Pages:
Author

Topic: Ultra Under-overclock image for A2 Innosilicon by Emdje - V5.0 - page 14. (Read 79746 times)

legendary
Activity: 1596
Merit: 1000
Thanks for the image. Smiley

Not free, but pretty cheap at about 0.07c USD. I'd have to figure out if it would be worth it to run them overvolted, and for how long.
hero member
Activity: 686
Merit: 500
Updated cgminer, with requested  clock options:
https://mega.co.nz/#!rUETAYaK!BMTxVusrO5rnZK5eYbDeS7XcFdYV4MyZGBn0BCatFOg

Full image will be updated later when compressed and uploaded.

Edit: full image uploaded: https://mega.co.nz/#!qUdwgC7T!FjieYOsePmkStsygU_l7vqNUiCdEvuifOwgT7Jv5jlU
hero member
Activity: 686
Merit: 500
That are still a lot of HW errors  Sad . It is quite possible that the two boards that have more HW errors have slightly different resistors (lower value) 'controlling' the voltage regulators. I don't know if you run these with free electricity, but if you do it is very worth it to overvolt them. I am running the board on 1400 MHz with 0.40% HW error:
Code:
ASC	Name	ID	Enabled	Status	Temperature	MHS av	MHS 5s	Accepted	Rejected	Hardware Errors	Utility	Last Share Pool	Last Share Time	Total MH	Diff1 Work	Difficulty Accepted	Difficulty Rejected	Last Share Difficulty	Last Valid Work	Device Hardware%	Device Rejected%	Device Elapsed
0 BA1 0 Y Alive         17.03 17.10 49,528             352             210          17.82/m 0 18:28:42 2,840,015.0589 43,284,885 43,016,736.84 257,847.06 735.17 -2s 0.00% 0.60% 166796

legendary
Activity: 1596
Merit: 1000
Here's a pic to give you an idea of the 2 flaky boards:



Do you think I should bother lowering their speeds to try and minimize HW errors? In that screenshot, 0 & 1 are running 1300MHz, while 2 & 3 are running 1280MHz.

Here's my 90MH, for the hell of it:



All 6 blades are running at 1300MHz.

What pisses me off the most about the 2 flaky boards is that I was screwed out of an RMA from pcfli, he told me a new firmware (which never came) would fix the issue.  Roll Eyes
legendary
Activity: 1596
Merit: 1000
Well, I could use them but I don't want to be a pain. My 60MH unit has 2 blades that like to spit out a whole lot of HW errors for some reason, so I used to run them at a slower speed than the other 2 which run at 1300MHz.

If you have to remove options, would removing 1320 and up be more logical? I don't think most blades will run properly above 1300 without increasing voltage which most people likely won't be able to do.

It's not a pain. I like to provide something that is most usefull.
I am at work tonight and the raspberry pi I use for compiling is not on right now. But I'll change the code for version 3.1, and put the options between 1200 and 1280 back in. And then in the next version I will take out the indeed more logical higher 'in between' levels. (And probably try again to leave them all in as well Tongue )

Sounds good, take you time. Don't rush yourself on my account!
hero member
Activity: 686
Merit: 500
Well, I could use them but I don't want to be a pain. My 60MH unit has 2 blades that like to spit out a whole lot of HW errors for some reason, so I used to run them at a slower speed than the other 2 which run at 1300MHz.

If you have to remove options, would removing 1320 and up be more logical? I don't think most blades will run properly above 1300 without increasing voltage which most people likely won't be able to do.

It's not a pain. I like to provide something that is most usefull.
I am at work tonight and the raspberry pi I use for compiling is not on right now. But I'll change the code for version 3.1, and put the options between 1200 and 1280 back in. And then in the next version I will take out the indeed more logical higher 'in between' levels. (And probably try again to leave them all in as well Tongue )
legendary
Activity: 1596
Merit: 1000
Well, I could use them but I don't want to be a pain. My 60MH unit has 2 blades that like to spit out a whole lot of HW errors for some reason, so I used to run them at a slower speed than the other 2 which run at 1300MHz.

If you have to remove options, would removing 1320 and up be more logical? I don't think most blades will run properly above 1300 without increasing voltage which most people likely won't be able to do.
hero member
Activity: 686
Merit: 500
No they weren't broken, but when I was working on the underclock version I needed to sacrifice some options, because with to much options cgminer doesn't start at all (don't know why).

But if you need them I will put them back in for 3.1.
legendary
Activity: 1596
Merit: 1000
Here's hoping you hear from him soon!

I flashed the new Image on my 60 and 90 MH units, and for some reason they're getting a lot less hw errors. Not that I'm complaining, but what could have caused that? I was previously running version 2.0 on both.

I guess you were never able to get cgminer 4.9 working properly on the A2s? Also, did you ever figure a way of seeing the cgminer screen through ssh/putty?

I think it has to do with the updated stratum code having the miner run more smoothly.
And no I can't get cgminer 4.9 get to work. And if you stop cgminer in the browser you can start cgminer in the terminal when you are logged into the pi, that way you can see the cgminer screen:
Code:
sudo ./cgminer -o stratum+tcp://eu-01.miningrigrentals.com:3333 -u emdje.3166 -p x --A1Pll1 1300 --A1Pll2 1300 --A1Pll3 1300 --A1Pll4 1300 --A1Pll5 1300 --A1Pll6 1300 --diff 4096 --api-listen --api-network --cs 8 --stmcu 0 --hwreset --no-submit-stale --lowmem

Cool, thanks. Going to give that a try later!

I noticed 3.1 is missing a few speed steps between 1200MHz and 1280MHz. Were they broken in 2.0, thus removed?
hero member
Activity: 686
Merit: 500
Here's hoping you hear from him soon!

I flashed the new Image on my 60 and 90 MH units, and for some reason they're getting a lot less hw errors. Not that I'm complaining, but what could have caused that? I was previously running version 2.0 on both.

I guess you were never able to get cgminer 4.9 working properly on the A2s? Also, did you ever figure a way of seeing the cgminer screen through ssh/putty?

I think it has to do with the updated stratum code having the miner run more smoothly.
And no I can't get cgminer 4.9 get to work. And if you stop cgminer in the browser you can start cgminer in the terminal when you are logged into the pi, that way you can see the cgminer screen:
Code:
sudo ./cgminer -o stratum+tcp://eu-01.miningrigrentals.com:3333 -u emdje.3166 -p x --A1Pll1 1300 --A1Pll2 1300 --A1Pll3 1300 --A1Pll4 1300 --A1Pll5 1300 --A1Pll6 1300 --diff 4096 --api-listen --api-network --cs 8 --stmcu 0 --hwreset --no-submit-stale --lowmem
legendary
Activity: 1596
Merit: 1000
Here's hoping you hear from him soon!

I flashed the new Image on my 60 and 90 MH units, and for some reason they're getting a lot less hw errors. Not that I'm complaining, but what could have caused that? I was previously running version 2.0 on both.

I guess you were never able to get cgminer 4.9 working properly on the A2s? Also, did you ever figure a way of seeing the cgminer screen through ssh/putty?
hero member
Activity: 686
Merit: 500
Thanks emdje, much appreciated. Sent you some beer money!  Smiley

Have you got your replacement chip back on the board?

Thnx Smiley

No I have not yet heard back from the person who offered his broken board (in return for underclock options, yet to be released).
If I won't hear anything I might try to connect pins 13 and 14 and try the board then.
legendary
Activity: 1596
Merit: 1000
Thanks emdje, much appreciated. Sent you some beer money!  Smiley

Have you got your replacement chip back on the board?
hero member
Activity: 686
Merit: 500
hero member
Activity: 686
Merit: 500
Sweet, thanks for the updated cgminer emdje. Are you planning on releasing a new image with the latest updates? I suck at linux command line.

Sure I can. Will do that today.

Edit: compressing image now, then uploading and posting the link
legendary
Activity: 1596
Merit: 1000
Sweet, thanks for the updated cgminer emdje. Are you planning on releasing a new image with the latest updates? I suck at linux command line.
hero member
Activity: 686
Merit: 500
Emdje, could you please compile new binary, where this parametere is 1min? Thanks

[2015-03-23 13:47:54] Pool 0 XXX stable for 5 mins
 [2015-03-23 13:47:54] Switching to pool 0 XXX


I will look at that BakSAj.

Nice, I kindly ask only for cgminer binary, not the whole image :-)

Rar file with the cgminer where the switching parameter has been set to 1 minute instead of 5, and the correct index file (made some mistakes in the previous versions with the difficulty setting and some obsolete options have been removed) Also all the stratum code of cgminer 4.9.1 has been implemented.:
https://www.dropbox.com/s/f3t9eykuv1x2b1v/cgminer3.9.0-A2-v3.1.rar?dl=0

Don't forget to mark the cgminer as an executable by using the command "chmod +x cgminer"
hero member
Activity: 686
Merit: 500
I remembered having a kitchen hotplate lying around, so I used that to heat the board to 90 Degrees Celsius:


Then the hot air soldering gun was enough to remove the chip (albeit a little bit to soon, should have let it heat up a tiny bit more. Now you can see a little damage to one of the grounds on the chip base, but since it is the ground, and that is very large I think it will still work if it gets a new chip)


full member
Activity: 157
Merit: 100
Emdje, could you please compile new binary, where this parametere is 1min? Thanks

[2015-03-23 13:47:54] Pool 0 XXX stable for 5 mins
 [2015-03-23 13:47:54] Switching to pool 0 XXX


I will look at that BakSAj.

Nice, I kindly ask only for cgminer binary, not the whole image :-)
hero member
Activity: 686
Merit: 500
Progress  Grin

Code:
 [2015-03-24 20:42:47] Started cgminer 3.9.0                    
 [2015-03-24 20:42:47] Run Reset=1                    
 [2015-03-24 20:42:47] ST MCU hardware reset start                    
 [2015-03-24 20:42:51] SPI Speed 4000 kHz                    
 [2015-03-24 20:42:51] ST MCU - Enable (Pre-header)                    
 [2015-03-24 20:42:51] A1 = 500,14                    
 [2015-03-24 20:42:51] A1 PLL Clock = 500MHz                    
 [2015-03-24 20:42:51] A1 = 600,13                    
 [2015-03-24 20:42:51] A1 PLL Clock = 600MHz                    
 [2015-03-24 20:42:51] A1 = 700,12                    
 [2015-03-24 20:42:51] A1 PLL Clock = 700MHz                    
 [2015-03-24 20:42:51] A1 = 800,11                    
 [2015-03-24 20:42:51] A1 PLL Clock = 800MHz                    
 [2015-03-24 20:42:51] A1 = 900,10                    
 [2015-03-24 20:42:51] A1 PLL Clock = 900MHz                    
 [2015-03-24 20:42:51] A1 = 1000,9                    
 [2015-03-24 20:42:51] A1 PLL Clock = 1000MHz                    
 [2015-03-24 20:42:51] AUTO GPIO CS                    
 [2015-03-24 20:42:51] SPI(cs0) no device                    
 [2015-03-24 20:42:51] ACK(cs0) timeout:cmd_RESET_BCAST-0.0355s                    
 [2015-03-24 20:42:52] spidev0.0(cs1): Found 8 A1 chips                    
 [2015-03-24 20:42:52] Found chip 1 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 2 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 3 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 4 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 5 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 6 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 7 with 54 active cores                    
 [2015-03-24 20:42:52] Found chip 8 with 54 active cores                    
 [2015-03-24 20:42:52] Found 8 chips with total 432 active cores  

When using 500 MHz I did not get an accepted share for the 10 minutes I tried it. 600 MHz started accepting shares right away. But might be the pool I was on.
Pages:
Jump to: