Bitcoin Forum
December 07, 2016, 04:39:43 PM *
News: Latest stable version of Bitcoin Core: 0.13.1  [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 44 45 46 [47] 48 49 50 »
  Print  
Author Topic: FPGA development board "Icarus" - DisContinued/ important announcement  (Read 184213 times)
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
June 28, 2012, 05:05:19 AM
 #921

Got a xilinx kit on the way, will test the v3,v4 and 200mhz streams to see if there are any improvements.

At the moment they are v2,v3,v3,v3.  I think a cooling upgrade is in order, the default fans have a bit to be desired.

It gets about 40c here in summer so i think they are going to struggle in that kind of heat.

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
1481128783
Hero Member
*
Offline Offline

Posts: 1481128783

View Profile Personal Message (Offline)

Ignore
1481128783
Reply with quote  #2

1481128783
Report to moderator
1481128783
Hero Member
*
Offline Offline

Posts: 1481128783

View Profile Personal Message (Offline)

Ignore
1481128783
Reply with quote  #2

1481128783
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
1481128783
Hero Member
*
Offline Offline

Posts: 1481128783

View Profile Personal Message (Offline)

Ignore
1481128783
Reply with quote  #2

1481128783
Report to moderator
1481128783
Hero Member
*
Offline Offline

Posts: 1481128783

View Profile Personal Message (Offline)

Ignore
1481128783
Reply with quote  #2

1481128783
Report to moderator
luffy
Hero Member
*****
Offline Offline

Activity: 606



View Profile
June 28, 2012, 09:14:36 PM
 #922

waiting for your results!
thanks Smiley
kano
Legendary
*
Offline Offline

Activity: 1932


Linux since 1997 RedHat 4


View Profile
June 29, 2012, 05:25:01 AM
 #923

Don't forget if you use cgminer, to use "--icarus-timing short" (or "--icarus-timing long")
... since a faster bitstream will of course have a different timing
The result of the timing (if it's correct due to being run on a quiet computer) can be fed into "--icarus-timing nanosectime" every time after that so you don't need to get cgminer to work it out ('short' or 'long') and thus cgminer will use the proper timing from the start.

Pool: https://kano.is BTC: 1KanoiBupPiZfkwqB7rfLXAzPnoTshAVmb
CKPool and CGMiner developer, IRC FreeNode #ckpool and #cgminer kanoi
Help keep Bitcoin secure by mining on pools with Stratum, the best protocol to mine Bitcoins with ASIC hardware
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
June 29, 2012, 05:37:30 AM
 #924


The result of the timing (if it's correct due to being run on a quiet computer) can be fed into "--icarus-timing nanosectime" every time after that so you don't need to get cgminer to work it out ('short' or 'long') and thus cgminer will use the proper timing from the start.


I believe this was worked out to be 11.3 from reading this thread.

Is it still ok to just leave cgminer and it will work it out even with the 200mhz bitstream as apposed to the 190mhz default?

I currently run cgminer with no extra timing options and after an hour it seems to settle on the correct speed to report.

I was under the impression that cgminer works it out over time even with no timing switches.

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
kano
Legendary
*
Offline Offline

Activity: 1932


Linux since 1997 RedHat 4


View Profile
June 29, 2012, 08:32:43 AM
 #925

I wrote all the Icarus timing code in cgminer and all the code that adjusts it (and redesigned all of that in the Icarus code)
So yeah the following is correct Smiley

The current code defaults to 11.2 seconds (not 11.3s) before aborting the work and starting again
It gives a little lea way of 1 unit of time (1/10sec) in case there are any processing delays

However, 11.2s is calculated based on the Icarus mining at 2.6316ns per hash
(which of course means 5.2632ns per pair of hashes)

i.e. (2^32 * 2.6316 * 10^-9) - 0.1 = 11.20...

What it will also mean is: if you are hashing faster, and complete the nonce range without finding a valid nonce before 11.2s, then the icarus will be idle for "11.2 - however long it takes to complete the nonce range" each time that happens.

So the correct value in place of 2.6316ns, will ensure it isn't ever completing and waiting for new work.
It would only have to be 1% faster to end up waiting.

The second thing that the correct nanosecond value does, is help cgminer to calculate the correct MH/s value.
With the correct value, it knows how many hashes it has done when it aborts the work (at 11.2s in the current settings)
If it doesn't have the correct nanosecond value, it will report the MH/s incorrectly (hi or low depending on the speed of the bitstream)

The option: --icarus-timing short (or long) tells the code to work out the correct values and use them.
N.B. you will get a message on the screen each time it re-estimates the values, like:
Code:
Icarus %d Re-estimate: Hs=%e W=%e read_count=%d fullnonce=%.3fs
The first %d is just the Icarus ID number
The Hs=%e is the hash nanosecond time it has calculated (in place of 2.6316 * 10^-9)
The W=%e is the code overhead per work
The read_count=%d is the new replacement value for 11.2 (times ten)
The fullnonce=%.3fs is how long it actually takes to complete a full nonce range

You can also ask the API for the current values it is using:
Code:
java API stats

Once you have a new correct nanosecond value for your faster bitstream (lets say it was 2.2222ns), you simply start cgminer with the option:
--icarus-timing 2.2222

Of course if you have any trouble with this, come visit FreeNode IRC #cgminer - I'm almost always logged into it as 'kanoi' (but not always available to help but I will see the messages later if I'm asleep or elsewhere)

Pool: https://kano.is BTC: 1KanoiBupPiZfkwqB7rfLXAzPnoTshAVmb
CKPool and CGMiner developer, IRC FreeNode #ckpool and #cgminer kanoi
Help keep Bitcoin secure by mining on pools with Stratum, the best protocol to mine Bitcoins with ASIC hardware
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
June 29, 2012, 08:44:20 AM
 #926

Thanks kano, I appreciate the in-depth reply.

Since the only other bistream is a 200mhz as compared to a 190mhz, any problem with just a down and dirty 5% alteration to the timing rather than a recalc by the miner software ?

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
kano
Legendary
*
Offline Offline

Activity: 1932


Linux since 1997 RedHat 4


View Profile
June 29, 2012, 11:16:13 AM
 #927

Thanks kano, I appreciate the in-depth reply.

Since the only other bistream is a 200mhz as compared to a 190mhz, any problem with just a down and dirty 5% alteration to the timing rather than a recalc by the miner software ?

kind regards
Well the complete recalc in 'short' mode should only take about an hour or so Smiley
and you can look at the 2nd value it spits out after about 5 minutes and that will tell you roughly what it is.

However, if you don't care about the MH/s display being correct, you can simply tell it the abort time.
e.g. --icarus-timing 2.5=80
That tells it to assume 2.5ns per hash and abort at 8.0 seconds.
2.5ns would actually come up as 10.6seconds, but giving it less will ensure you are under the required limit.
The wrong nanoseconds wont affect the number of shares you get (U:) if you also have a valid abort time, but it will display the MH/s wrong if 2.5 isn't close to the correct value.

The abort time can be as low as you like.
The only loss from having the abort value too low is the overhead of starting extra work.
When a miner checks a nonce range it doesn't matter if it aborts early - there is no statistical loss of shares or blocks when you don't check the full nonce range.
You will of course expect to find (on average) one share every time you check 2^32 nonce values, but it doesn't matter if they are from different getworks - it's simply how many nonce values you check, not which ones you check.

Pool: https://kano.is BTC: 1KanoiBupPiZfkwqB7rfLXAzPnoTshAVmb
CKPool and CGMiner developer, IRC FreeNode #ckpool and #cgminer kanoi
Help keep Bitcoin secure by mining on pools with Stratum, the best protocol to mine Bitcoins with ASIC hardware
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
June 29, 2012, 11:23:19 AM
 #928

Thanks again kano, very informative.

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
DrHaribo
Legendary
*
Offline Offline

Activity: 1960


Bitminter.com Operator


View Profile WWW
June 29, 2012, 01:09:21 PM
 #929

Adding Icarus support to my miner. But I can't get it to ever return a nonce starting with three null bytes. I feed it data for which there is such a solution, but it's not working.

Is that normal? Does it start scanning from nonce 0x00000100 ?

▶▶▶ Bitminter.com - Your trusted mining pool since 2011.
DrHaribo
Legendary
*
Offline Offline

Activity: 1960


Bitminter.com Operator


View Profile WWW
June 29, 2012, 04:28:52 PM
 #930

Nonces starting with three nullbytes (0x000000??) are never reported. Nonces starting with three FF (0xFFFFFF??) always get 01 for the fourth byte, which is usually an incorrect result. This had me confused for a while. Error in the bitstream?

▶▶▶ Bitminter.com - Your trusted mining pool since 2011.
Energizer
Sr. Member
****
Offline Offline

Activity: 274



View Profile
July 01, 2012, 03:24:23 AM
 #931

I've created a new thread to sell my Icarus boards:
https://bitcointalk.org/index.php?topic=90956.0
ngzhang
Hero Member
*****
Offline Offline

Activity: 592


We will stand and fight.


View Profile
July 01, 2012, 11:55:21 AM
 #932

Nonces starting with three nullbytes (0x000000??) are never reported. Nonces starting with three FF (0xFFFFFF??) always get 01 for the fourth byte, which is usually an incorrect result. This had me confused for a while. Error in the bitstream?


Adding Icarus support to my miner. But I can't get it to ever return a nonce starting with three null bytes. I feed it data for which there is such a solution, but it's not working.

Is that normal? Does it start scanning from nonce 0x00000100 ?


yes, it didn't scan the nonce range form 0. approx: 132

CEO of Canaan-creative, Founder of Avalon project.
https://canaan.io/
Business contact: love@canaan.io
All PMs will be unread.
luffy
Hero Member
*****
Offline Offline

Activity: 606



View Profile
July 01, 2012, 01:24:52 PM
 #933

i tried to run icarus with cgminer in an ATOM laptop but cgminer said "no opencl.ddl available" something like that.  what can i do?
i tried to load opencl sdk from microsoft but i see the message "no opencl capable GPU, instalation abort"! Sad
randomguy7
Hero Member
*****
Offline Offline

Activity: 528


View Profile
July 01, 2012, 01:27:37 PM
 #934

i tried to run icarus with cgminer in an ATOM laptop but cgminer said "no opencl.ddl available" something like that.  what can i do?
i tried to load opencl sdk from microsoft but i see the message "no opencl capable GPU, instalation abort"! Sad

Do you GPU mine on that box, too? If not, I'd recommend to compile cgminer from source disabling GPU support.
tnkflx
Sr. Member
****
Offline Offline

Activity: 346


View Profile
July 01, 2012, 01:53:06 PM
 #935

i tried to run icarus with cgminer in an ATOM laptop but cgminer said "no opencl.ddl available" something like that.  what can i do?
i tried to load opencl sdk from microsoft but i see the message "no opencl capable GPU, instalation abort"! Sad

Do you GPU mine on that box, too? If not, I'd recommend to compile cgminer from source disabling GPU support.

Crazy idea and all, but maybe read the cgminer thread and the cgminer README?

The option you want is "--disable-gpu|-G: Disable GPU mining even if suitable devices exist"

| Operating electrum.be & us.electrum.be |
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
July 01, 2012, 02:30:13 PM
 #936

Crazy idea and all, but maybe read the cgminer thread and the cgminer README?

The option you want is "--disable-gpu|-G: Disable GPU mining even if suitable devices exist"

Reading the thread will show you that other ppl cant run without the SDK installed regardless of the --disable-gpu

Recompiling without OpenCL is the only way.

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
tnkflx
Sr. Member
****
Offline Offline

Activity: 346


View Profile
July 01, 2012, 03:02:57 PM
 #937

Crazy idea and all, but maybe read the cgminer thread and the cgminer README?

The option you want is "--disable-gpu|-G: Disable GPU mining even if suitable devices exist"

Reading the thread will show you that other ppl cant run without the SDK installed regardless of the --disable-gpu

Recompiling without OpenCL is the only way.

kind regards

I did not know that and now I've learned something, thanks! (SDK is always installed on my machines...)

| Operating electrum.be & us.electrum.be |
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
July 01, 2012, 03:16:08 PM
 #938

(SDK is always installed on my machines...)

Same, I have a desktop GPU miner and an e350 with integrated.
When I tried to run an FPGA only setup on a p3 tablet I got the OpenCL issue crop up.

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
kano
Legendary
*
Offline Offline

Activity: 1932


Linux since 1997 RedHat 4


View Profile
July 03, 2012, 01:05:32 PM
 #939

Nonces starting with three nullbytes (0x000000??) are never reported. Nonces starting with three FF (0xFFFFFF??) always get 01 for the fourth byte, which is usually an incorrect result. This had me confused for a while. Error in the bitstream?


Adding Icarus support to my miner. But I can't get it to ever return a nonce starting with three null bytes. I feed it data for which there is such a solution, but it's not working.

Is that normal? Does it start scanning from nonce 0x00000100 ?


yes, it didn't scan the nonce range form 0. approx: 132
Can you give an exact answer?
Since that will slightly affect the timing for when the nonce range is aborted and also how many nonces were actually processed when a nonce is found.
i.e. also details of how it affects both FPGAs

Pool: https://kano.is BTC: 1KanoiBupPiZfkwqB7rfLXAzPnoTshAVmb
CKPool and CGMiner developer, IRC FreeNode #ckpool and #cgminer kanoi
Help keep Bitcoin secure by mining on pools with Stratum, the best protocol to mine Bitcoins with ASIC hardware
wildemagic
Member
**
Offline Offline

Activity: 112



View Profile
July 04, 2012, 08:02:50 PM
 #940

OK, ive got my xilinx cable to try and update the icarus firmware. BUT, I want to try the .bit file bitstreams before I permanently flash the spi.

Just not sure how to utilise the .bit file to temporarily hash with the icarus unit instead of using the xilinx tools to flash the mcs permanently.

ngzhang outlines in the icarus wiki that you can test the .bit files without writing anything permanently but gives no details.

If anyone can provide more information I would appreciate it, otherwise I only have the v3 bitstream mcs file that works with xilinx tools.

kind regards

.,-._|\     Offgrid 1.7kW Solar and 3G wireless internet powering my mining rig.
/ .Oz. \
\_,--.x/     [219.5btc of successful trades total] with : rastapool, miernik, flatronw & OneFixt
       o
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 44 45 46 [47] 48 49 50 »
  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!