU Mining problems with a Mac Pro and Radeon 5870

nushie

Sorry for posting to multiple sites, I found this forum after posting in the MacMiner thread on bitcointalk and figured it might be better off here.

I'm trying to use MacMiner on a Mac Pro with Radeon 5870 to do GPU scrypt
mining. My pool settings all work (tested and working with the CPU
miner). When I try to use GPU, the output lists new blocks, but it
doesn't seem to actually be doing any work. No mh listing and no
accepted/rejected or number of threads running or anything. Also, my RAM
usage spikes out of control and eats up like 8GB ram in 2 minutes and
then continues on into swap, and I have to reboot to fix everything. I'm
just using the recommended settings button (intensity at 4, also tried
at 11, everything else auto).


Log here if it helps (debug output on):
 [2014-03-07 16:26:13] Timers: Using gettimeofday
 [2014-03-07 16:26:13] Global quota greatest common denominator set to 1
 [2014-03-07 16:26:13] setrlimit: Changed soft fd limit from 2560 to 1024 (FD_SETSIZE=1024; hard limit=unlimited)
 [2014-03-07 16:26:13] Started bfgminer 3.10.0
 [2014-03-07 16:26:13] hidapi_try_lib: Successfully loaded libhidapi.dylib
 [2014-03-07 16:26:13] CL Platform 0 vendor: Apple
 [2014-03-07 16:26:13] CL Platform 0 name: Apple
 [2014-03-07 16:26:13] CL Platform 0 version: OpenCL 1.1 (Apr  9 2012 19:41:45)
 [2014-03-07 16:26:13] Platform 0 devices: 1
 [2014-03-07 16:26:13]     0    ATI Radeon HD 5870
 [2014-03-07 16:26:13] Unable to load ati adl library
 [2014-03-07 16:26:13] OCL 0 : Set temperature config: target=89 cutoff=95
 [2014-03-07 16:26:13] Probing for an alive pool
 [2014-03-07 16:26:13] Testing pool stratum+tcp://us.clevermining.com:3333
 [2014-03-07 16:26:13] HTTP request failed: Protocol stratum+tcp not supported or disabled in libcurl
 [2014-03-07 16:26:13] Failed to connect in json_rpc_call
 [2014-03-07 16:26:13] initiate_stratum with sockbuf=0x0
 [2014-03-07 16:26:16] Failed to get sessionid in initiate_stratum
 [2014-03-07 16:26:16] Pool 0 stratum bdifficulty set to 512.000000
 [2014-03-07 16:26:16] Received stratum notify from pool 0 with job_id=b81e
 [2014-03-07 16:26:16] Stratum authorisation success for pool 0
 [2014-03-07 16:26:16] Pool 0 stratum+tcp://us.clevermining.com:3333 alive
 [2014-03-07 16:26:16] Received no-transaction-list response for pool 0 job b81e
 [2014-03-07 16:26:16] Generated stratum header 0000000428f994b5bca316de2f7f6a3a8d806be122b7b5f4ff42164c02c4a00c00000000d8326d831c4d64e79055c85161458f426b285dff9e830e245082f209c62f6a0d531a63371c0464b500000000
 [2014-03-07 16:26:16] Work job_id b81e nonce2 00000000
 [2014-03-07 16:26:16] Generated target 0000007fff800000000000000000000000000000000000000000000000000000
 [2014-03-07 16:26:16] Network difficulty changed to 58 (417.1Mh/s)
 [2014-03-07 16:26:16] New block: ...bca316de28f994b5 diff 58 (417.1Mh/s)
 [2014-03-07 16:26:16] Stratum from pool 0 detected new block
 [2014-03-07 16:26:16] Pool 0 is hiding block contents from us
 [2014-03-07 16:26:17] Init GPU thread 0 GPU 0 virtual GPU 0
 [2014-03-07 16:26:17] CL Platform vendor: Apple
 [2014-03-07 16:26:17] CL Platform name: Apple
 [2014-03-07 16:26:17] CL Platform version: OpenCL 1.1 (Apr  9 2012 19:41:45)
 [2014-03-07 16:26:17] List of devices:
 [2014-03-07 16:26:17]     0    ATI Radeon HD 5870
 [2014-03-07 16:26:17] Selected 0: ATI Radeon HD 5870
 [2014-03-07 16:26:17] Preferred vector width reported 4
 [2014-03-07 16:26:17] Max work group size reported 1024
 [2014-03-07 16:26:17] Max mem alloc size is 268435456
 [2014-03-07 16:26:17] Selecting scrypt kernel
 [2014-03-07 16:26:17] GPU 0: selecting lookup gap of 2
 [2014-03-07 16:26:17] GPU 0: selecting thread concurrency of 1984
 [2014-03-07 16:26:17] OCL 0: Configured OpenCL kernel name: scrypt130511ATIRadeonHD5870glg2tc1984w256l8pOpenCL1_1Apr9_2012_19_41_45
 [2014-03-07 16:26:17] Setting worksize to 256
 [2014-03-07 16:26:17] cl_amd_media_ops not found, will not set BITALIGN
 [2014-03-07 16:26:17] BFI_INT patch requiring device not found, will not BFI_INT patch
 [2014-03-07 16:26:17] CompilerOptions: -D LOOKUP_GAP=2 -D CONCURRENT_THREADS=1984 -D WORKSIZE=256 -D GOFFSET
 [2014-03-07 16:26:47] Received stratum notify from pool 0 with job_id=7c67
 [2014-03-07 16:26:47] Generated stratum header 00000002cf2d12b957ff50da2ec1a33d2bd3010aff1b1dc8ff5ff84a1f58b5bfa0f157b3017e6d7c5acee40c3ab50c258a23484d1ee070bb720e87aa3c493dd4ccf5b95c531a63d31c2059f900000000
 [2014-03-07 16:26:47] Work job_id 7c67 nonce2 00000000
 [2014-03-07 16:26:47] Generated target 0000007fff800000000000000000000000000000000000000000000000000000
 [2014-03-07 16:26:47] Learned that block id cf2d12b9 is height 155308
 [2014-03-07 16:26:47] Network difficulty changed to 7 (56.64Mh/s)
 [2014-03-07 16:26:47] New block: ...cf2d12b9 #155308 diff 7 (56.64Mh/s)
 [2014-03-07 16:26:47] Stratum from pool 0 detected new block


It continues printing new block info until I stop it, but nothing seems to be happening, and I very quickly start swapping RAM, as mentioned above.

Any ideas?

FabulousPanda

Can you try mining a different coin type please? They have different backends so we could identify which is causing the problem this way. Are you mining BTC there?

nushie

It's all scrypt mining (all litecoin alts, but not sure what coin since clevermining switches between coin types).

Here's more debug log output, this time from scrypt on the fabulous panda pool:

 [2014-03-07 22:03:36] Timers: Using gettimeofday
 [2014-03-07 22:03:36] Global quota greatest common denominator set to 1
 [2014-03-07 22:03:36] setrlimit: Changed soft fd limit from 2560 to 1024 (FD_SETSIZE=1024; hard limit=unlimited)
 [2014-03-07 22:03:36] Started bfgminer 3.10.0
 [2014-03-07 22:03:36] hidapi_try_lib: Successfully loaded libhidapi.dylib
 [2014-03-07 22:03:36] CL Platform 0 vendor: Apple
 [2014-03-07 22:03:36] CL Platform 0 name: Apple
 [2014-03-07 22:03:36] CL Platform 0 version: OpenCL 1.1 (Apr  9 2012 19:41:45)
 [2014-03-07 22:03:36] Platform 0 devices: 1
 [2014-03-07 22:03:36]     0    ATI Radeon HD 5870
 [2014-03-07 22:03:36] Unable to load ati adl library
 [2014-03-07 22:03:36] OCL 0 : Set temperature config: target=89 cutoff=95
 [2014-03-07 22:03:36] Probing for an alive pool
 [2014-03-07 22:03:36] Testing pool http://pool.fabulouspanda.co.uk:9327
 [2014-03-07 22:03:37] JSON-RPC call failed: {
   "data": null,
   "code": -32601,
   "message": "Method not found"
}
 [2014-03-07 22:03:37] Failed to connect in json_rpc_call
 [2014-03-07 22:03:37] X-Roll-Ntime expiry set to 100
 [2014-03-07 22:03:37] Testing pool 0 stratum stratum+tcp://pool.fabulouspanda.co.uk:9327
 [2014-03-07 22:03:37] initiate_stratum with sockbuf=0x0
 [2014-03-07 22:03:37] Failed to get sessionid in initiate_stratum
 [2014-03-07 22:03:37] Switching pool 0 http://pool.fabulouspanda.co.uk:9327 to stratum+tcp://pool.fabulouspanda.co.uk:9327
 [2014-03-07 22:03:37] initiate_stratum with sockbuf=0x10082fe00
 [2014-03-07 22:03:38] Failed to get sessionid in initiate_stratum
 [2014-03-07 22:03:38] Pool 0 stratum bdifficulty set to 139.110289
 [2014-03-07 22:03:38] Received stratum notify from pool 0 with job_id=291013492878958558041839507627642448073
 [2014-03-07 22:03:38] Stratum authorisation success for pool 0
 [2014-03-07 22:03:38] Pool 0 http://pool.fabulouspanda.co.uk:9327 alive
 [2014-03-07 22:03:38] Pool 0 stratum bdifficulty set to 139.110289
 [2014-03-07 22:03:38] Init GPU thread 0 GPU 0 virtual GPU 0
 [2014-03-07 22:03:38] CL Platform vendor: Apple
 [2014-03-07 22:03:38] CL Platform name: Apple
 [2014-03-07 22:03:38] CL Platform version: OpenCL 1.1 (Apr  9 2012 19:41:45)
 [2014-03-07 22:03:38] List of devices:
 [2014-03-07 22:03:38]     0    ATI Radeon HD 5870
 [2014-03-07 22:03:38] Selected 0: ATI Radeon HD 5870
 [2014-03-07 22:03:38] Generated stratum header 000000022504118fada5257dde326989144d02331dc80a100288b2baac9cc2d7ced07d962051152e1271e518a4c706f4bfdde447b8913d211a4e74fc47cb1659bbd3623f531ab2b71b10b60e00000000
 [2014-03-07 22:03:38] Work job_id 291013492878958558041839507627642448073 nonce2 00000000
 [2014-03-07 22:03:38] Generated target 000001d719dcf0c4ea7000000000000000000000000000000000000000000000
 [2014-03-07 22:03:38] Learned that block id 2504118f is height 527189
 [2014-03-07 22:03:38] Network difficulty changed to 3.92k (28.07Gh/s)
 [2014-03-07 22:03:38] New block: ...2504118f #527189 diff 3.92k (28.07Gh/s)
 [2014-03-07 22:03:38] Stratum from pool 0 detected new block
 [2014-03-07 22:03:38] Pool 0 is hiding block contents from us
 [2014-03-07 22:03:38] Preferred vector width reported 4
 [2014-03-07 22:03:38] Max work group size reported 1024
 [2014-03-07 22:03:38] Max mem alloc size is 268435456
 [2014-03-07 22:03:38] Selecting scrypt kernel
 [2014-03-07 22:03:38] GPU 0: selecting lookup gap of 2
 [2014-03-07 22:03:38] GPU 0: selecting thread concurrency of 1984
 [2014-03-07 22:03:38] OCL 0: Configured OpenCL kernel name: scrypt130511ATIRadeonHD5870glg2tc1984w256l8pOpenCL1_1Apr9_2012_19_41_45
 [2014-03-07 22:03:38] Setting worksize to 256
 [2014-03-07 22:03:38] cl_amd_media_ops not found, will not set BITALIGN
 [2014-03-07 22:03:38] BFI_INT patch requiring device not found, will not BFI_INT patch
 [2014-03-07 22:03:38] CompilerOptions: -D LOOKUP_GAP=2 -D CONCURRENT_THREADS=1984 -D WORKSIZE=256 -D GOFFSET
 [2014-03-07 22:03:38] Received stratum notify from pool 0 with job_id=90232324576174523963980650577103635060
 [2014-03-07 22:03:38] Generated stratum header 000000022504118fada5257dde326989144d02331dc80a100288b2baac9cc2d7ced07d96eefe373e899fcb33c69175139995596f013025841e9192ad9c5a62ff62f21298531ab2b71b10b60e00000000
 [2014-03-07 22:03:38] Work job_id 90232324576174523963980650577103635060 nonce2 00000000
 [2014-03-07 22:03:38] Generated target 000001d719dcf0c4ea7000000000000000000000000000000000000000000000
 [2014-03-07 22:03:38] Stratum from pool 0 requested work update
 [2014-03-07 22:03:38] Received no-transaction-list response for pool 0 job 291013492878958558041839507627642448073
 [2014-03-07 22:03:38] Received no-transaction-list response for pool 0 job 291013492878958558041839507627642448073
 [2014-03-07 22:03:39] Received no-transaction-list response for pool 0 job 90232324576174523963980650577103635060
 [2014-03-07 22:03:45] Pool 0 stratum bdifficulty set to 151.433129

nushie

Just realized you wanted me to switch algorithms... Switched to the panda pool BTC SHA-256 and it's working fine on GPU. Getting around 300Mh with standard settings. So it looks like scrypt is the problem.

FabulousPanda

This card seems to be a bit picky sometimes, if it's that I wonder whether specifying a few options in the miner settings panel would help - try setting work size to 256 and use the manual flag 

-g 1

then start over

It's weird that RAM issue though, if you have Activity Monitor open is the memory allocated to the  MacMiner process or the bfgminer process?

nushie

Same issues with work size256 and -g 1. RAM and CPU usage are being allocated to CVMCompiler actually, which looks like it's part of the OSX OpenGL framework?

FabulousPanda

looks like a system bug, any updates to run, what OS version is it?

might be worth trying this version which may be a bit buggy though:

http://downloads.fabulouspanda.co.uk/MacMiner-gridseed8.zip

nushie

OS update fixed it. Weird.... thanks for all the help! Went from 10.7.4 to 10.7.5 and suddenly is working. So far all shares have been rejected. but now I can at least play with the settings and see what works best instead of having to reboot every 5 minutes because it's swapping.