Klondike - 16 chip ASIC Open Source Board - Preliminary

[synapse]

As far as surplus parts go, one would think that TerraHash would lots of extra stuff.
They were going to build K16's before they got a refund on 20,000 ASIC's.
If any one wants to sale a few Avalon ASIC's (like maybe 4 parts) let me know.

Also is anyone interested in continuing this project and adapting it to the new gen 2 Avalon parts?

yeah same here, after parts for 1 k16 and 4 k1's
might be more pain than gain, but i'm keen to make some of these suckers now that i have the chips in hand.

[Vl4dim1r]

As far as surplus parts go, one would think that TerraHash would lots of extra stuff.
They were going to build K16's before they got a refund on 20,000 ASIC's.
If any one wants to sale a few Avalon ASIC's (like maybe 4 parts) let me know.

Also is anyone interested in continuing this project and adapting it to the new gen 2 Avalon parts?

yeah same here, after parts for 1 k16 and 4 k1's
might be more pain than gain, but i'm keen to make some of these suckers now that i have the chips in hand.

I found k16 components through a member on this forum, and chips through another member. If you want I'll ask him if its alright to send you his info, so you can ask about parts.

I am going to be assembling a few of them just to make up some of what I've already spent, and I plan to try a few extreme cooling/overclocking ideas.

[fasmax]

I would like to look into the high hardware error rate for the K16.
If anyone has a few surplus Avalon chips for sale let me know.

[freeworm]

I would like to look into the high hardware error rate for the K16.
If anyone has a few surplus Avalon chips for sale let me know.

steamboat claimed he found the hardware issue. I really hope he can share and contribute to this project

[fasmax]

The consensus is that it is a firmware problem.
But until the actual problem is found no knows for sure.

[TheDroid]

Hi all,

The firmware issue was found and fixed a few hours ago!   I've confirmed it on my test unit as have others.  ZERO HW errors and over 5gh/s @350MHz.  The K16's are a go!

I think there will be some code cleanup, boot loader and a few other more minor things yet, but I suspect all the assemblers will start pumping them out now.

Thank-you to those who kept working on the project and getting things working!!

Message me if you need K16 assembly.  I'll have parts/PCB's, not ASIC's.  I'm in Canada (west coast) and can do assembly for USA/CAN customers.

[BitCsByBit]

Hi all,

The firmware issue was found and fixed a few hours ago!   I've confirmed it on my test unit as have others.  ZERO HW errors and over 5gh/s @350MHz.  The K16's are a go!

I think there will be some code cleanup, boot loader and a few other more minor things yet, but I suspect all the assemblers will start pumping them out now.

Thank-you to those who kept working on the project and getting things working!!

Message me if you need K16 assembly.  I'll have parts/PCB's, not ASIC's.  I'm in Canada (west coast) and can do assembly for USA/CAN customers.

Great news.

It is awesome to see that even with all the obstacles this project has been successfully finalised.

[freeworm]

Hi all,

The firmware issue was found and fixed a few hours ago!   I've confirmed it on my test unit as have others.  ZERO HW errors and over 5gh/s @350MHz.  The K16's are a go!

I think there will be some code cleanup, boot loader and a few other more minor things yet, but I suspect all the assemblers will start pumping them out now.

Thank-you to those who kept working on the project and getting things working!!

Message me if you need K16 assembly.  I'll have parts/PCB's, not ASIC's.  I'm in Canada (west coast) and can do assembly for USA/CAN customers.

It's exciting. We have been always waiting for this good news

[savetree]

Hi all,

The firmware issue was found and fixed a few hours ago!   I've confirmed it on my test unit as have others.  ZERO HW errors and over 5gh/s @350MHz.  The K16's are a go!

I think there will be some code cleanup, boot loader and a few other more minor things yet, but I suspect all the assemblers will start pumping them out now.

Thank-you to those who kept working on the project and getting things working!!

Message me if you need K16 assembly.  I'll have parts/PCB's, not ASIC's.  I'm in Canada (west coast) and can do assembly for USA/CAN customers.

Great
Where can we find newest firmware sourcecode?

[Ente]

Hi all,

The firmware issue was found and fixed a few hours ago!   I've confirmed it on my test unit as have others.  ZERO HW errors and over 5gh/s @350MHz.  The K16's are a go!

I think there will be some code cleanup, boot loader and a few other more minor things yet, but I suspect all the assemblers will start pumping them out now.

Thank-you to those who kept working on the project and getting things working!!

Message me if you need K16 assembly.  I'll have parts/PCB's, not ASIC's.  I'm in Canada (west coast) and can do assembly for USA/CAN customers.

Absolutely awesome! :-)
That's great news indeed!

Out of curiosity: What exactly is the problem? In simple terms? :-)

Ente

[emanymton]

I would like to look into the high hardware error rate for the K16.
If anyone has a few surplus Avalon chips for sale let me know.

I have 22 chips left from my sample pack. Would be happy to let them go for a nominal fee + shipping since their going to waste.

[zipiju]

You can download sources here https://github.com/zipiju/k16-firmware. For now just sourcefiles, I'll add compiled hex as soon as git starts cooperating with me .
The issue was probably some kind of overflow in PIC, moving a function to different place in a code helped to get rid of all hw errors + there were some little modifications handling receiver reset and baud-rate generator.
It was running with 333MHz here for about 6 hours, the stats were:

KLN 0: 333MHz 100% 55C | 4.472G/5.345Gh/s | A:29434 R:50 HW:1 WU:74.7/m

I needed to take a rest, so no further tests so far, but I'm planning to push it up to 400MHz now.

[freeworm]

You can download sources here https://github.com/zipiju/k16-firmware. For now just sourcefiles, I'll add compiled hex as soon as git starts cooperating with me .
The issue was probably some kind of overflow in PIC, moving a function to different place in a code helped to get rid of all hw errors + there were some little modifications handling receiver reset and baud-rate generator.
It was running with 333MHz here for about 6 hours, the stats were:

KLN 0: 333MHz 100% 55C | 4.472G/5.345Gh/s | A:29434 R:50 HW:1 WU:74.7/m

I needed to take a rest, so no further tests so far, but I'm planning to push it up to 400MHz now.

Great work! Thanks a lot.

[jasinlee]

Wait for the updated f/w the group of devs were up all night working on it. I think they solved it around 3 a.m. let them sleep then we can see what they come up with now that they have a functioning source.

[cardcomm]

Very, very cool. Nice work guys.

Even though I refunded my chips, I'm still glad to see progress on the boards. After following the thread for so long, it'll be nice to know the design is working, even if I never see any.

[synapse]

Latest steamboat email:

Update:

Assembly status:
We have completed both board level and final assembly. Unit testing has been delayed due to firmware issues and lack of ktest utility. We have been testing with rough values using the old firmware, which will assist us in expediting final testing, which expect to complete within two days.

The firmware bug causing high error rates has been discovered and corrected as of early this morning. We are continuing to run tests, but believe the base firmware is now stable. We have also greatly reduced the performance issues when connecting multiple units within CGminer. Currently, we are running stable at 300mhz at 4.9gh/s, with less than 0.5% errors.

We are continuing to work on the USB bootloader, which we hope to have finalized this week. Once the USB bootloader has been completed, we will begin work on the I2C connection and optimizing the firmware, while continuing to optimize CGminer.

We have received requests to ship the units in their current state, without the ability to update the firmware via USB. We are now accepting requests to ship units as-is, but advise against it. This option should only be considered if you have a large order and are comfortable with programming. Currently, the only method of flashing the units is to use a Pickit3 Pic programmer with a suggested pogo pin adapter, connected to a host running MPLAB IPE/IDE. This is a fairly complex process for those unfamiliar with programming, and materials costs are ~$80. Units shipped without the USB bootloader will require initially flashing the units using the on board ICSP pads to update the firmware to allow USB updates.

If you would like to request your units shipped as-is, please ensure the shipping address provided on your liquidsyndesigns.com account is correct, and reply to this email with the subject “Please ship my units without USB bootloader”.

Assembly refunds:

We have been working on the assembly refund process as time permits, and will be sending an email with instructions in the next few days. We appreciate your patience in this matter.

[kano]

Or even a little better with 333 MHz
On my local p2pool coz p2pool is easy to do 1diff for testing with low variance.

Code:

 cgminer version 3.6.4+ - Started: [2013-10-24 09:44:23]
--------------------------------------------------------------------------------
 (5s):12.18G (avg):10.55Gh/s | A:105775  R:1436  HW:8  WU:147.4/m
 ST: 2  SS: 0  NB: 106  LW: 125926  GF: 0  RF: 0
 Connected to z diff 1 with stratum as user z/+1
 Block: 000f77f2ed978a3d...  Diff:268M  Started: [21:50:21]  Best share: 65.8K
--------------------------------------------------------------------------------
 [P]ool management [S]ettings [D]isplay options [Q]uit
 KLN 0: 333MHz  0% 47C | 6.219G/5.274Gh/s | A:52904 R:692 HW:5 WU: 73.7/m
 KLN 1: 333MHz  0% 44C | 7.036G/5.276Gh/s | A:52873 R:744 HW:3 WU: 73.7/m
--------------------------------------------------------------------------------
 [2013-10-24 21:51:48] Accepted fd785d9f Diff 1/1 KLN 0
 [2013-10-24 21:51:48] Accepted bd4757bf Diff 1/1 KLN 1
 [2013-10-24 21:51:49] Accepted d1596286 Diff 1/1 KLN 1
 [2013-10-24 21:51:49] Accepted dde792b4 Diff 1/1 KLN 1

Doing my good old "Paid GH/s" calculation:

Elapsed = 12hr 7min 26sec = 43646 sec

Paid H/s = 2^32 * Diff A / Elapsed = 2^32 * 105775 / 43646 = 10,408,746,866 or 10.4 Paid GH/s

So each KLN is 5.2 Paid GH/s at 333 Mhz

Also note HW% = 8 / (8 + 1436 + 105775) = a tiny 0.0075%

Edit: of course that Rejects would be lower on a normal pool - they are of course higher due to p2pool 30s LP time.

Edit2: N.B. the code is current of my git and I've committed it to ckolivas' git master also - for anyone interested
BKK's code with lots of little changes ... all over the place

[cardcomm]

Or even a little better with 333 MHz
On my local p2pool coz p2pool is easy to do 1diff for testing with low variance.

Code:

 cgminer version 3.6.4+ - Started: [2013-10-24 09:44:23]
--------------------------------------------------------------------------------
 (5s):12.18G (avg):10.55Gh/s | A:105775  R:1436  HW:8  WU:147.4/m
 ST: 2  SS: 0  NB: 106  LW: 125926  GF: 0  RF: 0
 Connected to z diff 1 with stratum as user z/+1
 Block: 000f77f2ed978a3d...  Diff:268M  Started: [21:50:21]  Best share: 65.8K
--------------------------------------------------------------------------------
 [P]ool management [S]ettings [D]isplay options [Q]uit
 KLN 0: 333MHz  0% 47C | 6.219G/5.274Gh/s | A:52904 R:692 HW:5 WU: 73.7/m
 KLN 1: 333MHz  0% 44C | 7.036G/5.276Gh/s | A:52873 R:744 HW:3 WU: 73.7/m
--------------------------------------------------------------------------------
 [2013-10-24 21:51:48] Accepted fd785d9f Diff 1/1 KLN 0
 [2013-10-24 21:51:48] Accepted bd4757bf Diff 1/1 KLN 1
 [2013-10-24 21:51:49] Accepted d1596286 Diff 1/1 KLN 1
 [2013-10-24 21:51:49] Accepted dde792b4 Diff 1/1 KLN 1

Doing my good old "Paid GH/s" calculation:

Elapsed = 12hr 7min 26sec = 43646 sec

Paid H/s = 2^32 * Diff A / Elapsed = 2^32 * 105775 / 43646 = 10,408,746,866 or 10.4 Paid GH/s

So each KLN is 5.2 Paid GH/s at 333 Mhz

Also note HW% = 8 / (8 + 1436 + 105775) = a tiny 0.0075%

Edit: of course that Rejects would be lower on a normal pool - they are of course higher due to p2pool 30s LP time.

Edit2: N.B. the code is current of my git and I've committed it to ckolivas' git master also - for anyone interested
BKK's code with lots of little changes ... all over the place

Very, very sweet! I'm excited to see this, even though I refunded for my Avalon order :/

Thanks from everyone for your hard work on this, Kano.

[eroxors]

I'd like to see these go FASTER! The Hex16a's do 7.2, correct? at 450mhz?

[loshia]

I'd like to see these go FASTER! The Hex16a's do 7.2, correct? at 450mhz?

No way dude power supply of klondike can not stand to such high load and over voltage is needed
Let me tell you hex16a clocked to 500 with voltage set to 1.48 is making about 7700