Klondike - 16 chip ASIC Open Source Board - Preliminary

[steamboat]

BkkCoins will be receiving at minimum 16 chips from Batch 1. If Batch 2 samples ship concurrently, he will receive those as well.

[1714881410]

1714881410

[1714881410]

1714881410

[1714881410]

1714881410

[1714881410]

1714881410

[1714881410]

1714881410

[1714881410]

1714881410

[wrenchmonkey]

Ditto. Make it fit in a rack with a nice clean look. I want IT hardware, not a 1960s entertainment center.

Makes me think of:

 

How about this



This picture is not my room nor is this what I want it to look like, but you get the idea now. Something I can be proud to display.

Something tells me that the people you're hoping to impress with this are the same types who are impressed by skull shifter knobs, with blue LEDs in them? None of my friends are still in high school, so they wouldn't be very impressed with a $600 shiny box. Especially not with wood trim.

As for me, I've become unimpressed with the tawdry in my 'old' age. I see elegance in simplicity and neatness. Give me a clean, organized 'squared away' look, over wood-paneled, or jewel encrusted, or disco-lighted, any day of the week. That extra $600, in my eyes, is another 40 or so chips, including populated boards. I'll take an extra 11 Ghash of function, over vast amounts of "form" any day. As we used to say in the mechanic business, "The chrome won't get you home." That is to say, you're better off putting your money into making your vehicle more functional than you are into making it more pretty. The same holds true for an investment like bitcoin mining hardware. Are you in it to make money, or are you in it to impress people?

This makes money:


This 'impresses' people:


Most people will be more proud of hashing power in the long run. Especially since that's the part that pays you back. Will you still be proud of the after six months, when you've lost $500-$3000/month (depending on difficulty) that you could have made by spending that $600 on hashing power? The answer for me is a resounding HELL NO. To each his own though. I'm not hating. In the end, I'm sure you will build something you and your friends can be proud of.

[BkkCoins]

@zefir,
@steamboat
Thank you guys! That's much anticipated.

@wrenchmonkey
Whoa. Jules Verne does mining.

I thought you could put only 4 chips in a K16 board and it would still work? The board kits themselves don't appear to be that expensive [based on preliminary info]

Yes. The code I'm working with now allows for partial populating. It will be best if both banks are balanced since I count chips and use the highest bank count as divisor. This keeps the work pushing code simpler. If you populate more on one bank than the other, then some range will go untested as the more empty bank will be cut short and get work pushed before done. This isn't so bad.

By adding in pairs 2,4,6,8... they stay fully utilized. This is primarily due to the fact I push work to both banks at the same time so as to not take twice as long. Even with unbalanced banks no chips do repeat work so this is still optimal.

[kano]

@zefir,
@steamboat
Thank you guys! That's much anticipated.

@wrenchmonkey
Whoa. Jules Verne does mining.

I thought you could put only 4 chips in a K16 board and it would still work? The board kits themselves don't appear to be that expensive [based on preliminary info]

Yes. The code I'm working with now allows for partial populating. It will be best if both banks are balanced since I count chips and use the highest bank count as divisor. This keeps the work pushing code simpler. If you populate more on one bank than the other, then some range will go untested as the more empty bank will be cut short and get work pushed before done. This isn't so bad.

By adding in pairs 2,4,6,8... they stay fully utilized. This is primarily due to the fact I push work to both banks at the same time so as to not take twice as long. Even with unbalanced banks no chips do repeat work so this is still optimal.

Hopefully this information chips per bank, for each bank, is available, so calculating the hash rate can be done properly

(Edit: I will add I can calculate this using specific nonce test cases if I knew the layout but since the chips are only ~300MH/s it might take a while to test ... i.e. 'seconds')

[BkkCoins]

Hopefully this information chips per bank, for each bank, is available, so calculating the hash rate can be done properly

(Edit: I will add I can calculate this using specific nonce test cases if I knew the layout but since the chips are only ~300MH/s it might take a while to test ... i.e. 'seconds')

My plan is to code a detect function that sequentially feeds a known "golden" data set to each chip and listen for the nonce returned. So this <midstate,data,nonce> would be hard coded into the firmware. Mostly this is to test chip presence and function. However, by using a nonce that is X hashes away from golden, a brief timing of the result could give you actual speed. I wasn't planning this for now as the calculations would be more difficult on a PIC than using pre-calculated tables.

What I have partly coded is an array [8] of nonce base count, timer divider, and tick count values. So when detecting eg. 5 chips per bank I use that value to choose that I'll send N = 2^32 / 2 / 5 as base count for each range. So nonce1 starts 0, nonce 2 starts N, nonce 3 starts N+N, etc. Then I have a counter that runs at 12MHz / 256 / X value where X is set based on hash clock config. This would be chosen to tick very close to every 256,000 hashes and should be reliable. Now I count these ticks, and for current chip count I compare to the table value to decide when to push more work. This should find me pushing new work just before the end of each nonce range is reached, and have a tick count value for work aborts that allows reporting how far we got within 256,000 hashes.

What I could use is a reference <midstate,data,nonce> to code into the firmware. I realize that any new block would have valid values. I thought it may be nice to code in a special one. eg. midstate all zeros or data 0xdeadbeef... etc.

[crazyates]

As for me, I've become unimpressed with the tawdry in my 'old' age. I see elegance in simplicity and neatness. Give me a clean, organized 'squared away' look, over wood-paneled, or jewel encrusted, or disco-lighted, any day of the week. That extra $600, in my eyes, is another 40 or so chips, including populated boards. I'll take an extra 11 Ghash of function, over vast amounts of "form" any day. As we used to say in the mechanic business, "The chrome won't get you home." That is to say, you're better off putting your money into making your vehicle more functional than you are into making it more pretty. The same holds true for an investment like bitcoin mining hardware. Are you in it to make money, or are you in it to impress people?

I just turned 25, but I must be getting "old", as I 100% agree! Never saw the importance of bling if it's an inferior product. Pearls on a pig, they used to say.

Now, I have no problem skimping out on those extra 40 chips, if it means spending that $600 on a proper casing, cabling, grounding, clean power, proper cooling/airflow/venting, etc. Some of those home made zip tie rigs bug me to no end, as while it's cheap, it's also just asking for failure one of these days.

[wrenchmonkey]

As for me, I've become unimpressed with the tawdry in my 'old' age. I see elegance in simplicity and neatness. Give me a clean, organized 'squared away' look, over wood-paneled, or jewel encrusted, or disco-lighted, any day of the week. That extra $600, in my eyes, is another 40 or so chips, including populated boards. I'll take an extra 11 Ghash of function, over vast amounts of "form" any day. As we used to say in the mechanic business, "The chrome won't get you home." That is to say, you're better off putting your money into making your vehicle more functional than you are into making it more pretty. The same holds true for an investment like bitcoin mining hardware. Are you in it to make money, or are you in it to impress people?

I just turned 25, but I must be getting "old", as I 100% agree! Never saw the importance of bling if it's an inferior product. Pearls on a pig, they used to say.

Now, I have no problem skimping out on those extra 40 chips, if it means spending that $600 on a proper casing, cabling, grounding, clean power, proper cooling/airflow/venting, etc. Some of those home made zip tie rigs bug me to no end, as while it's cheap, it's also just asking for failure one of these days.

I'm 29.  I put "old" in scare quotes for a reason. I 100% agree. Spending the money to make it a solid system is certainly worth pursuing. I think you can build a decent rackmount case for under $100 though.

[villex]

Returning to the topic, I would love seeing a design for a case on thingiverse.

[Bicknellski]

BkkCoins will be receiving at minimum 16 chips from Batch 1. If Batch 2 samples ship concurrently, he will receive those as well.

Excellent Steam! +1
and thanks Zefir for the updates and doing a FAIR distribution of dev chips... wish I had been in your chip order or steamboats +1

I wish my group buy would get on this as well... however I think he is keeping all the chips for his own development... not sure yet.

[kano]

Hopefully this information chips per bank, for each bank, is available, so calculating the hash rate can be done properly

(Edit: I will add I can calculate this using specific nonce test cases if I knew the layout but since the chips are only ~300MH/s it might take a while to test ... i.e. 'seconds')

My plan is to code a detect function that sequentially feeds a known "golden" data set to each chip and listen for the nonce returned. So this <midstate,data,nonce> would be hard coded into the firmware. Mostly this is to test chip presence and function. However, by using a nonce that is X hashes away from golden, a brief timing of the result could give you actual speed. I wasn't planning this for now as the calculations would be more difficult on a PIC than using pre-calculated tables.

What I have partly coded is an array [8] of nonce base count, timer divider, and tick count values. So when detecting eg. 5 chips per bank I use that value to choose that I'll send N = 2^32 / 2 / 5 as base count for each range. So nonce1 starts 0, nonce 2 starts N, nonce 3 starts N+N, etc. Then I have a counter that runs at 12MHz / 256 / X value where X is set based on hash clock config. This would be chosen to tick very close to every 256,000 hashes and should be reliable. Now I count these ticks, and for current chip count I compare to the table value to decide when to push more work. This should find me pushing new work just before the end of each nonce range is reached, and have a tick count value for work aborts that allows reporting how far we got within 256,000 hashes.

What I could use is a reference <midstate,data,nonce> to code into the firmware. I realize that any new block would have valid values. I thought it may be nice to code in a special one. eg. midstate all zeros or data 0xdeadbeef... etc.

The Icarus golden nonce I found last year by going through the blockchain (with a php script I wrote)
(I needed one that was fast to test the presence of the Icarus - the Icarus one is 0.53ms on an Icarus)

On the other hand if you want specific case results (but not specific values like you mentioned) it's probably easier to add a few lines of code into cgminer looking for a result "type" you want and spit it out if it ever finds one while mining normally.

My Jalapeno would be good for this since they return all the nonce values (well up to a max of 8 ) for a given work item.
Anyway if you want to be specific about the chip layout and the expected nonce range (I'm not too clear from what you said) I could look for a work item that returns values effectively for each chip and thus I'd know if it only returned say 3 of 4 values, which chip was missing ... as an example.

However, I'd also suggest it's probably easier to divide up the nonce range with a fixed N
As you know, it doesn't matter if you don't use the whole range, and simply sending the work 1/8 to each chip (if there are a max of 8 per bank) and obviously not using all 8 of 1/8 if there is a chip missing, should make things even simpler I'd imagine for you ... and more 'constants', less 'variables'
Again, the only reason I'd need to know the chip count would be if some are missing/faulty to determine the MH/s
Breaking the nonce range up depending on the chip count is OK if it's easy for you to do, but I was just thinking that breaking it up on a constant size would be easier - but that's up to you.

[BkkCoins]

So...
Just ordered first run of 10 test boards for K16.

And SDS7102 Oscilloscope ordered today too, direct from China. They tell me it'll ship by EMS so it should only be about a week to get here if all goes well.

So the way things are looking now I'll have enough to test PIC firmware in a few days. Should have parts needed for testing K1 boards when they arrive in a week, and maybe scope will be here for that. That gives me a week to play with that stuff and get kinks worked out before K16 boards come in, and then I should have everything here to get that scoped and functional.

And if the planets align that will all be just in time for the samples to start arriving, and the real fireworks happen.

To aid with inspection of test and assembled boards I'm going to order one of these USB Microscopes. It looks pretty nice for checking PCB traces and IC pins, and my eyesight is not what it was (as I realized while looking at some boards that I got yesterday). For $60 it seems like it'll be worthwhile.

http://www.youtube.com/watch?feature=player_embedded&v=2te13V_YsPU

Anyone ever use one of these?

[Biomech]

As for me, I've become unimpressed with the tawdry in my 'old' age. I see elegance in simplicity and neatness. Give me a clean, organized 'squared away' look, over wood-paneled, or jewel encrusted, or disco-lighted, any day of the week. That extra $600, in my eyes, is another 40 or so chips, including populated boards. I'll take an extra 11 Ghash of function, over vast amounts of "form" any day. As we used to say in the mechanic business, "The chrome won't get you home." That is to say, you're better off putting your money into making your vehicle more functional than you are into making it more pretty. The same holds true for an investment like bitcoin mining hardware. Are you in it to make money, or are you in it to impress people?

I just turned 25, but I must be getting "old", as I 100% agree! Never saw the importance of bling if it's an inferior product. Pearls on a pig, they used to say.

Now, I have no problem skimping out on those extra 40 chips, if it means spending that $600 on a proper casing, cabling, grounding, clean power, proper cooling/airflow/venting, etc. Some of those home made zip tie rigs bug me to no end, as while it's cheap, it's also just asking for failure one of these days.

I can see the appeal of both sides of the argument, though I fall on the side of the "old" guys on utiitarian vs. ornate. (I'm 44). But I always did prefer simple and clean to cluttered. Don't see it as necessarily an age thing, just taste.

[wrenchmonkey]

As for me, I've become unimpressed with the tawdry in my 'old' age. I see elegance in simplicity and neatness. Give me a clean, organized 'squared away' look, over wood-paneled, or jewel encrusted, or disco-lighted, any day of the week. That extra $600, in my eyes, is another 40 or so chips, including populated boards. I'll take an extra 11 Ghash of function, over vast amounts of "form" any day. As we used to say in the mechanic business, "The chrome won't get you home." That is to say, you're better off putting your money into making your vehicle more functional than you are into making it more pretty. The same holds true for an investment like bitcoin mining hardware. Are you in it to make money, or are you in it to impress people?

I just turned 25, but I must be getting "old", as I 100% agree! Never saw the importance of bling if it's an inferior product. Pearls on a pig, they used to say.

Now, I have no problem skimping out on those extra 40 chips, if it means spending that $600 on a proper casing, cabling, grounding, clean power, proper cooling/airflow/venting, etc. Some of those home made zip tie rigs bug me to no end, as while it's cheap, it's also just asking for failure one of these days.

I can see the appeal of both sides of the argument, though I fall on the side of the "old" guys on utiitarian vs. ornate. (I'm 44). But I always did prefer simple and clean to cluttered. Don't see it as necessarily an age thing, just taste.

You're probably right that it has more to do with personal taste than age. I think it's a maturity thing though. I have to admit, there was a time when I thought ridiculous shit (like chrome shifter knobs with LEDs in them) was cool. That time was 10th-11th grade. Most people grow out of it quicker than some, some never do.  

[nekonos]

Anyone know how the chips are supplied in reel or tray?
In avalon specs do not mention this   

[steamboat]

While I appreciate case mod conversations, can we please keep the off topic chatter to a minimum?

[sensei]

Anyone know how the chips are supplied in reel or tray?
In avalon specs do not mention this  

I'm sure it won't be in a reel; most likely a tray,

The samples might be in a tube.

[Exoskeleton]

I'm not hating.

You have the wrong idea about what I am designing and you are attempting to tear it down for being "bling" when you have no idea whatsoever what it will look like. You imply that I like "ridiculous shit (like chrome shifter knobs with LEDs in them)". I don't drive on 20" rims. I don't have a chrome skull shift knob either. Im not going to encrust my case with diamonds. Im simply making it out of quality parts that happen to look good too.

I can make a case out of cardboard, milk crates, fans and zip ties for $100. If you want to buy anything rack mounted like your picture you will spend at least a few hundred on the case, parts, and the rack to store it in. My design has both a rackable 19" case (metal, not plastic) and a 19" 8U rack to store it in. Just the case alone and the rails for each bank of cards cost me over $300. The 19" rack to store it in is another $100+.

If you are going to keep posting again and again about case design, lets make a thread for that then. As I said before, I don't want to clutter up this thread talking about my case ideas. This is BKKCoins topic for Klondike boards. Sorry BKKCoins, I didn't mean to derail your thread at all. I really appreciate what you are doing for everyone here.

[wrenchmonkey]

I'm not hating.

You have the wrong idea about what I am designing and you are attempting to tear it down for being "bling" when you have no idea whatsoever what it will look like. You imply that I like "ridiculous shit (like chrome shifter knobs with LEDs in them)". I don't drive on 20" rims. I don't have a chrome skull shift knob either. Im not going to encrust my case with diamonds. Im simply making it out of quality parts that happen to look good too.

I can make a case out of cardboard, milk crates, fans and zip ties for $100. If you want to buy anything rack mounted like your picture you will spend at least a few hundred on the case, parts, and the rack to store it in. My design has both a rackable 19" case (metal, not plastic) and a 19" 8U rack to store it in. Just the case alone and the rails for each bank of cards cost me over $300. The 19" rack to store it in is another $100+.

If you are going to keep posting again and again about case design, lets make a thread for that then. As I said before, I don't want to clutter up this thread talking about my case ideas. This is BKKCoins topic for Klondike boards. Sorry BKKCoins, I didn't mean to derail your thread at all. I really appreciate what you are doing for everyone here.

Generally, a rackmount case is considered to be separate from the rack, in terms of cost...

Cardboard and zip ties for $100 dollars? I'm starting to think it's actually your shopping skills that suck if anything. You can pick up any number of 4U rackmount cases off newegg in the $50 range. A few brackets, standoffs, etc, shouldn't cost anywhere near $50, but I'll allow it in the $100 total, and call it good.

I based my assessment of 'ridiculous shit' on the picture you posted as an example of "Things you can be proud of", and the fact that it was actually, in fact, nothing more than some ridiculous shit. Like I said, to each his own. Maybe what you're envisioning, and what you're describing are two separate things, and we'll actually all be blown away by it. Regardless, I agree, perhaps we need a separate thread for discussing case design. I was merely responding in the thread in which you, against your own 'protestations', or course, brought it up, and continue to respond.

[LaserHorse]

[…]
To aid with inspection of test and assembled boards I'm going to order one of these USB Microscopes. It looks pretty nice for checking PCB traces and IC pins, and my eyesight is not what it was (as I realized while looking at some boards that I got yesterday). For $60 it seems like it'll be worthwhile.

http://www.youtube.com/watch?feature=player_embedded&v=2te13V_YsPU

Anyone ever use one of these?

I have used a very similar model.  They're certainly adequate for spotting solder bridges on SMT devices.  
I found the LEDs a bit harsh and the image quality a bit lacking, though this is likely unimportant for your usage.

EDIT: Also this model looks quite lovely & capable:
http://www.ebay.com/itm/Mini-2MP-USB-Digital-Microscope-Circuit-board-Inspection-Camera-Aluminium-Stand-/190847055453?pt=LH_DefaultDomain_0&hash=item2c6f5ef65d
likely same as this one: http://youtu.be/YDCicFZEINo?t=15s

[wrenchmonkey]

[…]
To aid with inspection of test and assembled boards I'm going to order one of these USB Microscopes. It looks pretty nice for checking PCB traces and IC pins, and my eyesight is not what it was (as I realized while looking at some boards that I got yesterday). For $60 it seems like it'll be worthwhile.

http://www.youtube.com/watch?feature=player_embedded&v=2te13V_YsPU

Anyone ever use one of these?

I have used a very similar model.  They're certainly adequate for spotting solder bridges on SMT devices.  
I found the LEDs a bit harsh and the image quality a bit lacking, though this is likely unimportant for your usage.

EDIT: Also this model looks quite lovely & capable:
http://www.ebay.com/itm/Mini-2MP-USB-Digital-Microscope-Circuit-board-Inspection-Camera-Aluminium-Stand-/190847055453?pt=LH_DefaultDomain_0&hash=item2c6f5ef65d
likely same as this one: http://youtu.be/YDCicFZEINo?t=15s

That one looks awesome. I need it.

BKK, if you agree that this one will work for your purposes, send me your address, and I'll purchase one and have it shipped to you.