|
notlist3d
Legendary
Offline
Activity: 1456
Merit: 1000
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 02:39:42 AM |
|
I think part of reason of making this thread is to worry other makers. They were smart to keep same nm and master it. I think they will be able to make a lot by doing this. I still wonder on time frame though. Once they get producing them I predict a lot coming fast of the new product.
|
|
|
|
-droid-
Legendary
Offline
Activity: 1050
Merit: 1001
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 06:44:52 AM |
|
I think part of reason of making this thread is to worry other makers. They were smart to keep same nm and master it. I think they will be able to make a lot by doing this. I still wonder on time frame though. Once they get producing them I predict a lot coming fast of the new product. Well I think with the smaller nm chips there's a huge problem of power consumption.. Intel is dealing with this now on their new gen stuff in the works.. So Bitmain is the winner here even with LKETC and their 14nm.. The power draw from their new chip will be comparable to bitmains fully custom 28nm
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 06:48:20 AM |
|
I think part of reason of making this thread is to worry other makers. They were smart to keep same nm and master it. I think they will be able to make a lot by doing this. I still wonder on time frame though. Once they get producing them I predict a lot coming fast of the new product. Well I think with the smaller nm chips there's a huge problem of power consumption.. Intel is dealing with this now on their new gen stuff in the works.. So Bitmain is the winner here even with LKETC and their 14nm.. The power draw from their new chip will be comparable to bitmains fully custom 28nm Especially when you consider the difference in production cost. It's probably of different magnitude. 14nm could definitively be able to laugh at 28nm in W/GH ratio if properly optimized but i don't think the development cost is going to be able to compete. As usual, China win with cheap mass production.
|
|
|
|
elrippo
Legendary
Offline
Activity: 1008
Merit: 1001
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 07:20:46 AM |
|
I think part of reason of making this thread is to worry other makers. They were smart to keep same nm and master it. I think they will be able to make a lot by doing this. I still wonder on time frame though. Once they get producing them I predict a lot coming fast of the new product. Well I think with the smaller nm chips there's a huge problem of power consumption.. Intel is dealing with this now on their new gen stuff in the works.. So Bitmain is the winner here even with LKETC and their 14nm.. The power draw from their new chip will be comparable to bitmains fully custom 28nm Especially when you consider the difference in production cost. It's probably of different magnitude. 14nm could definitively be able to laugh at 28nm in W/GH ratio if properly optimized but i don't think the development cost is going to be able to compete. As usual, China win with cheap mass production.Please do not forget TSMC from Taiwan, others from Singapour and south Korea. China is not the biggest....
|
For Advertisement. PM me to discuss.
|
|
|
notlist3d
Legendary
Offline
Activity: 1456
Merit: 1000
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 08:03:31 AM |
|
I think part of reason of making this thread is to worry other makers. They were smart to keep same nm and master it. I think they will be able to make a lot by doing this. I still wonder on time frame though. Once they get producing them I predict a lot coming fast of the new product. Well I think with the smaller nm chips there's a huge problem of power consumption.. Intel is dealing with this now on their new gen stuff in the works.. So Bitmain is the winner here even with LKETC and their 14nm.. The power draw from their new chip will be comparable to bitmains fully custom 28nm Especially when you consider the difference in production cost. It's probably of different magnitude. 14nm could definitively be able to laugh at 28nm in W/GH ratio if properly optimized but i don't think the development cost is going to be able to compete. As usual, China win with cheap mass production.Please do not forget TSMC from Taiwan, others from Singapour and south Korea. China is not the biggest.... There are lot's of places with cheap to do the production. But not all the countries really have the buisness's to do it. Producing chips is something not all of the countries you listed I think can do. China has kinda taken over on a lot. They are just able to make it cheaper with many factors effecting it. One thing with BTC so low I don't see a reason for them to be rushing to production.
|
|
|
|
QuintLeo
Legendary
Offline
Activity: 1498
Merit: 1030
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 09:15:07 AM |
|
On the box fan thing, just get a standard 20x20 furnace filter and fit it to the input of the fan - suction from the fan will hold it in place and it's a GOOD fit for most common box fans like that Lasko one poster linked. No need for special "fan filters".
The one issue with putting fan filters on the S5 fans it that they reduce airflow noticeably. Good for dust reduction IN the unit, but NOT GOOD for keeping them cool. Better to keep the dust out of the room entirely IMO.
|
I'm no longer legendary just in my own mind! Like something I said? Donations gratefully accepted. LYLnTKvLefz9izJFUvEGQEZzSkz34b3N6U (Litecoin) 1GYbjMTPdCuV7dci3iCUiaRrcNuaiQrVYY (Bitcoin)
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 09:22:19 AM |
|
On the box fan thing, just get a standard 20x20 furnace filter and fit it to the input of the fan - suction from the fan will hold it in place and it's a GOOD fit for most common box fans like that Lasko one poster linked. No need for special "fan filters".
The one issue with putting fan filters on the S5 fans it that they reduce airflow noticeably. Good for dust reduction IN the unit, but NOT GOOD for keeping them cool. Better to keep the dust out of the room entirely IMO.
That sound a bit expensive, i don't know if you can get those furnace filter for cheap. On ebay you can get boxfan filter net for a few dollars for a pack of them lasting 1month+. If you're going to spend significant money on filter, i'd go with a washable filter.
|
|
|
|
elrippo
Legendary
Offline
Activity: 1008
Merit: 1001
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 10:20:18 AM |
|
I think part of reason of making this thread is to worry other makers. They were smart to keep same nm and master it. I think they will be able to make a lot by doing this. I still wonder on time frame though. Once they get producing them I predict a lot coming fast of the new product. Well I think with the smaller nm chips there's a huge problem of power consumption.. Intel is dealing with this now on their new gen stuff in the works.. So Bitmain is the winner here even with LKETC and their 14nm.. The power draw from their new chip will be comparable to bitmains fully custom 28nm Especially when you consider the difference in production cost. It's probably of different magnitude. 14nm could definitively be able to laugh at 28nm in W/GH ratio if properly optimized but i don't think the development cost is going to be able to compete. As usual, China win with cheap mass production.Please do not forget TSMC from Taiwan, others from Singapour and south Korea. China is not the biggest.... There are lot's of places with cheap to do the production. But not all the countries really have the buisness's to do it. Producing chips is something not all of the countries you listed I think can do.China has kinda taken over on a lot. They are just able to make it cheaper with many factors effecting it. One thing with BTC so low I don't see a reason for them to be rushing to production. Well the biggest countries are China, US, Taiwan, Singapour ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) EU is not really a potential ![Grin](https://bitcointalk.org/Smileys/default/grin.gif)
|
For Advertisement. PM me to discuss.
|
|
|
ZenFr
Legendary
Offline
Activity: 1260
Merit: 1046
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 10:45:40 AM |
|
eh eh... and a USB stick with this BM1385 ? For a few dollars :-).
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 10:54:50 AM |
|
eh eh... and a USB stick with this BM1385 ? For a few dollars :-).
That'd be nice. I will get one of the BM1384 one soon. Hopefully sidehack gets his hand on some 1385s.
|
|
|
|
sidehack
Legendary
Offline
Activity: 3360
Merit: 1859
Curmudgeonly hardware guy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 02:16:16 PM |
|
There's pretty much no way Bitmain will let any BM1385 into the wild before their S7 is shipping. I also won't build a stick off it until I'm out of BM1384 sticks. That said, it'd be pretty cool to have a 40GH stickminer. We'll probably focus on a 4-6 chip pod, maybe one you can stick a fairly generic CPU cooler on.
Bitmain - thanks for posting the BM1385 datasheet with actually good info. I'm liking the pinout and integrated features so far. It's gonna be good to see these things working.
|
|
|
|
philipma1957
Legendary
Offline
Activity: 4228
Merit: 8383
'The right to privacy matters'
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 02:23:51 PM |
|
eh eh... and a USB stick with this BM1385 ? For a few dollars :-).
That'd be nice. I will get one of the BM1384 one soon. Hopefully sidehack gets his hand on some 1385s. A sidehack 1385 miner in time for christmas 2015 would be nice. Even in time for my birthday in Jan 2016 works for me
|
|
|
|
VirosaGITS
Legendary
Offline
Activity: 1302
Merit: 1068
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 02:33:40 PM |
|
eh eh... and a USB stick with this BM1385 ? For a few dollars :-).
That'd be nice. I will get one of the BM1384 one soon. Hopefully sidehack gets his hand on some 1385s. A sidehack 1385 miner in time for christmas 2015 would be nice. Even in time for my birthday in Jan 2016 works for me That would be dope. However i believe sidehack couldn't get more than a prototype going by then assuming he gets the BM1385 fairly soon. ![Sad](https://bitcointalk.org/Smileys/default/sad.gif)
|
|
|
|
sidehack
Legendary
Offline
Activity: 3360
Merit: 1859
Curmudgeonly hardware guy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 02:47:54 PM |
|
I plan to have a working prototype (hardware at least) within 30 days from receiving sample chips.
|
|
|
|
sidehack
Legendary
Offline
Activity: 3360
Merit: 1859
Curmudgeonly hardware guy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 03:04:18 PM |
|
I plan to have a working prototype of a multi-chip full board, not a stick.
I'm not sure yet how the protocol changes, but that's software-level. Node-level voltages change, inter-node level shifters change, no 0.9V LDO required per node, fewer bypass caps per chip required. The footprint is also quite different (and quite better, my opinion). If your question is "can you use the same PCB and drop the new chip on it" the answer is "definitely not". But the required changes aren't that big.
|
|
|
|
notlist3d
Legendary
Offline
Activity: 1456
Merit: 1000
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 03:34:42 PM |
|
I plan to have a working prototype of a multi-chip full board, not a stick.
I'm not sure yet how the protocol changes, but that's software-level. Node-level voltages change, inter-node level shifters change, no 0.9V LDO required per node, fewer bypass caps per chip required. The footprint is also quite different (and quite better, my opinion). If your question is "can you use the same PCB and drop the new chip on it" the answer is "definitely not". But the required changes aren't that big.
I would love to see you come out with some boards. Your thumb stick miner works great, love the feeling that it's from a growing company made out of this site. I hope to see you one day have your own models and hopefully range of products ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) . I would love a nice blade if price is right.
|
|
|
|
RichBC
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 06:13:01 PM |
|
I plan to have a working prototype of a multi-chip full board, not a stick.
I'm not sure yet how the protocol changes, but that's software-level. Node-level voltages change, inter-node level shifters change, no 0.9V LDO required per node, fewer bypass caps per chip required. The footprint is also quite different (and quite better, my opinion). If your question is "can you use the same PCB and drop the new chip on it" the answer is "definitely not". But the required changes aren't that big.
Nice that they have integrated the 0.9V LDO, although that was only a potential divider in the S5. I assume you will still need an external 1.8V LDO for the Analog PLL supply? I wonder, as I assume in S7 we will see another string design, if they considered incorporating the level shifters? Does at first glance the more comprehensive protocol spec agree with what you have found on the BM1384? Rich
|
|
|
|
Meech
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 06:13:10 PM |
|
I think most initial companies producing miners were forum members before asics arrived. Back then was more of a hobby with potential. But yes very pleasing to have open dialect with someone who starts at ground level.
|
|
|
|
sidehack
Legendary
Offline
Activity: 3360
Merit: 1859
Curmudgeonly hardware guy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
August 25, 2015, 07:08:40 PM |
|
It's not explicitly stated in the data sheet, but it looks like there are two sets of com pins - one for local potential, and one with integrated level shifting.
You'd still need the 1.8V LDO per node for IO voltages, and to power the 0.9V PLL LDO.
|
|
|
|
|