Topic: Bitmain Antminer S6 and S7 Speculation - page 10. (Read 33553 times)

I can see what you mean there.  Either they were hashing or they could be "refurbished."  Either case is not good.  If one pays for a "new" unit, they want a "new" unit!!!  A used or refurbished unit should be at a used or refurbished price.

This could also explain why the price of the rigs have come down some in the last couple of weeks.

First, inside many of S5 there was a lot of dust.

Second, if there were more or less large batches of S5s we should get the same versions of boards at the same time. Guys from bits.media told about v1.5, v1.7, v1.9 they received several days ago. IMHO hardware revisions should be sold according to the rule "first revision produced - first sold". So since all HW versions of hashing blades being sold these days, we can conclude that they all together were laying somewhere in Bitmain's data center and hashing (collecting the dust...), and now they sell it out by a "first HW version that came to hand" scheme.

Difficulty has officially changed...

Apr 05 2015   49,446,390,688         5.84% increase       353,951,052 GH/s

Like I mentioned earlier, I'm sure there are others on bitcointalk.org who understand engineering and design about hardware and chips more so than myself.  I was not quite sure what the BM1384 could do.  I have not taken the time to research it very well.  0.21 watts a GH would be great.  If this is the case, we might see Farm Wars back in business again and difficulty climbing again.

An interesting theory you have there.  I suppose it can make one wonder.  I only hope this is not the case.  Neither one sound good to me in the least.

I have not seen nor heard of these rumors you speak of.  How do you know the rigs received in Russia were used rigs?

You know the BM1384 can do under 0.35W/GH, right? I worked over some numbers a month ago speculating S2 upgrade kits, with suitable chip density they could make a 3.6TH miner off a 1KW PSU with the same power density (and therefore same fan volume) as the S2 if they really felt like it. It's possible to get 0.21W/GH from 28nm if done right, and pretty easy to do with 22nm.

What do you guys think about this my post in the similar speculation thread?

https://bitcointalksearch.org/topic/m.10985965

If the S7 includes .22nm chips, we could see .35 watts per GH.  If the S7 includes next generation 16nm or 14nm chips, we could see efficiency in the .21 to .14 watts per GH/s range.  It's really hard to say what chips Bitmain will use in the S7 to possibly be released in July.  This has me wondering why an S6 would be made with present chips used in the S5?  Are they going to do the same thing again like they did with the S4?  Meaning, come out with a rig like the S4 that's only on the market for 3 or 4 months then it's gone again because of new technology.  The S4 came out in October and only went through 2 batches.  We may see the same thing with the S6 if Bitmain comes out with an S6.

If we have .35 watts per GH/s in the S7, lets see if it is enough to create a "Farm War" which can affect difficulty:

.35 watts per GH/s

A 1,655 GH/s rig @ .35 watts per GH/s would total 579.25 watts.  I chose to keep the S7 at a similar total wattage as the S5.  It only makes sense to keep it there to avoid making consumers purchase new PSU's to power the new rig.  I've put the difficulty at 55,000,000,000 because that is where I believe it will be by the time July arrives.  






$112.80 in one month with difficulty at 55 Billion and BTC price @ $248.401 (BTC-e Exchange)
$  40.92 in one month with 10.5 cents per kWH for power costs.
$  71.88 in one month after power costs.

Present cost per GH/s without shipping for the S5 is $0.29437229 per GH/s.  If it were the same for the S7, the costs of the S7 would be $487.19 with a 22nm chip @ .35 watts per GH/s efficiency.  Again, this is only speculation to see if this would create a "Farm War."  ROI on this rig would be 6.7 months at present price of BTC with 10.5 cents per kWH power costs.  That ROI did not include shipping costs.  Not really much of a game changer to create a "Farm War" in my opinion.

However, if the next generation were 16nm or 14nm at [lets say] .21 watts per GH/s as a conservative estimate, this could be a game changer.  Lets see [This time I'm using a similar wattage to the S3 (347.55 watts) for the sake of the sheer size of the rig]:

An estimated 1,655 GH/s @ .21 watts per GH = 347.55 watts total for the rig.  The example below includes difficulty at 55 Billion and price of BTC @ $248.401








$112.80 in one month with difficulty at 55 Billion and BTC price @ $248.401 (BTC-e Exchange)
$  24.52 in one month with 10.5 cents per kWH for power costs.
$  88.28 in one month after power costs.

Present cost per GH/s without shipping for the S5 is $0.29437229 per GH/s.  If it were the same for the S7, the costs of the S7 would be $487.19 with a 16nm chip @ .21 watts per GH/s efficiency.  This would have ROI at 5.52 months if difficulty and price of BTC went up and/or down the same percentage during that time frame.  This ROI did not include shipping!  The ROI would be very close to 6 months with shipping.

This would definitely encourage more to invest into new hardware.  Especially, the big mining operations.

I think more people would be willing to buy a miner this efficient because they reason it will last a fairly good while after ROI.

I really don't see a "chance" for lower fan noise until 3rd generation that will require less watts per chip; meaning less power, meaning less heat; meaning less necessity for high RPM fans to get the heat out.

I have to agree with you.

I see the S6 keeping the existing 28nm chip and architecture as the S5 but in an S4 size form factor with built-in PSU.  It would be on the market no more than 6 months [Just before 3rd generation comes out in September/October].

.51 watts per GH/s for 4,000 GH/s = 2,040 watts total.  Remember, the efficiency rating for the S5 @ .51 watts per GH/s included the one fan it has.  If this rig has four fans with built-in PSU, it may be close in wattage to what I just speculated (2,040 watts).  The PSU may go up to 2,200 watts to allow some room for over-clocking.

The present cost of the S5 [Without shipping] is .29565217 cents per GH/s.  With built-in PSU, the price per GH/s is definitely higher.  I'm going with the same price for the S5 at the moment.

If the S6 was 4000 GH/s x .29565217 cents per GH/s = $1,182.61 for the rig without including costs for built-in PSU.  I can see this rig easily going for $1,450.00 if we also factor in the built-in PSU.  Now we would have to add in shipping costs.  For one single unit the shipping could be $150.00 if it's the same form factor as the S4.  Now we are looking at $1,600.00 after shipping for a 4,000 GH/s rig.

Here is what the "speculated" S6 could mine with 4,000 GH/s at an estimated difficulty of 50,000,000,000:



Here is the electricity costs [At 10.5 cents per kWH] for this rig if it were 4,000 GH/s @ .51 watts per GH/s = 2,040 watts [With 4 fans]:



If your power costs were lower, it would certainly be better regarding ROI.  I only provided these as an example.  I still have fourteen (14) IBM 2880W server grade PSU's with breakout boards waiting to be used.  I'm more likely to buy rigs without PSU's at the moment because of this.  Others may not be in my situation.  I don't know.  If you do not have the PSU's sitting around, like myself, and you are looking for more hash rate in a form factor designed for server racks, this may be the rig for you.  I will more than likely continue buying more S5's while I wait on the 3rd generation because of these PSU's I have sitting on racks taking up space.  

I don't think we will get 10 TH from S6.  I'm guessing it has the S5 chips with more efficient design.  I would guess 4-5 TH.

10TH I think will be have to be a big thing such as next gen chips.

Compact, I would think, is still quite a ways off.  Even 3rd generation will generate a lot of heat if they have a high hash rate.  It seems like to me if we are using less voltage [less amperage (meaning less wattage)], I would think the rigs would generate less heat.  I imagine it depends on the size of the rig in relation to its hash rate (number of chips).  I'm sure others on here are much more knowledgeable in this area of engineering and design than I am regarding the subject of heat generation with mining rigs.

I actually have a typo in my previous post as I actually wanted to say 10 Th/s. U3 is good but only if it can deliver higher hash so the total hash per space is small.

lets wait to june and see how the s6 will perform. since the difficulty level has taken a big rise because miners did not enjoy getting little. difficulty has rise. satoshi predicted that

they should instead try to use better fan to reduce the noise, it should be about time already, or at least make it regulable with 4 pin on the circuit board

This may and may not be true. Yes, it s correct, we do not see those 40% monthly diff jumps anymore and I hope we ll not see this with current available hardware. However, I m sure companies are working on next gen and next next gen which might increase their efficiency big time. Probably the only limiting factor is this limping BTC price. If it starts picking up, more money will be available for development as well.

The more I think about it.  Now is a better time to get into bitcoin mining than it was a year ago.  One does not have to be concerned with as large of an increase in difficulty now as they did a year ago.  I can't imagine having to increase my hash rate more than 700 times what it was a year prior to mine the same amount of bitcoin that I mined a year prior.  The main thing holding most back at the moment is the price of bitcoin in relation to their power costs.  Those who move to an area with extremely low power costs stand a better chance of surviving.  Especially, when the blocks halve.

If Bitmain comes out with another compact unit, it would more than likely begin with the letter "U" like the U3 is presently.  The U3 is at 63 GH/s.  I could see a U4 at 100 GH/s with approximately 1/2 the power of the U3.

I don't see it becoming much of a race either at the moment.  Yes, block erupters did not last very long.  The S1 really didn't last very long either.  When did the S3 first come out?  The S3 has actually lasted the longest thus far among ASIC's because the difficulty has not increased as much year over year as each difficulty change passes by.  I can see the S5 lasting even longer because of the decrease in difficulty year over year compared to months prior.  The 3rd generation rigs will last even longer before they are no longer profitable if the price of BTC remains the same.

The good thing is this:  I can see the difficulty the end of August sitting at approximately 67,000,000,000.  At the end of August last year the difficulty was 27,428,630,902.  This means we will have had an increase of 2.44 times more than what it was a year prior.  This also means one would need to little more than double their hashing power from what they had a year prior.  It used to be as much as 700 times what it was a year prior.  Can you imagine having to increase your hash rate 700 times what it was a year prior to make the same thing as what you did then?  I'm very happy I'm not having to try to do that anymore.  I hope to see it get down as low as 1.5 times more than what it was a year prior.  Will we ever see that kind of increase year over year?  I'm sure we will some day.  However, I don't see it anytime soon.

Was hopping S6 to be really compact and have really high hash like 10Gh/s for each. Compact is awesome!!