Pages:
Author

Topic: T17 OC/OV. S17 Series OC/OV, Antminer B7 OC/OV, S9k, Z11 + others 100% FREE - page 19. (Read 10245 times)

jr. member
Activity: 559
Merit: 4
@chipless

Are you able make firmware with lower clocks? Is there hidden page minerAdvanced.cgi where you can choose turbo or balance mode but they doesnt do anything. Are you able add here for example clocks? 750MHz 700MHz 650MHz and 600MHz? Of course with volts mods also so miner can choose some value and finally I can replace fans.

Thanks.

I am currently working on the autotuning which should help some. Eventually clocking can will be able to be done from the web gui like other miners. There will be more updates as time goes on. I currently have the clock set at 775 the next release should be a little higher until I get the rest together. This week updates may be slow I have a lot to do and may not get much time to tweak the firmware some more.

My biggest goal was to get something out there that the majority of users can use right away, now I can work on the tweaks.


As far as duplicate shares or rejected shares that may be a pool problem, I have a low reject rate from flypool but will add it to the list of things to work with. I have heard that even with the stock fw the reject rate was higher then other miners.
newbie
Activity: 2
Merit: 0
Beta 3 Update:

Luxor admin is telling me that my Z11 is submitting many duplicate shares, likely the cause of most of my rejected shares.

Elapsed           KSol/S(RT)   KSol/S(avg)
21h18m56s   145.63           146.41

Accepted shares: 7860
Shares rejected: 1738
Stale shares: 101

7860 / (7860+1738+101) = 81.01% share acceptance rate / efficiency

HW Errors: 14,586

Fan Speed: 100%

Chip Temps: 82c-86c
----------------------------------------------------------------------------------------------------------------------------------------
From the Luxor Discord:

NickH
Today at 2:14 PM

I see them on the server
that's the reject reason

2019/05/05 19:31:53 Error Message: {"id":9369,"version":"2.0","result":false,"error":[22,"Duplicate share from miner at address abcdefghijklmnopqrstuvwxyz",null]}
---------------------------------------------------------------------------------------------------------------------------------------

EDIT: I just swapped pools to see if I can replicate the issue.
full member
Activity: 176
Merit: 100
@chipless

Are you able make firmware with lower clocks? Is there hidden page minerAdvanced.cgi where you can choose turbo or balance mode but they doesnt do anything. Are you able add here for example clocks? 750MHz 700MHz 650MHz and 600MHz? Of course with volts mods also so miner can choose some value and finally I can replace fans.

Thanks.
jr. member
Activity: 559
Merit: 4
@chipless -  

9h40m16s   158.57   151.34

I give you props on bypassing the voltage limits you thought were in the PIC -- now we just gotta figure out what the voltage maximums are.

-j

Anyone having throttle back down problems should try to set the fan manually to 75% or higher. I run mine at 90%, many problems with the miners is the voltage chip gets warmer and cuts out. The only way to keep it cooler is to turn up the fan. This scenario can happen without any errors being set in the log. Bitmain needed small heatsinks on the voltage regulator.

I guess I have to say thanks....I guess...Coming from you thats odd but.. you are also welcome for finding the spot so you can now volt/clock per asic. I would like to know the file format for the data in chips_freq.config, i was considering trying to enable it to read the volt or freq thru the file...I had it as high as 1200mV so far running in 75-90c range. I have all my miners in an air-conditioned 12x12 building with all the units vented out the side. So the heat may be a problem for some until freq and voltage are added back to the config file for some control. I am in the deep south where it has been in the high 80's and humid.

Anything below 870mV you lose speed

I am going to get it to 156k for free then after that paids users....I have a plan that will be fair and work for everyone's budget. The details will be laid out on the first post when it is all ready

Eh, the chips_freq.config is stale code. I futzed with it some.

I reverse engineered the voltage control communications to the PIC and ISL this evening and by passed the factory stuff in favor of my own initialization code for frequency and voltage on a per-ASIC (not just per hashboard) basis (I already have the frequency stuff done from my previous work).

I am still not happy with the predictability of results, unfortunately. While it is possible I am testing with a machine that just isn't up to par (which would change the equation if so), the more likely scenario is that holding a stable overclock across a broad range of systems is improbable.

Who knows...

-j


Look into the voltage throttling the speed back down. It seams when I hit an area there is when it stopped throttling back to around 142 and instead started going up.It is either there or the scan freq that happening behind the scenes. That is why mine is in the 2 minute loop when it is not in the scan loop it will throttle back down but as long as it is in the 2 minute loop it holds steady. Weird...maybe resetting the 142 limit which is coming from somewhere I dont know of.

I cant explain none of it all I can really tell you is that it is a mess. If you are ever kind throw me a factory image with the voltage --bitmain-voltage and the --bitmain-freq re-enabled. I would like to try a couple things there but am busy doing too many other things right now to enable it.

If I read this correctly, you said "scanning keeps things fast". The scan loop and the 'turbo' thread each just do some calculations and ultimately set a frequency on each asic, which takes a couple of clock cycles at most.

The scan takes a split second at most to run to completion for what it is worth.

-j




Yes it was weird aparently the way I went into it the scan kept it higher then without. I am trying the full auto-tune currently have it running and  needs some tweaking for sure but have sustained good rates with it.
member
Activity: 504
Merit: 51
@chipless -  

9h40m16s   158.57   151.34

I give you props on bypassing the voltage limits you thought were in the PIC -- now we just gotta figure out what the voltage maximums are.

-j

Anyone having throttle back down problems should try to set the fan manually to 75% or higher. I run mine at 90%, many problems with the miners is the voltage chip gets warmer and cuts out. The only way to keep it cooler is to turn up the fan. This scenario can happen without any errors being set in the log. Bitmain needed small heatsinks on the voltage regulator.

I guess I have to say thanks....I guess...Coming from you thats odd but.. you are also welcome for finding the spot so you can now volt/clock per asic. I would like to know the file format for the data in chips_freq.config, i was considering trying to enable it to read the volt or freq thru the file...I had it as high as 1200mV so far running in 75-90c range. I have all my miners in an air-conditioned 12x12 building with all the units vented out the side. So the heat may be a problem for some until freq and voltage are added back to the config file for some control. I am in the deep south where it has been in the high 80's and humid.

Anything below 870mV you lose speed

I am going to get it to 156k for free then after that paids users....I have a plan that will be fair and work for everyone's budget. The details will be laid out on the first post when it is all ready

Eh, the chips_freq.config is stale code. I futzed with it some.

I reverse engineered the voltage control communications to the PIC and ISL this evening and by passed the factory stuff in favor of my own initialization code for frequency and voltage on a per-ASIC (not just per hashboard) basis (I already have the frequency stuff done from my previous work).

I am still not happy with the predictability of results, unfortunately. While it is possible I am testing with a machine that just isn't up to par (which would change the equation if so), the more likely scenario is that holding a stable overclock across a broad range of systems is improbable.

Who knows...

-j


Look into the voltage throttling the speed back down. It seams when I hit an area there is when it stopped throttling back to around 142 and instead started going up.It is either there or the scan freq that happening behind the scenes. That is why mine is in the 2 minute loop when it is not in the scan loop it will throttle back down but as long as it is in the 2 minute loop it holds steady. Weird...maybe resetting the 142 limit which is coming from somewhere I dont know of.

I cant explain none of it all I can really tell you is that it is a mess. If you are ever kind throw me a factory image with the voltage --bitmain-voltage and the --bitmain-freq re-enabled. I would like to try a couple things there but am busy doing too many other things right now to enable it.

If I read this correctly, you said "scanning keeps things fast". The scan loop and the 'turbo' thread each just do some calculations and ultimately set a frequency on each asic, which takes a couple of clock cycles at most.

The scan takes a split second at most to run to completion for what it is worth.

-j


jr. member
Activity: 559
Merit: 4
@chipless -  

9h40m16s   158.57   151.34

I give you props on bypassing the voltage limits you thought were in the PIC -- now we just gotta figure out what the voltage maximums are.

-j

Anyone having throttle back down problems should try to set the fan manually to 75% or higher. I run mine at 90%, many problems with the miners is the voltage chip gets warmer and cuts out. The only way to keep it cooler is to turn up the fan. This scenario can happen without any errors being set in the log. Bitmain needed small heatsinks on the voltage regulator.

I guess I have to say thanks....I guess...Coming from you thats odd but.. you are also welcome for finding the spot so you can now volt/clock per asic. I would like to know the file format for the data in chips_freq.config, i was considering trying to enable it to read the volt or freq thru the file...I had it as high as 1200mV so far running in 75-90c range. I have all my miners in an air-conditioned 12x12 building with all the units vented out the side. So the heat may be a problem for some until freq and voltage are added back to the config file for some control. I am in the deep south where it has been in the high 80's and humid.

Anything below 870mV you lose speed

I am going to get it to 156k for free then after that paids users....I have a plan that will be fair and work for everyone's budget. The details will be laid out on the first post when it is all ready

Eh, the chips_freq.config is stale code. I futzed with it some.

I reverse engineered the voltage control communications to the PIC and ISL this evening and by passed the factory stuff in favor of my own initialization code for frequency and voltage on a per-ASIC (not just per hashboard) basis (I already have the frequency stuff done from my previous work).

I am still not happy with the predictability of results, unfortunately. While it is possible I am testing with a machine that just isn't up to par (which would change the equation if so), the more likely scenario is that holding a stable overclock across a broad range of systems is improbable.

Who knows...

-j


Look into the voltage throttling the speed back down. It seams when I hit an area there is when it stopped throttling back to around 142 and instead started going up.It is either there or the scan freq that happening behind the scenes. That is why mine is in the 2 minute loop when it is not in the scan loop it will throttle back down but as long as it is in the 2 minute loop it holds steady. Weird...maybe resetting the 142 limit which is coming from somewhere I dont know of.

I cant explain none of it all I can really tell you is that it is a mess. If you are ever kind throw me a factory image with the voltage --bitmain-voltage and the --bitmain-freq re-enabled. I would like to try a couple things there but am busy doing too many other things right now to enable it.
member
Activity: 504
Merit: 51
@chipless -  

9h40m16s   158.57   151.34

I give you props on bypassing the voltage limits you thought were in the PIC -- now we just gotta figure out what the voltage maximums are.

-j

Anyone having throttle back down problems should try to set the fan manually to 75% or higher. I run mine at 90%, many problems with the miners is the voltage chip gets warmer and cuts out. The only way to keep it cooler is to turn up the fan. This scenario can happen without any errors being set in the log. Bitmain needed small heatsinks on the voltage regulator.

I guess I have to say thanks....I guess...Coming from you thats odd but.. you are also welcome for finding the spot so you can now volt/clock per asic. I would like to know the file format for the data in chips_freq.config, i was considering trying to enable it to read the volt or freq thru the file...I had it as high as 1200mV so far running in 75-90c range. I have all my miners in an air-conditioned 12x12 building with all the units vented out the side. So the heat may be a problem for some until freq and voltage are added back to the config file for some control. I am in the deep south where it has been in the high 80's and humid.

Anything below 870mV you lose speed

I am going to get it to 156k for free then after that paids users....I have a plan that will be fair and work for everyone's budget. The details will be laid out on the first post when it is all ready

Eh, the chips_freq.config is stale code. I futzed with it some.

I reverse engineered the voltage control communications to the PIC and ISL this evening and by passed the factory stuff in favor of my own initialization code for frequency and voltage on a per-ASIC (not just per hashboard) basis (I already have the frequency stuff done from my previous work).

I am still not happy with the predictability of results, unfortunately. While it is possible I am testing with a machine that just isn't up to par (which would change the equation if so), the more likely scenario is that holding a stable overclock across a broad range of systems is improbable.

Who knows...

-j
full member
Activity: 1190
Merit: 108
Wow, i found many nice information here, i have antminer z11 too in my home, ill try what peoples said above, i hope it will increase my miners speed as well, thanks for the OP for this great information, bless for you.
I will try it and post it here to share with you all.
jr. member
Activity: 559
Merit: 4
@chipless -  

9h40m16s   158.57   151.34

I give you props on bypassing the voltage limits you thought were in the PIC -- now we just gotta figure out what the voltage maximums are.

-j

Anyone having throttle back down problems should try to set the fan manually to 75% or higher. I run mine at 90%, many problems with the miners is the voltage chip gets warmer and cuts out. The only way to keep it cooler is to turn up the fan. This scenario can happen without any errors being set in the log. Bitmain needed small heatsinks on the voltage regulator.

I guess I have to say thanks....I guess...Coming from you thats odd but.. you are also welcome for finding the spot so you can now volt/clock per asic. I would like to know the file format for the data in chips_freq.config, i was considering trying to enable it to read the volt or freq thru the file...I had it as high as 1200mV so far running in 75-90c range. I have all my miners in an air-conditioned 12x12 building with all the units vented out the side. So the heat may be a problem for some until freq and voltage are added back to the config file for some control. I am in the deep south where it has been in the high 80's and humid.

Anything below 870mV you lose speed

I am going to get it to 156k for free then after that paids users....I have a plan that will be fair and work for everyone's budget. The details will be laid out on the first post when it is all ready
member
Activity: 504
Merit: 51
@chipless - 

9h40m16s   158.57   151.34

I give you props on bypassing the voltage limits you thought were in the PIC -- now we just gotta figure out what the voltage maximums are.

-j
newbie
Activity: 2
Merit: 0
No problem man,overclocking an voltage adjustment is a touchy thing,one 1Mhz higher crashes when 1 lower is perfect,and I'm sure theirs more to it with asic chips. Bitmain locking them down was a pretty sad day,I was hoping they'd open source their FW,but that was dream. And thanks @weremy73 for the updates!
newbie
Activity: 2
Merit: 0
Just sharing some observations. The beta 3 firmware held over 151-153 ksols for most of the day, but settled back down to 145 ksols recently with chip temps in the 81c-85c with fans set to 100%. Another Z11 twelve inches away running Bitmain 4/23 FW is at 70c-73c with fans at 95%. Re-releasing an April 23 version to cut down on hardware errors at the very least (and any other positive unintended consequences) seems like a no-brainer.
--------------------------------------------------------------------------------------------------
Beta 3:

Elapsed          MH/S(RT)          MH/S(avg)
8h44m23s      146.45              145.66

HW Errors: 7,813

Power Draw: 1780w @ 120v
---------------------------------------------------------------------------------------------------
Bitmain 4/23:

Elapsed          MH/S(RT)          MH/S(avg)
8h59m26s     132.37              137.99

HW Errors: 649

Power Draw: 1492w @ 120v
----------------------------------------------------------------------------------------------------
jr. member
Activity: 559
Merit: 4
Keep up the good work Chipless,I see your just trying to get our aisc's control back,Shrug off anyone trying to put you down,or discredit your work. I will be following/using your CFW's.

Thank you, the latest tweak has it up over 154 k/sol ,,,,, Baby steps I dont want to screw a 2k machine by pusing it too far. Takes about an hour to see what it tops off at so every tweak can take as much as 1 hr running


Elapsed    KSol/S(RT)    KSol/S(avg)
1h2m17s    164.96               154.3

2 hrs runtime... Thats where that tweak peaks out on my machine
Elapsed   KSol/S(RT)   KSol/S(avg)
1h53m26s   158.9             154.41
newbie
Activity: 2
Merit: 0
Keep up the good work Chipless,I see your just trying to get our aisc's control back,Shrug off anyone trying to put you down,or discredit your work. I will be following/using your CFW's.
jr. member
Activity: 559
Merit: 4
epudd, I dont know what to tell you others are reporting to me that everything is working well. I guess if it is broken then fix it.So lets not sit here to rry to discredit the firmware because you know as well as I do it was valid stuff I asked you about your non-disclosed control. This firmware is nothing more then a modded binary and free to try or use. Why dont you go back to fixing your own problem and let people give it a try without all your bullshit child games.

Seriously not trying to discredit. Honestly asking. You asked for feedback and I am providing some on behalf of another individual.

He will happily pay 100 if you can give him 175k sol... so trying to see if what you have released is working for him.. and it has left a question as to why the same results are not being achieved.

Sometimes you have to read between the lines and realize not everything is an attack or troll. Nuance is a skill.

-j

I do not know why you are getting bad results I do know it take 1-2 hours for these to max out. The heat is due to overclocking and increased voltage. From some of my testing you cant get higher speeds without having some extra power to go with it. I am working on faster speeds as we speak trying a few different things I have found that no matter what you arent getting the next speed range without more power. But I may be missing something.

This is what 1 machine is reporting and you should have similar results
Elapsed   KSol/S(RT)   KSol/S(avg)
11h47m20s   156.28   151.62

I need to update the base to the 4/23/19 release this base is the first bitmain image release

Latest test after tweaking some more
Elapsed   KSol/S(RT)   KSol/S(avg)
1h4m32s   158.41             153.34
member
Activity: 504
Merit: 51
epudd, I dont know what to tell you others are reporting to me that everything is working well. I guess if it is broken then fix it.So lets not sit here to rry to discredit the firmware because you know as well as I do it was valid stuff I asked you about your non-disclosed control. This firmware is nothing more then a modded binary and free to try or use. Why dont you go back to fixing your own problem and let people give it a try without all your bullshit child games.

Seriously not trying to discredit. Honestly asking. You asked for feedback and I am providing some on behalf of another individual.

He will happily pay 100 if you can give him 175k sol... so trying to see if what you have released is working for him.. and it has left a question as to why the same results are not being achieved.

Sometimes you have to read between the lines and realize not everything is an attack or troll. Nuance is a skill.

-j
jr. member
Activity: 559
Merit: 4
epudd, I dont know what to tell you others are reporting to me that everything is working well. I guess if it is broken then fix it.So lets not sit here to rry to discredit the firmware because you know as well as I do it was valid stuff I asked you about your non-disclosed control. This firmware is nothing more then a modded binary and free to try or use. Why dont you go back to fixing your own problem and let people give it a try without all your bullshit child games.
member
Activity: 504
Merit: 51


Ahh, I see you finally listened to me when I told you that voltage was not actually limited and all you had to do was get rid of one check. Good job!

Although, I do have a question.

In your updated first post you showed an example of the capabilities as part of offering 'custom' images for a one time fee of $100, it showed 154kSol at 16minutes elapsed.

Quote
** Latest Speed Update  5/3/2019 8:59 CST ***
Elapsed   KSol/S(RT)   KSol/S(avg)
16m40s   154.48             154.05

I've tested that image on a Z11 and as close to 16m40s as I could sample, I see the following result. It appears to be the same as all of the other results. Chip temperatures increased by 7C over what this machine was running before.

16m42s   152.68   149.06

Also, I see constant "scan frequency" messages.
Can you please explain what is going on here? What would paying $100 get me when I can't get the results you already claim for free?

-j

Actually the power area yes I missed 1 spot during my going thru it. No big deal, then you are not reading it right because there was never any speed posted for the upgraded images. That speed was a example of what the machines can and are getting and so on. But the data was from a machine.

I suppose you are going to find every little word to comment on about this image. As far as heat of course there is going to be heat, anytime you start to clock up a machine over factory it has more heat. There are still tweaks to be done but it works and its free. THe scan message will get fixed also soon it is not a big deal and doesnt seem to bother the speed any.

I dont understand the excuse you gave people about your release. You dont have to wait for the pool admin crap or whatever. As I been told you were still stuck at the throttle down but other parts were ok.

In the end for free it works pretty dam good.

I also had 1 report the Beta 3 is drawing around 1800 watts time to tweak it a little I guess.



Ignoring all the noise in that response, it doesn’t seem to be working as you describe. I posted the measured results at the same time period you posted results from and I am not seeing any improvements. Trying to understand what is going wrong.

Literally running this to test and seeing different results.

Can you assist?

-j
jr. member
Activity: 559
Merit: 4


Ahh, I see you finally listened to me when I told you that voltage was not actually limited and all you had to do was get rid of one check. Good job!

Although, I do have a question.

In your updated first post you showed an example of the capabilities as part of offering 'custom' images for a one time fee of $100, it showed 154kSol at 16minutes elapsed.

Quote
** Latest Speed Update  5/3/2019 8:59 CST ***
Elapsed   KSol/S(RT)   KSol/S(avg)
16m40s   154.48             154.05

I've tested that image on a Z11 and as close to 16m40s as I could sample, I see the following result. It appears to be the same as all of the other results. Chip temperatures increased by 7C over what this machine was running before.

16m42s   152.68   149.06

Also, I see constant "scan frequency" messages.
Can you please explain what is going on here? What would paying $100 get me when I can't get the results you already claim for free?

-j

Actually the power area yes I missed 1 spot during my going thru it. No big deal, then you are not reading it right because there was never any speed posted for the upgraded images. That speed was a example of what the machines can and are getting and so on. But the data was from a machine.

I suppose you are going to find every little word to comment on about this image. As far as heat of course there is going to be heat, anytime you start to clock up a machine over factory it has more heat. There are still tweaks to be done but it works and its free. THe scan message will get fixed also soon it is not a big deal and doesnt seem to bother the speed any.

I dont understand the excuse you gave people about your release. You dont have to wait for the pool admin crap or whatever. As I been told you were still stuck at the throttle down but other parts were ok.

In the end for free it works pretty dam good.

I also had 1 report the Beta 3 is drawing around 1800 watts time to tweak it a little I guess.


After running for 10 hrs

Elapsed         KSol/S(RT)   KSol/S(avg)
10h36m31s   146.45   151.56
member
Activity: 504
Merit: 51
I dont dodge or chrysler products sorry I prefer GM


oh look here my other machine i just loaded

Elapsed   KSol/S(RT)   KSol/S(avg)   FoundBlocks   LocalWork   Utility   WU   BestShare
48m55s   162.01           151.83



17.1 KH/s   642 (100%)   0 (0%)   7 minutes ago
16.9 KH/s   633 (100%)   0 (0%)   7 minutes ago
15.8 KH/s   594 (100%)   1 (0%)   7 minutes ago
16.1 KH/s   604 (100%)   0 (0%)   7 minutes ago
55.4 KH/s   2079 (100%)   0 (0%)   7 minutes ago
54.5 KH/s   2045 (100%)   1 (0%)   7 minutes ago
149.4 KH/s 5604 (100%)   0 (0%)   7 minutes ago
150.3 KH/s 5637 (100%)   0 (0%)   7 minutes ago


edit...... I will update you again on the last one I loaded
Elapsed   KSol/S(RT)   KSol/S(avg)
1h19m42s   153.33           151.95

most recent flypool also for the rigs above
Hashrate

479.8 KH/s


another edit......the last one loaded is still climbing
Elapsed   KSol/S(RT)   KSol/S(avg)
1h35m57s   152.35             152.13


After 2 hrs this is where it will settle in at
Elapsed    KSol/S(RT)   KSol/S(avg)
1h54m16s    147.43              151.97


power consumption?

I am not sure I would say about 20-50 watts more with the little increase in voltage. It was increased by from 880 mV max to 920 mV and is locked there by me, the frequency is set to 793 which so far from members I have testing there have been no complaints. I am currently finishing up a faster release that will be available for a 1 time fee. The details are on the first post of the thread.

If someone does have the ability to let me know the power consumption that would be nice, mine is on a 220 ac line and I dont have anything to really test it with. I am drawing about 7500 watts with all my miner then another 750 for an air conditioner to keep the room cooler

This is where 1 machine settled at....The other machine is up and down as I tweak it some more.

Elapsed   KSol/S(RT)   KSol/S(avg)
4h30m1s   154.48            151.56



Ahh, I see you finally listened to me when I told you that voltage was not actually limited and all you had to do was get rid of one check. Good job!

Although, I do have a question.

In your updated first post you showed an example of the capabilities as part of offering 'custom' images for a one time fee of $100, it showed 154kSol at 16minutes elapsed.

Quote
** Latest Speed Update  5/3/2019 8:59 CST ***
Elapsed   KSol/S(RT)   KSol/S(avg)
16m40s   154.48             154.05

I've tested that image on a Z11 and as close to 16m40s as I could sample, I see the following result. It appears to be the same as all of the other results. Chip temperatures increased by 7C over what this machine was running before.

16m42s   152.68   149.06

Also, I see constant "scan frequency" messages.
Can you please explain what is going on here? What would paying $100 get me when I can't get the results you already claim for free?

-j
Pages:
Jump to: