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

[blissz]

no crazy fans on firmware update ... great
will be not easy to check watchdog for now
but starting is good , check option for reboot on low H. is here , how is calculate the 88% ?

Yeah the watchdog is a bit hard to test, so I hope for some feedback from miners that have problems with their boards.
The 88% is just a number I came up with actually. When you're running good you won't ever drop below 90%. But the mechanism needs to kick in if for example if half of a board starts failing. Therefore this seems like a good value. 

[1714719621]

1714719621

[1714719621]

1714719621

[1714719621]

1714719621

[Maicol792]

@blissz i have send PM to you with personal request thanks

[cryptonitro]

@blissz i have send PM to you with personal request thanks

Did you try change those file by ssh ?

[fecell]

now anyone can recompile cgminer for me .. from bitmain ghithub dash miner wit this mod chance xcoin.c in groest.c (delete xcoin.c and rename groest.c in xcoin.c)

wasting time. read basic of programming.
groest.c used in xcoin.c and can't work else.

thans but already have that , that not the hexa file for programmer

are U sure?

Code:

#define PIC16F1704_PROGRAM_NEW                      "/sbin/pic.txt"

Code:

int PIC1704_update_pic_app_program_new(void)
{
    unsigned char program_data[5000] = {0};
    FILE * pic_program_file;
    unsigned int filesize = 0,i=0,j;
    unsigned char data_read[5]= {0,0,0,0,'\0'}, buf[16]= {0};
    unsigned int data_int = 0;
    struct stat statbuff;
    unsigned char start_addr_h = PIC_FLASH_POINTER_START_ADDRESS_H_NEW, start_addr_l = PIC_FLASH_POINTER_START_ADDRESS_L_NEW;
    unsigned char end_addr_h = PIC_FLASH_POINTER_END_ADDRESS_H, end_addr_l = PIC_FLASH_POINTER_END_ADDRESS_L;
    unsigned int pic_flash_length=0;
    int ret=0;

    applog(LOG_NOTICE,"%s", __FUNCTION__);

    // read upgrade file first, if it is wrong, don't erase pic, but just return;
    pic_program_file = fopen(PIC16F1704_PROGRAM_NEW, "r");
    if(!pic_program_file)
    {
        applog(LOG_ERR,"%s: open pic16f1704_app_new.txt failed\n", __FUNCTION__);
        return;
    }
    fseek(pic_program_file,0,SEEK_SET);
    memset(program_data, 0x0, 5000);

    pic_flash_length = (((unsigned int)end_addr_h << 8) + end_addr_l) - (((unsigned int)start_addr_h << 8) + start_addr_l) + 1;
    applog(LOG_NOTICE,"%s: pic_flash_length = %d\n", __FUNCTION__, pic_flash_length);

    for(i=0; i<pic_flash_length; i++)
    {
        fgets(data_read, MAX_CHAR_NUM - 1, pic_program_file);
        //printf("data_read[0]=%c, data_read[1]=%c, data_read[2]=%c, data_read[3]=%c\n", data_read[0], data_read[1], data_read[2], data_read[3]);
        data_int = strtoul(data_read, NULL, 16);
        //printf("data_int = 0x%04x\n", data_int);
        program_data[2*i + 0] = (unsigned char)((data_int >> 8) & 0x000000ff);
        program_data[2*i + 1] = (unsigned char)(data_int & 0x000000ff);
        //printf("program_data[%d]=0x%02x, program_data[%d]=0x%02x\n\n", 2*i + 0, program_data[2*i + 0], 2*i + 1, program_data[2*i + 1]);
    }

    fclose(pic_program_file);

    // after read upgrade file correct, erase pic
    ret = reset_PIC16F1704_pic_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: reset pic error!\n\n", __FUNCTION__);
        return 0;
    }

    ret = erase_PIC16F1704_app_flash_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: erase flash error!\n\n", __FUNCTION__);
        return 0;
    }

    ret = set_PIC16F1704_flash_pointer_new(PIC_FLASH_POINTER_START_ADDRESS_H_NEW, PIC_FLASH_POINTER_START_ADDRESS_L_NEW);
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: set flash pointer error!\n\n", __FUNCTION__);
        return 0;
    }

    for(i=0; i<pic_flash_length/PIC_FLASH_SECTOR_LENGTH*4; i++)
    {
        memcpy(buf, program_data+i*16, 16);
        /**/
        applog(LOG_NOTICE,"send pic program time: %d",i);
        for(j=0; j<16; j++)
        {
            applog(LOG_DEBUG,"buf[%d] = 0x%02x", j, *(buf+j));
        }

        send_data_to_PIC16F1704_new(buf);
        write_data_into_PIC16F1704_flash_new();
    }

    ret = reset_PIC16F1704_pic_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: reset pic error!\n\n", __FUNCTION__);
        return 0;
    }

    return 1;
}

[wxxyz]

Thanks blissz  Keep it up. No more scandalous D3. Temp(PCB) 60-67 while Temp(Chip) 73-82. 815W on my wall with 16.4 GH/s 

Using a mining calc, this gives slightly less daily profit than running it at 19.3 GH/s with the stock 1200 watts - am I missing something?

[thistlecoins]

Thanks blissz  Keep it up. No more scandalous D3. Temp(PCB) 60-67 while Temp(Chip) 73-82. 815W on my wall with 16.4 GH/s 

Using a mining calc, this gives slightly less daily profit than running it at 19.3 GH/s with the stock 1200 watts - am I missing something?

Today's record high DASH prices might have something to do with the profitability change.  That or the user's personal utility costs.

PS, first post.  ;-)

[thistlecoins]

(real first post, with actual content...)

D3 was my first miner (due to deflated ebay prices).  Have enjoyed this thread so far and happy to see the progress with the firmware improvements.  Look forward to testing it on my D3, which, as an in-tiny-home machine will benefit greatly from reduced noise, even at a slight profit loss if the DASH price keeps going up.   After I get my break-even ROI I plan to experiment with immersion cooling so it can tolerate the warm summer months here.  Using it to heat the house right now.

Thank you Blissz!    I don't feel the dev-fee is too high.   Some wealthy donor (not me yet) needs to send you an L3+ and a S9 to expand on these improvements.

My wish-list...

 - now that you have auto-reboot maybe working, add a "record uptime" field to the |system| tab.

 - optional field to configure for emailed notices, if possible? sendmail and mail exist, but my test messages have not arrived (perhaps anti-spam measures).

 - optional notice after a reboot?

 - optional emailed notice if that machine mined a block!  (for those crazy solo miners)

 - fix whatever needs to be fixed to allow "foregrounding" of cgminer process in ssh.  (I get a term-type error of some kind when trying).

 - 2nd'ing another's request for cron style scheduling of mining "profiles"... one profile should be "idle" too if possible, minimum fan speeds, minimum energy (might require a reboot to get out of it).

 Edited to add:
 - "shutdown" button to the system tab.  I have a smart-plug I can remotely kill it with, but just pulling power is not ideal, as the shutdown sequence leaves the fans running to let it cool.

 - Alter this shutdown behavior to ramp down fans to minimal, (equivalent to requested 'idle' profile).

Will post others as they come to me.

[dndssc]

now anyone can recompile cgminer for me .. from bitmain ghithub dash miner wit this mod chance xcoin.c in groest.c (delete xcoin.c and rename groest.c in xcoin.c)

wasting time. read basic of programming.
groest.c used in xcoin.c and can't work else.

thans but already have that , that not the hexa file for programmer

are U sure?

Code:

#define PIC16F1704_PROGRAM_NEW                      "/sbin/pic.txt"

Code:

int PIC1704_update_pic_app_program_new(void)
{
    unsigned char program_data[5000] = {0};
    FILE * pic_program_file;
    unsigned int filesize = 0,i=0,j;
    unsigned char data_read[5]= {0,0,0,0,'\0'}, buf[16]= {0};
    unsigned int data_int = 0;
    struct stat statbuff;
    unsigned char start_addr_h = PIC_FLASH_POINTER_START_ADDRESS_H_NEW, start_addr_l = PIC_FLASH_POINTER_START_ADDRESS_L_NEW;
    unsigned char end_addr_h = PIC_FLASH_POINTER_END_ADDRESS_H, end_addr_l = PIC_FLASH_POINTER_END_ADDRESS_L;
    unsigned int pic_flash_length=0;
    int ret=0;

    applog(LOG_NOTICE,"%s", __FUNCTION__);

    // read upgrade file first, if it is wrong, don't erase pic, but just return;
    pic_program_file = fopen(PIC16F1704_PROGRAM_NEW, "r");
    if(!pic_program_file)
    {
        applog(LOG_ERR,"%s: open pic16f1704_app_new.txt failed\n", __FUNCTION__);
        return;
    }
    fseek(pic_program_file,0,SEEK_SET);
    memset(program_data, 0x0, 5000);

    pic_flash_length = (((unsigned int)end_addr_h << 8) + end_addr_l) - (((unsigned int)start_addr_h << 8) + start_addr_l) + 1;
    applog(LOG_NOTICE,"%s: pic_flash_length = %d\n", __FUNCTION__, pic_flash_length);

    for(i=0; i<pic_flash_length; i++)
    {
        fgets(data_read, MAX_CHAR_NUM - 1, pic_program_file);
        //printf("data_read[0]=%c, data_read[1]=%c, data_read[2]=%c, data_read[3]=%c\n", data_read[0], data_read[1], data_read[2], data_read[3]);
        data_int = strtoul(data_read, NULL, 16);
        //printf("data_int = 0x%04x\n", data_int);
        program_data[2*i + 0] = (unsigned char)((data_int >> 8) & 0x000000ff);
        program_data[2*i + 1] = (unsigned char)(data_int & 0x000000ff);
        //printf("program_data[%d]=0x%02x, program_data[%d]=0x%02x\n\n", 2*i + 0, program_data[2*i + 0], 2*i + 1, program_data[2*i + 1]);
    }

    fclose(pic_program_file);

    // after read upgrade file correct, erase pic
    ret = reset_PIC16F1704_pic_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: reset pic error!\n\n", __FUNCTION__);
        return 0;
    }

    ret = erase_PIC16F1704_app_flash_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: erase flash error!\n\n", __FUNCTION__);
        return 0;
    }

    ret = set_PIC16F1704_flash_pointer_new(PIC_FLASH_POINTER_START_ADDRESS_H_NEW, PIC_FLASH_POINTER_START_ADDRESS_L_NEW);
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: set flash pointer error!\n\n", __FUNCTION__);
        return 0;
    }

    for(i=0; i<pic_flash_length/PIC_FLASH_SECTOR_LENGTH*4; i++)
    {
        memcpy(buf, program_data+i*16, 16);
        /**/
        applog(LOG_NOTICE,"send pic program time: %d",i);
        for(j=0; j<16; j++)
        {
            applog(LOG_DEBUG,"buf[%d] = 0x%02x", j, *(buf+j));
        }

        send_data_to_PIC16F1704_new(buf);
        write_data_into_PIC16F1704_flash_new();
    }

    ret = reset_PIC16F1704_pic_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: reset pic error!\n\n", __FUNCTION__);
        return 0;
    }

    return 1;
}

The pic is reprogrammable without open the d3 and solder or use specific readers?

[Pedrojok]

Hey Blissz, got a couple of questions/suggestions:

- Why did you slow the auto-speed fan so much ? Any reason ? Could you raised it again as it was after your first mod ? Or I won't be able to use auto-fan anymore !  My chip temps never got that high before your last mod.  Or maybe you can make some auto-speed fan profile like "silent" and "normal" ? Auto-speed was perfect because it allows the fan to run faster during daytime, and slower at night when the temp drops a little. But running too slow and it become useless, at least where I live with high temp daytime.

- Could you switch to your devfee pool at the end of the 2h period instead of before ? When we reboot to try different setting, you get much more than 1,5%. Not that I complain about your fee, but let's say you reboot 5 times an hour, you get like 7,5% instead of 1,5% for this hour. And I don't think anyone would bother rebooting every 1h55 to avoid it !

Thanks in advance.

[dsldsl]

I second that with regards to the reduce auto fan speed. I have tested it on a batch 1 17Gh D3 and I have rolled back to stock for the time being.

Problem I have:

1) Following your instruction and setting speed at 400 and fan at 25% is risky. The moment I upgraded the firmware my temp went up to 80. I will suggest setting the fan to at least 59% (if ambient temp is around 20 degrees) or higher if your ambient temp is much higher.

2) I tested with lowest, 1,2 voltage at 487 and I am not getting and HW errors but it keeps stopping with no hashrate.

3) You dev fee pool keeps kicking in when I'm trying to test with different settings and it is better if you set it so that it kicks in at the end of the hour or so.

4) Maybe a way to manually tweak the auto fan speed will be useful.

5) A temp protection like the L3 can also help.

[chase.pxelplus]

Well this looks my stable peak  - https://imgur.com/a/v7X9J  im worried abou that last chip having 77 temperature do u think it will be alright ?

a really nice overclock. What are your voltage settings? Your temperatures are perfectly fine. I can't give you any guarantee, but if you have cheap electricity and can handle the noise I would let it run at these speeds.

If you're worried about the last chain, you can clock that chain down a bit to i.e 575 Mhz and lower the voltage a bit too to get it to ~73 degrees as well. Let chain 1 and 2 run at the settings as they are now.

I was playing with the chain a bit but when i change voltage of 1 chain it somehows affect the others to and i get HWS / high temps ? MY current voltage settings are 20 on all chains , i tryed nearly all of the  high combinations this one seems to be the most stable with 4-7 HW for hour on middle chain as i said only things that worries me is the last chain with 77 degree atm,   Might aslo meantion my D3 is in ventilated room with room temp around  0 degree

Hi,
I would reduce the frequency a little for the last chain which seems always to have higher temperatures (I have 4 D3s and I see the same phenomenon). This is probably due to the way the fans are pushing the air through the body of the miner, which causes more air to flow to one side, affecting the other side where the 3rd chain is located. 77 deg. is ok, but I would not let it run 24/7. The highest temp. I would run any of my D3 miners on a 24/7 basis is 75 deg. C. and even then, one would get more HW errors than with lower temperatures.

Thanks i wanted to know the peak ima get it to 75 also im not getting HW errors after 1 hour i got just 4 HW  only on middle chain

What is your voltage setting for the 3 chains?

[dsldsl]

I have got another problem after tested your firmware from page 1.

After rollback to the official firmware, my temp is really abnormal. It is much higher than usually.

I have 2 D3 in the same location and they normal run around the same 70 / sub 70, but the temp has suddenly shoot right up even after rolling back to the official firmware which is quite worrying.

D3 which tested with your firmware and rolled back to stock is running at 487(default) Board temp 59/59/61 and Chip temp 76/76/78 fan at 4560rpm

D3 right next to it with stock and not tested with your firmware is running at 487 (default) board temp 51/52/56 and Chip temp 66/67/71 fan at 4140 rpm.

My D3 have never had board temp over 57 and they are normally at around 56 max but after flashing your firmware I can't seems to get the temps down.

Please advise!

[may142012]

I have got another problem after tested your firmware from page 1.

After rollback to the official firmware, my temp is really abnormal. It is much higher than usually.

I have 2 D3 in the same location and they normal run around the same 70 / sub 70, but the temp has suddenly shoot right up even after rolling back to the official firmware which is quite worrying.

D3 which tested with your firmware and rolled back to stock is running at 487(default) Board temp 59/59/61 and Chip temp 76/76/78 fan at 4560rpm

D3 right next to it with stock and not tested with your firmware is running at 487 (default) board temp 51/52/56 and Chip temp 66/67/71 fan at 4140 rpm.

My D3 have never had board temp over 57 and they are normally at around 56 max but after flashing your firmware I can't seems to get the temps down.

Please advise!

I wonder why did you decide to rollback to Bitmain's firmware? Is it because of the dev fees? Was the modded firmware working well for you?

[dsldsl]

I rolled back because my D3 keeps going offline and I didn't like the auto fan speed had been adjusted to such low rpm.

I even tried to flash the Bitmain Nov firmware twice and temp still remain so high.

[dt8666]

I second that with regards to the reduce auto fan speed. I have tested it on a batch 1 17Gh D3 and I have rolled back to stock for the time being.

Problem I have:

1) Following your instruction and setting speed at 400 and fan at 25% is risky. The moment I upgraded the firmware my temp went up to 80. I will suggest setting the fan to at least 59% (if ambient temp is around 20 degrees) or higher if your ambient temp is much higher.

2) I tested with lowest, 1,2 voltage at 487 and I am not getting and HW errors but it keeps stopping with no hashrate.

3) You dev fee pool keeps kicking in when I'm trying to test with different settings and it is better if you set it so that it kicks in at the end of the hour or so.

4) Maybe a way to manually tweak the auto fan speed will be useful.

5) A temp protection like the L3 can also help.

Fully agreed with the Dev fee to be at the end of the hr. Also this justify the testing phase that the initial tester will be potentially facing loads of restart to get it right.

Will wait till the version with the above fixed to try. I am from the tropical climate so will also worries on the high temperature if it does not auto switch for higher speed to cool it.

@developer
will you consider next release to have the above 2 fixed? Thanks

[MantaMine]

I believe the fan settings need tweaked.  Seems that they default to ~50% if the manual setting box is unchecked and never move from there regardless of chip temp.  Problem with setting them higher is it seems to cause instability with the rig I'm testing on.

Like what you're doing here, but this feature definitely needs addressed IMO.

[isukabtc]

For Extreme performance that use power up to 1400 watt : If I push my room temperature to 10c, can I go further ? Let said YOLO performance ? So Im not wasting the 200 watt left in my PSU. Anyone done this ?

*I dont have problem with electric bill

Haha...this is funny, my mining room right now is currently at an ambient temp of 10c also as i'm piping cold air in from outside. My D3's  are not modded as per Blissz's method yet and they are so quiet as the chips barely get hotter than 55c-65c.(this should good for the miners right? the cold?)..so i'm sure I could turn these babies up as well. Let us know if you do. I have the 19.3gh/s batch

Yes please, I really appreciate your review for maximize your PSU up to 1600 watt (or just 1550 watt to be safe). I wish you can get to 22gh+ mate, really looking forward for it..

I got an email from Bitmain today, they're sending me 1 D3 and 6 PSU's.  I ordered 6 D3's and 6 PSUs.  Here we have some people getting D3s without PSUs and I'm getting only 1 D3 with all 6 PSUs...smh....

Shits happen, and you got 5 less shits lol
Hope it come ASAP mate   My miner still not in this dimension yet..

[mahouneko]

Today I get D3.(Submodel:19.3G and APW3++)

First original firm:
default(537MHz)
1226W 19490.41MH/S (100V)

Change 400MHz
892W 14351.22MH/s

Applied Blissz firmware:
400MHz
952W

Change lowest voltage.
737W at 400MHz
1000W at 537MHz

Applied your firm,I get silent.
Thank you Blissz.

[mcan]

now anyone can recompile cgminer for me .. from bitmain ghithub dash miner wit this mod chance xcoin.c in groest.c (delete xcoin.c and rename groest.c in xcoin.c)

wasting time. read basic of programming.
groest.c used in xcoin.c and can't work else.

thans but already have that , that not the hexa file for programmer

are U sure?

Code:

#define PIC16F1704_PROGRAM_NEW                      "/sbin/pic.txt"

Code:

int PIC1704_update_pic_app_program_new(void)
{
    unsigned char program_data[5000] = {0};
    FILE * pic_program_file;
    unsigned int filesize = 0,i=0,j;
    unsigned char data_read[5]= {0,0,0,0,'\0'}, buf[16]= {0};
    unsigned int data_int = 0;
    struct stat statbuff;
    unsigned char start_addr_h = PIC_FLASH_POINTER_START_ADDRESS_H_NEW, start_addr_l = PIC_FLASH_POINTER_START_ADDRESS_L_NEW;
    unsigned char end_addr_h = PIC_FLASH_POINTER_END_ADDRESS_H, end_addr_l = PIC_FLASH_POINTER_END_ADDRESS_L;
    unsigned int pic_flash_length=0;
    int ret=0;

    applog(LOG_NOTICE,"%s", __FUNCTION__);

    // read upgrade file first, if it is wrong, don't erase pic, but just return;
    pic_program_file = fopen(PIC16F1704_PROGRAM_NEW, "r");
    if(!pic_program_file)
    {
        applog(LOG_ERR,"%s: open pic16f1704_app_new.txt failed\n", __FUNCTION__);
        return;
    }
    fseek(pic_program_file,0,SEEK_SET);
    memset(program_data, 0x0, 5000);

    pic_flash_length = (((unsigned int)end_addr_h << 8) + end_addr_l) - (((unsigned int)start_addr_h << 8) + start_addr_l) + 1;
    applog(LOG_NOTICE,"%s: pic_flash_length = %d\n", __FUNCTION__, pic_flash_length);

    for(i=0; i<pic_flash_length; i++)
    {
        fgets(data_read, MAX_CHAR_NUM - 1, pic_program_file);
        //printf("data_read[0]=%c, data_read[1]=%c, data_read[2]=%c, data_read[3]=%c\n", data_read[0], data_read[1], data_read[2], data_read[3]);
        data_int = strtoul(data_read, NULL, 16);
        //printf("data_int = 0x%04x\n", data_int);
        program_data[2*i + 0] = (unsigned char)((data_int >> 8) & 0x000000ff);
        program_data[2*i + 1] = (unsigned char)(data_int & 0x000000ff);
        //printf("program_data[%d]=0x%02x, program_data[%d]=0x%02x\n\n", 2*i + 0, program_data[2*i + 0], 2*i + 1, program_data[2*i + 1]);
    }

    fclose(pic_program_file);

    // after read upgrade file correct, erase pic
    ret = reset_PIC16F1704_pic_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: reset pic error!\n\n", __FUNCTION__);
        return 0;
    }

    ret = erase_PIC16F1704_app_flash_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: erase flash error!\n\n", __FUNCTION__);
        return 0;
    }

    ret = set_PIC16F1704_flash_pointer_new(PIC_FLASH_POINTER_START_ADDRESS_H_NEW, PIC_FLASH_POINTER_START_ADDRESS_L_NEW);
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: set flash pointer error!\n\n", __FUNCTION__);
        return 0;
    }

    for(i=0; i<pic_flash_length/PIC_FLASH_SECTOR_LENGTH*4; i++)
    {
        memcpy(buf, program_data+i*16, 16);
        /**/
        applog(LOG_NOTICE,"send pic program time: %d",i);
        for(j=0; j<16; j++)
        {
            applog(LOG_DEBUG,"buf[%d] = 0x%02x", j, *(buf+j));
        }

        send_data_to_PIC16F1704_new(buf);
        write_data_into_PIC16F1704_flash_new();
    }

    ret = reset_PIC16F1704_pic_new();
    if(ret == 0)
    {
        applog(LOG_ERR,"%s: reset pic error!\n\n", __FUNCTION__);
        return 0;
    }

    return 1;
}

 deleting xhash would not but , if we return after groest result in xhash function ?
There is 11 hash steps , so if we return from first one ?

[Dmitriy2k17]

Hi Blissz, thanks for your work!
I use your firmware on 3 d3, it work more stable than on firmware from bitmain.

But I do not use the "auto-fan", because I'm afraid that long-term operation at high temperature can lead to a breakage of the boards.
Could you, please, take the target temperature into manual settings, so that everyone would decide what is more important for him, the silence or safety of the chips.
Thank you in advance!