Looks good on paper. 20nm ASIC's, 6THS, $3299, 0.26J/W and can be underclocked for greater efficiency.
Whether it exists or not I do not know? and this is the S6 speculation thread.....
Rich
Topic: Antminer S6 speculation - page 3. (Read 19270 times)
Never heard of it, is it something that actually exists or is it vaporware/scamware?
And define "6'th generation", as far as I know there is NO ASIC COMPANY that has built 6 generations of Bitcoin chip yet.
And define "6'th generation", as far as I know there is NO ASIC COMPANY that has built 6 generations of Bitcoin chip yet.
Get an Uranus miner . its 6th and is more energy eff.
Bitmain just never got around to building a "rack mount" datacenter-style machine based on the BM1384.
I would guess their sales on the S5 made them decide there wouldn't be enough sales of a S6 model to justify the design/test/build costs.
I would guess their sales on the S5 made them decide there wouldn't be enough sales of a S6 model to justify the design/test/build costs.
Is it only name change or do they produce S6 and don't like it? What's this skipping generation?
There is NO POINT to them releasing an S6 at this time. If they were GOING to release one, it would have been several months ago when the BM1384 was still a performance leader, not when they appear to have emptied all of their stocks of it building the last new S5 units and are moving on to the NEXT generation.
Water cooled would be a C series machine, NOT an S6.
I'll take the word out of a Bitmain representative over your "I insist they are not going to skip" wasting time and money on a machine that they know would generate almost ZERO sales.
Water cooled would be a C series machine, NOT an S6.
I'll take the word out of a Bitmain representative over your "I insist they are not going to skip" wasting time and money on a machine that they know would generate almost ZERO sales.
I still insist they are not going to skip S6.I believe there is a problem with cooling, they will overcome with water cooling.
S6 has been mentioned (I believe by Dogie, but might have been by BitMain directly) that it's not going to happen, will be "skipped".
It's the S7 we're all waiting on now....
It's the S7 we're all waiting on now....
Do any of you know when will they officially announce S6 specifications? When will this speculation end?
.249 is the CHIP spec, the actual usage in the S5 is a little higher due to the other stuff in the machine that uses power, like the fan controller support chips etc.
I HAD forgotten the chip spec went quite that low, but I'd still say .3 is probably about the best an S5 unit could manage even if you can get it fixed to run at 9v
It's closer to chip spec than machines that DO have programable voltage due to lack of losses in a voltage converter/regulator, but there's still SOME non-hash-chip power draw out of the other stuff.
I'd still like to see that .3 though, if I could ever find an EFFICIENT 9v power supply that can support the current draw.
I HAD forgotten the chip spec went quite that low, but I'd still say .3 is probably about the best an S5 unit could manage even if you can get it fixed to run at 9v
It's closer to chip spec than machines that DO have programable voltage due to lack of losses in a voltage converter/regulator, but there's still SOME non-hash-chip power draw out of the other stuff.
I'd still like to see that .3 though, if I could ever find an EFFICIENT 9v power supply that can support the current draw.
The S5 runs fine in chain mode at 12V, and even down to 11V. Something goes wrong somewhere between 11V & 10V. Sidehack has speculated that this is nothing to do with the voltage per say, but probably the level shifters for the signals between the chips?
The bottom line, for me, is that the previous Bitmain data sheets on the BM1380 & BM1382 have been pretty accurate on the under & over clocking & volting potential of the chips so I feel that the BM1384 sheet is also likely to be correct?
So I see no reason with the correct design why the BM1384 should not be capable of a 2.49J/GH at 0.6V. Whether this is best achieved by sticking with the more efficient chain supply mode moving from 15 to 20 chip pairs / chain (to enable 12V running) or returning to conventional DC-DC converters (please make them programmable) I do not know?
I plan as soon as I can pick up an S5 in the UK for a sensible price to sort out the best / easiest way to undervolt / underclock it and get as close as I can to the specified J/GH. Bitmain on the other hand have very little incentive to do this as their prime customers have very low price KW/H and are only interested in GH/$
Rich
The bottom line, for me, is that the previous Bitmain data sheets on the BM1380 & BM1382 have been pretty accurate on the under & over clocking & volting potential of the chips so I feel that the BM1384 sheet is also likely to be correct?
So I see no reason with the correct design why the BM1384 should not be capable of a 2.49J/GH at 0.6V. Whether this is best achieved by sticking with the more efficient chain supply mode moving from 15 to 20 chip pairs / chain (to enable 12V running) or returning to conventional DC-DC converters (please make them programmable) I do not know?
I plan as soon as I can pick up an S5 in the UK for a sensible price to sort out the best / easiest way to undervolt / underclock it and get as close as I can to the specified J/GH. Bitmain on the other hand have very little incentive to do this as their prime customers have very low price KW/H and are only interested in GH/$
Rich
Chip spec doesn't go quite that low - they did mention "run an S5 on 9v" early on but I think the figure was more like .3w/GH or .4w/GH range.
Problem with that is FINDING 9v power supplies with enough current to run an S5 - I did some digging early on and never did find anything suitable - REALLY poor efficiency way $$$ pricey industrial stuff was the only option I could dig up and even THAT stuff was going to be WAY iffy on providing enough current. The one supply I DID find that looked like it could handle one side of a S5 at 9volts was going to lose almost as much in the PS as would be gained from the undervolting.
Problem with that is FINDING 9v power supplies with enough current to run an S5 - I did some digging early on and never did find anything suitable - REALLY poor efficiency way $$$ pricey industrial stuff was the only option I could dig up and even THAT stuff was going to be WAY iffy on providing enough current. The one supply I DID find that looked like it could handle one side of a S5 at 9volts was going to lose almost as much in the PS as would be gained from the undervolting.
Didn't Dogie or someone direct from BitMain already announce they were skipping the S6?
No point in releasing anything new with the BM1384 any more unless (1) they have a BUNCH onhand, AND (2) they come out with a "S5 Lite" type design with 18-20 chips in the chain for better efficiency OR they come out with a design with the ability to set the voltage like the SP20E.
Given they are no longer selling "new" S5s, I'd bet that Bitmain doesn't have a significant number IF ANY of the BM1384 chips left in stock.
No point in releasing anything new with the BM1384 any more unless (1) they have a BUNCH onhand, AND (2) they come out with a "S5 Lite" type design with 18-20 chips in the chain for better efficiency OR they come out with a design with the ability to set the voltage like the SP20E.
Given they are no longer selling "new" S5s, I'd bet that Bitmain doesn't have a significant number IF ANY of the BM1384 chips left in stock.
They would need to design a better board, but if they had the chip on hands, maybe a version of the S5 that *actually* let you run it at 9v?
If i recall correctly, there was a mention of the S5 being able to run at 0.2j/w with a 9v PSU, that was later
Didn't Dogie or someone direct from BitMain already announce they were skipping the S6?
No point in releasing anything new with the BM1384 any more unless (1) they have a BUNCH onhand, AND (2) they come out with a "S5 Lite" type design with 18-20 chips in the chain for better efficiency OR they come out with a design with the ability to set the voltage like the SP20E.
Given they are no longer selling "new" S5s, I'd bet that Bitmain doesn't have a significant number IF ANY of the BM1384 chips left in stock.
No point in releasing anything new with the BM1384 any more unless (1) they have a BUNCH onhand, AND (2) they come out with a "S5 Lite" type design with 18-20 chips in the chain for better efficiency OR they come out with a design with the ability to set the voltage like the SP20E.
Given they are no longer selling "new" S5s, I'd bet that Bitmain doesn't have a significant number IF ANY of the BM1384 chips left in stock.
S6 will be released when Bitmain sells second hand S5 stock totally or a big proportion of it.
Its not a bad assumption but i wouldn't be surprised if that was held off for a while and we started just buying their used units. We're supposed to be getting an announcement from Bitmain soon but i get they are in no rush to do so.
With the S4+ there's possibly quite a bit of time before the S6 and mentions of S7 are unsure as well.
S6 will be released when Bitmain sells second hand S5 stock totally or a big proportion of it.
When are they releasing S6 and S7? we need bigger hash rate and compact to reduce the shipment charges per Gh/s
Not that sure it is happening until the late summer Sept 10th at the least.
Is there something significant about September 10th? It all looks very uncertain, and yet September 10th is a fairly specific date. Care to share your reasoning?
Good question I personally cannot think of anything special about the date. I was surprised they released a S4+.
I think it's good for miners we get a little extra time to ROI before next gen. That is a win in my book.
Nothing special about that, he just likes that date. I too think it might happen somewhere late summer.
When are they releasing S6 and S7? we need bigger hash rate and compact to reduce the shipment charges per Gh/s
Not that sure it is happening until the late summer Sept 10th at the least.
Is there something significant about September 10th? It all looks very uncertain, and yet September 10th is a fairly specific date. Care to share your reasoning?
Good question I personally cannot think of anything special about the date. I was surprised they released a S4+.
I think it's good for miners we get a little extra time to ROI before next gen. That is a win in my book.
When are they releasing S6 and S7? we need bigger hash rate and compact to reduce the shipment charges per Gh/s
Not that sure it is happening until the late summer Sept 10th at the least.
Is there something significant about September 10th? It all looks very uncertain, and yet September 10th is a fairly specific date. Care to share your reasoning?
When are they releasing S6 and S7? we need bigger hash rate and compact to reduce the shipment charges per Gh/s
Not that sure it is happening until the late summer Sept 10th at the least.
Good, my gear should ROI around late summer
Jump to: