OLD: BFGMiner 3.10.0: modular ASIC+FPGA, GBT+Strtm, RPC, Mac/Lnx/W64, AntU1, DRB

[Luke-Jr]

Update: HUPed after 3 hours operation.

What do you mean by HUPed?
The HUP signal is sent when you close the terminal...

[1714794774]

1714794774

[Xer0]

runs very well on a TP-Link MR3020 (WR703N)

Thanks!

(now i just need come watchdog stuff for failsafe things)

[Zeronic]

Results of working on getting BFGMiner to work with X6500.

Using libusb-win32_1.2.6.0
http://sourceforge.net/projects/libusb-win32/?source=dlp

http://pastebin.com/bn8tecQT

Luke can we got a known working version or a better suggestion please?

All current versions are known working - I wouldn't release them if they didn't work.
There is no sign of a X6500 on your USB bus in the pastebin, with or without the correct driver.
X6500 is 0403:6001

http://tinypic.com/r/26126g5/5

I'm kind of lost at this point.

[jml]

Update: HUPed after 3 hours operation.

What do you mean by HUPed?
The HUP signal is sent when you close the terminal...

HUP happens because I use SSH to open a terminal. When I try to SSH back in, I can't. This is why I need to power it down and then back up again.

[RoboCoder]

Results of working on getting BFGMiner to work with X6500.

Using libusb-win32_1.2.6.0
http://sourceforge.net/projects/libusb-win32/?source=dlp

http://pastebin.com/bn8tecQT

Luke can we got a known working version or a better suggestion please?

All current versions are known working - I wouldn't release them if they didn't work.
There is no sign of a X6500 on your USB bus in the pastebin, with or without the correct driver.
X6500 is 0403:6001

http://tinypic.com/r/26126g5/5

I'm kind of lost at this point.

I have exactly this same problem.  If i run 3.11 they are reconized and work properly, but if i run 3.13 i get the same error. (so i am running

[Luke-Jr]

Results of working on getting BFGMiner to work with X6500.

Using libusb-win32_1.2.6.0
http://sourceforge.net/projects/libusb-win32/?source=dlp

http://pastebin.com/bn8tecQT

Luke can we got a known working version or a better suggestion please?

All current versions are known working - I wouldn't release them if they didn't work.
There is no sign of a X6500 on your USB bus in the pastebin, with or without the correct driver.
X6500 is 0403:6001

http://tinypic.com/r/26126g5/5

I'm kind of lost at this point.

Be sure you're using Zadig the way described in README.FPGA

Update: HUPed after 3 hours operation.

What do you mean by HUPed?
The HUP signal is sent when you close the terminal...

HUP happens because I use SSH to open a terminal. When I try to SSH back in, I can't. This is why I need to power it down and then back up again.

I recommend running BFGMiner inside GNU Screen.

I have exactly this same problem.  If i run 3.11 they are reconized and work properly, but if i run 3.13 i get the same error. (so i am running

OH! I do have a fix for a X6500 regression slated for 3.1.4. Just didn't realize it manifested this way >_<

Stay tuned, probably just a few more days...

If you'd like to help test now, the latest git is always available from WeBisect.

[RoboCoder]

I have exactly this same problem.  If i run 3.11 they are reconized and work properly, but if i run 3.13 i get the same error. (so i am running

OH! I do have a fix for a X6500 regression slated for 3.1.4. Just didn't realize it manifested this way >_<

Stay tuned, probably just a few more days...

If you'd like to help test now, the latest git is always available from WeBisect.

I downloaded the win64 to test.  It worked with both the x6500's and the ModMiners just fine.  Yay!  Thanks.  Will let you know if i see any problems.

[jml]

Update: HUPed after 3 hours operation.

What do you mean by HUPed?
The HUP signal is sent when you close the terminal...

HUP happens because I use SSH to open a terminal. When I try to SSH back in, I can't. This is why I need to power it down and then back up again.

I recommend running BFGMiner inside GNU Screen.



I have been using GNU screen all along.

[Zeronic]

I followed the README.FPGA but using the WinUSB (v6.1.7600.16385) no version of BFGMiner will detect the X6500.

Update: After a Software Update and a reboot, I now get "FTDI reports 1 devices". So from this point I need to add something like the -S X6500:auto to the command string?

[Aahzman]

Howdy. program works great except for random crashing. I have two BFL 5Gh/s Jalapenos, mining on Slush's pool, Windows 7, 64bit.  It'll hum along just fine for a while, and then randomly just stop working. After reading a few pages of this thread, i've restarted it with -D -T args to see if i can get any info about the crashing.

Also, by default it starts up detects/mines with both BFL devices and my GPU. i've looked at the --device|-d command not sure how to get it to do something useful like only enable the BFLs.  The devices are listed as

OCL 0
BFL 0
BFL 1

so, would the numbering be as simple as 0, 1 and 2? so if i just want the BFL's used, i would add -d 1-2 to my command line?

[Zeronic]

Howdy. program works great except for random crashing. I have two BFL 5Gh/s Jalapenos, mining on Slush's pool, Windows 7, 64bit.  It'll hum along just fine for a while, and then randomly just stop working. After reading a few pages of this thread, i've restarted it with -D -T args to see if i can get any info about the crashing.

Also, by default it starts up detects/mines with both BFL devices and my GPU. i've looked at the --device|-d command not sure how to get it to do something useful like only enable the BFLs.  The devices are listed as

OCL 0
BFL 0
BFL 1

so, would the numbering be as simple as 0, 1 and 2? so if i just want the BFL's used, i would add -d 1-2 to my command line?

I think if you add -G to the command string that will disable the GPU's.

[Aahzman]

Howdy. program works great except for random crashing. I have two BFL 5Gh/s Jalapenos, mining on Slush's pool, Windows 7, 64bit.  It'll hum along just fine for a while, and then randomly just stop working. After reading a few pages of this thread, i've restarted it with -D -T args to see if i can get any info about the crashing.

Also, by default it starts up detects/mines with both BFL devices and my GPU. i've looked at the --device|-d command not sure how to get it to do something useful like only enable the BFLs.  The devices are listed as

OCL 0
BFL 0
BFL 1

so, would the numbering be as simple as 0, 1 and 2? so if i just want the BFL's used, i would add -d 1-2 to my command line?

I think if you add -G to the command string that will disable the GPU's.

Thanks. Will try that on next crash/restart. :-)

[Taugeran]

Howdy. program works great except for random crashing. I have two BFL 5Gh/s Jalapenos, mining on Slush's pool, Windows 7, 64bit.  It'll hum along just fine for a while, and then randomly just stop working. After reading a few pages of this thread, i've restarted it with -D -T args to see if i can get any info about the crashing.

Also, by default it starts up detects/mines with both BFL devices and my GPU. i've looked at the --device|-d command not sure how to get it to do something useful like only enable the BFLs.  The devices are listed as

OCL 0
BFL 0
BFL 1

so, would the numbering be as simple as 0, 1 and 2? so if i just want the BFL's used, i would add -d 1-2 to my command line?

I think if you add -G to the command string that will disable the GPU's.

Thanks. Will try that on next crash/restart. :-)

it depends on which version of bfgminer is in use.
3.1.2>= -G should work
3.1.3 forward will be -S OpenCL:noauto

[Aahzman]

ah. ok. i'm on 3.1.3, so i'll do that -S thing.

Thanks

[Zeronic]

I followed the README.FPGA but using the WinUSB (v6.1.7600.16385) no version of BFGMiner will detect the X6500.

Update: After a Software Update and a reboot, I now get "FTDI reports 1 devices". So from this point I need to add something like the -S X6500:auto to the command string?

Try the -S X6500:auto or just -S all and it still won't mine on the X6500.

[Dork.]

Hello!

Just letting you all know that I had been trying to get bfgminer 3.1.3  to work on an old Mac, to use some block erupters.
It has 10.5, and I have been having problems with homebrew on 10.5, so I downloaded and compiled all the libraries individually.
I managed to successfully compile it after a few strategic edits, but every time I tried to use it with my block erupters, it would fail on the scan, complaining about an "Unsupported baud rate".

I knew that my erupters and hub work fine because the bitminter client works fine for me.

On a whim, I decided to try compiling earlier versions, and it turns out Version 3.1.2 works fine for me. So I am good now, but I figured I should let you all know....

[Luke-Jr]

NEW VERSION 3.1.4, AUGUST 2 2013

Human readable changelog:

• Enable triggering "identify" (flash LED) function from Manage device TUI for devices that support it.
• modminer & x6500: Enable changing clock frequency from Manage devices TUI interface.
• Mac/BSD: Fix new IO speed support detection, so Icarus-compatible devices work by default.
• Device ailments (SICK, ERR, etc) are now displayed in red/pink.
• erupter: Add support for "identify" (flash LED) function. After some delay (up to 13 seconds), the LED will light for 3 seconds (hashing stops during this time). Thanks to Rotorgeek and Canary for helping test!
• bitforce: Display voltages in Manage device TUI.
• Reorganised TUI display, adding network transfer rate, and totals summary line including count of total devices/processors hashing as well as hottest temperature across all sensors.
• Improved timer functionality under the hood, taking advantage of QueryPerformanceCounter on Windows and clock_gettime on *nix. This should fix issues (eg, erupters reporting insane hashrates) related to system clock changes (such as NTP drift adjustment).
• Removed blank/wasted lines above/below log window in TUI mode.
• bitforce: Clear queues at startup to avoid spurious warnings.
• Stratum: Add support for rolling ntime header.

Full changelog

• Windows: Rebuild pdcurses with UTF-8 and wide character support
• Bugfix: Avoid using wide curses symbols/macros when USE_UNICODE is not defined
• Unicode: Use line drawing in TUI Help
• Use bfg_waddstr even with Unicode disabled, since it's needed for red highlight
• Colour bad conditions in red
• Unicode: Cross-tee intersecting lines
• Unicode: Use WACS_VLINE for vertical lines
• Unicode: If degrees symbol is available, add it to temperatures
• Unicode: bfg_waddstr wrapper to handle non-ASCII characters, currently used only by logging and statlines
• Unicode: Use WACS_HLINE for horizontal lines
• Add framework for using Unicode in TUI (can be disabled with --no-unicode)
• Avoid using potentially locale-dependent ctype functions in locale-independent contexts
• Refactor temperature in TUI statlines to share code nicer
• Bugfix: avalon: Fix applog formatting
• Bugfix: Align totals columns in per-processor view
• Bugfix: Fix curses-less build
• configure: Workaround buggy autoconf versions
• Bugfix: erupter: Include headers in order necessary for Windows
• Bugfix: Reimplement get_intrange using strtol instead of sscanf (which is broken on Windows)
• Bugfix: get_intrange: Check for extra garbage at the end, only after we know we have an end-position
• Bugfix: Fix Enter key in TUI on Windows
• erupter: Split identify-handling logic into handle_identify function
• Bugfix: erupter: Ensure identify is handled during no-once or firstrun
• erupter: After identify, check if a work restart is needed immediately
• erupter: Implement identify function by pausing hashing for 3 seconds
• Bugfix: icarus: Remember firstrun state in case it gets changed for the next run
• icarus: Move actual dynclock updates to icarus_job_start
• icarus: Split out icarus_job_prepare, and rename icarus_job_start
• Bugfix: ZeroStats: Reset column widths to 1
• miner.php: Include max temperature in device totals line
• Bugfix: Stratum Fix debug logging of initial mining.subscribe command
• Bugfix: Call pool_set_opaque from work_decode, so block content hiding/providing messages work for getwork/GBT
• Split block contents hiding/providing notices out from stratum code
• Add test suite for get_intrange
• Bugfix: Check for error conditions in get_intrange to not have weird --device behaviour when bad values are provided
• Bugfix: erupter: Take advantage of detectone_meta_info to handle Emerald autodetection
• TUI Help describing the various status fields (contributed by midnightmagic)
• Bugfix: ManageTUI: Allow 'I' key to be used by devices not supporting identify
• Bugfix: Prefer Sapphire over Emerald for -S erupter:*
• Bugfix: Clear total_bad_nonces when zeroing statistics
• Bugfix: modminer: Since we are not searching iManuf string for needles, only look for "ModMiner"
• Bugfix: sysfs autodetect: Recurse into tty/ subdirectory (necessary for CDC/ACM ttys)
• sysfs autodetect: Split tty* directory search into new _sysfs_find_tty function
• modminer: Reduce default clock to 190 MHz
• README: Update driver info to include Erupter driver
• README: FAQ about scrypt and difficulty
• Include count of working devices/processors in totals statline
• Format totals statline the same way as individual device/processor statlines
• Rearrange TUI a bit, including menu at the top (+1 log line) and hashrate total closer to device summaries
• Bugfix: setup_stratum_curl: Need to release stratum lock on connection failure too
• Bugfix: Avoid unnecessary locks inside curses_print_status, which is called with the console lock held
• Bugfix: setup_stratum_curl: Hold stratum lock until connection completes, to avoid potential races
• Bugfix: stratum_works: If stratum is already active, it works (avoid trying to initialise it again)
• Replace hashrate_to_bufstr/ti_hashrate_bufstr with format_unit/multi_format_unit_array
• New multi_format_unit_array to fill multiple buffers instead of building a delimited string
• multi_format_unit: Skip recounting length of fixed-length strings
• Shrink status line to fit in 80 columns
• Add network bandwidth rate to TUI
• New multi_format_unit variadic macro to handle formatting multiple numbers at once
• format_unit: Option to choose 3-digit integer display vs 5-character floating-point display
• Optimization: format_unit: Handle number first, to avoid having to restore suffix later
• Generalise hashrate_pick_unit/hashrate_to_bufstr into pick_unit/format_unit
• Extend hashrate_pick_unit/hashrate_to_bufstr to handle sub-kilo units
• Split total_bytes_xfer to total_bytes_rcvd and total_bytes_sent
• Bugfix: _decode_udev_enc_dup: Allocate enough space for full string
• Bugfix: Never use waddstr for logwin, since it would bypass special newline handling
• Bugfix: bitforce: Set kname on chip processors
• bitforce: Include voltages in Manage device TUI
• Defer newlines going to curses logwin, to avoid a useless blank line at the bottom of the window
• Ensure printing to logwin always goes through _wlog
• Remove blank line above log window
• bitforce: Identify parallel queue protocol distinctly from mere bulk queue
• ManageTUI: Include kernel name, when available
• Stratum: Roll ntime as we generate work
• Stratum: Make swork.ntime native-endian
• Stratum: Treat ntime as uint32_t (as it should be), still always big endian
• Debuglog ManageTUI actions/responses
• ManageTUI: Add generic Identify support
• Bugfix: Move serial_detect* and open_bitstream to DevAPI code so CPU/OpenCL can build properly without fpgautils
• Short-circuit logging sooner in quiet mode
• Write to both stderr and console within same console lock "session"
• Bugfix: Also hold the console lock when writing to stderr
• Use common console locking function for stdout in logging.c
• Move console lock and unlock functions (which also handle thread cancelstate) to miner.h
• Bugfix: bitforce: Only try to clear queues of SC devices, since FPGA MR boards interpret ZQX/ZOX differently
• Timer-based gettimeofday substitute for systems with poor time-of-day clocks (Windows)
• Use clock_gettime(CLOCK_MONOTONIC) for timers when available
• Use QueryPerformanceCounter for timers on Windows
• Generic refactoring for timer_set_now
• Replace all remaining uses of gettimeofday for timers, with timer_set_now (aka cgtime)
• Don't mix timers with timestamps (visual only)
• Always use struct timeval for timers, and don't mix timers with timestamps (functional only)
• get_datestamp: Change timeval parameter to time_t, and implement get_now_datestamp for common "current time" use case
• Use get_datestamp for (non-microsecond) log timestamps
• Bugfix: ztex: Allocate final processor names on the heap, so they survive when the stack for ztex_prepare is gone
• Bugfix: ztex: Copy serial number to device "name" before cloning it for other processors
• Bugfix: x6500: Use cgpu->temp directly since there is only one sensor per processor
• Bugfix: Actually show the highest temperature, not just calculate it
• x6500: Allow changing clock speed from TUI Manage device
• x6500: Port pgaset clock from modminer driver at 66d2a3ab072fcdbc3c7ed41a97f265afa917bbee
• modminer: Allow changing clock speed from TUI Manage device
• bitforce: Flush job and result queues at startup to avoid unnecessary warnings
• x6500: Reduce default clock to 190 MHz
• Bugfix: fpgautils: Close libusb handle after copying USB strings
• use BSD sed syntax to generate iospeed_local.h

[Aahzman]

Finally captured the error message i'm getting with 3.1.3 and two jalapenos...

BFL0: idle for more than 60 seconds, declaring sick!
BFL0: attempting to restart
BFL0: re-initialising

and that's when bfg stops working and throws up the "search for a solution" or "close the application" dialog. 

[JakeTri]

NEW VERSION 3.1.4, AUGUST 2 2013

...

• Reorganised TUI display, adding network transfer rate, and totals summary line including count of total devices/processors hashing as well as hottest temperature across all sensors.

...

I like the new TUI

So far 3.1.4 works fine from last night on 3 systems. I'll let you know if I see any issue.

[purelithium]

MMQ clock speed changes don't actually take effect. It appears they do, but revert to bfgminer-chosen clocks immediately after exiting Manage menu.

Combined clockrate at 900mhz, actual hashrate 676. not helpful.