Topic: The combined sidehack-novak usb stick review thread. AKA GekkoScience BM1384 - page 11. (Read 26372 times)

s-5 is not able to go under 10.8 volts  and needs to be dialed down to that after startup.

power savings is only 10%. and a plat psu at 12 volts will drop to a gold at 10.8 volt setting

so the 10% power savings may only be  5-6%

something in the string design.  A bit above my paygrade. as to the why of it.


Back to this stick.

I found an issue it does not like to go lower then freq 125.

My freq 100 setting was not very stable and a bit power hungry.

So I would say think :


freq 125 no fan and 0.305 watts per gh as the minimum.


freq 225 with fan and 0.35 watts per gh as the maximum.


   100                   5.00>>>>>>>>>>>>>not very stable

  
  125                   6.88
   150                   8.25  
   175                   9.63  
   193.75         10.66  
   196.88         10.83  
   200                 11.00  
   206.25         11.34  
   212.5                 11.69  
   218.75         12.03  
  225                 12.38

  
   237.5                 13.06  
   243.75         13.41  
   250                 13.75  

you cant change the volt on the s5 with the firmware

it also seems to not work they way they intended it to by reducing the input volts from the psu.

this thread makes me really curious about undervolting the S5. I've searched but cant find any posts about it. i'm running the nicehash firmware which allows the ability to change the voltage. but i havent seen anyone post about it. im afraid to mess with it because i dont the the correct syntax(or parameters) to put in. dont want to fry my miner on the first try....

About the best Bitmain could feasibly do with an S[odd] size miner with the BM1384 is probably a 17x3 string, getting 1.4TH at 500W and 70% higher chip cost. That's pretty dense. A new chip with 30-40% better top clock would do the job better and probably at lower cost. I'll be surprised if they roll out another S[odd] miner with the BM1384 instead of a new chip.

They could do another S[even] miner with it, and some 18s unregulated strings could get around 3.4TH off a 1KW PSU in a rack box which would be pretty darn sexy.

Even if we build with BM1384, 18-chip boards won't be available for months. If I had a design right now, I'd have to raise the money for a chip batch and then wait for them from the factory. If I had a design right now, S7 would probably be selling before I had the money all together. And since it'll be probably a month before I have the design ready...

But all that is discussion for the project discussion thread. Let's try and keep this one to stick facts instead of non-stick speculations.

guess it all depends how much they value it..

There is actually someting coming from july from Bitmain. A lot of good source have confirmed it .

Should be the S7. If we follow the pattern, it will be a new chip. S1 - S1  new chip S3 - S4 new chip S5 - Nothing new chip S7.

Hope so! 

im not sure about new chips.. right now its pure speculation.. not even an official announcement.

if the s5 chips can do much better then they are doing, why spend the resources to make newer chips?
it would be more cost effective to just redesign the boards and use the stock they have?


we have until 65 billion difficulty for s3s at 10c a kwh. can maybe get some more out of them if they are undervolted.

the s5 has until around 115 billion before we have to start worrying about that.

i was going to pick up a 2nd s5, but i found these threads and i think if the 18 chip boards can be made soon it would be better to get them.. even if they are made on the s5 chips they will be better then anything on the market.

Mine's still running great at stock speeds with no fan.

My opinion so far is that it's a great miner for beginners. It took all of 30 seconds to get it up and running. Efficiency is great, but you're going to have to string up a few of them in a hub to average out the draw from your controller.

I'll have some free time this weekend to get it hooked up to a cubian and rpi and start taking some more measurements.

18 chipboard with 0.18W/GH TO 0.30W/GH

Would really be nice.

New chips yeah.

So I had a really long and thoroughly quantitative post written up addressing points of discussion from the previous 10 or so posts but the forum ate it and I don't really feel like going back over all those numbers again. Call it a lack of motivation if you want. Novak and I got back to the shop this morning after a 37-hour marathon drive (1630 miles) for supplies to find the power had gone out, all the UPS were depleted so the servers were powered down, the internet wasn't working and the hosting router had eaten its own hard drive, and also I haven't really had anything to eat except Twizzlers and sunflower seeds in about 23 hours. All in all it was a pretty fun couple of days.

Anyway. The point was, the S5 was designed at a time when the hardware scene was pretty different than it is now. It's fairly optimal when you consider the size constraints of the S[odd] miners, prioritizing hashrate first and efficiency second (since top clock on the BM1384 is still more efficient than bottom clock on just about every other chip brought to market). With higher chip density they could do better, but at increased cost (17x3 like in the S4+ could get you 1.4TH at 500W with 70% more chips, whooptee doo).

The 18-chip board will be more efficient than 18 sticks, because you have 1x support hardware instead of 18x support hardware and the regulator should run closer to 95% than the 87-90% on the stick.

I will probably also not release an 18-chip board with the BM1384. I'm assuming, with the dev timeframe we have, that Bitmain's next gen chip will be out and if I can design around that for the larger stuff I certainly will. Maybe if the BM1384 chip prices go way down it'd still be viable, but it'd be starting off at a disadvantage when the next chip will probably see 0.3W/GH around top clock instead of bottom.

A lot depends on how you set a stick up.

my pc is really efficient . it runs a node at 25 watts and if I choose to run 1 stick at freq 125 .  I won't need a hub or a fan.

if sidehack makes a usb hub that allows a psu to power it the hub would be very efficient .

the s-5 is 1150gh and at .52 per gh  you get  598 watts.

the controller and the 2 fans are close to 20 watts of waste so 578/1150 = 0.502 watts a gh.


 your argument is valid  if the pc/rasp pi/cubie/bugle bone  were never going to run.  that means all power used by the controller/fans/hubs counts against the watts per gh.

my 1 stick 1 hub 1 fan is pulling more then 8 watts to do 11 gh.  i do not count the 25 watts for the pc since i was using the same 25 watts to run

the node.

 now if I add 2 more sticks and go to 16 watts to do 33gh I am at 16/33 = .4848 watts per gh not counting the pc's 25 watts.

buts a guy with a rasp pi will use 16 + 10 = 26 watts for 3 sticks a fan a hub a rasp pi   this is 26/33 or 0.7878 watts a gh


many people don't want big home mining.  26 watts vs  600 watts almost no noise vs lots of noise.

for sticks to really beat an s-5

 10 at 11 gh = 110gh  watts per gh at .33 = 36.3 watts + 10 watts for rasp pi = 46.3 watts add 3.7 watts for fan and hub waste you are doing

50 watts to get 110gh that is 0.4545 watts a gh and beats the s-5 in power use and sound use.  so the sticks show under clocked chips work well.

carry forward to an 18 chip board and maybe we have a special piece of gear.

210 gh at 75 watts using a hub and a rasp pi will bitch slap the s-5's  .52 watts a gh.

One item of note: The numbers that Phillip comes up with I assume are measured "at the USB connector"  as I understand it. The actual final "at the wall" efficiency won't look as good, since there has to be power supply, and hub losses. These numbers also, except for some early ones, don't include the PC that's running the mining software.

When compared to an S5 for example, where it's all pretty much bundled together,  you only really get to measure the "at the wall" figure. That's I think where the .52W/GH comes from .

Don't get me wrong, this is all really good work. It's just not quite as good as it seems is all.

so far there has been CrazyGuy and philipma1957, you pinged the other guys?

on that note, CrazyGuy, hows yours been going?

 1) Note all HW are way under 0.5%
 2) Note to set new freq a full restart of my windows 7 pc is needed.



tested are in bold

   100           5.5  
   125           6.88  
   150           8.25    .                   .31             Hw under 0.5% Hash rate 7.97gh-----biggest difference will retest it.
   175           9.63                        .31 to .32    Hw under 0.5% Hash rate 9.64gh

   193.75         10.66  
   196.88         10.83
  
   200         11.00                         .33 to .35  Hw under 0.5% Hash rate 10.97gh


   206.25      11.34                          0.34, 0.34, 0.36 Hw under 0.5%  Hash rate 11.37gh
 
I look at this set of three tests at freq 206.25  and  realize I did not go low enough on the pots's setting


                                            
   212.50         11.69                     0.335 watts per gh  Hw under 0.5%  hash rate 11.73gh  

  
   218.75          12.03                         0.3396 watts per gh Hw under 0.5% hash rate 11/97gh

  
   225.00      12.38                           0.3498 watts per gh HW under 0.5%   hash rate 12.40gh


I believe my hub will do all of these real easy  with my  2 port bridge.  Most hubs will not be able to do these below.  They will be tested last.
   237.5         13.06  
   243.75      13.41  
   250           13.75  

the freq 225 test went pretty well   0.3498 watts a gh  Hw 7/12000  that is 99.941% good   I hashed at about 12.40gh.

I will post photos and clean thread up a bit.

So from freq 150 to freq 225  I ran 0.31 to 0.35 watts a gh and was able to always be better then 99.5%

Once I get all photos posted I will do the lower freq's of 100 , 125

I will see if 0.28 watts a gh with no fan can work on 100 and or 125 freq

Yup, the 18-chip boards look very promising for what they may produce, can't wait to see those come along.  I've got an S1 and S3 that I would use as donor to upgrade to this design.

I would look to replace my 2GH/s nanofury sticks with these stick as well, 4x the hash is alway welcomed.

maybe  the 18 chip is a little worse then the stick .  but if it does 12.0gh a chip at freq 218.75 that is   216gh  at 80-90 watts.

I just finished my 218.75 test   I got .84 amps x 4.84 volts and 11.97gh  with 99.5%  good share rate.

this 4.0656 watts and divided by 11.97 = 0.3396 watts per gh  errors were higher at 0.5% but a little tweak on the pot would have dropped the error rate and bumped the power a little bit.
 so at 0.341 watts errors would be 0.1%

I am now doing freq 225 I will stop here as I needed to push amps to .900 with volts at 4.82 watts = 4.338

most hubs at maxxed at this point .  so I will stop at this point. I will post numbers but errors are decent with the small turn of the pot errors are well under 0.5%

It's approximately 1"x1.75".  AntMiner U2 is ~1"x1.57", so a bit shorter - and also flatter.

Keep in mind, though, these are not the final heat sinks;

so at 218.75 which is the s3 stock freq... you should get like 234ghs at less then 100 watts?

i would love a few of those 18 chip boards.. maybe ill get 4 or 6 to replace my s3s..

you really are putting bitmains techs to shame..

you can see they went for speed, instead of efficiency.

Do these have thermal paste or tape between/to the heatsink? It looks very close in size to the U2+ heatsink but the fins are thinner, sorta like something I've seen before.