113 Commits

Author	SHA1	Message	Date
Con Kolivas	eeac9d81fc	Replace all references to the old n*sleep functions with the equivalent cgsleep_*s replacements.	2013-08-20 19:23:49 +10:00
Con Kolivas	bedd8eeead	Revert "Sleep for half a work period only in bflsc_get_results if no dev is suitable." ... This reverts commit 9743194547. Fine now that sleep works properly with the longer sleep.	2013-08-19 21:52:07 +10:00
Con Kolivas	fb66e8b7b7	Only grab a queued work item if we successfully grab the lock to submit work in bflsc_send_work	2013-08-19 21:13:23 +10:00
Con Kolivas	dddc2d1e60	Revert "Reuse times extracted from the clock in bflsc_get_results." ... This reverts commit 1d8235334d. Monotonic clock does not return the same times as gettimeofday.	2013-08-19 17:59:59 +10:00
Con Kolivas	1d8235334d	Reuse times extracted from the clock in bflsc_get_results.	2013-08-19 17:55:32 +10:00
Con Kolivas	9743194547	Sleep for half a work period only in bflsc_get_results if no dev is suitable.	2013-08-19 16:36:01 +10:00
Con Kolivas	a61e41a070	Check for thr->work_restart in restart_wait.	2013-08-19 15:58:22 +10:00
Con Kolivas	89f571f527	We should be using que_low to decrease scan sleep time in bflsc.	2013-08-19 15:51:14 +10:00
Con Kolivas	09dbd616e2	Prepare sleep time on bflsc if no dev needs work yet to avoid busy waiting.	2013-08-19 13:57:36 +10:00
Con Kolivas	09e9091dd7	Simplify cgsleep code for windows by using a typedef for cgtimer_t that resolves to clock resolution, using that internally.	2013-08-19 13:54:56 +10:00
Con Kolivas	e56c9a5d3b	Use the cgsleep reentrant function to sleep for bflsc between read results to account for time taken to perform reads.	2013-08-18 16:49:15 +10:00
Con Kolivas	3c707202cf	Use 100ms delay between checking for results on all bflsc devices as the buffering of results mean checking more frequently just wastes CPU and causes more lock contention for only marginally better latencies.	2013-08-18 16:46:27 +10:00
Con Kolivas	d71e9f91f4	Add a --bflsc-overheat command which allows you to set the throttling temperature for BFLSC devices or disable it.	2013-08-09 06:43:25 +10:00
Con Kolivas	812852cc36	Move bflsc defines to a header file.	2013-08-09 06:31:57 +10:00
Kano	18e4edc776	bflsc nonce per work item stats	2013-07-08 20:28:03 +10:00
Kano	89d0062ac8	bflsc increase flush count to handle parallel work	2013-07-08 19:41:23 +10:00
Kano	ad1572f77f	size check get_datestamp/get_timestamp and remove unused cgpu->init	2013-07-06 20:17:17 +10:00
Kano	385a70bea7	make all statline overflow safe	2013-07-06 19:15:59 +10:00
Con Kolivas	da680f5162	Don't count invalid nonces as hashrate for bflsc.	2013-07-01 14:44:27 +10:00
Kano	2beb30f13d	bflsc get volts stats needs its own GETVOLTS	2013-06-28 22:55:55 +10:00
Kano	e8bbb49b5b	bflsc x-link header different to documentation	2013-06-27 00:39:25 +10:00
ckolivas	9dc2f3f175	Redo +1 fix in bflsc.	2013-06-25 11:21:28 +10:00
Con Kolivas	a7f9bf3c52	Merge pull request #454 from kanoi/master ... bflsc - handle xlink timeouts by having generic IO functions	2013-06-23 15:37:00 -07:00
Con Kolivas	f5911528bf	Demote bflsc hw error messages to verbose logging only.	2013-06-24 08:17:27 +10:00
Kano	f2bbfee704	bflsc - handle xlink timeouts by having generic IO functions	2013-06-23 18:14:12 +10:00
Con Kolivas	d741410f6d	sc_count shouldn't be +1 in bflsc.	2013-06-23 10:13:18 +10:00
Con Kolivas	0cd2c5c3c8	Use the info timeout for read_nl in getidentify bflsc.	2013-06-23 10:12:44 +10:00
Kano	e993cf3644	bflsc try getinfo twice	2013-06-23 10:01:18 +10:00
Con Kolivas	af0ae8aa5b	Decrease BAL and BAS latency to be just larger than one result read.	2013-06-19 12:09:30 +10:00
Con Kolivas	045e105dcc	BFLSC latest firmware has its own thermal cutoff set to 90, so use the same value in case we have an old firmware that isn't throttling by itself.	2013-06-19 11:42:43 +10:00
Con Kolivas	7b42b83d3f	Drop watermark low limits for bflsc.	2013-06-19 11:38:45 +10:00
Con Kolivas	e172c987f3	Set the fanspeed on bflsc to max if we don't know the temperature.	2013-06-19 11:29:53 +10:00
Con Kolivas	7be081d81b	Use a low watermark for queueing mandatory work on bflsc instead of zero.	2013-06-19 11:24:04 +10:00
Con Kolivas	0b5eb43696	Revert "Only wait in the bflsc scanwork loop if the queues are satisfactorily filled." ... This reverts commit a49969908c. Makes the driver busy wait.	2013-06-19 10:46:28 +10:00
Con Kolivas	a49969908c	Only wait in the bflsc scanwork loop if the queues are satisfactorily filled.	2013-06-19 10:40:03 +10:00
Con Kolivas	fa881badf4	Only mandatorily grab the bflsc mutex on submitting work when the queue is empty.	2013-06-19 10:22:54 +10:00
Con Kolivas	0fbd375ea3	Adjust bflsc v2 watermarks.	2013-06-19 09:43:36 +10:00
Con Kolivas	05adbbf1e9	Only increase sleep time on bflsc if the queue isn't emptying at all over the sleep duration.	2013-06-19 09:42:32 +10:00
Con Kolivas	fe1a3752b7	Fix warning.	2013-06-19 08:57:08 +10:00
Kano	a9fef661b9	bflsc yet more API stats	2013-06-19 02:50:03 +10:00
Kano	711328762f	bflsc add some more API stats	2013-06-19 02:30:06 +10:00
Kano	3ac09fa2d9	bflsc correct firmware matching	2013-06-19 01:02:20 +10:00
Kano	84a5da0377	bflsc correct comment	2013-06-19 00:58:35 +10:00
Kano	8f18d4c8b0	bflsc driver support for v2 firmware	2013-06-19 00:52:13 +10:00
Con Kolivas	6d8d44c233	Fanauto on bflsc is Z9X according to the source code, not 5 as per the draft protocol document.	2013-06-17 09:57:13 +10:00
Kano	b8b137aeef	USB make device_path handled by usbutils	2013-06-10 15:59:02 +10:00
Kano	749fa78d8a	tidy up free in device detect functions	2013-06-10 15:39:44 +10:00
Kano	9287ee90f6	USB control creation and free of cgpu	2013-06-10 15:22:37 +10:00
Con Kolivas	24d26d9949	Throttling the BFLSC at 80 seems to prevent generating garbled responses of higher temps.	2013-06-10 02:54:18 +10:00
Con Kolivas	6fdc068f1b	Revert "Give a generous timeout whenever we expect an OK message for working bflsc communications." ... This reverts commit 3d8cf60b96.	2013-06-09 23:00:36 +10:00