You could probably do that - if you don't mind voiding their joke of a warrenttee.
Ha, that will be the least of my concerns, as you say it's a joke. I'll refuse to create an enclosure though, what's next, program my own controller? Fuck them. 
Topic: Avalon Quality vs Antminer Quality (Read 523 times)
You could probably do that - if you don't mind voiding their joke of a warrenttee.
Do you really need the chassis Bitmain uses? Could you buy an S9 and pull the boards along with the control board and put them in your own chassis with better cooling? Aren’t they all just sitting in there?
Based on the issues with the first few S9 batches, and the existence of the T9, the primary issue they seem to be having is "chip to chip variation" is VERY WIDE on 14/16nm node processes when one is trying to push the chip as hard as Bitmain is doing.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
So in your opinion is this a QC issue or a development issue with the chip just can’t handle the load?
Combination - they try to run the chip too close to it's "lowest voltage" and the variations cause them issues - which their design QC hasn't gotten a good handle on.
They also have a very poor design on their airflow management.
They are too happy with their "way compact extreme heat density" design, and would probably have a lot fewer issues if they went to a larger chassis to mount everything in, with GOOD airflow management design - something closer to the old Dragon miners or the Innosilicon A2.
I can't speak to the physical QC of the S9 as I've never owned one - but I've NEVER been fond of "glue-on heatsinks" in ANYTHING ELSE I've suffered through that had those things.
Based on the issues with the first few S9 batches, and the existence of the T9, the primary issue they seem to be having is "chip to chip variation" is VERY WIDE on 14/16nm node processes when one is trying to push the chip as hard as Bitmain is doing.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
So in your opinion is this a QC issue or a development issue with the chip just can’t handle the load?
many believe that due to size of chip 14/16nm and high power sent to it that it can not help but have power leakages.
there are 189 chips doing 13.5 th
there are 189 chips doing 1400 watts
Intel has the i7 6700k and it had a higher fail rate then previous
that was their first 14nm chip and it did have a higher failure rate then previous cpu's
it was the first intel cpu I ever fried out of 300 or more used in builds.
I would love the next design to use less power and less chips and run cooler.
Based on the issues with the first few S9 batches, and the existence of the T9, the primary issue they seem to be having is "chip to chip variation" is VERY WIDE on 14/16nm node processes when one is trying to push the chip as hard as Bitmain is doing.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
So in your opinion is this a QC issue or a development issue with the chip just can’t handle the load?
Based on the issues with the first few S9 batches, and the existence of the T9, the primary issue they seem to be having is "chip to chip variation" is VERY WIDE on 14/16nm node processes when one is trying to push the chip as hard as Bitmain is doing.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
Based on the issues they had with the S7, they may also be trying to push the thermal density more than they should be.
So what’s the deal with Bitmain? Is it the design or poor assembly practices? Maybe the manufacture rate is affecting the quality.
I never had a significant issue with the S5, but my sample size was small.
I've seen quite a few folks with medium-sized S9 deployments talk about 20-30% failure rates (sometimes "one board fail" is counted as a miner fail by them though, even though it's actually a PARTIAL fail).
I've not seen ANY folks with more than a few Avalon miners talk about that kind of failure rate - normally they talk about 5% or less (and again, ONE board failure = miner failure to those folks).
I've seen quite a few folks with medium-sized S9 deployments talk about 20-30% failure rates (sometimes "one board fail" is counted as a miner fail by them though, even though it's actually a PARTIAL fail).
I've not seen ANY folks with more than a few Avalon miners talk about that kind of failure rate - normally they talk about 5% or less (and again, ONE board failure = miner failure to those folks).
Good luck mining solo. You’ll be dancing to the bank if you hit a block.
... I feel like it’s not as easy now to get into it like it was last year...
Same shit, different year. When I first started mining all of my stress came from Black Arrow (and a little bit from BFL). The only stable gear I was able to get my hands on were ASICMINER blades and cubes... and a whole lot of U1 sticks
Getting back into it after a 4 year hiatus and am shocked at the concentration of power in Bitmain's hands, so will not do the S9 thing. I am only running an S3 solo right now, but hope to get serious if I can get my hands on an Avalon 8x or two (or hit a block!)
I’m guessing it’s probably easier and cheaper to fix miners that are doa than invest in higher quality right from the get go. Plus some customers may try to repair it themselves or may have purchased through a reseller which won’t be covered under warranty with Bitmain at least anyway.
What are you talking about? They rarely get an S9 back for repairs, quality must be awesome! Or maybe their process around the "warranty" is so fucking terrible, that people prefer getting them fixed in other ways, instead of shipping the entire unit back, for a fortune, wishing to see it back some day not more broken than before sending it Edit: Yeah, easier to don't give a shit if you can get away with it...
From personal experience I never had an issue with the S5 or S7 series miners. Can't say for the S9's, but apparently a lot of people hate them
I’m guessing it’s probably easier and cheaper to fix miners that are doa than invest in higher quality right from the get go. Plus some customers may try to repair it themselves or may have purchased through a reseller which won’t be covered under warranty with Bitmain at least anyway.
I don't think past experience is a predictor. Since the 8 series is new there's no way to gauge it's reliability before the fact.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
S-9 suck
S-7 not that good
S- 5 a little better
S-3 really really good
Avalon 741 not that good
Avalon 721 good
Avalon 6 good
Avalon 4.1 really really really good
I wonder how are we going to keep mining SHA256 if things keep going in that direction.
Power isn’t too bad. Under .10 cents per Kw. Sounds like Avalon is it. Even if it’s failure rate was no better than an Antminers. Better customer support makes it much better. I really like mining. I feel like I can really be a part of this by mining. I feel like it’s not as easy now to get into it like it was last year. I will press on though. I am waiting for the 821’s to be released. If I find an S9 that is working and isnt too overly priced I may buy one. I’m just not going to get crazy with them I think.
If you have very very cheap power, then the avalon is a better choice. Avalon burns more power and costs more per hash(even when the device itself costs less), other than that there's no problem.
Unlike in Antminers, failure rate has increased considerably. On top of this, customer support is almost non-existent. Whereas Canaan's support would most likely solve(or at least try) your problems within hours.
Unlike in Antminers, failure rate has increased considerably. On top of this, customer support is almost non-existent. Whereas Canaan's support would most likely solve(or at least try) your problems within hours.
I don't think past experience is a predictor. Since the 8 series is new there's no way to gauge it's reliability before the fact.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
S-9 suck
S-7 not that good
S- 5 a little better
S-3 really really good
Avalon 741 not that good
Avalon 721 good
Avalon 6 good
Avalon 4.1 really really really good
I wonder how are we going to keep mining SHA256 if things keep going in that direction. I don't think past experience is a predictor. Since the 8 series is new there's no way to gauge it's reliability before the fact.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
S-9 suck
S-7 not that good
S- 5 a little better
S-3 really really good
Avalon 741 not that good
Avalon 721 good
Avalon 6 good
Avalon 4.1 really really really good
I don't think past experience is a predictor. Since the 8 series is new there's no way to gauge it's reliability before the fact.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
My experience is that S9s are very poor. 6 months in we are down maybe 30% in hash rate.
L3+ have been much more reliable though.
Jump to: