Bitcoin Forum
November 17, 2017, 05:55:31 PM *
News: Latest stable version of Bitcoin Core: 0.15.1  [Torrent].
 
   Home   Help Search Donate Login Register  
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 [14] 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 »
  Print  
Author Topic: Fury/Blizzard tuning and mods  (Read 103992 times)
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 21, 2014, 10:56:30 AM
 #261

1,598 kH/s poolside 24hr avg when I looked just now.  381 clock. So kh/s is getting very close to 0.7 * Freq on one Fury.  2nd one is not quite as fast & needs 4 more hrs for a clean 24hr run.

LongAndShort posted the Zeus Miner Requests For Terry doc https://bitcointalk.org/index.php?topic=644921.msg7401931#msg7401931 and I read through it:  https://docs.google.com/document/d/19Vp6tG_INmcVtV4qo1Uhg9KzFm7VqjIKx95iGX9s58w/edit?pli=1  Question 11 indicates we should be able to get to 1,600 kH/s or a bit more.
1510941331
Hero Member
*
Offline Offline

Posts: 1510941331

View Profile Personal Message (Offline)

Ignore
1510941331
Reply with quote  #2

1510941331
Report to moderator
1510941331
Hero Member
*
Offline Offline

Posts: 1510941331

View Profile Personal Message (Offline)

Ignore
1510941331
Reply with quote  #2

1510941331
Report to moderator
1510941331
Hero Member
*
Offline Offline

Posts: 1510941331

View Profile Personal Message (Offline)

Ignore
1510941331
Reply with quote  #2

1510941331
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
1510941331
Hero Member
*
Offline Offline

Posts: 1510941331

View Profile Personal Message (Offline)

Ignore
1510941331
Reply with quote  #2

1510941331
Report to moderator
1510941331
Hero Member
*
Offline Offline

Posts: 1510941331

View Profile Personal Message (Offline)

Ignore
1510941331
Reply with quote  #2

1510941331
Report to moderator
1510941331
Hero Member
*
Offline Offline

Posts: 1510941331

View Profile Personal Message (Offline)

Ignore
1510941331
Reply with quote  #2

1510941331
Report to moderator
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 21, 2014, 11:19:43 AM
 #262

nwoolls & Luke-Jr BFGMiner for Windows build here: https://bitcointalk.org/index.php?topic=626361.msg7427246#msg7427246
They'd probably like some feedback from testers.

happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 21, 2014, 11:59:20 AM
 #263

So for the time being we've reached a limit.  J4bberwock is going try to break through on the hardware side.  A few guys are working on software.

Can we improve on the kh/s to freq ratio?

Can anyone explain the effect of raising or lowering readcount timeout? 

Quote
4.3   Readcount Timeout
After we send one command packet to the chained chips, that BIG chip will cover all the 32 bits nonce in a certain TIME. So we must send a new work to it before that, otherwise its hash power is wasted.

Cgminer’s readcount timeout is used for that. Every count means 0.1 second. We now set the readcount to be 3/8 of the covering time. It may be 3/4 later.

Current software builds use 3/4 for 6 chips or 96 chips etc.  Can this be optimized for 6 chips?

Small improvements with the kh/s to freq ratio would produce substantial gains.

381 * .7 * 6 = 1600
381 * .75 * 6 = 1714
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 21, 2014, 01:21:35 PM
 #264

1,602 kH/s 24hr avg on pool now.   60 hours running, no restarts etc. The longer it runs the better it gets.  
nst6563
Sr. Member
****
Offline Offline

Activity: 252


View Profile
June 21, 2014, 01:31:15 PM
 #265

1,602 kH/s 24hr avg on pool now.   60 hours running, no restarts etc. The longer it runs the better it gets.  

Is that Nwoolls build?  I cloned and compiled it last night and started it running. 
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 21, 2014, 01:37:59 PM
 #266

No, it's Darkwinde code, Cryptomining Blog build.

I think the newest nwoolls code is: https://github.com/nwoolls/bfgminer/tree/feature/zeusminer-support-cleaner
Next oldest: https://github.com/nwoolls/bfgminer/tree/feature/zeusminer-support-clean
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 21, 2014, 02:07:54 PM
 #267

Zeusminer makes the chips and the devices.

GAWMiners would be an OEM distributor (Fury)
ZoomHash would be a distributor (Blizzard)

Zeusminer also sells direct to anyone (Blizzard)
Kokok
Jr. Member
*
Offline Offline

Activity: 54


View Profile
June 21, 2014, 04:05:41 PM
 #268

This is a total NOOB question, but I am brand new to working in LINUX and have switched my miners to being controlled by RPIs.

I just need a little help with installing CGMINER 3.1.1 (for my Fury's) onto a new install of Raspbian. If someone can help and post the instructions I would be very appreciative.

Thank you in advanced.
tuanvie
Full Member
***
Offline Offline

Activity: 182


View Profile WWW
June 21, 2014, 06:28:45 PM
 #269

hi sir,
sorry i want to ask you about this post,
because I do not understand this,
This tutorial to maximize the performance of fury?
if there is damage that will we get?

thank you for your answer

J4bberwock
Hero Member
*****
Offline Offline

Activity: 840


View Profile
June 21, 2014, 07:11:15 PM
 #270

hi sir,
sorry i want to ask you about this post,
because I do not understand this,
This tutorial to maximize the performance of fury?
if there is damage that will we get?

thank you for your answer

If your know what you are doing. If your know how to solder a 0603 SMD resistor, there is almost no risk with the 7.5k resistor instead of the 9.1k. The chip will run around 1.35v. That is the upper limit in Zeus datasheet.
Extra cooling of the chips will help also.
As usual, if your aren't sure you can fix it if you fail, don't do it. Ask someone with good soldering knowledge to do the resistor swap.

Custom Server PSU breakout boards, 1200w, 1300w, 2000w, 2880w https://bitcointalk.org/index.php?topic=738527.0
Btc: 1J4bberWs6f6XVQ53gn3BNc8qHAicmm7wu
Ltc: LNDmMBtzD3cduwjfxANm2wDrxt9jXRk5ZP
worldlybedouin
Sr. Member
****
Offline Offline

Activity: 376


View Profile WWW
June 21, 2014, 11:01:23 PM
 #271

I've not had time to read the entire thread since finding it...but I wanted to ask for those folks who applied new TIM to their Blizzards, are you seeing any benefits other than temps?  I see some folks are pushing the clocks pretty high (~380-ish)...but is that with volt-mods or just new TIM applied.

Cheers!

LTC:  LXrAe2E6cBsK52GvUsYraeXkc2s7Ti7R5X
BTC:  1FLTMqVjTZ5MTdCF4npNZGFMEUGyBV4zcj
nst6563
Sr. Member
****
Offline Offline

Activity: 252


View Profile
June 22, 2014, 12:25:53 AM
 #272

I've not had time to read the entire thread since finding it...but I wanted to ask for those folks who applied new TIM to their Blizzards, are you seeing any benefits other than temps?  I see some folks are pushing the clocks pretty high (~380-ish)...but is that with volt-mods or just new TIM applied.

Cheers!

Yes, you can increase your clocks quite a bit from my experience just from using good thermal paste and applying heatsinks to the tops of the chips.  Adding a small fan to blow through the chamber over the circuitboard also helps.

Without the addition of the extra fan, I was able to raise my clocks from 333 with a 4%-5% hw error rate to 350 with a 4%-5% hw error rate.  (roughly about 1.4Mhs).

The addition of the voltmod does allow all the way up to the max clock of 381Mhz though.  We're hoping we can actually overcome that 381Mhz barrier.
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 22, 2014, 01:07:38 AM
 #273

looks like Zeus has changed readcount in the latest code

https://github.com/zeusminer/cgminer_zeus

Code:
if(info->read_count>opt_zeus_readcount){
info->read_count = opt_zeus_readcount;//send a new work every 10 seconds

no more

Code:
info->read_count = info->read_count * 3/4;

I don't understand it well enough to know if we can optionally set readcount in the new code or not.

There's other new stuff in there also.

I'm hoping this new code lets us push through the .7 freq to kh/s ratio.
nst6563
Sr. Member
****
Offline Offline

Activity: 252


View Profile
June 22, 2014, 01:15:14 AM
 #274

looks like Zeus has changed readcount in the latest code

https://github.com/zeusminer/cgminer_zeus

Code:
if(info->read_count>opt_zeus_readcount){
info->read_count = opt_zeus_readcount;//send a new work every 10 seconds

no more

Code:
info->read_count = info->read_count * 3/4;

I don't understand it well enough to know if we can optionally set readcount in the new code or not.

There's other new stuff in there also.

I'm hoping this new code lets us push through the .7 freq to kh/s ratio.

Just for the hell of it I changed that line of code in my last build of bfgminer to
Code:
info->read_count = info->read_count * 8/10;
so that it was .8 instead of .75...didn't notice much of a difference really.  I only let it run for half a day but the poolside numbers didn't really look much higher than normal, the local hashrate was in the 1.7-1.8 range though.  Maybe higher gives better results? 
worldlybedouin
Sr. Member
****
Offline Offline

Activity: 376


View Profile WWW
June 22, 2014, 01:37:28 AM
 #275

I've not had time to read the entire thread since finding it...but I wanted to ask for those folks who applied new TIM to their Blizzards, are you seeing any benefits other than temps?  I see some folks are pushing the clocks pretty high (~380-ish)...but is that with volt-mods or just new TIM applied.

Cheers!

Yes, you can increase your clocks quite a bit from my experience just from using good thermal paste and applying heatsinks to the tops of the chips.  Adding a small fan to blow through the chamber over the circuitboard also helps.

Without the addition of the extra fan, I was able to raise my clocks from 333 with a 4%-5% hw error rate to 350 with a 4%-5% hw error rate.  (roughly about 1.4Mhs).

The addition of the voltmod does allow all the way up to the max clock of 381Mhz though.  We're hoping we can actually overcome that 381Mhz barrier.

I looked at one of my Blizzards and the entire back side of the circuit board looks like it was coated in TIM...doesn't seem quite right.  Should I clean that off and apply some high quality TIM just behind the 6 chips or spread it over the whole board like Zeus did?

LTC:  LXrAe2E6cBsK52GvUsYraeXkc2s7Ti7R5X
BTC:  1FLTMqVjTZ5MTdCF4npNZGFMEUGyBV4zcj
nst6563
Sr. Member
****
Offline Offline

Activity: 252


View Profile
June 22, 2014, 02:24:16 AM
 #276

I've not had time to read the entire thread since finding it...but I wanted to ask for those folks who applied new TIM to their Blizzards, are you seeing any benefits other than temps?  I see some folks are pushing the clocks pretty high (~380-ish)...but is that with volt-mods or just new TIM applied.

Cheers!

Yes, you can increase your clocks quite a bit from my experience just from using good thermal paste and applying heatsinks to the tops of the chips.  Adding a small fan to blow through the chamber over the circuitboard also helps.

Without the addition of the extra fan, I was able to raise my clocks from 333 with a 4%-5% hw error rate to 350 with a 4%-5% hw error rate.  (roughly about 1.4Mhs).

The addition of the voltmod does allow all the way up to the max clock of 381Mhz though.  We're hoping we can actually overcome that 381Mhz barrier.

I looked at one of my Blizzards and the entire back side of the circuit board looks like it was coated in TIM...doesn't seem quite right.  Should I clean that off and apply some high quality TIM just behind the 6 chips or spread it over the whole board like Zeus did?

Definitely clean the old stuff off first. 
You can either cover the entire back with a nice coat (not really needed in my opinion) or just do the spots with the chips.  Personally, I put three lines across the back of the board under the chips (2 rows for the chips, 1 row for the rest of the components).  I however have a huge 30gram tube of Prolimatek pk-2...so 3 thin rows wasn't really a lot.  Whether you do three rows, cover the whole back, or just do dabs on the backsides of the chips...it will make a huge difference because the stock setup was done for cost efficiency...not thermal transfer efficiency.
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 22, 2014, 02:31:38 AM
 #277

New BFGMiner 4.2.2 using new Zeus code https://litecointalk.org/index.php?topic=20477.0
nst6563
Sr. Member
****
Offline Offline

Activity: 252


View Profile
June 22, 2014, 02:35:42 AM
 #278

New BFGMiner 4.2.2 using new Zeus code https://litecointalk.org/index.php?topic=20477.0

Damnit...seems like every time I clone and compile a new build another one comes out.

I'm glad it's actively being worked on though....  better than the stagnate state things were in when we first got the hardware  Wink
icekohl
Full Member
***
Offline Offline

Activity: 126

Gamer, wannabee bitillionare


View Profile WWW
June 22, 2014, 03:42:43 AM
 #279

New BFGMiner 4.2.2 using new Zeus code https://litecointalk.org/index.php?topic=20477.0

I'll have to try this as 4.2 never ran for me.

Member of Hashclub.org Hodler of ALTS
happydaze
Full Member
***
Offline Offline

Activity: 140


View Profile
June 22, 2014, 11:02:51 AM
 #280

Just for the hell of it I changed that line of code in my last build of bfgminer to
Code:
info->read_count = info->read_count * 8/10;
so that it was .8 instead of .75...didn't notice much of a difference really.  I only let it run for half a day but the poolside numbers didn't really look much higher than normal, the local hashrate was in the 1.7-1.8 range though.  Maybe higher gives better results?  

The Zeus doc indicated it was 3/8 and maybe 3/4 later.
All the code I looked at showed 3/4.
So how do we figure out what it does?  Which is faster readcount 3/8 or 7/8?
I guessed the smaller fraction would be faster readcount timeout but don't know for sure.

Newer code lets us specify readcount like we do chips and clock etc.  Default is 300.

So would 250 be a quicker readcount?  How does diff affect readcount?

Lots of questions ...........  Grin

Edit: just started Darkwinde's new build with default rc at 300 and will see how it does.
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 [14] 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 »
  Print  
 
Jump to:  

Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!