Topic: BITMAIN Antminer S3 support and OverClocking thread - page 43. (Read 158132 times)
Anyone have packages available yet with a newer cgminer? S3's come with 3.1.2.. old and crusty...
(Or instructions on how to get a build/dev/chain working, and I'll compile my own on something with horsepower)
For the ones who'd like to try other frequencies, they are in the data sheet available on bitmain site.
Or maybe the old table from the antminer over clocking thread would work after swaping the last digit from "1" to "2".
It seems to be the only difference between the settings.
Or maybe the old table from the antminer over clocking thread would work after swaping the last digit from "1" to "2".
It seems to be the only difference between the settings.
Link?
You are too fast, I was editing
For the ones who'd like to try other frequencies, they are in the data sheet available on bitmain site.
Or maybe the old table from the antminer over clocking thread would work after swaping the last digit from "1" to "2".
It seems to be the only difference between the settings.
Or maybe the old table from the antminer over clocking thread would work after swaping the last digit from "1" to "2".
It seems to be the only difference between the settings.
Link?
For the ones who'd like to try other frequencies, they are in the data sheet available on bitmain site.
Or maybe the old table from the antminer over clocking thread would work after swaping the last digit from "1" to "2".
It seems to be the only difference between the settings.
https://bitcointalksearch.org/topic/m.7544198
Direct link to data sheet
https://bitmaintech.com/files/download/BM1382_Datasheet_v3.0.pdf
I guess we can calculate the acceptable frequencies the same way as in the old thread, but I'm on my phone, so not that easy.
Or maybe the old table from the antminer over clocking thread would work after swaping the last digit from "1" to "2".
It seems to be the only difference between the settings.
https://bitcointalksearch.org/topic/m.7544198
Direct link to data sheet
https://bitmaintech.com/files/download/BM1382_Datasheet_v3.0.pdf
I guess we can calculate the acceptable frequencies the same way as in the old thread, but I'm on my phone, so not that easy.
Quick power #'s, measured with my best unit.. and the only one that will hit 500GH.
Corsair TX850 80plus bronze PSU - single S3:
218 mhz -- 450 GH -- 346 watt at the wall
225 mhz -- 480 GH -- 370 watt at the wall
237 mhz -- 500 GH -- 394 watt at the wall
My first order was 10, and here's an average over the last few hours:
Overall, kinda disappointed with the lack of overclocking. (Which was warned about due to a bad batch of DC-DC..) While I do have one fast unit, One of the units won't even run at the STOCK clocks without X'ing out at least 1-2 of the ASIC's, and most of them can't survive even bumping up to 225. At this stage 2 or 4 power connectors doesn't seem to matter at all. The 500GH unit is running on (2), and the 420-units won't go faster on (4). I'm running 3 S3's per 1300W power supply, so it's not a power issue.
Feeling a little deju'vu to KNC here.. First batch got bad dies, 400GH, second batch got fixes, 550GH.
Corsair TX850 80plus bronze PSU - single S3:
218 mhz -- 450 GH -- 346 watt at the wall
225 mhz -- 480 GH -- 370 watt at the wall
237 mhz -- 500 GH -- 394 watt at the wall
My first order was 10, and here's an average over the last few hours:
Quote
miner01 501.87 GH/s 3404032 (99.91%) 3072 / 0 / 0 0:00:03
miner02 456.70 GH/s 3092736 (99.93%) 2304 / 0 / 0 0:00:03
miner03 422.43 GH/s 3056128 (99.88%) 3584 / 0 / 0 0:00:03
miner04 442.37 GH/s 3094016 (99.62%) 11648 / 0 / 0 0:00:03
miner05 423.99 GH/s 3081984 (99.87%) 4096 / 0 / 0 0:00:03
miner06 429.29 GH/s 3043328 (99.68%) 9728 / 0 / 0 0:00:03
miner07 430.22 GH/s 2934272 (99.64%) 10496 / 0 / 0 0:00:04
miner08 451.09 GH/s 3077632 (99.88%) 3584 / 0 / 0 0:00:04
miner09 441.12 GH/s 2973696 (99.51%) 14592 / 0 / 0 0:00:04
miner10 454.52 GH/s 3096832 (99.69%) 9728 / 0 / 0 0:00:04
miner02 456.70 GH/s 3092736 (99.93%) 2304 / 0 / 0 0:00:03
miner03 422.43 GH/s 3056128 (99.88%) 3584 / 0 / 0 0:00:03
miner04 442.37 GH/s 3094016 (99.62%) 11648 / 0 / 0 0:00:03
miner05 423.99 GH/s 3081984 (99.87%) 4096 / 0 / 0 0:00:03
miner06 429.29 GH/s 3043328 (99.68%) 9728 / 0 / 0 0:00:03
miner07 430.22 GH/s 2934272 (99.64%) 10496 / 0 / 0 0:00:04
miner08 451.09 GH/s 3077632 (99.88%) 3584 / 0 / 0 0:00:04
miner09 441.12 GH/s 2973696 (99.51%) 14592 / 0 / 0 0:00:04
miner10 454.52 GH/s 3096832 (99.69%) 9728 / 0 / 0 0:00:04
Overall, kinda disappointed with the lack of overclocking. (Which was warned about due to a bad batch of DC-DC..) While I do have one fast unit, One of the units won't even run at the STOCK clocks without X'ing out at least 1-2 of the ASIC's, and most of them can't survive even bumping up to 225. At this stage 2 or 4 power connectors doesn't seem to matter at all. The 500GH unit is running on (2), and the 420-units won't go faster on (4). I'm running 3 S3's per 1300W power supply, so it's not a power issue.
Feeling a little deju'vu to KNC here.. First batch got bad dies, 400GH, second batch got fixes, 550GH.
Strangely on both of mine. their conf is
option 'freq_value' '0982' #225M in comment instead of #250
option 'chip_freq' '250'
option 'timeout' '16'
The 225 settings are (currently using this setting @ 453 gh/s avg from 440gh/s at #218.75M)
option 'freq_value' '0882' #225M
option 'chip_freq' '225'
option 'timeout' '16'
I was looking for the #250 option which wasnt there and then i spotted the chip freq value of 250 in the #225 settings
option 'freq_value' '0982' #225M in comment instead of #250
option 'chip_freq' '250'
option 'timeout' '16'
The 225 settings are (currently using this setting @ 453 gh/s avg from 440gh/s at #218.75M)
option 'freq_value' '0882' #225M
option 'chip_freq' '225'
option 'timeout' '16'
I was looking for the #250 option which wasnt there and then i spotted the chip freq value of 250 in the #225 settings
Same happened with me. Seems to just be a type o on a very early version of the software.
Is there a go-to power supply any of you have been going to power 2-4 of these units?
Most seem limited on PCI-e cables, I would love to know.
Most seem limited on PCI-e cables, I would love to know.
I'm using an OCZ 1000W ZX. Can cope with a slight overclock looking at the numbers, but as yet unable to verify. Runs perfect on stock. Has 6 PCIE cables.
Strangely on both of mine. their conf is
option 'freq_value' '0982' #225M in comment instead of #250
option 'chip_freq' '250'
option 'timeout' '16'
The 225 settings are (currently using this setting @ 453 gh/s avg from 440gh/s at #218.75M)
option 'freq_value' '0882' #225M
option 'chip_freq' '225'
option 'timeout' '16'
I was looking for the #250 option which wasnt there and then i spotted the chip freq value of 250 in the #225 settings
option 'freq_value' '0982' #225M in comment instead of #250
option 'chip_freq' '250'
option 'timeout' '16'
The 225 settings are (currently using this setting @ 453 gh/s avg from 440gh/s at #218.75M)
option 'freq_value' '0882' #225M
option 'chip_freq' '225'
option 'timeout' '16'
I was looking for the #250 option which wasnt there and then i spotted the chip freq value of 250 in the #225 settings
used putty, typed in s3 address login root password root but access denied !
some help pls
some help pls
managed to login, stupid me, forgot that i changed password.
got the freq table up. after enter "i" more tables showed up uncomment the freq i want but then !wq or :wq did not work out.
Make sure you hit esc then type in :wq
thx Beastlymac . . . just plugged in the other 2 more pci-e ... just wanna see if i got the lucky s3's
all 3 that i got is hasshing stable around 440'ish stable for more thn 12 hrs.
Is there a go-to power supply any of you have been going to power 2-4 of these units?
Most seem limited on PCI-e cables, I would love to know.
Most seem limited on PCI-e cables, I would love to know.
used putty, typed in s3 address login root password root but access denied !
some help pls
some help pls
managed to login, stupid me, forgot that i changed password.
got the freq table up. after enter "i" more tables showed up uncomment the freq i want but then !wq or :wq did not work out.
Make sure you hit esc then type in :wq
Remember to apply a good thermal paste after you've cleaned the original one 
The 4x cabling: seeing the paths on the board it look like both inputs are powering up exactly the same (i.e. this is not an additional board/unit, just a duplicate input) so I guess the "use 2 cables per hashing board" requirement is mostly due to possible cable/solder overheating due to high amperage.
Has anyone already tried sticking radiators to the R47 elements? does this do the same magic as it did in the S1's? (lower HW which seems to be the issue here)
The 4x cabling: seeing the paths on the board it look like both inputs are powering up exactly the same (i.e. this is not an additional board/unit, just a duplicate input) so I guess the "use 2 cables per hashing board" requirement is mostly due to possible cable/solder overheating due to high amperage.
Has anyone already tried sticking radiators to the R47 elements? does this do the same magic as it did in the S1's? (lower HW which seems to be the issue here)
And how do you clean the paste off the chip ?? With alcohol ??
Isopropyl alcohol works the best, usually for thermal paste.
Found out why one of my units was not hashing and having X's on the chips. Paste was everywhere on the inside. Even the screws holding the heat sink on had thermal paste on them. If you have a unit that is not hashing correctly, I suggest taking it apart and cleaning them.
And how do you clean the paste off the chip ?? With alcohol ?? used putty, typed in s3 address login root password root but access denied !
some help pls
some help pls
managed to login, stupid me, forgot that i changed password.
got the freq table up. after enter "i" more tables showed up uncomment the freq i want but then !wq or :wq did not work out.
Found out why one of my units was not hashing and having X's on the chips. Paste was everywhere on the inside. Even the screws holding the heat sink on had thermal paste on them. If you have a unit that is not hashing correctly, I suggest taking it apart and cleaning them.
[IM]http://i.imgur.com/8tuhdpR.jpg?1[/img]
[IM]http://i.imgur.com/fASrdWs.jpg[/img]
[IM]http://i.imgur.com/8tuhdpR.jpg?1[/img]
[IM]http://i.imgur.com/fASrdWs.jpg[/img]
Your image alerts everyone to unscrew the heat sink, clean and put some thermal paste again properly.
In your image the paste is not properly placed to exchange heat from ASICs.
Do it even if its hashing well if it should be cared more. Else, do it if, the temperature is abnormal or hashrate is lower.
Found out why one of my units was not hashing and having X's on the chips. Paste was everywhere on the inside. Even the screws holding the heat sink on had thermal paste on them. If you have a unit that is not hashing correctly, I suggest taking it apart and cleaning them.
Anybody could post high quality pics of the back side of the PCB under the 53355 chips?
I can't read the values on the high res pics from Dogie guide, and it seems it's the place they have put the divider resistors that we'll need for overvolting. I believe I already identified the resistors, but I'll need the value to start the work while waiting for my S3 to arrive..
I can't read the values on the high res pics from Dogie guide, and it seems it's the place they have put the divider resistors that we'll need for overvolting. I believe I already identified the resistors, but I'll need the value to start the work while waiting for my S3 to arrive..
Windows user doing unix hacker things...
Where is my popcorn?
Where is my popcorn?
used putty, typed in s3 address login root password root but access denied !
some help pls
some help pls
Jump to: