Block Eruptor USB Overclocked

[BitcoinFX]

EDIT: All aspects considered, using:

 --icarus-options 115200:1:1 --icarus-timing short=90

instead of:

    --icarus-options 115200:1:1 --icarus-timing 3.0=100

Should provide most BE users with a slightly higher overall hash rate, more Accepted shares and a higher WU value to match.

...

Inspired by this amazing thread (and some other homebrew projects that I'm currently working on): https://bitcointalk.org/index.php?topic=241652.0 - Block Erupter USB - Overclocking / hacking ? - which basically involves moding 2 chips and settings to acheive higher hash rates - I started testing the recommended cgminer default values. Is this mearly an illusion ? I think not...





These are 4 'stock' USB Block Eruptors running off one 4-port USB hub with the 'best' tested settings I could find.

The avg: An all time average hash rate - appears to be doubled in some respects, which is somewhat questionable, however U: The Utility defined as the number of shares / minute - and - WU:  The Work Utility defined as the number of diff1 shares work / minute (accepted or rejected) - appears to be quite a bit higher. Mine has increased from 18.8 avg to a steady 21.5 (up from 19.3 in this early screen shot). The U value starts lower, but has now exceeded the previous avg.

  --icarus-options 115200:2:2 --icarus-timing short

Can be used on a dedicated mining PC with good CPUs for increased performance.

  --icarus-options 115200:2:2 --icarus-timing short=90

Can be used on slower PCs that are also in-use.

EDIT: The --icarus-options 115200:2:2 is unlikely to give most users more Accepted shares, depending on your set-up and the pool.

N.B. I've tested these settings on windows and linux on both Bitcoin and Peercoin Pools. A single Block Eruptor mining PPC on a pool is currently reporting 768 MH/s (although it seems to range from around 448 MH/s to 665 MH/s) with the above settings.

IMPORTANT: I'm currently unsure if these settings will decrease the life of a Block Eruptor USB using standard cgminer.

Free CPU is essential for these settings. I'm also still testing if the number of accepted shares against rejects and hardware errors is really beneficial long-term.

I have a moded build of cgminer 3.1.1 (not the one pictured) and a moded BE that appears to be running at almost 1.5 GH/s - stable. If the device survives into early January 2014 - then I'll consider documenting the modifications for others to use.

Cheers!

Additional tip for windows users: go to Computer Management (My Computer) > Device Management > Ports (COM & LPT) > Silicon Labs ... (COM1) , (COM2) etc. > 2nd mouse button click > Properties Tab > Select: Bits per Second 115200 > OK and repeat for each mining device. Then start cginer again.

I've set my windows miner ports to the maximum 128000 - cgminer overrides these 'default' windows settings, but with the ports default being 9600 - one might as well at least match it. You can't and don't need to go above 115200 in cgminer as this setting is the maximum, but it is windows afterall.

[BitcoinFX]

Reserved for something special.         

[cctekh]

You doing anything special to cool them?  They seem to run really hot even at the default speed.

[Raptor2213]

I don't know about OP, but I've found that one of these makes them cool enough to touch without any difficulty.  From what I've seen from others, its a pretty standard fan to use on these setups.  Newegg is currently out of stock, but they're also available on amazon - for about twice as much.

Mine are using the OP's configuration (minus the custom cgminer) and are holding steady at about 397MH/s.




Most setups look more like this:

Mine will too - eventually.

[jimsonunique]

[BitcoinFX]

You doing anything special to cool them?  They seem to run really hot even at the default speed.

Nope. I probably should do though. I've just made sure that they have adequate spacing in my hubs.

I do have a fan on my main set-up and after running overnight on 2 sets of 4 BE's, one with a fan and one without, the 4 with the fan do have slightly less hardware errors, which could just be a coincidence of course.

These seem to be the best heatsinks that I can find:

http://www.frozencpu.com/products/14951/vid-177/ModMyToys_Solid_Copper_Passive_Micro-Heatsink_-_8mm_x_8mm_x_8mm_Pack_of_4_-_Black_Electro-Plated-MMT-HS-8X8-CU-BK.html

One of them on the main BE 100 chip and a good fan is probably all you really need to do. Although the main heat source is actually from the back of the BE chip, so replacing the main heatsink on the back is best when using moded chips.

[BitcoinFX]

I don't know about OP, but I've found that one of these makes them cool enough to touch without any difficulty.  From what I've seen from others, its a pretty standard fan to use on these setups.  Newegg is currently out of stock, but they're also available on amazon - for about twice as much.

Mine are using the OP's configuration (minus the custom cgminer) and are holding steady at about 397MH/s.



...

Are you running with: --icarus-options 115200:2:2 --icarus-timing short=90 on those ?

Has your WU value increased ? Are you getting more hardware errors than stock settings ?

Thanks.

[aeddan1]

I'm using an Arctic Breeze fan for my block erupters too. It works great!

[Raptor2213]

Are you running with: --icarus-options 115200:2:2 --icarus-timing short=90 on those ?

Has your WU value increased ? Are you getting more hardware errors than stock settings ?

Thanks.

I am running with those values.
Before this, I was using bitminter's stock software, as I found it gave be better results than BFGminer, so I wasn't tracking my WU.
It looks like I am getting more hardware errors than stock, but tracking shows that my overall stats have gone up by about .1-.2 GH/s.

I'm interested in seeing what it will look like with a modded cgminer.

[DoubleMoo]

I don't know about OP, but I've found that one of these makes them cool enough to touch without any difficulty.  From what I've seen from others, its a pretty standard fan to use on these setups.  Newegg is currently out of stock, but they're also available on amazon - for about twice as much.

Mine are using the OP's configuration (minus the custom cgminer) and are holding steady at about 397MH/s.


Most setups look more like this:

Mine will too - eventually.

That is a sick set up! (The 4 USB blocks with a USB fan)

[lajz99]

Have you compared what's being reported in cgminer to the actual hash rate on your pool?

What you're seeing in your mining software is not your true hash rate.

[qwertyuiopmnbvcxz]

A little off topic regarding the fan pictures, but would a long metal heatsink with the lower part sitting in water work better than a fan? Or would there be water/electrical issues?

[doxavg]

Have you compared what's being reported in cgminer to the actual hash rate on your pool?

What you're seeing in your mining software is not your true hash rate.

Agreed - my miners are still showing the correct hash rate at my pool not the 'overclocked' rate and I left them running for a good 8 hours with these settings. 

[BitcoinFX]

Have you compared what's being reported in cgminer to the actual hash rate on your pool?

What you're seeing in your mining software is not your true hash rate.

Agreed - my miners are still showing the correct hash rate at my pool not the 'overclocked' rate and I left them running for a good 8 hours with these settings.  

Depending on your host / set-up and the mining pool I'm fairly sure that there is a slight increase in the overall hash rate.

It is certainly not as high as the hash rate that appears to be shown in my original post, hence why I refered to it as 'an illusion'.

The actual speed is probably similar or less than that posted by Raptor2213

...

Mine are using the OP's configuration (minus the custom cgminer) and are holding steady at about 397MH/s.

...

The real question is can these settings generate more Accepted shares. I'm fairly certain that with some set-up's using stock block eruptors - they do, but again it is highly dependent on the host and the individual set-up.

Some of my own rigs are seeing more Accepted shares at various pools with a higher WU rate. I'll continue testing and moding the block eruptors and software. I'm certain its possible to get faster stock speeds with BE's, because of something else that I'm working on.

[BitcoinFX]

All aspects considered, using:

 --icarus-options 115200:1:1 --icarus-timing short=90

instead of:

    --icarus-options 115200:1:1 --icarus-timing 3.0=100

Should provide most BE users with a slightly higher overall hash rate, more Accepted shares and a higher WU value to match.

[Raptor2213]

How long do you think it will be until we get to try out your modded software?

[Raptor2213]

I took one of mine and changed the timing to

--icarus-options 115200:2:2 --icarus-timing 2.0=75

Now it's averaging about 520Mh/s
Stock hardware.

Edit: Five of them are running at 500-550 Mh/s.  Hardware errors are still in acceptable ranges.

[lajz99]

I took one of mine and changed the timing to

--icarus-options 115200:2:2 --icarus-timing 2.0=75

Now it's averaging about 520Mh/s
Stock hardware.

Edit: Five of them are running at 500-550 Mh/s.  Hardware errors are still in acceptable ranges.

Is that according to the pool you're using or the mining software?

[Raptor2213]

I took one of mine and changed the timing to

--icarus-options 115200:2:2 --icarus-timing 2.0=75

Now it's averaging about 520Mh/s
Stock hardware.

Edit: Five of them are running at 500-550 Mh/s.  Hardware errors are still in acceptable ranges.

Is that according to the pool you're using or the mining software?

That's an interesting question.  I'm only able to see reporting from one of them (mining for other ppl right now too).

Miner shows 500-550 Mh
Pool live stats show .5 Gh but exactly 2x the submits the software is reporting
Pool history shows payouts for .3
Pool used: Bitminter

Tried another pool, and it's showing 100-250 Mh/s, so I don't think it's accurate.
Pool used: miningpool.co

I'm going to wait a couple days and try more of them in a larger batch.

[Raptor2213]

Unfortunately, this appears to be wrong.  While the port speed settings did improve things slightly, the timings only affect the reported speed, not the actual.