Title: trouble scrypt mining on linux - 0 accepted/0 rejected, no shares at all
Post by: mat5x on May 06, 2013, 07:46:54 PM
Having an impossible time scrypt mining on linux. I have compiled BOTH bfg-3.0.2 and cgminer 3.1.0 on my box here. Both of them act totally wierd on multiple different AMD/ATI cards. I have zero problems SHA256 mining. If i mine CNC solo on 7970, i get 56MH/s reported and never find any blocks. If I mine CNC pooled, iget 26MH/s reported and no shares. At least the block difficulty is reported ok. On cgminer the block difficulty for CNC and FTC is like 60,000x higher than the actual network block diffs (which are in the 100s range, they report 12M for ftc for eg). And I either get 1Kh/s on CNC pool, or multiple MH/s reported. I switched to solo FTC and get a reasonable hashrate with a 5830 (600Kh/s?) but never anything found - for a faster test check, i moved to featherpool and still getting 570-600KH/s for a 5830 (which seems a bit fast, tho), but the GPU stats in BFG show 20-30% usage and temp 40C (much lower than the 80C i usually get with my 40% fan tolerable noise level setting). Still 0 accepted, and 0 rejected at all times. Ie no work traffic at all. Pretty sure none of it is working right. Here's a bit of the debug output from bfg on featherpool [2013-05-06 15:43:15] [thread 0: 604672 hashes, 601.8 khash/sec] [2013-05-06 15:43:15] 1s:592.9 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:16] OCL 0: Popping work from get queue to get work [2013-05-06 15:43:16] OCL 0: Got work from get queue to get work for thread 0 [2013-05-06 15:43:16] Successfully rolled time header in work [2013-05-06 15:43:16] Successfully rolled time header in work [2013-05-06 15:43:16] Pushing cloned available work to stage thread [2013-05-06 15:43:16] Pushing work from pool 1 to hash queue [2013-05-06 15:43:16] Cloned getwork work [2013-05-06 15:43:16] [thread 0: 570880 hashes, 567.9 khash/sec] [2013-05-06 15:43:16] 1s:584.0 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:17] Temperature below target, increasing clock speed [2013-05-06 15:43:17] 46.0 C F: 41%(1601RPM) E: 600MHz M: 222MHz V: 1.063V A: 29% P: 0% [2013-05-06 15:43:17] [thread 0: 558592 hashes, 555.9 khash/sec] [2013-05-06 15:43:17] 1s:573.9 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:18] [thread 0: 607104 hashes, 605.3 khash/sec] [2013-05-06 15:43:18] 1s:585.5 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:19] [thread 0: 606720 hashes, 604.2 khash/sec] [2013-05-06 15:43:19] 1s:592.6 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:20] Temperature below target, increasing clock speed [2013-05-06 15:43:20] 45.0 C F: 41%(1606RPM) E: 600MHz M: 222MHz V: 1.063V A: 29% P: 0% [2013-05-06 15:43:20] [thread 0: 569088 hashes, 567.0 khash/sec] [2013-05-06 15:43:20] 1s:583.2 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:21] [thread 0: 564736 hashes, 561.9 khash/sec] [2013-05-06 15:43:21] 1s:575.7 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:22] OCL 0: Popping work from get queue to get work [2013-05-06 15:43:22] OCL 0: Got work from get queue to get work for thread 0 [2013-05-06 15:43:22] DBG: sending http://featherpool.com:9999 get RPC call: {"method": "getwork", "params": [], "id":0}
[2013-05-06 15:43:22] JSON protocol request: {"method": "getwork", "params": [], "id":0}
[2013-05-06 15:43:22] Pool 1: Found bundle for host featherpool.com: 0xe687c0 [2013-05-06 15:43:22] Pool 1: Connection 36 seems to be dead! [2013-05-06 15:43:22] Pool 1: Closing connection 36 [2013-05-06 15:43:22] Pool 1: About to connect() to featherpool.com port 9999 (#37) [2013-05-06 15:43:22] Pool 1: Trying 72.20.53.5... [2013-05-06 15:43:22] Pool 1: TCP_NODELAY set [2013-05-06 15:43:22] Pool 1: Adding handle: conn: 0x7ffbf40353a0 [2013-05-06 15:43:22] Pool 1: Adding handle: send: 0 [2013-05-06 15:43:22] Pool 1: Adding handle: recv: 0 [2013-05-06 15:43:22] Pool 1: Curl_addHandleToPipeline: length: 1 [2013-05-06 15:43:22] Pool 1: - Conn 37 (0x7ffbf40353a0) send_pipe: 1, recv_pipe: 0 [2013-05-06 15:43:22] Pool 1: Connected to featherpool.com (72.20.53.5) port 9999 (#37) [2013-05-06 15:43:22] Pool 1: Server auth using Basic with user 'clarpfog.c5830-2' [2013-05-06 15:43:22] HTTP hdr(Content-Type): application/json [2013-05-06 15:43:22] HTTP hdr(X-Long-Polling): /LP [2013-05-06 15:43:22] HTTP hdr(X-Roll-NTime): expire=120 [2013-05-06 15:43:22] X-Roll-Ntime expiry set to 120 [2013-05-06 15:43:22] HTTP hdr(Date): Mon, 06 May 2013 19:42:42 GMT [2013-05-06 15:43:22] HTTP hdr(Content-Length): 621 [2013-05-06 15:43:22] Pool 1: Connection #37 to host featherpool.com left intact [2013-05-06 15:43:22] JSON protocol response: { "id": 0, "result": { "target": "ffffffffffffffffffffffffffffffffffffffffffffffffffffffffff070000", "midstate": "9d6716bef1503d539a8922ddc69d6119c13dc7cffa6d41c27d329b20160ba85a", "hash1": "00000000000000000000000000000000000000000000000000000000000000000000008000000000000000000000000000000000000000000000000000010000", "data": "0000000158fc81cc4ec71fb4428f3d6b0e4357714cb142e93fe48e10353058204e5a775098394f647e7b73fb19b71f783ecf36970205d5d6c22aacfa28d8bded54864d24518807bb1c015c3d00000000000000800000000 000000000000000000000000000000000000000000000000000000000000000000000000080020000", "algorithm": "scrypt:1024,1,1" }, "error": null } [2013-05-06 15:43:22] Generated getwork work [2013-05-06 15:43:22] Pushing work from pool 1 to hash queue [2013-05-06 15:43:23] [thread 0: 598528 hashes, 596.0 khash/sec] [2013-05-06 15:43:23] 1s:583.3 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:23] Temperature below target, increasing clock speed [2013-05-06 15:43:23] 45.0 C F: 41%(1604RPM) E: 600MHz M: 222MHz V: 1.063V A: 31% P: 0% [2013-05-06 15:43:24] [thread 0: 574080 hashes, 572.0 khash/sec] [2013-05-06 15:43:24] 1s:579.2 avg:577.3 u: 0.0 kh/s | A:0 R:0 S:0 HW:0 U:0.0/m BS:0 [2013-05-06 15:43:25] [thread 0: 583808 hashes, 580.5 khash/sec]
anyone (luke-jr? lol.) got a hint there? compiled totally wrong? this is debian sid with all the right libraries (jansson, etc etc). Are there any debian packages for cg or bfg out there? Ive actually tried someone else's binary build of bfg-3.0.0 (debian wheezy compile), same issues. Is it my AMD libraries or something? Why would SHA256 work then?
Title: Re: trouble scrypt mining on linux - 0 accepted/0 rejected, no shares at all
Post by: larkstongues on May 06, 2013, 08:40:27 PM
If i mine CNC solo on 7970, i get 56MH/s reported and never find any blocks. If I mine CNC pooled, iget 26MH/s reported and no shares. At least the block difficulty is reported ok.
I'm pretty sure your Hashrate is actually in the KH/s range not MH/s. Scrypt hashing is about 1000 times slower than sha256 hashing. On cgminer the block difficulty for CNC and FTC is like 60,000x higher than the actual network block diffs
Yup, that's because it was implemented for bitcoin mining. You must devide by 2^16 (=65535) to get the right Difficulty. Still 0 accepted, and 0 rejected at all times. Ie no work traffic at all.
This is because it takes much luck and time to solo mine. You only get a Accept if you succeed in solving a block. You can use this calculator to check how long this will take on average: https://www.litecoinpool.org/calc?hashrate=350&difficulty=20 Just set the difficulty to whatever the current difficulty of the scrypt-coin you want to mine is. Also I suggest using cgminer for scrypt coins. I guess I can't be of much more help because I don't use bfg or debian linux.
Title: Re: trouble scrypt mining on linux - 0 accepted/0 rejected, no shares at all
Post by: mat5x on May 06, 2013, 11:07:32 PM
I'm pretty sure your Hashrate is actually in the KH/s range not MH/s. Scrypt hashing is about 1000 times slower than sha256 hashing.
yes quite, all the numbers I pasted are exactly as I saw them with the right metric prefixes. So yes, 56MH/s is WAY wrong. And so is 1KH/s. 200-900KH/s for my 5830s 50s and 7970s is what i expect. On cgminer the block difficulty for CNC and FTC is like 60,000x higher than the actual network block diffs
Yup, that's because it was implemented for bitcoin mining. You must devide by 2^16 (=65535) to get the right Difficulty. ah ok, yeah, that number looked a little suspicious - i actually did the math and saw like 66020 or something and i was like "close to 2^16 hmm". Still 0 accepted, and 0 rejected at all times. Ie no work traffic at all.
This is because it takes much luck and time to solo mine. You only get a Accept if you succeed in solving a block. You can use this calculator to check how long this will take on average: https://www.litecoinpool.org/calc?hashrate=350&difficulty=20 Just set the difficulty to whatever the current difficulty of the scrypt-coin you want to mine is. Also I suggest using cgminer for scrypt coins. I guess I can't be of much more help because I don't use bfg or debian linux. yeah but i mentioned i also join a CNC and FTC pool for 5830s and for my 7970s and never even got a rejected share, never mind an accepted one. im not solomining those, im doing shares but seeing no work being done at all.
|