What Do you Guys think about the number of HW errors Is it good or bad Personally I don't like it but if they are good what the hell at this point ?
Topic: ANTMINER S7 is available at bitmaintech.com with 4.86TH/s, 0.25J/GH - page 190. (Read 527809 times)
November 24, 2015, 09:21:29 PM
What Do you Guys think about the number of HW errors Is it good or bad Personally I don't like it but if they are good what the hell at this point ?
November 24, 2015, 03:37:06 PM
Very interesting and not what I had expected. They have got rid of the bottom 9 chips in the chain, as opposed to 9 out of the middle I had predicted. What I do not understand is what and why that additional component (Large Inductor?) and the additional capacitors are for? Unless perhaps there is a buck converter chip on the other side of the board & we have moved away from the string design? Need a Sidehack opinion here? 
Or perhaps it's a hybrid with a buck converter to a slightly lower, Adjustable? voltage & then still a string for the chips?
Also a picture of the other side of the board would be great?
Rich
Or perhaps it's a hybrid with a buck converter to a slightly lower, Adjustable? voltage & then still a string for the chips?Also a picture of the other side of the board would be great?
Rich
In some cases, inductors are installed to offset the affects of capacitance. Especially, when high frequencies are involved. Those round components near the inductor looks like capacitors. It has me curious as well.
this modification is necessary, because bitmain chip voltage and data voltages are different, Serial chip feeding becomes need more stable voltage.
when chip goes down to under 0,61 Volt then chip fried. (under 11V) if overclock your miner you must to have a lot of amperes, otherwise your voltages goes down immidiately. i add 2200uF/16V capacitor each hashboard. This capacitors filter SMPS's peak and 500Khz 12V charge time. we are talk about more than 100A. This hungry machine suck 12V quicly. and ferrit inductor job is choke inductor too. this method is more stable for silly PSU's.
November 24, 2015, 12:57:47 PM
it was the case with SP20/SP30/S1, but not really the case of S7. It stays pretty much close to 0.25J/Gh, in overclock and underclock.
The little difference is probably all due to fans. As far as HW errors are concerned, B5 went from 0.0000 (13 errors in two days) at 600 to 0.0030% at 625, which is fine.
I tried the 625 setting on one of my S7s and I got great results too. I'm going to wait till the 90 warranty is up before I turn the other few up though, just in case I run into any problems with them in the next few months.
November 24, 2015, 12:35:22 PM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
why not, looks okay to me.
sorry, don't know why I wrote German^^
Because it's less hashpower and more power consumption previous batches. I would like something like batch 5, kinda like that one.
hahahha i had to google translate that. i understand people say its an overpriced already overclocked model, but people with batch 1-5, how high can you overclock it without too much hw ? im only going to try that after 90 days though for warranty purposes
not only HW errors, but about an extra 100w for a few hundred GH.... in my experience it is almost ALWAYS better to underclock then overclock
it was the case with SP20/SP30/S1, but not really the case of S7. It stays pretty much close to 0.25J/Gh, in overclock and underclock.
The little difference is probably all due to fans. As far as HW errors are concerned, B5 went from 0.0000 (13 errors in two days) at 600 to 0.0030% at 625, which is fine.
November 24, 2015, 12:27:09 PM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
why not, looks okay to me.
sorry, don't know why I wrote German^^
Because it's less hashpower and more power consumption previous batches. I would like something like batch 5, kinda like that one.
hahahha i had to google translate that. i understand people say its an overpriced already overclocked model, but people with batch 1-5, how high can you overclock it without too much hw ? im only going to try that after 90 days though for warranty purposes
not only HW errors, but about an extra 100w for a few hundred GH.... in my experience it is almost ALWAYS better to underclock then overclock
November 24, 2015, 12:23:27 PM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
why not, looks okay to me.
sorry, don't know why I wrote German^^
Because it's less hashpower and more power consumption previous batches. I would like something like batch 5, kinda like that one.
hahahha i had to google translate that. i understand people say its an overpriced already overclocked model, but people with batch 1-5, how high can you overclock it without too much hw ? im only going to try that after 90 days though for warranty purposes
November 24, 2015, 11:45:50 AM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
why not, looks okay to me.
sorry, don't know why I wrote German^^
Because it's less hashpower and more power consumption previous batches. I would like something like batch 5, kinda like that one.
November 24, 2015, 11:45:44 AM
My B4 at 575-1156W at the wall with EVGA1300 G2 gold (0.2513 J/Gh), 4600Gh 3960/3840rpm, 77F (25C) ambient
MY B5 at 625-1289W (0.2577 J/Gh) at the wall with EVGA1300 G2 Gold, 5001Gh, 4200/3960rpm, 77F
B5 was 1250, not 1210W at 600 with fan at 4400/4080
I mention fan speed because it affects W at the wall, naturally.
Was not with 220V and 40% Fan
no, 110/120V and auto fan for both. Did they specify 40% fan for 0.25? I tried lowering fan on B1 and B2, but results were that they somehow overheated even when sensor was reading OK temp, and as a result had some software glitches, so I stopped lowering fan below 65. In any case, I do auto for a few days to let them "settle in".
the difference from expected J/Gh is quite small and so far, I am quite happy.
November 24, 2015, 11:35:48 AM
My B4 at 575-1156W at the wall with EVGA1300 G2 gold (0.2513 J/Gh), 4600Gh 3960/3840rpm, 77F (25C) ambient
MY B5 at 625-1289W (0.2577 J/Gh) at the wall with EVGA1300 G2 Gold, 5001Gh, 4200/3960rpm, 77F
B5 was 1250, not 1210W at 600 with fan at 4400/4080
I mention fan speed because it affects W at the wall, naturally.
Was not with 220V and 40% Fan
November 24, 2015, 11:23:53 AM
Batch 5. Not too shabby.
did anyone do a wattage at the wall at 625? I was pulling 1170 at 593 if memory serves.
My B4 at 575-1156W at the wall with EVGA1300 G2 gold (0.2513 J/Gh), 4600Gh 3960/3840rpm, 77F (25C) ambient
MY B5 at 625-1289W (0.2577 J/Gh) at the wall with EVGA1300 G2 Gold, 5001Gh, 4200/3960rpm, 77F
B5 was 1250, not 1210W at 600 with fan at 4400/4080
I mention fan speed because it affects W at the wall, naturally.
November 24, 2015, 11:23:13 AM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
why not, looks okay to me.
It is not OK, pay more BTC for overclocked miner.
Basic version of this miner was ANTMINER S7 BATCH 6 4TH/s and 4 BTC . I can overclock miner himself, If it's possible.
And Power Consumption: 1293W + 10% (at the wall, with APW3, 93% efficiency, 25C ambient temp)
with Hash Rate: 4.73 TH/s ±5% does not in any way Power Efficiency: 0.25 J/GH + 10%
more like 0.273 W/Ghs versus 0,249 W/GHs patch 3 and 5
So overclocking do not give better miners, only more sales revenues for Bitmain.
I skip this batch and look forward to something better and not overclocked.
November 24, 2015, 11:09:25 AM
Batch 5. Not too shabby.
did anyone do a wattage at the wall at 625? I was pulling 1170 at 593 if memory serves.
November 24, 2015, 10:47:47 AM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
why not, looks okay to me.
November 24, 2015, 08:43:27 AM
Batch 8 ist mir nicht sympathisch... also halt bis nächstes Jahr warten, was scheisse ist..
November 24, 2015, 07:38:04 AM
looks like we're going to have a + version next year.
nothing new until end of this year tho
mail from Bitmain :
Hi,
new model Not this year! Plus version we are considering, but no exact info we can provide now!
Best Regards
nothing new until end of this year tho
mail from Bitmain :
Hi,
new model Not this year! Plus version we are considering, but no exact info we can provide now!
Best Regards
November 24, 2015, 02:19:43 AM
@not.you: nice error rate at that speed!
Did anyone went above 5gh? or the design of the power for the asics doesn't allow it?
thx
Did anyone went above 5gh? or the design of the power for the asics doesn't allow it?
thx
the chip can go much higher according to the datasheet
Typical hash rate and power
Voltage(V) Hash Rate(GH/S) Current(A) Total power(W) W/GH
0.71 38.75 14.350 10.189 0.263
0.66 32.50 10.760 7.102 0.219
0.60 21.25 6.410 3.846 0.181
need adjustable voltage to up the power, pcie cables + connectors have to be able to take the heat, better cooling, depends on chip production batch, etc etc ...
of course efficiency, is it worthwhile to oc that much ?
basic rule is : the more u oc = the shorter the lifespan, less stable, more heat
November 24, 2015, 02:18:48 AM
I can't wait to see patch 9, I have a good feeling about it 
.
We have 700m frequency with patch 8. I think the new patch could hit 5.6Th-6Th easily if they added the same number of chips as patch 7 (162 chips) + the same frequency as patch 8.
.We have 700m frequency with patch 8. I think the new patch could hit 5.6Th-6Th easily if they added the same number of chips as patch 7 (162 chips) + the same frequency as patch 8.
November 24, 2015, 12:40:32 AM
@not.you: nice error rate at that speed!
Did anyone went above 5gh? or the design of the power for the asics doesn't allow it?
thx
Did anyone went above 5gh? or the design of the power for the asics doesn't allow it?
thx
November 24, 2015, 12:09:05 AM
Batch 6 and 8 135 45 chip blade
those caps + coil .. used to "help" boost the voltage
so it can be clocked to 700mhz n gives out the "advertised" hashrate ?i think i read somewhere if anything more than +12V will help to increase hashrate & more stable
hmmm, up it to 12.5V n see how it goes ? i think that might slow down or stop the wild hashrate fluctuations ? November 23, 2015, 11:30:11 PM
There is different firmware for some frequencies though. I think that is what he is talking about.
Take a look at bitmain S7 firmwares: https://bitmaintech.com/support.htm?pid=00720151023101739550pOBiK5sO06A0
Yes I had seen that. I wonder if they are essentially the same with the exception of which freq is default. My batch 5 came with the latest firmware on it.
interesting...my batch 5 still has the Sept 15 software.
It gets to 5001Gh at 625 "Like a BOSS" with 0.0029% error so far. We shall see.
Jump to: