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

[blissz]

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.

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.

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 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.

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.

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!

Sorry guys, seems like the lower auto fan was not such a good idea. I will release a version soon that will add AUTO fan profiles:

- AUTO silent (more or less like it is in the latest firmware)
- AUTO blissz v1 (as it was in the first custom firmware)
- AUTO bitmain stock (bitmain stock, slightly higher fan speeds then blissz v1)
- AUTO cool (faster fan speeds then stock, tries to keep your chips below 70 deg)

[fecell]

Sorry for my bad Eglish.

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 ?

May be. Without miner I can only inspect sources and make hypothesis. 5 dec my D3 must arrive

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

I think, if PIC programming function present in cgminer.c, this means BeagleBoard can flash PIC directly without external programming device.
This function called from "check_whether_need_update_pic_program", and it calls from "bitmain_DASH_init".
And "bitmain_DASH_init" (we can see it in log) really called at booting.

[blastar]

Sorry guys, seems like the lower auto fan was not such a good idea. I will release a version soon that will add AUTO fan profiles:

- AUTO silent (more or less like it is in the latest firmware)
- AUTO blissz v1 (as it was in the first custom firmware)
- AUTO bitmain stock (bitmain stock, slightly higher fan speeds then blissz v1)
- AUTO cool (faster fan speeds then stock, tries to keep your chips below 70 deg)

Great! Because I also have the same problem with low fan speed and high chip temps.

[cryptonitro]

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;
}

is it in d3 firmware ?
we need for l3+, seem to be a bit different
https://github.com/bitmaintech/cgminer-ltc/blob/master/driver-btm-L3.h

i don't really understand all of that and where you want to go with that . but if you can have some explication (how to) on this topic designed about the problem of PIC for L3+ , it would be nice from you and better that doing that here

thanks for your way

[fecell]

is it in firmware ?

Yes, placed in "driver-btm-DASH.c"
For L3+ it's named "update_pic_program" (in "driver-btm-L3.c")
but without hardware I can't do debug, tests and experiments. only inspect sources for future.

ps: my D3 must arrive at 5 dec.

psps: did U know 'L3+' is a 'S8' (internal bitmain's name) (from D3 drivers sources (driver-btm-L3.c) - printf("\n%s: open hash_s8_app.txt failed\n", __FUNCTION__);)

[cryptonitro]

i have higher temp with the 23/11 version ...

[kristikun]

Heya so i started using your firmware to OC mine D3 so far iv been able to get stable 21gh with few hw on 1 chain , fan on 90% but im not sure about the chip temperature for 24/7 usage  my hottest chain has chip  74 +-  would this be alrigh for nonstop   mining ?

Of course the chances of breaking something increases when OCing, but I think your temperatures are fine, so it should not be an issue. The hashrate of 21Gh.s us really nice if you don't mind the noise and electricity Let us know how it goes.

OK so this is my final peak for OC   running nearly a day https://imgur.com/a/nhZga   looks good huh ?  temp around 73-76 max  on chips ,  hash 21.6- 22.2 gh

[cryptonitro]

Heya so i started using your firmware to OC mine D3 so far iv been able to get stable 21gh with few hw on 1 chain , fan on 90% but im not sure about the chip temperature for 24/7 usage  my hottest chain has chip  74 +-  would this be alrigh for nonstop   mining ?

Of course the chances of breaking something increases when OCing, but I think your temperatures are fine, so it should not be an issue. The hashrate of 21Gh.s us really nice if you don't mind the noise and electricity Let us know how it goes.

OK so this is my final peak for OC   running nearly a day https://imgur.com/a/nhZga   looks good huh ?  temp around 73-76 max  on chips ,  hash 21.6- 22.2 gh

that s bad , we need everyone lowering hasrate to get more profit instead of electricity bill
but no it s not good for life of your miner

[blissz]

Hi Blissz

1) on version from 21/11 for some reason i dont have your logo

Here are my parameters on overclocking

Voltage to max, fan speed 100% frequency 650 mhz

https://imgur.com/a/dCJx4

about the logo: press ctrl + f5

But man that's some serious hashrate a new record guys?
I hope your electricity is free as it will draw at least 1500 watts of power I guess.
I don't advice to use it 24/7 at this voltage as chances increases something will break.
Furthermore chain 1 is really struggling (look at the HW errors). I advice to lower the frequency on that chain.

My advice: try to aim for 575Mhz / 600Mhz max as daily driver. It will still give you 21Mh/s + speeds, and it's better on the long run

[blissz]

23/11 test UPDATE (2)

Hi guys, I need some feedback on this version before I put it to the topic start. It adds quite some experimental features, so please let me know if it works ok for you or not.


https://mega.nz/#!nlRjQJQR!QSLT_LCz6JAeUdzX4vHtz2aI5Kr1i724TrZLLGaXzYk


Changes 23/11: (compared to 21/11)
- fast mining startup after changing settings / reboot (~19 seconds vs ~50 seconds in the previous firmwares). Safes you some time when experimenting with settings.
- adds the reboot on low hashrate option (checks every 2 minutes and kicks in when the hashrate is measured below 88% of the expected hashrate)
- ~ 1% improvement on effective average hashrate *jeeej*
- fixed flash upgrade speed to new versions
- added AUTO FAN profiles. The default profile is the one that's in since the start, the performance is more like the stock bitmain auto fan profile and the silent profile keeps things nice and quiet. Manual mode is for setting a static fan speed.



know issues:
- switching pools can sometimes cause the miner to wait for work for 10 seconds.

happy mining!

[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!

Hi  Blissz,

Any chance you can give me some ideas on why my temp seems messed up after trying your firmware and rollback to the Bitmain one? I am currently underclocking it to keep the temps down.

[dalat]

how I can fix the fan at 10 or 25% before install the beta firmware?

[kristikun]

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?

I changed to those - https://imgur.com/a/VBkMS  - but the voltage was 20 before  i lowered it for less temp  its 17 atm

[Jup1t3rR]

23/11 test UPDATE (2)

Hi guys, I need some feedback on this version before I put it to the topic start. It adds quite some experimental features, so please let me know if it works ok for you or not.


https://mega.nz/#!nlRjQJQR!QSLT_LCz6JAeUdzX4vHtz2aI5Kr1i724TrZLLGaXzYk


Changes 23/11: (compared to 21/11)
- fast mining startup after changing settings / reboot (~19 seconds vs ~50 seconds in the previous firmwares). Safes you some time when experimenting with settings.
- adds the reboot on low hashrate option (checks every 2 minutes and kicks in when the hashrate is measured below 88% of the expected hashrate)
- ~ 1% improvement on effective average hashrate *jeeej*
- fixed flash upgrade speed to new versions
- added AUTO FAN profiles. The default profile is the one that's in since the start, the performance is more like the stock bitmain auto fan profile and the silent profile keeps things nice and quiet. Manual mode is for setting a static fan speed.



know issues:
- switching pools can sometimes cause the miner to wait for work for 10 seconds.

happy mining!

Hi

i know this question has been asked many times but i will ask directly mate

what are the chances of making this miner work with other algo ?

the reason I ask is that I'm getting D3 end of this month and the current profitability is so poor that I'm thinking of just getting rid of it....

thank you

[banzsolt1]

23/11 test UPDATE (2)

Hi guys, I need some feedback on this version before I put it to the topic start. It adds quite some experimental features, so please let me know if it works ok for you or not.


https://mega.nz/#!nlRjQJQR!QSLT_LCz6JAeUdzX4vHtz2aI5Kr1i724TrZLLGaXzYk


Changes 23/11: (compared to 21/11)
- fast mining startup after changing settings / reboot (~19 seconds vs ~50 seconds in the previous firmwares). Safes you some time when experimenting with settings.
- adds the reboot on low hashrate option (checks every 2 minutes and kicks in when the hashrate is measured below 88% of the expected hashrate)
- ~ 1% improvement on effective average hashrate *jeeej*
- fixed flash upgrade speed to new versions
- added AUTO FAN profiles. The default profile is the one that's in since the start, the performance is more like the stock bitmain auto fan profile and the silent profile keeps things nice and quiet. Manual mode is for setting a static fan speed.



know issues:
- switching pools can sometimes cause the miner to wait for work for 10 seconds.

happy mining!

cheers for this!

[kristikun]

Hi Blissz

1) on version from 21/11 for some reason i dont have your logo

Here are my parameters on overclocking

Voltage to max, fan speed 100% frequency 650 mhz

https://imgur.com/a/dCJx4

U goona burn it pretty fast like dat check mine  - https://imgur.com/a/RrAYh  - stable

[Jup1t3rR]

Hi Blissz

1) on version from 21/11 for some reason i dont have your logo

Here are my parameters on overclocking

Voltage to max, fan speed 100% frequency 650 mhz

https://imgur.com/a/dCJx4

U goona burn it pretty fast like dat check mine  - https://imgur.com/a/RrAYh  - stable

what power usage mate?

[kristikun]

Hi Blissz

1) on version from 21/11 for some reason i dont have your logo

Here are my parameters on overclocking

Voltage to max, fan speed 100% frequency 650 mhz

https://imgur.com/a/dCJx4

U goona burn it pretty fast like dat check mine  - https://imgur.com/a/RrAYh  - stable

what power usage mate?

chain 1 - 20
chain 2 -  19
chain 3 - 17

[BartoszN1990]

guys i try to unpack the packed with 7 zip. i make edit fiew setings and i pack again by 7 zip and i change format when i try to instal packed i have this msg

tar: invalid magic tar: short read

anybody know how to pack this packed again ?

1* Unpack Antminer-D3-20171120_3-blissz.tar.gz -  by 7zip
2* Unpack initramfs.bin.SD -  by 7zip
3* Unpack Angstrom-antminer_m-eglibc-ipk-v2013.06-beaglebone.rootfs.cpio -  by 7zip

edit fiew things on setings for test and now

1* How to pack folder:

Angstrom-antminer_m-eglibc-ipk-v2013.06-beaglebone.rootfs to keep all setings CPIO ? to take

Angstrom-antminer_m-eglibc-ipk-v2013.06-beaglebone.rootfs.cpio

2* How to pack fille:

Angstrom-antminer_m-eglibc-ipk-v2013.06-beaglebone.rootfs.cpio to keep all setings .bin.SD ? to take

initramfs.bin.SD

and last

3* How to pack fille:

am335x-boneblack-bitmainer.dtb
initramfs.bin.SD
runme.sh to keep all setings .tar.gz ? to take

Antminer-D3-20171120_3-blissz.tar.gz


REGARDS !

[blastar]

Thanks for the update Blissz! I will check it later today...

For all those overclocking their D3s: The increase in profits with overclocking vs electricity usage is generally mutually exclusive! Unless you have free or extremely cheap electricity, you won't gain much with overclocking because you will use more electricity. Also, pushing the miner to its limits and let it run like this for a prolonged periods of time might cause it to fail completely! It is much better to reduce the hashing rate a little while reducing energy consumption a lot (and here is where Blissz's version shines!).

I you want to check what makes sense for you, use this site to calculate the most profitable X11 hashrate vs power usage for your location, taking into consideration your local electricity price:
https://www.coinwarz.com/cryptocurrency

Select X11 only and de-select other algorithms. Here is an example for 16Gh hashing rate / 780 Wh. electricity usage / 0.10 cents per kWh electricity price:

https://i.imgur.com/K7Ll5sc.jpg


The result appears like this (Bitcoin is always there for compasrison):
https://i.imgur.com/C9DHx39.jpg