One of my Z9 mini fails at 675, max for good works is 650.
At 650 i see some rejects, after 4hours 24 reject... mining on Antpool,
that is normal?
Topic: Antminer Z9 mini overclocked - page 18. (Read 26129 times)
my 12 batch 1s can all run at 750Mhz and in my setup range from a stable 63C to 74C across them; a lot of variance.
2 of mine run fastest at 725Mhz.
All of them run "best" at 700Mhz or less -- best is a combination of heat vs. hashrate.
At max (750), mine do a stable 192kSol/sec with variance up to ~195.
At 700Mhz, mine do a stable 186kSol/sec with temps in a managable 72C or less.
My setup: https://imgur.com/a/6SBUt2S
(copying my setup as is would be a bad idea for reasons I don't have time to get into at the moment, but you cannot egress 1000cfm out of your house without other problems...).
The most ideal configuration is for 650Mhz across all of mine -- they are stable and cool with minimal (relative) airflow across them (and thus minimal impact to the house environment).
My point here is... even though I can make the batch1s not crash at 750Mhz, that isn't their ideal point. Any unit alone, I can get to 725/750, but when I combine them all together and their heat impacts each other, 650-700 is ideal with minimal 'loss' in hashrate.
... Which puts them in line with the ~681mhz folk are seeing now.
Once my Z9s come in, I'll see how those chips work in comparison....
-j
2 of mine run fastest at 725Mhz.
All of them run "best" at 700Mhz or less -- best is a combination of heat vs. hashrate.
At max (750), mine do a stable 192kSol/sec with variance up to ~195.
At 700Mhz, mine do a stable 186kSol/sec with temps in a managable 72C or less.
My setup: https://imgur.com/a/6SBUt2S
(copying my setup as is would be a bad idea for reasons I don't have time to get into at the moment, but you cannot egress 1000cfm out of your house without other problems...).
The most ideal configuration is for 650Mhz across all of mine -- they are stable and cool with minimal (relative) airflow across them (and thus minimal impact to the house environment).
My point here is... even though I can make the batch1s not crash at 750Mhz, that isn't their ideal point. Any unit alone, I can get to 725/750, but when I combine them all together and their heat impacts each other, 650-700 is ideal with minimal 'loss' in hashrate.
... Which puts them in line with the ~681mhz folk are seeing now.
Once my Z9s come in, I'll see how those chips work in comparison....
-j
Very nice set up
Now there are specs. on Z9 big, as was not shown earlier.
Before power consumption was stated as 1250watt,now 970-1162watt. for big and 300watt for mini, now 247-266watt, so my conclusion that is another chip they using, probably changed the spec. and made a batch rerun, the instability in OC mode ithink is a secundary outcome, the primary goal was to lower energy consumption? or cheaper chip? Any way the first one is more stable and could probably run higher than 750 is changed manualy.
We can also see that there is 3 boards, total 48 chips (vs. mini's 12) so its a "full version", it also says it runs 42ksol/s normal and 44kSol/s in the supposed "Turbo mode"
Thanks
I'm not sure that they have shipped yet -- but should "soon" per their support. As of yesterday there was not a certificate of conformity posted -- they might actually need that in order to ship through US customs, .. not sure.
-j
Take a look at mine i am just waiting for 3 more to arrive on Tuesday https://twitter.com/mrtraderman
Now there are specs. on Z9 big, as was not shown earlier.
Before power consumption was stated as 1250watt,now 970-1162watt. for big and 300watt for mini, now 247-266watt, so my conclusion that is another chip they using, probably changed the spec. and made a batch rerun, the instability in OC mode ithink is a secundary outcome, the primary goal was to lower energy consumption? or cheaper chip? Any way the first one is more stable and could probably run higher than 750 is changed manualy.
We can also see that there is 3 boards, total 48 chips (vs. mini's 12) so its a "full version", it also says it runs 42ksol/s normal and 44kSol/s in the supposed "Turbo mode"
Thanks
I'm not sure that they have shipped yet -- but should "soon" per their support. As of yesterday there was not a certificate of conformity posted -- they might actually need that in order to ship through US customs, .. not sure.
-j
my 12 batch 1s can all run at 750Mhz and in my setup range from a stable 63C to 74C across them; a lot of variance.
2 of mine run fastest at 725Mhz.
All of them run "best" at 700Mhz or less -- best is a combination of heat vs. hashrate.
At max (750), mine do a stable 192kSol/sec with variance up to ~195.
At 700Mhz, mine do a stable 186kSol/sec with temps in a managable 72C or less.
My setup: https://imgur.com/a/6SBUt2S
(copying my setup as is would be a bad idea for reasons I don't have time to get into at the moment, but you cannot egress 1000cfm out of your house without other problems...).
The most ideal configuration is for 650Mhz across all of mine -- they are stable and cool with minimal (relative) airflow across them (and thus minimal impact to the house environment).
My point here is... even though I can make the batch1s not crash at 750Mhz, that isn't their ideal point. Any unit alone, I can get to 725/750, but when I combine them all together and their heat impacts each other, 650-700 is ideal with minimal 'loss' in hashrate.
... Which puts them in line with the ~681mhz folk are seeing now.
Once my Z9s come in, I'll see how those chips work in comparison....
-j
2 of mine run fastest at 725Mhz.
All of them run "best" at 700Mhz or less -- best is a combination of heat vs. hashrate.
At max (750), mine do a stable 192kSol/sec with variance up to ~195.
At 700Mhz, mine do a stable 186kSol/sec with temps in a managable 72C or less.
My setup: https://imgur.com/a/6SBUt2S
(copying my setup as is would be a bad idea for reasons I don't have time to get into at the moment, but you cannot egress 1000cfm out of your house without other problems...).
The most ideal configuration is for 650Mhz across all of mine -- they are stable and cool with minimal (relative) airflow across them (and thus minimal impact to the house environment).
My point here is... even though I can make the batch1s not crash at 750Mhz, that isn't their ideal point. Any unit alone, I can get to 725/750, but when I combine them all together and their heat impacts each other, 650-700 is ideal with minimal 'loss' in hashrate.
... Which puts them in line with the ~681mhz folk are seeing now.
Once my Z9s come in, I'll see how those chips work in comparison....
-j
Now there are specs. on Z9 big, as was not shown earlier.
Before power consumption was stated as 1250watt,now 970-1162watt. for big and 300watt for mini, now 247-266watt, so my conclusion that is another chip they using, probably changed the spec. and made a batch rerun, the instability in OC mode ithink is a secundary outcome, the primary goal was to lower energy consumption? or cheaper chip? Any way the first one is more stable and could probably run higher than 750 is changed manualy.
We can also see that there is 3 boards, total 48 chips (vs. mini's 12) so its a "full version", it also says it runs 42ksol/s normal and 44kSol/s in the supposed "Turbo mode"
Thanks
Now there are specs. on Z9 big, as was not shown earlier.
Before power consumption was stated as 1250watt,now 970-1162watt. for big and 300watt for mini, now 247-266watt, so my conclusion that is another chip they using, probably changed the spec. and made a batch rerun, the instability in OC mode ithink is a secundary outcome, the primary goal was to lower energy consumption? or cheaper chip? Any way the first one is more stable and could probably run higher than 750 is changed manualy.
We can also see that there is 3 boards, total 48 chips (vs. mini's 12) so its a "full version", it also says it runs 42ksol/s normal and 44kSol/s in the supposed "Turbo mode"
I think it would be appropiate for the company to release an explantion to clear all the speculations that florish and demonize their company and damaging the Brand. That would be in their best intrest!
Before power consumption was stated as 1250watt,now 970-1162watt. for big and 300watt for mini, now 247-266watt, so my conclusion that is another chip they using, probably changed the spec. and made a batch rerun, the instability in OC mode ithink is a secundary outcome, the primary goal was to lower energy consumption? or cheaper chip? Any way the first one is more stable and could probably run higher than 750 is changed manualy.
We can also see that there is 3 boards, total 48 chips (vs. mini's 12) so its a "full version", it also says it runs 42ksol/s normal and 44kSol/s in the supposed "Turbo mode"
I think it would be appropiate for the company to release an explantion to clear all the speculations that florish and demonize their company and damaging the Brand. That would be in their best intrest!
Mine was working fine at 662 for 36 hours, then one of the board went to zero... I have to lower to 650. So maybe then run fine only at 500M finally...
I also strongly disagree with such marketing practices. They made sure the first batch would be perfect so all the reviewers would make free publicity for them about a 750M stable Z9mini then the orders should have flow in from people expecting the Z9 mini to all be like the ones they saw in the reviews. Also the first batch came in with new promises of transparency from Bitmain with the published log of each unit shipping but for batch 2 it'S radio silence again.
I also strongly disagree with such marketing practices. They made sure the first batch would be perfect so all the reviewers would make free publicity for them about a 750M stable Z9mini then the orders should have flow in from people expecting the Z9 mini to all be like the ones they saw in the reviews. Also the first batch came in with new promises of transparency from Bitmain with the published log of each unit shipping but for batch 2 it'S radio silence again.
My first bath all pull 750 Frequency with 16 to 17k Sols. Temp can be in high 50s with full fan or mid 60s on auto fan.
Second batch came with the balanced firmware where you cant adjust. I experimented with changing it manually and found anything over 700 a board would cut out. So im running second bath at 700 all stable and running at 15.5Sols and low 60s temp on auto fan.
Overall i would say they react to heat and fan speed exactly the same as the first batch. But second batch issnt as forgiving above 700 frequency.
L
Power supply is 1 APW3++ two every two z9 minis. Ambient temperature in the data centre is around 68 deg right now.
Second batch came with the balanced firmware where you cant adjust. I experimented with changing it manually and found anything over 700 a board would cut out. So im running second bath at 700 all stable and running at 15.5Sols and low 60s temp on auto fan.
Overall i would say they react to heat and fan speed exactly the same as the first batch. But second batch issnt as forgiving above 700 frequency.
L
Power supply is 1 APW3++ two every two z9 minis. Ambient temperature in the data centre is around 68 deg right now.
You dont seem to know alot about ASIC production, things .......
Yeah. I don't know a lot about ASIC production. I'm just quite happy with the second batch performance. Hope I'm happy about the profitability long enough to break even though!
I don’t think it’s limited to one board. I’ve had hardware failures on two out of the three boards on my machines.
snip ......wafers, cut dyes do not meet specfications .... snip
Frankly I think this is a massive overreaction. No chip designed to be run at 10k and running comfortably at 14.5k can be rationally labelled as "not meeting specifications" just because some other chips previously ran at 16.5k.
snip ......wafers, cut dyes do not meet specfications .... snip
Frankly I think this is a massive overreaction. No chip designed to be run at 10k and running comfortably at 14.5k can be rationally labelled as "not meeting specifications" just because some other chips previously ran at 16.5k.
You dont seem to know alot about ASIC production, things can wrong and then you have to discard the whole batch. We can all agree that the problem sits in ONE board. The Chips on them are not capable to run on higher freqvency. You DONT change the specs. in different batches, becuase 99% of the cost is in the making specs and production setup. The so called non-recurring engineering process, the production of singel chips does not cost many pennies each, 1-5%. So who belives that they redesigned some chips to lower OC posibility and put in ONE board, and who thinks there migth be faulty chips caused by production errors and instead of discard them, they made boards of them wich was systematically put in each devive? The answer is obvious isnt it? or lets have have vote? Note the PR-guys are working hard here to find the rigth arguments to save the brand!!
But actually the above said is irrelevant becuase the delivered devices meets whats promised, 10k/sol. The OC expectations is something else. So in this perspective there is nothing wrong with the chips or the device it self. BUT the "irrelevant" expectations was any way a product of the company.
snip ......wafers, cut dyes do not meet specfications .... snip
Frankly I think this is a massive overreaction. No chip designed to be run at 10k and running comfortably at 14.5k can be rationally labelled as "not meeting specifications" just because some other chips previously ran at 16.5k.
snip ......wafers, cut dyes do not meet specfications .... snip
Frankly I think this is a massive overreaction. No chip designed to be run at 10k and running comfortably at 14.5k can be rationally labelled as "not meeting specifications" just because some other chips previously ran at 16.5k.
You dont seem to know alot about ASIC production, things can wrong and then you have to discard the whole batch. We can all agree that the problem sits in ONE board. The Chips on them are not capable to run on higher freqvency. You DONT change the specs. in different batches, becuase 99% of the cost is in the making specs and production setup. The so called non-recurring engineering process, the production of singel chips does not cost many pennies each, 1-5%. So who belives that they redesigned some chips to lower OC posibility and put in ONE board, and who thinks there migth be faulty chips caused by production errors and instead of discard them, they made boards of them wich was systematically put in each devive? The answer is obvious isnt it? or lets have have vote? Note the PR-guys are working hard here to find the rigth arguments to save the brand!!
To unlock back the way we are complaining about not having, go to the Bitmain website and get the firmware. It should have a 500 at the end of it. Back your old one up and flash the new one. Problem solved.
snip ......wafers, cut dyes do not meet specfications .... snip
Frankly I think this is a massive overreaction. No chip designed to be run at 10k and running comfortably at 14.5k can be rationally labelled as "not meeting specifications" just because some other chips previously ran at 16.5k.
Folks, it's pretty obvious what goning on with batch 2, namely "container garbage". Some times something goes wrong in the asic manufacturing process, i.e masking problems i.e adhesion, lithographic mishappenings i.e operating misstakes, etc the result, wafers, cut dyes do not meet specfications, the whole batch is ruined, and is to be discarded (container garbage, some thimes picket up at nigths and used in "cheap apps, toys) BUT in this case BITMAIN descided not to discard them but use the to build "crappy" boards, and systematically place one in each Z9 mini, (i orded 2 of them and each has faulty board!) i would like call this as deviant behaviour unexpected from such a large company! Even if the they sold the device as with 10k specs is not the same product as the 1 batch as there ara unexpected changes i.e Balance and also to come a "turbo upgrade" as noteciced in config file, becuase of the failty batch they had to design a "new concept" compared to the first batch to "compensate". What worries my is that the big Z9 may be build on these chips, and if i'm rigth who nows how these devices will behave on short and long perspective?!?!
I also belive that the mini was to be discontiued inorder to push z9 big, and have cannibalization from running the mini in parallell (as every one would buy 3 minis 3XOV42Ksol/$2550 to achive the same performance but for less money, or 4xOC=64kSol/$3400 plus ability to use more more cupons as well, one for each device), as, the mini suddenly was reinstated, (what every marketing manager would call bad market strategy), underpins my theory that they had to come up with an emergency strategy! Note that their PR department beeing very active on this thread defending the 10k deal!
I also belive that the mini was to be discontiued inorder to push z9 big, and have cannibalization from running the mini in parallell (as every one would buy 3 minis 3XOV42Ksol/$2550 to achive the same performance but for less money, or 4xOC=64kSol/$3400 plus ability to use more more cupons as well, one for each device), as, the mini suddenly was reinstated, (what every marketing manager would call bad market strategy), underpins my theory that they had to come up with an emergency strategy! Note that their PR department beeing very active on this thread defending the 10k deal!
I summarized the steps for overclocked a locked batch 2 Mini:
Use Google Chrome. Log into your Antminer and go to the tab where you would change the frequency (now you only see 'balanced' there). Press F12 which opens the console to see the website code. Press Ctrl+f to search this for ant_freq. You will see that the balanced profile is equal to a clock frequency of 500Mhz. Here you can edit the number (in the code) to whatever you like (675 seems to be a safe bet) and then make sure you select the balanced profile in the antminer gui and click save/apply. You got yourself an overclocked batch 2 Z9 Mini.
Use Google Chrome. Log into your Antminer and go to the tab where you would change the frequency (now you only see 'balanced' there). Press F12 which opens the console to see the website code. Press Ctrl+f to search this for ant_freq. You will see that the balanced profile is equal to a clock frequency of 500Mhz. Here you can edit the number (in the code) to whatever you like (675 seems to be a safe bet) and then make sure you select the balanced profile in the antminer gui and click save/apply. You got yourself an overclocked batch 2 Z9 Mini.
How did you guys managed to change the clock setting? For the new batch I only see "balanced" in the advanced options. My previous miner from batch one I can simply change to 750Mhz in the drop down. The new batch 2 miner do not have this option. Is it locked at 500Mhz?
This was discussed in this thread before. Check https://bitcointalksearch.org/topic/m.45002416
How did you guys managed to change the clock setting? For the new batch I only see "balanced" in the advanced options. My previous miner from batch one I can simply change to 750Mhz in the drop down. The new batch 2 miner do not have this option. Is it locked at 500Mhz?
Yep about the same here, thankfully summer is ending.
Jump to: