Antminer D3 Blissz firmware (10/12 v1.12 update)

[dndssc]

has anyone tried to bottle the D3 Miner with the original flash firmware from the L3 +?

chip on l3+ ---> BM1485
Chip on d3 ----> BM1760

You can't put l3 fw on d3!

[1714725896]

1714725896

[1714725896]

1714725896

[1714725896]

1714725896

[1714725896]

1714725896

[Gladimor]

To the few individuals still saying Baikal is an FPGA:

The Baikal hashing units are ASIC chips utilizing 28nm cores with subdivided arrays of nodes that hash specific links in the chained X11 algo (eg; Groestl, Echo, Blue-Midnight-Wish). Same thing for Bitmain

[may142012]

Can someone report the results after 24h of the updated firmware?

[taserz]

To the few individuals still saying Baikal is an FPGA:

The Baikal hashing units are ASIC chips utilizing 28nm cores with subdivided arrays of nodes that hash specific links in the chained X11 algo (eg; Groestl, Echo, Blue-Midnight-Wish). Same thing for Bitmain

No you have no idea what you are talking about.

[ruptan]

20/11 recommended UPDATE



https://mega.nz/#!ngwhgaBS!bflAJghChk1pOMtWIyjmzFi7a9UNM7Itz4VkvYcnMq0

Changes 20/11:
- ASIC frequency adjustable per chain
- Voltage settings adjustable per chain
- new graphics. (You might need to clear your browser cache)
- Switching pools overhead lowered (now mines for fee every 2 hours).


know issues:
- when using the AUTO fan profile, the fan starts at a too high RPM after powering up.
- switching pools can sometimes cause the miner to wait for work for 10 seconds.

when you can't change the fan / pool settings:
Please download and flash again as the first link to this version doesn't allow changing the fan speed. Sorry for the inconvenience.

Thanks a lot for the info. Just for an understanding...if for any reason I want to go back and start using the BITMAIN version Farmware. Can I revert back to that? Or once we start using it, we have to use the new firmware and there is no going back? Thank you.

[blissz]

20/11 recommended UPDATE



https://mega.nz/#!ngwhgaBS!bflAJghChk1pOMtWIyjmzFi7a9UNM7Itz4VkvYcnMq0

Changes 20/11:
- ASIC frequency adjustable per chain
- Voltage settings adjustable per chain
- new graphics. (You might need to clear your browser cache)
- Switching pools overhead lowered (now mines for fee every 2 hours).


know issues:
- when using the AUTO fan profile, the fan starts at a too high RPM after powering up.
- switching pools can sometimes cause the miner to wait for work for 10 seconds.

when you can't change the fan / pool settings:
Please download and flash again as the first link to this version doesn't allow changing the fan speed. Sorry for the inconvenience.

Thanks a lot for the info. Just for an understanding...if for any reason I want to go back and start using the BITMAIN version Farmware. Can I revert back to that? Or once we start using it, we have to use the new firmware and there is no going back? Thank you.

You can just flash back to stock bitmain firmware

[taserz]

Can someone report the results after 24h of the updated firmware?

Yeah 1.5% lost in profit due to dev fees.. Aka text editor and compile.

[may142012]

Can someone report the results after 24h of the updated firmware?

Yeah 1.5% lost in profit due to dev fees.. Aka text editor and compile.

Well, gotta support him some how, who knows he could make D3 work on other algorithms in the future. Let's us pray!

[mcan]

The D3 cgminer driver (released by Bitmain) doesn't even mention X11 or subalgorithms. You can check:

• https://github.com/bitmaintech/cgminer-dash/blob/master/driver-btm-DASH.c
• https://github.com/bitmaintech/cgminer-dash/blob/master/driver-btm-DASH.h

There is struct work_dash that contains interesting type field. However it has fixed unknown value 0x20.

Code:

    
struct work_dash
{
    uint8_t type;                           // Bit[7:5]: Type,fixed as 0x01. Bit[4]:sno valid   Bit[3:0]:reserved
    uint8_t wc;                             // bit[7]: reserved, bit[6:0]: work count base
    uint8_t work[WORK_DATA_INPUT_LENGTH];   // 0-75 bytes are dash work; 76-79 are start nonce field
    uint16_t crc16;                         // crc, but not include the header 0x55, 0xaa
};
//...

// BM1760 ASIC input data length
#define WORK_INPUT_TYPE_WITH_SNO                    0x30
#define WORK_INPUT_TYPE_WITHOUT_SNO                 0x20

//...

void *DASH_fill_work(void *usrdata)
{
struct work_dash workdata;
//...
workdata.type = WORK_INPUT_TYPE_WITHOUT_SNO;
//...
DASH_write(info->dev_fd[chainid], (uint8_t *)&workdata, WORK_INPUT_LENGTH_WITH_CRC);

It would be interesting to fuzz work_dash.type a bit, to see if BM1760 supports some other types. However probably this wouldn't do anything useful.

So IMHO this means unless there is secret golden switch inside chip BM1760, there is no way to choose algorithm.
BM1760 is Bitmain private chip. There is no public spec available, all they say about it is "BM1760 custom-made chip using 28nm technology."  

thank you for infos!
I think we are close to the end of the discussion about mining other algos!

Where is Xhash() function defined ? It has called from driver-btm-DASH.c but i could not see anywhere that the function decribed declared ..

[ruptan]

20/11 recommended UPDATE



https://mega.nz/#!ngwhgaBS!bflAJghChk1pOMtWIyjmzFi7a9UNM7Itz4VkvYcnMq0

Changes 20/11:
- ASIC frequency adjustable per chain
- Voltage settings adjustable per chain
- new graphics. (You might need to clear your browser cache)
- Switching pools overhead lowered (now mines for fee every 2 hours).


know issues:
- when using the AUTO fan profile, the fan starts at a too high RPM after powering up.
- switching pools can sometimes cause the miner to wait for work for 10 seconds.

when you can't change the fan / pool settings:
Please download and flash again as the first link to this version doesn't allow changing the fan speed. Sorry for the inconvenience.

Thanks a lot for the info. Just for an understanding...if for any reason I want to go back and start using the BITMAIN version Farmware. Can I revert back to that? Or once we start using it, we have to use the new firmware and there is no going back? Thank you.

You can just flash back to stock bitmain firmware

Thanks a lot. You mean just to upload the firmware again right?

[lramishvili]

there is some 2 more stratum with 2 workers there how can i remove them?

stratum+tcp://dash.suprnova.cc:9995
devFeeMiner.2
Alive
32.0000
31
3
48
3,056
1,536
0
0
0
115
0
32
0:02:48
4
stratum+tcp://dash-eu.coinmine.pl:6099
devFeeMiner.1
Alive
20.0000
1
4
0
0
0
0
0
0
0
0
0
Never

or it is must to keep? if so how can i revert back to original firmware?

[fecell]

there is some 2 more stratum with 2 workers there how can i remove them?
(...)
or it is must to keep? if so how can i revert back to original firmware?

I think it's a fee of developer.
If you dont like to mine for him, revert to original firmware (dowload from bitmain support site).
(or just patch cgminer with hexeditor and change his worker's name to yours.. it's a some kind of cheat)

[whoisthismanoO]

Thank you for your efforts.

- Dev fee lowered to 1,5%

I would like to know exactly how Devfee works?
Does it mine to Dev 1 minute every hour?
Or is it only mine to Dev when it's turned on at first time until we change the pool?
Or in some other way?

Thanks!

[cryptonitro]

Thank you for your efforts.

- Dev fee lowered to 1,5%

I would like to know exactly how Devfee works?
Does it mine to Dev 1 minute every hour?
Or is it only mine to Dev when it's turned on at first time until we change the pool?
Or in some other way?

Thanks!

Hello , please had some time to read, devfee was mining 54sec/hour , if i'm right with last updates it's mining 108 secondes / 2 Hours

[blissz]

Thank you for your efforts.

- Dev fee lowered to 1,5%

I would like to know exactly how Devfee works?
Does it mine to Dev 1 minute every hour?
Or is it only mine to Dev when it's turned on at first time until we change the pool?
Or in some other way?

Thanks!

Hello , please had some time to read, devfee was mining 54sec/hour , if i'm right with last updates it's mining 108 secondes / 2 Hours

You are right, this will be in there until I solved the issue where switching the pool makes the miner go idle for 10 seconds sometimes. Once that's solved I prefer to put it back on hourly base.

[cryptonitro]

thinking basicaly but the best way to have subalgo seem to be CCMINER , the look like have better implementation for it

[aktarus82]

Regarding sub-X11 algo mining, I also quickly read through the code and my feeling is that the controler board sends an X11 work package to the ASIC boards and the ASIC boards return the final X11-Hash to the main controller board. I guess there is no way to trigger single algo mining out of the linux controller board, even if someone tweeks the miner software.
BUT... (I haven't investigated that yet) it seems that each ASIC board has a PIC µC on it, which would handle communication between main Linux controler board and the ASIC chips. So if there is one place where it may be possible to trigger single algo mining, it would be inside the PIC.

What do you guys think about this?

[Ubersonic]

Awesome work Blissz

Just a quick question if you would be so kind, if I am aiming for silence > all (rewiring garage ATM so miners inside the house) would I be correct in thinking that the best solution is to set the frequency at "400mhz", set the voltage to "lowest", and then lower the fan % until chip temp rises to between 75 - 80 degrees?

Thank you.

[blissz]

Awesome work Blissz

Just a quick question if you would be so kind, if I am aiming for silence > all (rewiring garage ATM so miners inside the house) would I be correct in thinking that the best solution is to set the frequency at "400mhz", set the voltage to "lowest", and then lower the fan % until chip temp rises to between 75 - 80 degrees?

Thank you.

Exactly the strategy I would suggest for silence After that you can try to play around with increasing the frequency a bit as long as you can keep it cool.

[blissz]

Regarding sub-X11 algo mining, I also quickly read through the code and my feeling is that the controler board sends an X11 work package to the ASIC boards and the ASIC boards return the final X11-Hash to the main controller board. I guess there is no way to trigger single algo mining out of the linux controller board, even if someone tweeks the miner software.
BUT... (I haven't investigated that yet) it seems that each ASIC board has a PIC µC on it, which would handle communication between main Linux controler board and the ASIC chips. So if there is one place where it may be possible to trigger single algo mining, it would be inside the PIC.

What do you guys think about this?

That's exactly how it works now indeed. The PIC is in between the controller board and the ASIC's so let's hope bitmain added some secret switch. I looked into it quite some hours now and didn't succeed yet. But I still have hope there is a command to trigger it, but some inside information from Bitmain would make things a bit more simple...