[ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 824.

dcct

sr. member

Activity: 280

Merit: 250

Quote from: carlos on September 14, 2014, 03:09:11 PM

Quote from: dcct on September 14, 2014, 03:02:26 PM

Give it a try. I think its a lot faster than 4500 nonces/min.

Thank you very much..

And what about that formula? Should staggered nonces fit in RAM?
e.g. your stagger number 50000*0.256= 12.8GB
So you've got 12.8 GB available RAM?

Sorry I'm totally in the dark

Thats the amount of memory you need to create a plot with that stagger size. Using my C-plotter. If you merge if afterwards, it can be created with a lot less memory.

Yes its 256kb per nonce

carlos

member

Activity: 107

Merit: 10

Quote from: dcct on September 14, 2014, 03:02:26 PM

Give it a try. I think its a lot faster than 4500 nonces/min.

Thank you very much..

And what about that formula? Should staggered nonces fit in RAM?
e.g. your stagger number 50000*0.256= 12.8GB (1 nonce = 256kb)
So you've got 12.8 GB available RAM?

Sorry I'm totally in the dark

dcct

sr. member

Activity: 280

Merit: 250

Quote from: carlos on September 14, 2014, 02:59:25 PM

Quote from: dcct on September 14, 2014, 02:55:47 PM

Merging from a small stagger size takes quite some time - but it depends on how fast your HDD is. But its worth it!

I have just one file on them. As long as its not 100's of small files it does not matter.

Do you have some estimation? Isn't faster to generate them all over? I'm doing 4500 nonces/min

Give it a try. I think its a lot faster than 4500 nonces/min.

carlos

member

Activity: 107

Merit: 10

Quote from: dcct on September 14, 2014, 02:55:47 PM

Merging from a small stagger size takes quite some time - but it depends on how fast your HDD is. But its worth it!

I have just one file on them. As long as its not 100's of small files it does not matter.

Do you have some estimation? Isn't faster to generate them all over? I'm doing 4500 nonces/min

carlos

member

Activity: 107

Merit: 10

Quote from: fanepatent on September 14, 2014, 02:36:26 PM

Quote from: fanepatent on September 14, 2014, 02:00:06 PM

http://burstdice.ddns.net/

80%, 60%, 40% and 20% chances.
2% house edge since minimum bet is 10 and fee is 1 (10%)
Just send to one address and wait for the result.

bump

tested and working for me...

dcct

sr. member

Activity: 280

Merit: 250

Quote from: carlos on September 14, 2014, 02:53:56 PM

Wow .. 20 seconds is really fast for 4TB.. I'm nowhere near this...
I've created it using GPU Plotter and thought its mainly for optimizing the plotting generation. I've used some 800 or what according to some recommendation to set it whatever is MB RAM of graphics card...

Clearly its the issue...

How long does it take to say merge 50 files into one 1TB with different stagger using your utility? Is it feasible?

How many files do you have on that usb3 drive? Does also the number of files have some impact or its just matter of stagger...

Merging from a small stagger size takes quite some time - but it depends on how fast your HDD is. Its worth it!

I have just one file on them. As long as its not 100's of small files it does not matter.

carlos

member

Activity: 107

Merit: 10

Quote from: dcct on September 14, 2014, 02:15:28 PM

Quote from: carlos on September 14, 2014, 01:58:15 PM

Quote from: carlos on September 14, 2014, 10:47:37 AM

This seems to me that also size of plot & disk i/o are limiting factor. (Sometimes blocks comes sooner than my miner evaluates whole plot). How long does it take your miner to evaluate yours?

I'm currently plotted to multiple 25GB files... Is it problem? Do you use one file per HDD partition (e.g. 1 TB file)? Is there much of improvement with disk IO?

I'm using ext2, noatime... according to vmstat there is max 2000 kB_read/s disk reading during mining (so very mild without any significant CPU load).

I don't want to repost but... Do you have please any thoughts on this about mining performance? Thanks!

It depends on the stagger size you use. My 4TB usb3-drives take ~ 20s to evaluate, with 50k stagger size. If you go with 1000 or less - thats 64kb/read - you should merge them using the merge utility.

Wow .. 20 seconds is really fast for 4TB.. I'm nowhere near this...
I've created it using GPU Plotter and thought its mainly for optimizing the plotting generation. I've used some 800 or what according to some recommendation to set it whatever is MB RAM of graphics card...

Clearly its the issue...

How long does it take to say merge 50 files into one 1TB and change stagger to your recommendation using your utility? Is it feasible?

How can I come with exact stagger for optimized mining? Is there some formula?

How many files do you have on that usb3 drive? Does also the number of files have some impact or its just matter of stagger...

dcct

sr. member

Activity: 280

Merit: 250

Quote from: Avaahnaa on September 14, 2014, 02:41:53 PM

Where is this merge utility. The only thing that I can think of that you could be referring to has been taken down.

Right on the OP.

https://bchain.info/dcct_miner.tgz

It also includes the merge tool

Avaahnaa

member

Activity: 67

Merit: 10

Quote from: dcct on September 14, 2014, 02:15:28 PM

Quote from: carlos on September 14, 2014, 01:58:15 PM

Quote from: carlos on September 14, 2014, 10:47:37 AM

This seems to me that also size of plot & disk i/o are limiting factor. (Sometimes blocks comes sooner than my miner evaluates whole plot). How long does it take your miner to evaluate yours?

I'm currently plotted to multiple 25GB files... Is it problem? Do you use one file per HDD partition (e.g. 1 TB file)? Is there much of improvement with disk IO?

I'm using ext2, noatime... according to vmstat there is max 2000 kB_read/s disk reading during mining (so very mild without any significant CPU load).

I don't want to repost but... Do you have please any thoughts on this about mining performance? Thanks!

It depends on the stagger size you use. My 4TB usb3-drives take ~ 20s to evaluate, with 50k stagger size. If you go with 1000 or less - thats 64kb/read - you should merge them using the merge utility.

Where is this merge utility. The only thing that I can think of that you could be referring to has been taken down.

fanepatent

full member

Activity: 224

Merit: 100

Quote from: fanepatent on September 14, 2014, 02:00:06 PM

http://burstdice.ddns.net/

80%, 60%, 40% and 20% chances.
2% house edge since minimum bet is 10 and fee is 1 (10%)
Just send to one address and wait for the result.

bump

Stanist

newbie

Activity: 2

Merit: 0

Quote from: m3ta on September 14, 2014, 01:32:31 PM

Quote from: Stanist on September 14, 2014, 01:07:05 PM

Is it possible to use a synology diskstation to store plot files?

Yes. This has already been said in this thread.

Quote from: Stanist on September 14, 2014, 01:07:05 PM

How to link them to miner it thats work? tnx for answer.

Any way you want. Synology has SMB, NFS export on its shared dirs, just use whatever.

From your question, it seems to me you didn't even try before asking.

You right i have not tried it but wanted to know before start ploting. Only have an old nvida card so i cpu plot slow. I thank you for fast and good answer. Will give it a try now
/ tnx

go6ooo1212

legendary

Activity: 1512

Merit: 1000

quarkchain.io

Quote from: unsoindovo on September 14, 2014, 02:11:55 PM

Quote from: go6ooo1212 on September 14, 2014, 02:09:08 PM

Quote from: unsoindovo on September 14, 2014, 01:52:33 PM

Quote from: go6ooo1212 on September 14, 2014, 01:29:08 PM

Could someone help me with the number of noons per HDD. I taught the count is the * (which is usually 4096 with NTFS under WIn).
Example: I set the number of noons 7360512 which should be arount 1800GB, but the gpuplotgenerator 2.1.1 made the plot around 1500GB. How to properly clac the number of noons ?

maybe this xls can help you.

https://www.dropbox.com/s/qajs4fj9iyc927l/Burst_PLOT.xlsx?dl=0

settings for cpu/gpu plotter

have fun

Thanks you for this export , I have something more to understand. When I set the number of noons into the gpuplotgen... , ,it alway change the number of noons itself with some small value around 2000 noons.
Once its 832 noons , the other is 2048 noons of difference , and I cant dont know where it is comeing from ...

you need to plot in CPU mode or GPU mode?
if gpu mode AMD or Nvidia?

Im plotting with AMD R9 280X, with gpuplotgenenrator 2.1.1; 14.7 drivers and AMD SDK 2.9.1
EDIT: My command line is:
gpuPlotGenerator.exe generate 0 0 "localpath\plots" account 25955332 1515520 3072 128 1024

pause

dcct

sr. member

Activity: 280

Merit: 250

Quote from: carlos on September 14, 2014, 01:58:15 PM

Quote from: carlos on September 14, 2014, 10:47:37 AM

This seems to me that also size of plot & disk i/o are limiting factor. (Sometimes blocks comes sooner than my miner evaluates whole plot). How long does it take your miner to evaluate yours?

I'm currently plotted to multiple 25GB files... Is it problem? Do you use one file per HDD partition (e.g. 1 TB file)? Is there much of improvement with disk IO?

I'm using ext2, noatime... according to vmstat there is max 2000 kB_read/s disk reading during mining (so very mild without any significant CPU load).

I don't want to repost but... Do you have please any thoughts on this about mining performance? Thanks!

It depends on the stagger size you use. My 4TB usb3-drives take ~ 20s to evaluate, with 50k stagger size. If you go with 1000 or less - thats 64kb/read - you should merge them using the merge utility.

unsoindovo

legendary

Activity: 1932

Merit: 1042

https://locktrip.com/?refId=40964

Quote from: go6ooo1212 on September 14, 2014, 02:09:08 PM

Quote from: unsoindovo on September 14, 2014, 01:52:33 PM

Quote from: go6ooo1212 on September 14, 2014, 01:29:08 PM

Could someone help me with the number of noons per HDD. I taught the count is the * (which is usually 4096 with NTFS under WIn).
Example: I set the number of noons 7360512 which should be arount 1800GB, but the gpuplotgenerator 2.1.1 made the plot around 1500GB. How to properly clac the number of noons ?

maybe this xls can help you.

https://www.dropbox.com/s/qajs4fj9iyc927l/Burst_PLOT.xlsx?dl=0

settings for cpu/gpu plotter

have fun

Thanks you for this export , I have something more to understand. When I set the number of noons into the gpuplotgen... , ,it alway change the number of noons itself with some small value around 2000 noons.
Once its 832 noons , the other is 2048 noons of difference , and I cant dont know where it is comeing from ...

you need to plot in CPU mode or GPU mode?
if gpu mode AMD or Nvidia?

go6ooo1212

legendary

Activity: 1512

Merit: 1000

quarkchain.io

Quote from: unsoindovo on September 14, 2014, 01:52:33 PM

Quote from: go6ooo1212 on September 14, 2014, 01:29:08 PM

Could someone help me with the number of noons per HDD. I taught the count is the * (which is usually 4096 with NTFS under WIn).
Example: I set the number of noons 7360512 which should be arount 1800GB, but the gpuplotgenerator 2.1.1 made the plot around 1500GB. How to properly clac the number of noons ?

maybe this xls can help you.

https://www.dropbox.com/s/qajs4fj9iyc927l/Burst_PLOT.xlsx?dl=0

settings for cpu/gpu plotter

have fun

Thanks you for this export , I have something more to understand. When I set the number of noons into the gpuplotgen... , ,it alway change the number of noons itself with some small value around 2000 noons.
Once its 832 noons , the other is 2048 noons of difference , and I cant dont know where it is comeing from ...

unsoindovo

legendary

Activity: 1932

Merit: 1042

https://locktrip.com/?refId=40964

Quote from: Amph on September 14, 2014, 02:05:08 PM

for those who want to plot with 750ti

i'm getting a good result with 8 128(last two numbers, thread and hashes)

7200n/m

amph
no news about c miner for windows?

Amph

legendary

Activity: 3248

Merit: 1070

for those who want to plot with 750ti

i'm getting a good result with 8 128(last two numbers, thread and hashes)

7200n/m

timk225

hero member

Activity: 955

Merit: 1004

Quote from: timk225 on September 14, 2014, 09:41:09 AM

I have one PC running one solo mining instance, and another running two. All three instances randomly close. Every now and then the mining screen will just close and I have to restart it. This is very frustrating, as I never know when mining has just stopped, seemingly for no reason. WTF is this problem?

I don't think the wallet would affect it, but this never happened before I put the 1.1.1 wallet on.

Are there newer versions of the solo miner?

How can I re-code the miner config file to make it auto restart? Someone had mentioned doing that. The miner just randomly shutting off was never a problem until yesterday.

fanepatent

full member

Activity: 224

Merit: 100

http://burstdice.ddns.net/

80%, 60%, 40% and 20% chances.
2% house edge since minimum bet is 10 and fee is 1 (10%)
Just send to one address and wait for the result.

carlos

member

Activity: 107

Merit: 10

Quote from: carlos on September 14, 2014, 10:47:37 AM

This seems to me that also size of plot & disk i/o are limiting factor. (Sometimes blocks comes sooner than my miner evaluates whole plot). How long does it take your miner to evaluate yours?

I'm currently plotted to multiple 25GB files... Is it problem? Do you use one file per HDD partition (e.g. 1 TB file)? Is there much of improvement with disk IO?

I'm using ext2, noatime... according to vmstat there is max 2000 kB_read/s disk reading during mining (so very mild without any significant CPU load).

I don't want to repost but... Do you have please any thoughts on this about mining performance? Thanks!

Topic: [ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 824. (Read 2170900 times)