Klondike - 16 chip ASIC Open Source Board - Preliminary

[zipiju]

55 nm is starting auction tomorrow.  I wonder if they're pin compatible and if anyone will trust Avalon again or if we'll just let greed take over.

They're not, Avalon released datasheet for them a while ago. And they are using different clock frequency and different voltages.

[1714047515]

1714047515

[1714047515]

1714047515

[1714047515]

1714047515

[1714047515]

1714047515

[1714047515]

1714047515

[vs3]

Guys - thank you for working on and seeing this project to completion!

By the way - did anyone get any news on Chris (a.k.a. BKKCoins)?


And to sort of continue his legacy and since my design was inspired to a great extent from the K1 Nano - I've also released the NanoFury designs as Open Source : https://bitcointalk.org/index.php?topic=321287

[TomKeddie]

They're not, Avalon released datasheet for them a while ago. And they are using different clock frequency and different voltages.

There's still a lot of components to rescue though perhaps?  New PCB, a few resistors to set the voltage and a different oscillator?

From glancing at it, it looked like the comms protocol was similar but I might be wrong.

The real problem is that the rapid growth in the network makes anything like this too risky.  It will be interesting to see what happens to this auction.

[zipiju]

They're not, Avalon released datasheet for them a while ago. And they are using different clock frequency and different voltages.

There's still a lot of components to rescue though perhaps?  New PCB, a few resistors to set the voltage and a different oscillator?

From glancing at it, it looked like the comms protocol was similar but I might be wrong.

The real problem is that the rapid growth in the network makes anything like this too risky.  It will be interesting to see what happens to this auction.

Yes it seems to be identical (at least timing is and anything else can be changed in firmware).
So basically only Avalon footprint (pinout) has to be changed (so small modifications to the board around ASICs), then you will have to replace the oscillator for 25MHz one and power regs for PLL (because 55nm Avalon now uses 1.0 instead of 1.2V).
Core voltage is now 0.9V, but since IR3895 can go as low as 0.5V output, no problem with that (you will only change one/two resistors as you're saying).
I'm planning to modify the project, but I'm probably not going to build one (maybe just one board if I'll be able to get a few chips from somewhere - and it wont be an auction).

[Unacceptable]

They're not, Avalon released datasheet for them a while ago. And they are using different clock frequency and different voltages.

There's still a lot of components to rescue though perhaps?  New PCB, a few resistors to set the voltage and a different oscillator?

From glancing at it, it looked like the comms protocol was similar but I might be wrong.

The real problem is that the rapid growth in the network makes anything like this too risky.  It will be interesting to see what happens to this auction.

Exactly,anything larger than 28nm,will be obsolete in a few months  

Why didn't Avalon go straight to 28nm    Not very smart,if you ask me    Plus,I will not give Avalon any money/BTC ever due thier lying  

[roybitcoin]

Hi!

Some time ago I buy some samples from Zefir in order to begin with a personal project based in K1 desing single ASIC. After the Avalon fail, I stoped this project.

Some weeks ago I returned to work in this project with a friend (just for fun  ) and we want to know if the firmware released by Zipiju for the K16 and the last Cgminer Klondike driver have a chance to work with a K1 design with some changes?

We have the Hardware working. We have reviewed the last firmware and made some changes in the klondike.c file in order to work with only one Asic and it´s working with the ktest utility from Bkkcoins but it´s not working with Cgminer.

We will release all the design files if someone interested.

Best regards!

Roy

[fasmax]

Hi!

Some time ago I buy some samples from Zefir in order to begin with a personal project based in K1 desing single ASIC. After the Avalon fail, I stoped this project.

Some weeks ago I returned to work in this project with a friend (just for fun  ) and we want to know if the firmware released by Zipiju for the K16 and the last Cgminer Klondike driver have a chance to work with a K1 design with some changes?

We have the Hardware working. We have reviewed the last firmware and made some changes in the klondike.c file in order to work with only one Asic and it´s working with the ktest utility from Bkkcoins but it´s not working with Cgminer.

We will release all the design files if someone interested.

Best regards!

Roy

What does it do in CGM ?

[zipiju]

Hi!

Some time ago I buy some samples from Zefir in order to begin with a personal project based in K1 desing single ASIC. After the Avalon fail, I stoped this project.

Some weeks ago I returned to work in this project with a friend (just for fun  ) and we want to know if the firmware released by Zipiju for the K16 and the last Cgminer Klondike driver have a chance to work with a K1 design with some changes?

We have the Hardware working. We have reviewed the last firmware and made some changes in the klondike.c file in order to work with only one Asic and it´s working with the ktest utility from Bkkcoins but it´s not working with Cgminer.

We will release all the design files if someone interested.

Best regards!

Roy

Yes, it should work out of the box with just two modifications - nonce range (one in this case) and no work splitting between two banks.
But if you're using different PIC port and pins than the ones that are used in K16 for ASIC config, you have to edit also SEND_ONE and SEND_ZERO defines in klondike.h (to which value it depends on port and pins used - K1 design maybe uses different pins as in K16 - it can be looked up in schematic).
I have couple of things on to-do list now with K1 firmware, but I'm quite busy last few days, so no progress on that so far.
And btw, what does the CGM says with that board connected?

[chadtn]

I just pushed up updated firmware with 333MHz default clock. It should be more stable at the start, so try that. Then you can use cgm config to clock it higher.

I've got "klondike-options" : "350,50" in my config file and everything works fine under Windows.  Averaging about 5.6 gigahashes with no hardware errors and temps hovering around 49-50C.  If I change it from 350 to 375 my K16's won't hash.  They show up in cgminer, but sit at 0 doing nothing.  Any idea what's going on there?  I just tried flashing the October 31st firmware from your git, but that didn't seem to change anything.

Thanks!

Chad

[zipiju]

I just pushed up updated firmware with 333MHz default clock. It should be more stable at the start, so try that. Then you can use cgm config to clock it higher.

I've got "klondike-options" : "350,50" in my config file and everything works fine under Windows.  Averaging about 5.6 gigahashes with no hardware errors and temps hovering around 49-50C.  If I change it from 350 to 375 my K16's won't hash.  They show up in cgminer, but sit at 0 doing nothing.  Any idea what's going on there?  I just tried flashing the October 31st firmware from your git, but that didn't seem to change anything.

Thanks!

Chad

The current draw from the 1.2V regulators is probably too high and they are shutting down as soon as clock signal is enabled to the chips - which is with first work item. So try something lower, like 370 or 365 and it should work.

[Bogart]

I just pushed up updated firmware with 333MHz default clock. It should be more stable at the start, so try that. Then you can use cgm config to clock it higher.

I've got "klondike-options" : "350,50" in my config file and everything works fine under Windows.  Averaging about 5.6 gigahashes with no hardware errors and temps hovering around 49-50C.  If I change it from 350 to 375 my K16's won't hash.  They show up in cgminer, but sit at 0 doing nothing.  Any idea what's going on there?  I just tried flashing the October 31st firmware from your git, but that didn't seem to change anything.

Thanks!

Chad

I ran into that problem running the git master somewhere between cgminer 3.6.6 and 3.7.0, where no speed other than 350 would work.  The bug may have persisted into 3.7.0 (I haven't tested it yet).

If I use 3.6.6, I am able to adjust the clockspeed.

[chadtn]

I just pushed up updated firmware with 333MHz default clock. It should be more stable at the start, so try that. Then you can use cgm config to clock it higher.

I've got "klondike-options" : "350,50" in my config file and everything works fine under Windows.  Averaging about 5.6 gigahashes with no hardware errors and temps hovering around 49-50C.  If I change it from 350 to 375 my K16's won't hash.  They show up in cgminer, but sit at 0 doing nothing.  Any idea what's going on there?  I just tried flashing the October 31st firmware from your git, but that didn't seem to change anything.

Thanks!

Chad

I ran into that problem running the git master somewhere between cgminer 3.6.6 and 3.7.0, where no speed other than 350 would work.  The bug may have persisted into 3.7.0 (I haven't tested it yet).

If I use 3.6.6, I am able to adjust the clockspeed.

I think zipiju sorted out my problem with my rpi crashing.  I'm experimenting with clock speed right now with cgminer 3.7.0.  Anything over 360 and my K16 won't hash.  360 and under my K16 will hash, but very slowly.  At 360 I was getting around 700 megahashes.  I've dropped it down 1mhz at time until I got to 355 where the hash rate climbed up to a little over 2 gigahashes.  Right now I"m just testing 351mhz and it appears to be working at full speed.

Chad

[chadtn]

Right now it looks like 352mhz is as fast as I can go before the hash rate starts to drop.  On very short runs (~around five minutes) the K16 was hashing around 5.8 gh to 5.9gh at 352mhz.  At 353mhz that drops down to about 4.8 gh.  I just noticed cgminer 3.7.2 has been released.  I'll start building that here in a minute and see if that changes anything.  I'm running 3.7.0 right now.

Chad

[Vl4dim1r]

Does anyone have 10 x MCP16321T-330E/NGCT-ND they could sell me? PM me your price with rush shipping to USA-Arkansas, 72118.

[roybitcoin]

Hi!

Some time ago I buy some samples from Zefir in order to begin with a personal project based in K1 desing single ASIC. After the Avalon fail, I stoped this project.

Some weeks ago I returned to work in this project with a friend (just for fun   ) and we want to know if the firmware released by Zipiju for the K16 and the last Cgminer Klondike driver have a chance to work with a K1 design with some changes?

We have the Hardware working. We have reviewed the last firmware and made some changes in the klondike.c file in order to work with only one Asic and it´s working with the ktest utility from Bkkcoins but it´s not working with Cgminer.

We will release all the design files if someone interested.

Best regards!

Roy

Yes, it should work out of the box with just two modifications - nonce range (one in this case) and no work splitting between two banks.
But if you're using different PIC port and pins than the ones that are used in K16 for ASIC config, you have to edit also SEND_ONE and SEND_ZERO defines in klondike.h (to which value it depends on port and pins used - K1 design maybe uses different pins as in K16 - it can be looked up in schematic).
I have couple of things on to-do list now with K1 firmware, but I'm quite busy last few days, so no progress on that so far.
And btw, what does the CGM says with that board connected?

Thank you very much for your reply!

We changed the defines according the schematic because they are different from K16 design. Another thing is we don´t use the inverters in the clock line, just use the NOR gates, but it´s wroking ok.

We have this from CGM:

[2013-11-04 01:21:30] Pool 0 http://stratum.btcguild.com:3333 alive
 [2013-11-04 01:21:30] KLN0:0 Send cmd [S ]
 [2013-11-04 01:21:30] KLN0:0 Reply status [S ] dev=0 chips=1 slaves=0 workcq=3 workid=0 temp=0 fan=0 errors=0 hashes=16163 max=0 noise=0
 [2013-11-04 01:21:30] KLN0:0 Send config [C] dev=0 clock=0 temptarget=0 tempcrit=0 fan=0
 [2013-11-04 01:21:30] KLN0:0 Reply config [C] dev=0 clock=128 temptarget=60 tempcrit=80 fan=255
 [2013-11-04 01:21:31] KLN0:0 config (0: Clk: 128, T:17, C:25, F:99)
 [2013-11-04 01:21:31] KLN0:0 Send cmd [S ]
 [2013-11-04 01:21:31] KLN0:0 Reply status [S ] dev=0 chips=1 slaves=0 workcq=3 workid=0 temp=0 fan=0 errors=0 hashes=16265 max=0 noise=0
 [2013-11-04 01:21:31] KLN0:0 Send enable [E] enable=1
 [2013-11-04 01:21:31] KLN0:0 Reply status [E] dev=0 chips=1 slaves=0 workcq=3 workid=0 temp=0 fan=0 errors=0 hashes=16316 max=0 noise=0
 [2013-11-04 01:21:31] Work update message received
 [2013-11-04 01:21:31] Network diff set to 391M
 [2013-11-04 01:21:31] KLN0:0 Send cmd [A]
 [2013-11-04 01:21:31] New block: 000636636dbb6fd9... diff 391M
 [2013-11-04 01:21:31] KLN0:0 Reply status [A] dev=0 chips=1 slaves=0 workcq=0 workid=0 temp=0 fan=0 errors=0 hashes=16316 max=0 noise=0
 [2013-11-04 01:21:31] KLN0:0 Send work [W] dev=0 workid=0x00 ...
 [2013-11-04 01:21:31] KLN0:0 Reply status [W] dev=0 chips=1 slaves=0 workcq=0 workid=0 temp=0 fan=0 errors=0 hashes=0 max=0 noise=0
 [2013-11-04 01:21:31] KLN0:0 Send work [W] dev=0 workid=0x01 ...
 [2013-11-04 01:21:31] KLN0:0 Reply status [W] dev=0 chips=1 slaves=0 workcq=1 workid=0 temp=0 fan=0 errors=0 hashes=49 max=0 noise=0
 [2013-11-04 01:21:31] Popping work from get queue to get work
 [2013-11-04 01:21:31] Got work from get queue to get work for thread 0
 [2013-11-04 01:21:31] Selecting pool 0 for work
 [2013-11-04 01:21:31] Generated stratum merkle 7618c59587bd4f535f6199068ad972819c9f8dd87ec8d48e9460ff1e4f9f661d
 [2013-11-04 01:21:31] Generated stratum header 00000002d44f95c3062241d9984ce5b44dbcba426fd9a9cf36636dbb00000006000000007618c59 587bd4f535f6199068ad972819c9f8dd87ec8d48e9460ff1e4f9f661d52788106190afc85000000 000000008000000000000000000000000000000000000000000000000000000000000000
 [2013-11-04 01:21:31] Work job_id 337957 nonce2 4 ntime 52788106
 [2013-11-04 01:21:31] WORKDATA: 5700022d4a6e55ba38cb634cd0fc216485d7b1f2f6b7d872cfec7505a1792d7b7ace656a7767555 2788106190afc85
 [2013-11-04 01:21:31] Generated target 0000000000000000000000000000000000000000000000000080ff7f00000000
 [2013-11-04 01:21:31] KLN0:0 sending work (0:02)
 [2013-11-04 01:21:31] Generated stratum work
 [2013-11-04 01:21:31] KLN0:0 Send work [W] dev=0 workid=0x02 ...
 [2013-11-04 01:21:31] Pushing work from pool 0 to hash queue
 [2013-11-04 01:21:31] KLN0:0 Reply status [W] dev=0 chips=1 slaves=0 workcq=2 workid=0 temp=0 fan=0 errors=0 hashes=106 max=0 noise=0
 [2013-11-04 01:21:31] Popping work from get queue to get work
 [2013-11-04 01:21:31] Got work from get queue to get work for thread 0
 [2013-11-04 01:21:31] Selecting pool 0 for work
 [2013-11-04 01:21:31] Generated stratum merkle f9dd78e4eb7dbfc13efc05f32397d00569f481bf21e1b205f03120de62afd52e
 [2013-11-04 01:21:31] Generated stratum header 00000002d44f95c3062241d9984ce5b44dbcba426fd9a9cf36636dbb0000000600000000f9dd78e 4eb7dbfc13efc05f32397d00569f481bf21e1b205f03120de62afd52e52788106190afc85000000 000000008000000000000000000000000000000000000000000000000000000000000000
 [2013-11-04 01:21:31] Work job_id 337957 nonce2 5 ntime 52788106
 [2013-11-04 01:21:31] Generated target 0000000000000000000000000000000000000000000000000080ff7f00000000
 [2013-11-04 01:21:31] WORKDATA: 570003955bd73295580dcc55b321716ea7c54fa9ec3e9c35547e710bf165cfcdec6540683003a85 2788106190afc85
 [2013-11-04 01:21:31] Generated stratum work
 [2013-11-04 01:21:31] KLN0:0 sending work (0:03)
 [2013-11-04 01:21:31] Pushing work from pool 0 to hash queue
 [2013-11-04 01:21:31] KLN0:0 Send work [W] dev=0 workid=0x03 ...
 [2013-11-04 01:21:31] KLN0:0 Reply status [W] dev=0 chips=1 slaves=0 workcq=3 workid=0 temp=0 fan=0 errors=0 hashes=159 max=0 noise=0
 [2013-11-04 01:21:31] Popping work from get queue to get work
 [2013-11-04 01:21:31] Selecting pool 0 for work
 [2013-11-04 01:21:31] Generated stratum merkle ef6010385620b8111ba19a66a7d8e0e37d7cf1eed1f3d59f829baa52f3545483
 [2013-11-04 01:21:31] Got work from get queue to get work for thread 0
 [2013-11-04 01:21:31] Generated stratum header 00000002d44f95c3062241d9984ce5b44dbcba426fd9a9cf36636dbb0000000600000000ef60103 85620b8111ba19a66a7d8e0e37d7cf1eed1f3d59f829baa52f354548352788106190afc85000000 000000008000000000000000000000000000000000000000000000000000000000000000
 [2013-11-04 01:21:31] Work job_id 337957 nonce2 6 ntime 52788106
 [2013-11-04 01:21:31] Generated target 0000000000000000000000000000000000000000000000000080ff7f00000000
 [2013-11-04 01:21:31] Generated stratum work
 [2013-11-04 01:21:31] Pushing work from pool 0 to hash queue
 [2013-11-04 01:21:31] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:32] [thread 0: 0 hashes, 0.0 khash/sec]
 [2013-11-04 01:21:32] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:32] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:32] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:32] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:32] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:33] KLN0: GetResults (amt=0 err=-7 ern=25)
 [2013-11-04 01:21:33] KLN0: getting status
 [2013-11-04 01:21:33] KLN0:0 Send cmd [S ]
 [2013-11-04 01:21:33] KLN0:0 Reply status [S ] dev=0 chips=1 slaves=0 workcq=3 workid=0 temp=0 fan=0 errors=0 hashes=927 max=0 noise=0

And some time later:

[2013-11-04 01:21:50] Work stale due to stratum job_id mismatch
 [2013-11-04 01:21:50] Discarded work
 [2013-11-04 01:21:50] Work stale due to stratum job_id mismatch
 [2013-11-04 01:21:50] Discarded work
 [2013-11-04 01:21:50] Selecting pool 0 for work
 [2013-11-04 01:21:50] Discarded 2 stales that didn't match current hash
 [2013-11-04 01:21:50] Generated stratum merkle bf419eb5061c716515e84e44a62db782e31b7f26e2af7538f17a03c89ef1f675
 [2013-11-04 01:21:50] Generated stratum header 00000002d44f95c3062241d9984ce5b44dbcba426fd9a9cf36636dbb0000000600000000bf419eb 5061c716515e84e44a62db782e31b7f26e2af7538f17a03c89ef1f67552788124190afc85000000 000000008000000000000000000000000000000000000000000000000000000000000000
 [2013-11-04 01:21:50] Work job_id 337960 nonce2 7 ntime 52788124
 [2013-11-04 01:21:50] Generated target 0000000000000000000000000000000000000000000000000080ff7f00000000

Thanks in advance for any advise!

Roy

[LiveJay]

I just submitted a pull request at GitHub for a fix I made to the klondike driver in cgminer.  It was incorrectly reporting the target fan speed on the status line (always 0% due to a math error).  Here is the diff if any of you adventurous souls out there want to try it out:

https://github.com/LiveJay/cgminer/commit/0124217db80cd4506d4fe1c921a6fd49bba59efb

Now it looks like this (it's capped at 99% because there isn't room for a fourth character):

https://photos-3.dropbox.com/t/0/AABLHwHkiYxSag3F8H6E3NUiyuUxGv8kqWewpDi4MHqkDA/12/44288815/png/1024x768/3/1384027200/0/2/Screenshot%202013-11-09%2010.25.55.png/eip0CZO0no5uOh5M9nwMA62GLBvLG4MNn1qJHuyykIE

My question is this:  The firmware aways seems to be returning 255 or 100% for the fan target speed, even when the fan is obviously running slower.  Is there any current activity around this in the firmware?  Is the firmware source available for review?  And where are the firmware devs hanging out.  I wouldn't mind listening in or helping out.

Thanks,

-Jay

[ScaryHash]

For whatever reason, my klondikes say

KLN2 went idle before work was sent

over and over again.

No idea what is going on, using CGminer 3.7.2

[zipiju]

I just submitted a pull request at GitHub for a fix I made to the klondike driver in cgminer.  It was incorrectly reporting the target fan speed on the status line (always 0% due to a math error).  Here is the diff if any of you adventurous souls out there want to try it out:

https://github.com/LiveJay/cgminer/commit/0124217db80cd4506d4fe1c921a6fd49bba59efb

Now it looks like this (it's capped at 99% because there isn't room for a fourth character):



My question is this:  The firmware aways seems to be returning 255 or 100% for the fan target speed, even when the fan is obviously running slower.  Is there any current activity around this in the firmware?  Is the firmware source available for review?  And where are the firmware devs hanging out.  I wouldn't mind listening in or helping out.

Thanks,

-Jay

Fantarget was dropped in latest firmware and cgm driver, since there is a temptarget. Because of compatibility with older firmware, the fantarget is set by cgm driver to 255.
If you want to monitor an actual fan speed (or current pwm level, since fan rpm measurement is not implemented in firmware yet), you have to take a look at klninfo->status[dev].kline.ws.fanspeed variable.
You can review K16 firmware at github.com/zipiju/k16-firmware.

[zipiju]

For whatever reason, my klondikes say

KLN2 went idle before work was sent

over and over again.

No idea what is going on, using CGminer 3.7.2

Take a look at http://projectklondike.org/how-to-run#went-idle

[ScaryHash]

For whatever reason, my klondikes say

KLN2 went idle before work was sent

over and over again.

No idea what is going on, using CGminer 3.7.2

Take a look at http://projectklondike.org/how-to-run#went-idle

Thank you for the suggestion Zipiju, I had thought that only applied to raspberry-pi, but I tried it, and it did help quite a bit, on Linux, where I can compile it easily.

On my windows machine, it's another story, but it seems to be working ok.

It seems if you add too may K16 too fast, cgminer gets confused. Best to add them one at a time, while keeping cgminer waiting on hotplug. At least, that works for me.