Author

Topic: ANTMINER S3+ Discussion and Support Thread - page 376. (Read 710164 times)

hero member
Activity: 784
Merit: 504
It appears the company has no intention to actually refund us 7.7% for the miners that did not meet spec.
Is there a single honest miner vendor out there?

If you want to trust any of the ASIC manufacturer, BitmainTech should be in the first place.
They will return your money much before Aug 2.
jr. member
Activity: 50
Merit: 1
It appears the company has no intention to actually refund us 7.7% for the miners that did not meet spec.
Is there a single honest miner vendor out there?
sr. member
Activity: 392
Merit: 250
Can u put the screenshot?
hero member
Activity: 784
Merit: 504
Quote
VIM Editor Commands

Vim is an editor to create or edit a text file.

There are two modes in vim. One is the command mode and another is the insert mode.

In the command mode, user can move around the file, delete text, etc.

In the insert mode, user can insert text.

Changing mode from one to another

From command mode to insert mode   type a/A/i/I/o/O ( see details below)

From insert mode to command mode   type Esc (escape key)

Some useful commands for VIM

Text Entry Commands (Used to start text entry)

a Append text following current cursor position

A Append text to the end of current line

i Insert text before the current cursor position

I Insert text at the beginning of the cursor line

o Open up a new line following the current line and add text there

O Open up a new line in front of the current line and add text there

The following commands are used only in the commands mode.

Cursor Movement Commands

h Moves the cursor one character to the left

l Moves the cursor one character to the right

k Moves the cursor up one line

j Moves the cursor down one line

nG or :n Cursor goes to the specified (n) line

(ex. 10G goes to line 10)

^F (CTRl F) Forward screenful

^B Backward screenful

^f One page forward

^b One page backward

^U Up half screenful

^D Down half screenful

$ Move cursor to the end of current line

0 (zero) Move cursor to the beginning of current line

w Forward one word

b Backward one word

Exit Commands

:wq Write file to disk and quit the editor

:q! Quit (no warning)

:q Quit (a warning is printed if a modified file has not been saved)

ZZ Save workspace and quit the editor (same as :wq)

: 10,25 w temp

write lines 10 through 25 into file named temp. Of course, other line

numbers can be used. (Use :f to find out the line numbers you want.

 

Text Deletion Commands

x Delete character

dw Delete word from cursor on

db Delete word backward

dd Delete line

d$ Delete to end of line

d^ (d caret, not CTRL d) Delete to beginning of line

Yank (has most of the options of delete)-- VI's copy commmand

yy yank current line

y$ yank to end of current line from cursor

yw yank from cursor to end of current word

5yy yank, for example, 5 lines

Paste (used after delete or yank to recover lines.)

p paste below cursor

P paste above cursor

"2p paste from buffer 2 (there are 9)

u Undo last change

U Restore line

J Join next line down to the end of the current line

File Manipulation Commands

:w Write workspace to original file

:w file Write workspace to named file

:e file Start editing a new file

:r file Read contents of a file to the workspace

To create a page break, while in the insert mode, press the CTRL key

And l. ^L will appear in your text and will cause the printer to start

A new page.

 

 

Other Useful Commands

Most commands can be repeated n times by typing a number, n, before

the command. For example 10dd means delete 10 lines.

. Repeat last command

cw Change current word to a new word

r Replace one character at the cursor position

R Begin overstrike or replace mode – use ESC key to exit

:/ pattern Search forward for the pattern

:? pattern Search backward for the pattern

n (used after either of the 2 search commands above to

continue to find next occurrence of the pattern.

:g/pat1/s//pat2/g replace every occurrence of pattern1 (pat1) with

pat2

Example :g/tIO/s//Ada.Text_IO/g

This will find and replace tIO by Ada.text_IO everywhere in the file.

:g/a/s// /g replace the letter a, by blank

:g/a/s///g replace a by nothing

note: Even this command be undone by u

 

Examples

Opening a New File

Step 1   type   vim filename   (create a file named filename)

Step 2   type   i   ( switch to insert mode)

Step 3   enter text   (enter your Ada program)

Step 4   hit   Esc key   (switch back to command mode)

Step 5   type   :wq   (write file and exit vim)

 

Editing the Existing File

Step 1   type   vim filename   (edit the existing file named filename)

Step 2   move around the file using h/j/k/l key or any appropriate command

h Moves the cursor one character to the left

l Moves the cursor one character to the right

k Moves the cursor up one line

j Moves the cursor down one line

nG or :n Cursor goes to the specified (n) line

(ex. 10G goes to line 10)

Step 3   edit required text (replace or delete or insert)

Step 4   hit Esc key (exit from insert mode if you insert or replace text)

Step 5   type   :wq
Courtesy/Source : http://www.radford.edu/~mhtay/CPSC120/VIM_Editor_Commands.htm
hero member
Activity: 744
Merit: 514
gotta let a coin be a coin
I was able to get the discarded amount lowered as well as the load. I changed the queue to 0 from 4096 and added --scan-time 1 --expiry 1 to the cgminer file.

***IF YOU ARE NOT SURE WHAT THIS MEANS DO NOT TRY AS YOU WILL BRICK YOUR UNIT***

This is the first attempt, not sure if 0 is the optimal queue value for all pools but if you are a P2P miner you may want to try this out. I use a smaller pool and my numbers have gone up. This is along the same tune you could do for an S1 but a far cry from what we did in the GPU days. I have another unit running at a lower speed but have the same results.
  
SSH in then:
Code:
vim /etc/init.d/cgminer
--scan-time 1 --expiry 1

Find line 75 PARAMS...at the end change --queue 0 and add --scan-time 1 --expiry 1 before the quotation.

***DO NOT TRY THIS UNLESS YOU FULLY UNDERSTAND AS YOU WILL BRICK YOUR UNIT****

Before Config Change (I added a line so it is easy to see, I took it out when I finished)



After Config Change



Ant Before



Ant After CHECK THE LOAD VALUES



I appreciate you posting this  Grin I too have no idea why they set such a high queue on the units

Has anyone else had any luck reducing discards with this mod. I've been learning about how to SSH and I think I want to try it. I'm getting a lot of discards and some of the pools are reporting strange hash rates and other feedback, after I played around with load balancing and balancing.

I don't want to speak on Duce's behalf, but experience seems to show at this point that there is possible detriment to changing the queue value from 4096. BMT seems to have done this for a reason.

I am seeing no difference in performance using modified expiry and scan-time settings. I'm working with about 36-hours worth of data at this point.
I made these changes yesterday and my discards dropped x10 and the load is around 1.8 for both. My discards were x15 of accepted shares now then are only x.10. You have to restart cgminer for the changes to take effect.

If you want you can use sed to replace the values in the file (it only occurs once):

Code:
sed -i 's/--queue 4096"/--queue 0 --scan-time 1 --expiry 1"/g' /etc/init.d/cgminer

From what I've seen (not from the above posts) is a lot of people posting really are missing some technical fundamentals to just get the boxes online. Where did they hear about bitcoin mining anyway? It's weird.
sr. member
Activity: 392
Merit: 250
I was able to get the discarded amount lowered as well as the load. I changed the queue to 0 from 4096 and added --scan-time 1 --expiry 1 to the cgminer file.

***IF YOU ARE NOT SURE WHAT THIS MEANS DO NOT TRY AS YOU WILL BRICK YOUR UNIT***

This is the first attempt, not sure if 0 is the optimal queue value for all pools but if you are a P2P miner you may want to try this out. I use a smaller pool and my numbers have gone up. This is along the same tune you could do for an S1 but a far cry from what we did in the GPU days. I have another unit running at a lower speed but have the same results.
  
SSH in then:
Code:
vim /etc/init.d/cgminer
--scan-time 1 --expiry 1

Find line 75 PARAMS...at the end change --queue 0 and add --scan-time 1 --expiry 1 before the quotation.

***DO NOT TRY THIS UNLESS YOU FULLY UNDERSTAND AS YOU WILL BRICK YOUR UNIT****

Before Config Change (I added a line so it is easy to see, I took it out when I finished)


After Config Change


Ant Before


Ant After CHECK THE LOAD VALUES

What's the code to close the editing for ssh? I use windown and putty
And lowering the discarded will benefit what?
:wq
If I understood you correctly.
Oh it's the same like sshing the s1
Thanks I will give it a try.  I am sure many people of your s3 has a huge big pie of discarded work or shares
legendary
Activity: 1148
Merit: 1000
I was able to get the discarded amount lowered as well as the load. I changed the queue to 0 from 4096 and added --scan-time 1 --expiry 1 to the cgminer file.

***IF YOU ARE NOT SURE WHAT THIS MEANS DO NOT TRY AS YOU WILL BRICK YOUR UNIT***

This is the first attempt, not sure if 0 is the optimal queue value for all pools but if you are a P2P miner you may want to try this out. I use a smaller pool and my numbers have gone up. This is along the same tune you could do for an S1 but a far cry from what we did in the GPU days. I have another unit running at a lower speed but have the same results.
 
SSH in then:
Code:
vim /etc/init.d/cgminer
--scan-time 1 --expiry 1

Find line 75 PARAMS...at the end change --queue 0 and add --scan-time 1 --expiry 1 before the quotation.

***DO NOT TRY THIS UNLESS YOU FULLY UNDERSTAND AS YOU WILL BRICK YOUR UNIT****

Before Config Change (I added a line so it is easy to see, I took it out when I finished)


After Config Change


Ant Before


Ant After CHECK THE LOAD VALUES

What's the code to close the editing for ssh? I use windown and putty
And lowering the discarded will benefit what?
:wq
If I understood you correctly.
sr. member
Activity: 392
Merit: 250
Dear Bitmain CEO is there something wrong with huge discarded shares ? Can you work something out for a new firmware to address this problem?
sr. member
Activity: 280
Merit: 250
Yeah, as much as I want to overclock these guys, I don't really want to pioneer it.  That and I don't really have spare 2x 6 pcie cords to shove into my psus, though I'm sure I could buy them.  Mine all run between 430-450.  I have one that seems to dip to 410 every once and a while, im not sure why, but for the most part stable.

The way I look it is, we are getting MORE than double the hashpower for the same power usage.  As a casual at home miner, that's good enough for me.  Hell, I might even buy some of their USB units when they come out Cheesy
sr. member
Activity: 392
Merit: 250
I was able to get the discarded amount lowered as well as the load. I changed the queue to 0 from 4096 and added --scan-time 1 --expiry 1 to the cgminer file.

***IF YOU ARE NOT SURE WHAT THIS MEANS DO NOT TRY AS YOU WILL BRICK YOUR UNIT***

This is the first attempt, not sure if 0 is the optimal queue value for all pools but if you are a P2P miner you may want to try this out. I use a smaller pool and my numbers have gone up. This is along the same tune you could do for an S1 but a far cry from what we did in the GPU days. I have another unit running at a lower speed but have the same results.
 
SSH in then:
Code:
vim /etc/init.d/cgminer
--scan-time 1 --expiry 1

Find line 75 PARAMS...at the end change --queue 0 and add --scan-time 1 --expiry 1 before the quotation.

***DO NOT TRY THIS UNLESS YOU FULLY UNDERSTAND AS YOU WILL BRICK YOUR UNIT****

Before Config Change (I added a line so it is easy to see, I took it out when I finished)



After Config Change



Ant Before



Ant After CHECK THE LOAD VALUES


What's the code to close the editing for ssh? I use windown and putty
And lowering the discarded will benefit what?
full member
Activity: 207
Merit: 112
Nope. Too impatient. Heading to the office to try and run them naked. I think having this DC/DC modules couped up without a heatsink is no good. I'm heading out for the day afterwards so I'll report back when I have useful data and/or warnings.
full member
Activity: 207
Merit: 112
I was able to get the discarded amount lowered as well as the load. I changed the queue to 0 from 4096 and added --scan-time 1 --expiry 1 to the cgminer file.

***IF YOU ARE NOT SURE WHAT THIS MEANS DO NOT TRY AS YOU WILL BRICK YOUR UNIT***

This is the first attempt, not sure if 0 is the optimal queue value for all pools but if you are a P2P miner you may want to try this out. I use a smaller pool and my numbers have gone up. This is along the same tune you could do for an S1 but a far cry from what we did in the GPU days. I have another unit running at a lower speed but have the same results.
 
SSH in then:
Code:
vim /etc/init.d/cgminer
--scan-time 1 --expiry 1

Find line 75 PARAMS...at the end change --queue 0 and add --scan-time 1 --expiry 1 before the quotation.

***DO NOT TRY THIS UNLESS YOU FULLY UNDERSTAND AS YOU WILL BRICK YOUR UNIT****

Before Config Change (I added a line so it is easy to see, I took it out when I finished)



After Config Change



Ant Before



Ant After CHECK THE LOAD VALUES



I appreciate you posting this  Grin I too have no idea why they set such a high queue on the units

Has anyone else had any luck reducing discards with this mod. I've been learning about how to SSH and I think I want to try it. I'm getting a lot of discards and some of the pools are reporting strange hash rates and other feedback, after I played around with load balancing and balancing.

I don't want to speak on Duce's behalf, but experience seems to show at this point that there is possible detriment to changing the queue value from 4096. BMT seems to have done this for a reason.

I am seeing no difference in performance using modified expiry and scan-time settings. I'm working with about 36-hours worth of data at this point.
member
Activity: 119
Merit: 10
I was able to get the discarded amount lowered as well as the load. I changed the queue to 0 from 4096 and added --scan-time 1 --expiry 1 to the cgminer file.

***IF YOU ARE NOT SURE WHAT THIS MEANS DO NOT TRY AS YOU WILL BRICK YOUR UNIT***

This is the first attempt, not sure if 0 is the optimal queue value for all pools but if you are a P2P miner you may want to try this out. I use a smaller pool and my numbers have gone up. This is along the same tune you could do for an S1 but a far cry from what we did in the GPU days. I have another unit running at a lower speed but have the same results.
 
SSH in then:
Code:
vim /etc/init.d/cgminer
--scan-time 1 --expiry 1

Find line 75 PARAMS...at the end change --queue 0 and add --scan-time 1 --expiry 1 before the quotation.

***DO NOT TRY THIS UNLESS YOU FULLY UNDERSTAND AS YOU WILL BRICK YOUR UNIT****

Before Config Change (I added a line so it is easy to see, I took it out when I finished)



After Config Change



Ant Before



Ant After CHECK THE LOAD VALUES



I appreciate you posting this  Grin I too have no idea why they set such a high queue on the units

Has anyone else had any luck reducing discards with this mod. I've been learning about how to SSH and I think I want to try it. I'm getting a lot of discards and some of the pools are reporting strange hash rates and other feedback, after I played around with load balancing and balancing.
full member
Activity: 207
Merit: 112
Has anyone tried running the units uncovered, S1 style? Im guessing the only reason the cover exists is to protect the front heat sink and allow air from the fan to stay within that small space. Since the control board sits on top, and the fans never even seem to come close to max RPM, I'm thinking the units might still operate well without the cover.

In fact, without the cover they look very similar in design to the new Rockbox blade miners which only use one push fan.

http://minersource.net/products/r3-box-450-480gh-bitcoin-miner-in-stock-now

Granted these are ASICminer chip, I feel like this might allow the DC/DC modules to at least breathe.

EDIT: I dont have access to my S3s until Monday, so I can't try any of this yet...
newbie
Activity: 19
Merit: 0
Here's my McGuyver setup in my loft! lol

http://i61.tinypic.com/2lnet5i.jpg
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
I actually did this mod for all of my S1s and saw a 1GH inprovement across each unit. I have enough for one unit coming in on Monday so I'll go ahead and try it myself to verify.

The part you are looking for are the eight grey square "blocks" above the heat sink. There were only four-per-blade on the S1

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

The layout of the S3 board seems to allow for a single long heat sink to cover all of the dc-dc converters?  The S1 layout had caps in between preventing the use of a single large heat sink.  I bet the larger surface area will work much better than the smaller individual heat sinks.

Anyone know where to find something that will fit, and how to properly attach?

This is a terrific idea. I can check our old stocks and find some perfectly straight/flat heat sinks from old server video cards. I have some dual-head Quadros in mind that I might be able to cut up easily.

Either way won't be able to test until Monday. My fiancée will kill me if I head back in this weekend.
Definitely not on call.

Anyone know a good source for 3M thermal adhesive strips?

Looking forward to seeing this with a stripe across the 8 of them.

Yes, this is great design if it works.
In S2 bitmain figured out to incorporate a steel plate in latest batch. I think that that steel plate helped in heat distribution as well (apart from structural integrity).
I doubt that they can incorporate such heat sink in their design at this stage, but who knows.
Edit: they (Bitmain) can probably use this at a minimal cost ($0.3/piece):
http://www.alibaba.com/product-detail/led-strip-aluminium-heat-sink_1996606714.html

Might be able to easily cut and trim something like this. http://www.ebay.com/itm/1x-290x120x20-mm-Aluminium-Heatsink-8x3w-14x1w-High-Power-leds-w-o-fan-/290745977625?pt=LH_DefaultDomain_0&hash=item43b1cf8b19

Bump for any knowledge on where to pick up strips of 3M thermal adhesive.

EDIT: Obviously this particular part is way over-priced, but you get my drift.


 if the case had 3 holes with 3 screws and 3 springs attached to the new heat sink.

 the new heat sink can be flat aluminum stock !

with 3 of these soldered to it

http://www.ebay.com/itm/1-4-Hex-Standoff-1-11-16-Long-Female-Aluminum-6-32-5-8-Deep-Threads-50-Piece-/111403256918?pt=LH_DefaultDomain_0&hash=item19f026cc56

 then the same on the other side  you would only need some thermal paste.

setting  the heat sink would be done after case is closed. to open the case the six new heat sink screws would need to released first.

this would:

 a )  be under 2 usd a machine in the factory

b ) no need for new screw holes in the pcb

c)  retro fit  in home kit could be sold

this would prove cheaper then the 7.7% refunds they gave.


note as you tighten  the exterior screws  it lifts the heatsink away from the 8 r47.

as you loosen the screws it pushes the heatsink tighter
full member
Activity: 174
Merit: 100
16Q3R8NAfK63DvkTUGgLdPScyMU8uSAJUH
I have a problem. My "Authorized Distributor" has done the following.

I (stupidly) agreed to overpay on batch 1 S3 because I wanted a batch 1. They usually make money especially batch 1 where you have good clarity on what difficulty might be when they ship. I paid very early on July 1.

I received my confirmation emails. I re-checked my account after end of business later that day. So far everything was in order. Batch 1 S3 etc.

2 weeks about go by and I pop in to check my account to find out when my Batch 1 order will ship to discover it's a magical Batch 2. I have no idea when it was ordered, what point in the queue it was at or anything. At this point I have received 0 contact from distributor. Order was altered 25.5 hours after I placed it. I wish I had known I needed to babysit this order to keep things on track and at a price I could live with.

I notice in my account there is ask us a question. So I ask if they can fix my Batch 1 order to actually be a Batch 1. I really wasn't thinking that someone intentionally changed anything but that there was a clerical error.

24 hours pass still no contact from distributor.

I send a message to the person who has their email on my receipt (I dug it up just to ask again). I was not nice about it. Reply was received in 3 hours. Nothing too helpful just an offer for fair compensation. What is fair for bumping me into a cheaper batch, failing to ask me about changing my order, failing to ask about re-pricing my order? At this point I didn't want a batch 2. I never wanted a batch 2.
 
I asked within 3 hours for the following, after apologizing, What my total refund would be AND what my "Fair Compensation" would be.

Within 12 hours the disputed order shipped.

No answer Friday, Saturday, Sunday, Monday or Tuesday. Email was sent late Thursday. As of Monday late my order finally finished its warehousing. Tuesday I send another email asking if distributor would please reply to my last email. Last email included in current email.

Wednesday I get a reply. Here is your tracking #. Batch 1 and 2 shipped about the same time. I have to refund the 7.7% for specs and I have to refund 0.02 for the price difference. Let me know wallet address.

I reply I asked for full refund and fair compensation. Order is complete when package is accepted I have not and in 7 days UPS will return your package to you. I point out I still wouldn't have paid shipping on Batch 2 (ordering from bitmain) and told distributor I figured the 0.1115 over the cost of batch 2 directly was fair compensation at this point. I notified distributor that I was interested in either 0.1115  or a full refund and they have until the package goes back at that point I expect a full refund.

No reply in 48 hours. Send email to admin because I get an order complete email. I point out 1.48 is better then 0. Ask what they would prefer.

48 hours more still no replies.

I really would like a way to deal with this that doesn't leave me with a more then Batch 1 priced shipped same as batch 3 held in China for 4-5 days batch 2 then left at UPS because I have an issue with the questionable way this has been handled. Most of this is the price of business


1. Why did you not order directly from bitmain?
2. If you paid with paypal or a credit card you can file a dispute directly with them.  Although, filing a dispute due to "shipping time" might be a hard case to win.  "Batch 1" and "Batch 2" are identical products.
3. If you paid with BTC you are SOL
4. Best thing to do is identify them in the forum as a warning to others.  The bad publicity might make them act.
hero member
Activity: 981
Merit: 500
DIV - Your "Virtual Life" Secured and Decentralize
I have a problem. My "Authorized Distributor" has done the following.

I (stupidly) agreed to overpay on batch 1 S3 because I wanted a batch 1. They usually make money especially batch 1 where you have good clarity on what difficulty might be when they ship. I paid very early on July 1.

I received my confirmation emails. I re-checked my account after end of business later that day. So far everything was in order. Batch 1 S3 etc.

2 weeks about go by and I pop in to check my account to find out when my Batch 1 order will ship to discover it's a magical Batch 2. I have no idea when it was ordered, what point in the queue it was at or anything. At this point I have received 0 contact from distributor. Order was altered 25.5 hours after I placed it. I wish I had known I needed to babysit this order to keep things on track and at a price I could live with.

I notice in my account there is ask us a question. So I ask if they can fix my Batch 1 order to actually be a Batch 1. I really wasn't thinking that someone intentionally changed anything but that there was a clerical error.

24 hours pass still no contact from distributor.

I send a message to the person who has their email on my receipt (I dug it up just to ask again). I was not nice about it. Reply was received in 3 hours. Nothing too helpful just an offer for fair compensation. What is fair for bumping me into a cheaper batch, failing to ask me about changing my order, failing to ask about re-pricing my order? At this point I didn't want a batch 2. I never wanted a batch 2.
 
I asked within 3 hours for the following, after apologizing, What my total refund would be AND what my "Fair Compensation" would be.

Within 12 hours the disputed order shipped.

No answer Friday, Saturday, Sunday, Monday or Tuesday. Email was sent late Thursday. As of Monday late my order finally finished its warehousing. Tuesday I send another email asking if distributor would please reply to my last email. Last email included in current email.

Wednesday I get a reply. Here is your tracking #. Batch 1 and 2 shipped about the same time. I have to refund the 7.7% for specs and I have to refund 0.02 for the price difference. Let me know wallet address.

I reply I asked for full refund and fair compensation. Order is complete when package is accepted I have not and in 7 days UPS will return your package to you. I point out I still wouldn't have paid shipping on Batch 2 (ordering from bitmain) and told distributor I figured the 0.1115 over the cost of batch 2 directly was fair compensation at this point. I notified distributor that I was interested in either 0.1115  or a full refund and they have until the package goes back at that point I expect a full refund.

No reply in 48 hours. Send email to admin because I get an order complete email. I point out 1.48 is better then 0. Ask what they would prefer.

48 hours more still no replies.

I really would like a way to deal with this that doesn't leave me with a more then Batch 1 priced shipped same as batch 3 held in China for 4-5 days batch 2 then left at UPS because I have an issue with the questionable way this has been handled. Most of this is the price of business
full member
Activity: 207
Merit: 112
I actually did this mod for all of my S1s and saw a 1GH inprovement across each unit. I have enough for one unit coming in on Monday so I'll go ahead and try it myself to verify.

The part you are looking for are the eight grey square "blocks" above the heat sink. There were only four-per-blade on the S1

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

The layout of the S3 board seems to allow for a single long heat sink to cover all of the dc-dc converters?  The S1 layout had caps in between preventing the use of a single large heat sink.  I bet the larger surface area will work much better than the smaller individual heat sinks.

Anyone know where to find something that will fit, and how to properly attach?

This is a terrific idea. I can check our old stocks and find some perfectly straight/flat heat sinks from old server video cards. I have some dual-head Quadros in mind that I might be able to cut up easily.

Either way won't be able to test until Monday. My fiancée will kill me if I head back in this weekend.
Definitely not on call.

Anyone know a good source for 3M thermal adhesive strips?

Looking forward to seeing this with a stripe across the 8 of them.

Yes, this is great design if it works.
In S2 bitmain figured out to incorporate a steel plate in latest batch. I think that that steel plate helped in heat distribution as well (apart from structural integrity).
I doubt that they can incorporate such heat sink in their design at this stage, but who knows.
Edit: they (Bitmain) can probably use this at a minimal cost ($0.3/piece):
http://www.alibaba.com/product-detail/led-strip-aluminium-heat-sink_1996606714.html

Might be able to easily cut and trim something like this. http://www.ebay.com/itm/1x-290x120x20-mm-Aluminium-Heatsink-8x3w-14x1w-High-Power-leds-w-o-fan-/290745977625?pt=LH_DefaultDomain_0&hash=item43b1cf8b19

Bump for any knowledge on where to pick up strips of 3M thermal adhesive.

EDIT: Obviously this particular part is way over-priced, but you get my drift.
legendary
Activity: 3892
Merit: 4331
I actually did this mod for all of my S1s and saw a 1GH inprovement across each unit. I have enough for one unit coming in on Monday so I'll go ahead and try it myself to verify.

The part you are looking for are the eight grey square "blocks" above the heat sink. There were only four-per-blade on the S1

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

The layout of the S3 board seems to allow for a single long heat sink to cover all of the dc-dc converters?  The S1 layout had caps in between preventing the use of a single large heat sink.  I bet the larger surface area will work much better than the smaller individual heat sinks.

Anyone know where to find something that will fit, and how to properly attach?

This is a terrific idea. I can check our old stocks and find some perfectly straight/flat heat sinks from old server video cards. I have some dual-head Quadros in mind that I might be able to cut up easily.

Either way won't be able to test until Monday. My fiancée will kill me if I head back in this weekend.
Definitely not on call.

Anyone know a good source for 3M thermal adhesive strips?

Looking forward to seeing this with a stripe across the 8 of them.

Yes, this is great design if it works.
In S2 bitmain figured out to incorporate a steel plate in latest batch. I think that that steel plate helped in heat distribution as well (apart from structural integrity).
I doubt that they can incorporate such heat sink in their design at this stage, but who knows.
Edit: they (Bitmain) can probably use this at a minimal cost ($0.3/piece):
http://www.alibaba.com/product-detail/led-strip-aluminium-heat-sink_1996606714.html
Jump to: