Braiins OS & Braiins OS+ custom ASIC firmware: optimize performance & efficiency

[jbillk]

Have been playing with OS+ tuning. Anybody know the most efficient setting for temperature and power. I can get close to 75w/th with 1050w setting at 75C target temperature. This is with my best hashboards. I find higher temps I start to lose efficiency, but I do like the lower sound around 80C or 85C target temp.

I find with the power setting of 900w or less, efficiency isn't that great.

[1715137674]

1715137674

[1715137674]

1715137674

[1715137674]

1715137674

[1715137674]

1715137674

[1715137674]

1715137674

[1715137674]

1715137674

[Artemis3]

Are you trying with the latest release from yesterday? There is apparently a new setting to limit the lowest frequency to try, but by default it will try all the way down to 155 Mhz (and all the way up to 950Mhz).

Maybe this answers Kano's question:

CGminer, BOSminer, and GPL Compliance

There has also been some discussion recently amongst miners about GPL compliance when it comes to CGminer. We want to be clear about how this applies to Braiins OS.

Up until our latest major release (March 2020), Braiins OS was built on top of CGminer, which was using Bitmain’s FPGA code as a backend. This component from Bitmain was closed. We built BOSminer from scratch to replace CGminer in order to make everything open-source, including replacing the previously closed FPGA IP core with our own open core.

In other words, the above paragraph simply means that Braiins OS users do not need to worry about this GPL matter because we are not using CGminer or Bitmain’s FPGA IP core.

Additionally, it should be noted that we own the IP for BOSminer and we have a contributor license agreement in place for it. Braiins OS is released free of charge under the GNU GPLv3, while Braiins OS+ has its own software license agreement. We reserve the right to use this software however we see fit, including providing commercial licensing to individuals should they be interested. If any hardware manufacturers want to implement BOSminer, we will be happy to accept their commits to the public repository and keep this open standard for the mining community through the contributor license agreement.

We are focused on continuing to develop our products and supporting our users as best as we can. What others in the industry do in regards to this matter is up to them.

Remember that if you want to directly address the Braiins team, you can do so in their Braiins OS Telegram group.

[faheshang]

Have been playing with OS+ tuning. Anybody know the most efficient setting for temperature and power. I can get close to 75w/th with 1050w setting at 75C target temperature. This is with my best hashboards. I find higher temps I start to lose efficiency, but I do like the lower sound around 80C or 85C target temp.

I find with the power setting of 900w or less, efficiency isn't that great.

I chose 900W in the configuration with default fan and temp settings and it auto tuned my S9i to 825W at 11.9TH. That's 69.3 W/TH
Another S9i I chose 800W with default fan and temp settings and it auto tuned to 723W at 10.4TH. That's 69.5W/TH. It always ends
up with a watt reading lower than what you choose on the configuration page.

[Pendrak]

Any idea why the CHIP temps report is not working? (i am using the latest version)

with the 2019 version have no problems...

Antminer S9J

[kano]

...
Remember that if you want to directly address the Braiins team, you can do so in their Braiins OS Telegram group.

Then you should close this thread about them if they aren't going to read any of this.
They created this thread - if you say they don't have anything to do with the forum any more, then I'll ask a mod to close it.

Their release - including their own rust code - is GPLv3 also - you seem to have ignored that.

[mikeywith]

[...]

And why would any mod close a topic that is active and fits the board's rules and requirements? the fact that they haven't responded to your questions doesn't give any mod the right to close the topic, even if they personally tell you "We are not going to answer any of your questions" that will change nothing.

If you think they are doing something against the law - sue them. And if you think they are doing something against the community you can add a flag to their profile and ask DT members to support your flag.

[kano]

then I'll ask a mod to close it.

And why would any mod close a topic that is active and fits the board's rules and requirements? the fact that they haven't responded to your questions doesn't give any mod the right to close the topic, even if they personally tell you "We are not going to answer any of your questions" that will change nothing.

It was Artemis3 who said I'd have to go to some other forum to address the Braiins team, not me.

If you think they are doing something against the law - sue them. And if you think they are doing something against the community you can add a flag to their profile and ask DT members to support your flag.

So where have I said any of that?
Sigh, people like to be drama queens just so they can post.

[philipma1957]

That kettle is an awfully dark black said the pot.

I am giving you ten merits for the laugh 😂 I just had.

[Artemis3]

I chose 900W in the configuration with default fan and temp settings and it auto tuned my S9i to 825W at 11.9TH. That's 69.3 W/TH
Another S9i I chose 800W with default fan and temp settings and it auto tuned to 723W at 10.4TH. That's 69.5W/TH. It always ends up with a watt reading lower than what you choose on the configuration page.

Can you give us some details, in method of measuring, which kind of mains voltage, and what exact S9 model/speed you used to achieve this? Its quite impressive to have under 70 W/TH actually

Any idea why the CHIP temps report is not working? (i am using the latest version)

with the 2019 version have no problems...

Antminer S9J

It would be nice if you could show us the logs so we could see if there are any hints, could you paste them into hastebin or similar text sharing site?

If you decide to paste them in the forum, make sure to use the [code][/code] bbcode for the log output text.

[Pendrak]

https://pasteboard.co/J7nAbqI.jpg

Code:

[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Linux version 4.4.0-xilinx-g15b13cb5ef12-dirty (build@runner-pEgeGjKa-project-418-concurrent-0) (gcc version 5.4.0 (LEDE GCC 5.4.0 r3560-79f57e422d) ) #0 SMP PREEMPT Thu Apr 30 16:57:12 2020
[    0.000000] CPU: ARMv7 Processor [413fc090] revision 0 (ARMv7), cr=18c5387d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine model: AntMiner S9 Miner Control Board
[    0.000000] cma: Reserved 16 MiB at 0x1f000000
[    0.000000] Memory policy: Data cache writealloc
[    0.000000] On node 0 totalpages: 131072
[    0.000000] free_area_init_node: node 0, pgdat c07c33c0, node_mem_map debf0000
[    0.000000]   Normal zone: 1024 pages used for memmap
[    0.000000]   Normal zone: 0 pages reserved
[    0.000000]   Normal zone: 131072 pages, LIFO batch:31
[    0.000000] PERCPU: Embedded 12 pages/cpu @debcc000 s18944 r8192 d22016 u49152
[    0.000000] pcpu-alloc: s18944 r8192 d22016 u49152 alloc=12*4096
[    0.000000] pcpu-alloc: [0] 0 [0] 1 
[    0.000000] Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 130048
[    0.000000] Kernel command line: console=ttyPS0,115200 noinitrd ubi.mtd=8 ubi.block=0,1 root=/dev/ubiblock0_1 r rootfstype=squashfs rootwait mtdparts=pl35x-nand:512k(boot),2560k(uboot),2m(fpga1),2m(fpga2),512k(uboot_env),512k(miner_cfg),22m(recovery),95m(firmware1),95m(firmware2) earlyprintk
[    0.000000] PID hash table entries: 2048 (order: 1, 8192 bytes)
[    0.000000] Dentry cache hash table entries: 65536 (order: 6, 262144 bytes)
[    0.000000] Inode-cache hash table entries: 32768 (order: 5, 131072 bytes)
[    0.000000] Memory: 495024K/524288K available (5568K kernel code, 240K rwdata, 1848K rodata, 260K init, 221K bss, 12880K reserved, 16384K cma-reserved, 0K highmem)
[    0.000000] Virtual kernel memory layout:
[    0.000000]     vector  : 0xffff0000 - 0xffff1000   (   4 kB)
[    0.000000]     fixmap  : 0xffc00000 - 0xfff00000   (3072 kB)
[    0.000000]     vmalloc : 0xe0800000 - 0xff800000   ( 496 MB)
[    0.000000]     lowmem  : 0xc0000000 - 0xe0000000   ( 512 MB)
[    0.000000]     pkmap   : 0xbfe00000 - 0xc0000000   (   2 MB)
[    0.000000]     modules : 0xbf000000 - 0xbfe00000   (  14 MB)
[    0.000000]       .text : 0xc0008000 - 0xc07463c0   (7417 kB)
[    0.000000]       .init : 0xc0747000 - 0xc0788000   ( 260 kB)
[    0.000000]       .data : 0xc0788000 - 0xc07c4320   ( 241 kB)
[    0.000000]        .bss : 0xc07c4320 - 0xc07fba44   ( 222 kB)
[    0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=2, Nodes=1
[    0.000000] Preemptible hierarchical RCU implementation.
[    0.000000] 	RCU restricting CPUs from NR_CPUS=4 to nr_cpu_ids=2.
[    0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=2
[    0.000000] NR_IRQS:16 nr_irqs:16 16
[    0.000000] slcr mapped to e0800000
[    0.000000] L2C: platform modifies aux control register: 0x02060000 -> 0x32460000
[    0.000000] L2C: DT/platform modifies aux control register: 0x02060000 -> 0x32460000
[    0.000000] L2C-310 erratum 769419 enabled
[    0.000000] L2C-310 enabling early BRESP for Cortex-A9
[    0.000000] L2C-310 full line of zeros enabled for Cortex-A9
[    0.000000] L2C-310 dynamic clock gating enabled, standby mode enabled
[    0.000000] L2C-310 cache controller enabled, 8 ways, 512 kB
[    0.000000] L2C-310: CACHE_ID 0x410000c8, AUX_CTRL 0x46460001
[    0.000000] zynq_clock_init: clkc starts at e0800100
[    0.000000] Zynq clock init
[    0.000011] sched_clock: 64 bits at 333MHz, resolution 3ns, wraps every 4398046511103ns
[    0.000035] clocksource: arm_global_timer: mask: 0xffffffffffffffff max_cycles: 0x4ce07af025, max_idle_ns: 440795209040 ns
[    0.000136] clocksource: ttc_clocksource: mask: 0xffff max_cycles: 0xffff, max_idle_ns: 537538477 ns
[    0.000165] timer #0 at e0808000, irq=17
[    0.000417] Console: colour dummy device 80x30
[    0.000439] Calibrating delay loop... 1332.01 BogoMIPS (lpj=6660096)
[    0.090306] pid_max: default: 32768 minimum: 301
[    0.090414] Mount-cache hash table entries: 1024 (order: 0, 4096 bytes)
[    0.090428] Mountpoint-cache hash table entries: 1024 (order: 0, 4096 bytes)
[    0.090929] CPU: Testing write buffer coherency: ok
[    0.091134] CPU0: thread -1, cpu 0, socket 0, mpidr 80000000
[    0.091223] Setting up static identity map for 0x8280 - 0x82d8
[    0.260294] CPU1: thread -1, cpu 1, socket 0, mpidr 80000001
[    0.260367] Brought up 2 CPUs
[    0.260386] SMP: Total of 2 processors activated (2664.03 BogoMIPS).
[    0.260395] CPU: All CPU(s) started in SVC mode.
[    0.264064] VFP support v0.3: implementor 41 architecture 3 part 30 variant 9 rev 4
[    0.264492] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[    0.265384] pinctrl core: initialized pinctrl subsystem
[    0.266662] NET: Registered protocol family 16
[    0.268705] DMA: preallocated 256 KiB pool for atomic coherent allocations
[    0.290286] cpuidle: using governor ladder
[    0.320272] cpuidle: using governor menu
[    0.330575] zynq-ocm f800c000.ocmc: ZYNQ OCM pool: 256 KiB @ 0xe0880000
[    0.330978] zynq-pinctrl 700.pinctrl: zynq pinctrl initialized
[    0.391091] GPIO IRQ not connected
[    0.391109] XGpio: /amba_pl/gpio@41200000: registered, base is 897
[    0.391298] GPIO IRQ not connected
[    0.391310] XGpio: /amba_pl/gpio@41210000: registered, base is 884
[    0.392169] vgaarb: loaded
[    0.392598] SCSI subsystem initialized
[    0.392953] usbcore: registered new interface driver usbfs
[    0.393037] usbcore: registered new interface driver hub
[    0.393295] usbcore: registered new device driver usb
[    0.393699] media: Linux media interface: v0.10
[    0.393780] Linux video capture interface: v2.00
[    0.393871] pps_core: LinuxPPS API ver. 1 registered
[    0.393883] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[    0.393930] PTP clock support registered
[    0.394030] EDAC MC: Ver: 3.0.0
[    0.400691] Advanced Linux Sound Architecture Driver Initialized.
[    0.401883] clocksource: Switched to clocksource arm_global_timer
[    0.410552] NET: Registered protocol family 2
[    0.411253] TCP established hash table entries: 4096 (order: 2, 16384 bytes)
[    0.411318] TCP bind hash table entries: 4096 (order: 3, 32768 bytes)
[    0.411403] TCP: Hash tables configured (established 4096 bind 4096)
[    0.411460] UDP hash table entries: 256 (order: 1, 8192 bytes)
[    0.411492] UDP-Lite hash table entries: 256 (order: 1, 8192 bytes)
[    0.411703] NET: Registered protocol family 1
[    0.422211] RPC: Registered named UNIX socket transport module.
[    0.422225] RPC: Registered udp transport module.
[    0.422234] RPC: Registered tcp transport module.
[    0.422242] RPC: Registered tcp NFSv4.1 backchannel transport module.
[    0.422273] PCI: CLS 0 bytes, default 64
[    0.423820] futex hash table entries: 512 (order: 3, 32768 bytes)
[    0.434870] squashfs: version 4.0 (2009/01/31) Phillip Lougher
[    0.437265] io scheduler noop registered
[    0.437286] io scheduler deadline registered
[    0.437338] io scheduler cfq registered (default)
[    0.439195] dma-pl330 f8003000.dmac: Loaded driver for PL330 DMAC-241330
[    0.439215] dma-pl330 f8003000.dmac: 	DBUFF-128x8bytes Num_Chans-8 Num_Peri-4 Num_Events-16
[    0.439888] Serial: 8250/16550 driver, 2 ports, IRQ sharing disabled
[    0.441178] e0001000.serial: ttyPS0 at MMIO 0xe0001000 (irq = 145, base_baud = 6249999) is a xuartps
[    1.069682] console [ttyPS0] enabled
[    1.073834] xdevcfg f8007000.devcfg: ioremap 0xf8007000 to e081c000
[    1.080496] [drm] Initialized drm 1.1.0 20060810
[    1.095879] brd: module loaded
[    1.104987] loop: module loaded
[    1.110885] CAN device driver interface
[    1.116060] libphy: MACB_mii_bus: probed
[    1.202044] macb e000b000.ethernet eth0: Cadence GEM rev 0x00020118 at 0xe000b000 irq 147 (12:b5:e6:c6:47:3b)
[    1.211927] macb e000b000.ethernet eth0: attached PHY driver [Generic PHY] (mii_bus:phy_addr=e000b000.etherne:01, irq=-1)
[    1.223093] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[    1.228843] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[    1.239109] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[    1.245603] ehci-pci: EHCI PCI platform driver
[    1.250165] usbcore: registered new interface driver usb-storage
[    1.257560] mousedev: PS/2 mouse device common for all mice
[    1.273485] i2c /dev entries driver
[    1.277495] cdns-i2c e0004000.i2c: 33 kHz mmio e0004000 irq 141
[    1.283792] cdns-i2c e0005000.i2c: 33 kHz mmio e0005000 irq 142
[    1.292195] EDAC MC: ECC not enabled
[    1.295910] Xilinx Zynq CpuIdle Driver started
[    1.300581] sdhci: Secure Digital Host Controller Interface driver
[    1.306703] sdhci: Copyright(c) Pierre Ossman
[    1.311026] sdhci-pltfm: SDHCI platform and OF driver helper
[    1.318021] sdhci-arasan e0100000.sdhci: No vmmc regulator found
[    1.323973] sdhci-arasan e0100000.sdhci: No vqmmc regulator found
[    1.363955] mmc0: SDHCI controller on e0100000.sdhci [e0100000.sdhci] using ADMA
[    1.372249] ledtrig-cpu: registered to indicate activity on CPUs
[    1.378653] usbcore: registered new interface driver usbhid
[    1.384172] usbhid: USB HID core driver
[    1.388883] nand: device found, Manufacturer ID: 0x2c, Chip ID: 0xda
[    1.395182] nand: Micron MT29F2G08ABAGAWP
[    1.399149] nand: 256 MiB, SLC, erase size: 128 KiB, page size: 2048, OOB size: 128
[    1.406817] nand: WARNING: pl35x-nand: the ECC used on your system is too weak compared to the one required by the NAND chip
[    1.418273] Bad block table found at page 131008, version 0x01
[    1.424480] Bad block table found at page 130944, version 0x01
[    1.430452] nand_read_bbt: bad block at 0x000003200000
[    1.435607] 9 cmdlinepart partitions found on MTD device pl35x-nand
[    1.441791] Creating 9 MTD partitions on "pl35x-nand":
[    1.446947] 0x000000000000-0x000000080000 : "boot"
[    1.452741] 0x000000080000-0x000000300000 : "uboot"
[    1.458526] 0x000000300000-0x000000500000 : "fpga1"
[    1.474338] 0x000000500000-0x000000700000 : "fpga2"
[    1.490109] 0x000000700000-0x000000780000 : "uboot_env"
[    1.506318] 0x000000780000-0x000000800000 : "miner_cfg"
[    1.522423] 0x000000800000-0x000001e00000 : "recovery"
[    1.538434] 0x000001e00000-0x000007d00000 : "firmware1"
[    1.554691] 0x000007d00000-0x00000dc00000 : "firmware2"
[    1.573460] nf_conntrack version 0.5.0 (7990 buckets, 31960 max)
[    1.581329] NET: Registered protocol family 10
[    1.586995] sit: IPv6 over IPv4 tunneling driver
[    1.592483] NET: Registered protocol family 17
[    1.596900] bridge: automatic filtering via arp/ip/ip6tables has been deprecated. Update your scripts to load br_netfilter if you need this.
[    1.609464] can: controller area network core (rev 20120528 abi 9)
[    1.615668] NET: Registered protocol family 29
[    1.620039] can: raw protocol (rev 20120528)
[    1.624311] can: broadcast manager protocol (rev 20120528 t)
[    1.629935] can: netlink gateway (rev 20130117) max_hops=1
[    1.635629] 8021q: 802.1Q VLAN Support v1.8
[    1.640118] Registering SWP/SWPB emulation handler
[    1.646350] ubi0: attaching mtd8
[    1.974633] ubi0: scanning is finished
[    1.985680] ubi0: attached mtd8 (name "firmware2", size 95 MiB)
[    1.991515] ubi0: PEB size: 131072 bytes (128 KiB), LEB size: 126976 bytes
[    1.998393] ubi0: min./max. I/O unit sizes: 2048/2048, sub-page size 2048
[    2.005161] ubi0: VID header offset: 2048 (aligned 2048), data offset: 4096
[    2.012121] ubi0: good PEBs: 760, bad PEBs: 0, corrupted PEBs: 0
[    2.018075] ubi0: user volume: 3, internal volumes: 1, max. volumes count: 128
[    2.025301] ubi0: max/mean erase counter: 2/1, WL threshold: 4096, image sequence number: 3488791415
[    2.034407] ubi0: available PEBs: 0, total reserved PEBs: 760, PEBs reserved for bad PEB handling: 40
[    2.043635] ubi0: background thread "ubi_bgt0d" started, PID 818
[    2.054538] block ubiblock0_1: created from ubi0:1(rootfs)
[    2.059956] hctosys: unable to open rtc device (rtc0)
[    2.069453] ALSA device list:
[    2.072378]   No soundcards found.
[    2.079281] VFS: Mounted root (squashfs filesystem) readonly on device 254:0.
[    2.086711] Freeing unused kernel memory: 260K (c0747000 - c0788000)
[    2.556803] init: Console is alive
[    2.560277] init: - watchdog -
[    2.961403] kmodloader: loading kernel modules from /etc/modules-boot.d/*
[    3.063436] kmodloader: done loading kernel modules from /etc/modules-boot.d/*
[    3.078025] init: - preinit -
[    3.660935] random: jshn urandom read with 0 bits of entropy available
[    3.856045] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[    5.222326] macb e000b000.ethernet eth0: link up (100/Full)
[    5.227835] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[    7.017798] UBIFS (ubi0:2): background thread "ubifs_bgt0_2" started, PID 894
[    7.045257] UBIFS (ubi0:2): recovery needed
[    7.115837] UBIFS (ubi0:2): recovery completed
[    7.120291] UBIFS (ubi0:2): UBIFS: mounted UBI device 0, volume 2, name "rootfs_data"
[    7.128066] UBIFS (ubi0:2): LEB size: 126976 bytes (124 KiB), min./max. I/O unit sizes: 2048 bytes/2048 bytes
[    7.137948] UBIFS (ubi0:2): FS size: 77963264 bytes (74 MiB, 614 LEBs), journal size 3936256 bytes (3 MiB, 31 LEBs)
[    7.148366] UBIFS (ubi0:2): reserved for root: 3682397 bytes (3596 KiB)
[    7.154962] UBIFS (ubi0:2): media format: w4/r0 (latest is w4/r0), UUID 23F70614-05B9-4699-B18C-4328AF8F962B, small LPT model
[    7.168341] mount_root: switching to ubifs overlay
[    7.179157] urandom-seed: Seeding with /etc/urandom.seed
[    7.308768] procd: - early -
[    7.311630] procd: - watchdog -
[    7.954466] procd: - watchdog -
[    7.957743] procd: - ubus -
[    8.116319] procd: - init -
[    8.260523] macb e000b000.ethernet eth0: link down
[    8.518534] kmodloader: loading kernel modules from /etc/modules.d/*
[    8.527614] tun: Universal TUN/TAP device driver, 1.6
[    8.532634] tun: (C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com>
[    8.541837] ip6_tables: (C) 2000-2006 Netfilter Core Team
[    8.554826] ip_tables: (C) 2000-2006 Netfilter Core Team
[    8.683280] xt_time: kernel timezone is -0000
[    8.693543] kmodloader: done loading kernel modules from /etc/modules.d/*
[   11.488073] Start post-upgrade process...
[   11.546208] device eth0 entered promiscuous mode
[   11.565301] IPv6: ADDRCONF(NETDEV_UP): br-lan: link is not ready
[   12.292183] macb e000b000.ethernet eth0: link up (100/Full)
[   12.297772] br-lan: port 1(eth0) entered forwarding state
[   12.303163] br-lan: port 1(eth0) entered forwarding state
[   12.320698] IPv6: ADDRCONF(NETDEV_CHANGE): br-lan: link becomes ready
[   12.588080] random: nonblocking pool is initialized
[   14.301907] br-lan: port 1(eth0) entered forwarding state

[faheshang]

I measured the watts at the wall with a standard watt meter. The miner is on a 240V line. Model is S9i 14TH.

UPDATE: I've tried two new power settings for auto-tuning. I put in 1200W on the configuration and it auto-tuned to 1111W at 14.8TH which is 75W/TH. I next entered 910W on the configuration page and it auto-tuned to 834W at 12.04TH, about 69.3W/TH. For my S9i-14TH miners, anything above 910W the efficiency starts to go down.

I have three S9i-14TH miners running at about 36TH and 2500W.

https://pasteboard.co/J7MYtI6.jpg

[Artemis3]

https://pasteboard.co/J7nAbqI.jpg

Pendrak, there are two logs. I think one is kernel and the other is system, please try pasting both of them.


faheshang Your results are awesome, thanks!

Lets hope we can get some more people to match your results.

[Seph213]

I just tuned a S9i 14th with power limit 1050. I got 13.6Th with 325W on all boards so that's 975W , 71.69W/Th.

The approx. Total Watts shows 1049W. If that's the accurate one it would be 77.13W/Th.

I did not plug my meter on that miner to test the actual consumption . I'll report back once i plug my meter on the miner and run the autotune to see which of these 2 reading is the one we wanna follow.

UPDATE: so i was able to check the consumption directly on the miner and it was 954W.

So this miner is running at 13.68th for 954W 69.73W/Th.

It seems the consumption showing on each board would be the closest one to reality.

[Tamerz]

How long does auto-tuning take? I installed Braiins OS+ on two S9's and let it run all night. I'm getting horrible hashrates and it still says "Tuning individual chips" on both of them. It has been over 10 hours.

So it ran all day and still, terrible hash rate and said it was tuning. I rolled back to the stock firmware by pointing it at the backup and now with stock firmware it is hashing terribly as well.

How do I completely reset back to factory?

bos-plus-toolbox.exe was used for both the install and uninstall, including the backup.

[faheshang]

How long does auto-tuning take? I installed Braiins OS+ on two S9's and let it run all night. I'm getting horrible hashrates and it still says "Tuning individual chips" on both of them. It has been over 10 hours.

Three factors are involved with answering this question. The first is how low or how high is your watt setting the higher the watt setting the longer auto-tuning takes. The second factor is ambient temperature. I have found that if the ambient temperature is cool and controlled the process is much quicker than if the mining environment is a hot garage with little ventilation, for example. The third factor in "how long" is the condition of your individual chips. Since auto-tuning tests and sets levels for each chip, if you have a well used S9 which maybe has lots of debris in the heat sinks, thus damaging some of your chips prematurely,  or if you have overclocked your S9 beyond factory specs, it is possible you could have weak or damaged chips which will take longer to tune because they intermittently fail thus causing the auto-tuning to change it's values.

What models and how old are the S9s you are tuning? What is the watt value you are putting on the configuration page? "Horrible hashrates": When the tuning process is on it's last steps it begins "tuning individual chips". The hashrate gets significantly lower when this process begins. I have had rigs take 5 hours to auto-tune at 1200W and 2 hours at 900W. If you are overclocking to 1400W or higher I am guessing this will take much longer. There is no set answer to your question. You just need to be patient until "tuning individual chips" changes to stable for all 3 boards. It could be that your S9s might not work as well with auto-tuning as others have reported here.

[Tamerz]

I'm using a BitMain APW5 PSU for each one which are rated at 2600 W each. So I set it to 2600 W since I am running on 240V. If this is incorrect, what should I be setting it to?

They were both bought in May, 2017. One is a 13.5TH and the other a 11.5TH.

[faheshang]

2600W is way to high for air-cooled S9s. If you are looking to overclock those older S9s, I would try 1400W-1600W for the 13.5 model and 1200-1400W for the 11.5 model and see if you are happy with the results of hashrate and efficiency.

Never run more than 90% of the rated power of your PSU by the way.

[Tamerz]

I thought the setting was more just to let the software know how much power the supply could provide, not how far to push the actual miner. I'll give it a shot with far less aggressive settings.

Thanks for the fast response!

OK, that made a huge difference. It finished tuning, and I went from around 25 TH/sec with stock firmware to 31 TH/sec total.

[Artemis3]

I'm using a BitMain APW5 PSU for each one which are rated at 2600 W each. So I set it to 2600 W since I am running on 240V. If this is incorrect, what should I be setting it to?

They were both bought in May, 2017. One is a 13.5TH and the other a 11.5TH.

I think you should leave some margin just in case. 2600W is a crazy lot for an S9, if you had that with immersion, you could even reach 20TH (if the chips are in good condition). I know this because i already know of such a case with a powerlimit setting of 1675W and its at 19TH already with the chips at 49°C, so it was only hold back by the PSU. Measure at the wall just in case, and try not to get close to 100% capacity. An S9 is unlikely to ever demand 2600W anyhow so i doubt it would be a problem, unlike the 1600 and 1800 PSUs who can be put to the max with a single unit.

I agree with faheshang about the factors for speedier autotune, i have observed the same thing with others. Hopefully there will be some sort of "saving state" of the autotune in a coming update so it doesn't have to do it all over again everytime you powercycle the unit or change some unrelated setting like the pool.

Of course the powerlimit also indirectly affects the temperature and speeds the autotune will try, if you use a lower setting, it results in a colder slower unit (as efficient as it can get within the allowed powerlimit).

[philipma1957]

hmm may try this on my dead heat sensor build.