Pages:
Author

Topic: [ANN][YAC] Yacminer GPU miner for Yacoin - page 10. (Read 57517 times)

hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
August 02, 2013, 03:52:28 PM
#78
what settings are you using for 7950?

"intensity" : "15",
"vectors" : "1",
"shaders" : "1792",
"worksize" : "256",
"lookup-gap" : "2",
"thread-concurrency" : "31712",
"gpu-threads" : "1",

Core clock and mem clock at 1000MHz. I think you can get more out of them with higher core clock frequency, but that's more than I'm happy overclocking (I'm a softie and want to keep them reasonably cool). Mem clock doesn't make a huge difference for me and clocking down should make a little difference to my electricity bill!
full member
Activity: 214
Merit: 100
August 02, 2013, 03:39:48 PM
#77
what settings are you using for 7950?
hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
August 02, 2013, 02:22:37 PM
#76
Yep 7950s.
1 Asus, 4 Gigabytes (the old awesome f43s)

OK, pm'ed with a config file that works for me. I'd have thought that you should get well in excess of 500 with those five GPUs. Let me know if it works out.
member
Activity: 103
Merit: 10
August 02, 2013, 12:53:32 PM
#75
Yep 7950s.
1 Asus, 4 Gigabytes (the old awesome f43s)
hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
August 02, 2013, 02:40:43 AM
#74
I really feel like I'm behind the curve on this.  I'm running 4 cards at only 75khps each. I just got another card and it's only running at 60. 

I tried --worksize 256 --lookup-gap 2 -I 15 --thread-concurrency 21712
but I only get HW errors.

Anyway, I'm willing to pay 200 YAC if you get my single card above 100, and 500 YAC if you get my 4 about 100.


If you're using 7950s I'd be happy to send you my cgminer config file. Just let me know.
member
Activity: 103
Merit: 10
August 02, 2013, 02:11:24 AM
#73
I really feel like I'm behind the curve on this.  I'm running 4 cards at only 75khps each. I just got another card and it's only running at 60. 

I tried --worksize 256 --lookup-gap 2 -I 15 --thread-concurrency 21712
but I only get HW errors.

Anyway, I'm willing to pay 200 YAC if you get my single card above 100, and 500 YAC if you get my 4 about 100.
full member
Activity: 196
Merit: 100
rise YAC rise !
hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
Hi,

can anyone post the optimal settings for a 7950 please?
With --worksize 256 --lookup-gap 2 -I 15 --thread-concurrency 21712 i got 110 Kh/s
It depends on your card so you have to have a play, though you're not far off the mark.  I can only get 123 kh/s out of mine.

115 Kh/s is about the best that I can get out of mine too (with those settings). I'm using 1000MHz core and memory clocks. Higher memory clock doesn't seem to help much for me and I'm a bit reluctant to push the core clock higher with my current cooling.
hero member
Activity: 820
Merit: 1000
Hi,

can anyone post the optimal settings for a 7950 please?
With --worksize 256 --lookup-gap 2 -I 15 --thread-concurrency 21712 i got 110 Kh/s
It depends on your card so you have to have a play, though you're not far off the mark.  I can only get 123 kh/s out of mine.
sr. member
Activity: 252
Merit: 250
settings for a 560ti at N=11 10kh

--scrypt --worksize 128 --lookup-gap 2 -I 11 --thread-concurrency 8448

i5 2500 is doing just under 6kh
newbie
Activity: 10
Merit: 0
Hi,

can anyone post the optimal settings for a 7950 please?
With --worksize 256 --lookup-gap 2 -I 15 --thread-concurrency 21712 i got 110 Kh/s
hero member
Activity: 693
Merit: 500
I've updated my settings on my 7850 with 1GB of RAM to the following :

yacminer --scrypt --worksize 256 --lookup-gap 3 -I 15 --thread-concurrency 18200 --gpu-engine 1100 --gpu-memclock 1250

this produces 61.15 KH/sec - higher intensities push it up slightly, but the WU/s go down.  I get no hardware errors with this configuration.  With a LUG of 2 best I could get was 76 KH/sec with a significant number of hardware errors. 

I've been tuning my thread-concurrency on each card by setting it to an outrageous number, and running cgminer.  It will then tell me what I'm trying to allocate memory wise (TRY), and what it could allocate(MAX).  I then use those two numbers to lower the TC down to (MAX/TRY) * TC and then subtract 256. 

My 5870 still gets about 51KH/sec using the same settings as N=10, just with a lower intensity.  Again, with no hardware errors.

In the code, what does the hardware error actually occur from?  is it running out of memory, hashes not adding up, or something else?

would you mind posting your settings for your 5870?  i am unable to get mine to accept a single share there all HW errors

yacminer --scrypt --worksize 64 --lookup-gap 2 -I 11 --thread-concurrency 8192

This is my active desktop machine with three monitors connected and set to give decent responsiveness.  If I up the intensity to 18 (as I same mumus was doing in another thread) I get well over 80 KH/s, but the machine is completely unusable for anything else, and gave more hardware errors than accepted shares.
legendary
Activity: 1197
Merit: 1000
my both pools are using stratum so GPU miners are welcome:

http://yac.coinmine.pl
http://yacp.coinmine.pl

feeleep
full member
Activity: 213
Merit: 100
Mining history: BTC->LTC->FTC->CRC:2200kh/s
I've updated my settings on my 7850 with 1GB of RAM to the following :

yacminer --scrypt --worksize 256 --lookup-gap 3 -I 15 --thread-concurrency 18200 --gpu-engine 1100 --gpu-memclock 1250

this produces 61.15 KH/sec - higher intensities push it up slightly, but the WU/s go down.  I get no hardware errors with this configuration.  With a LUG of 2 best I could get was 76 KH/sec with a significant number of hardware errors. 

I've been tuning my thread-concurrency on each card by setting it to an outrageous number, and running cgminer.  It will then tell me what I'm trying to allocate memory wise (TRY), and what it could allocate(MAX).  I then use those two numbers to lower the TC down to (MAX/TRY) * TC and then subtract 256. 

My 5870 still gets about 51KH/sec using the same settings as N=10, just with a lower intensity.  Again, with no hardware errors.

In the code, what does the hardware error actually occur from?  is it running out of memory, hashes not adding up, or something else?

would you mind posting your settings for your 5870?  i am unable to get mine to accept a single share there all HW errors
hero member
Activity: 693
Merit: 500
In the code, what does the hardware error actually occur from?  is it running out of memory, hashes not adding up, or something else?

It comes from two (or more) threads using the same bit of memory. If the buffer allocated on the GPU is not enough for the active threads, then the threads will start wrapping around in the buffer. The threads will start overwriting each others intermediary values stored in the memory which screws up the hash calculation for those threads. The wrongly calculated hashes pop up as HW errors when they meet the minimum share difficulty and are double-checked by the miner.

I'm so used to reddit - I really wanted to upvote this post.  Thanks for the answer!
sr. member
Activity: 301
Merit: 250
BTW. I think it would be useful to add a percentage of HW errors of all processed blocks. That way one could see the effective hash rate when taking HW errors into account.

Yup, I've been considering adding HW error percentage.
sr. member
Activity: 301
Merit: 250
In the code, what does the hardware error actually occur from?  is it running out of memory, hashes not adding up, or something else?

It comes from two (or more) threads using the same bit of memory. If the buffer allocated on the GPU is not enough for the active threads, then the threads will start wrapping around in the buffer. The threads will start overwriting each others intermediary values stored in the memory which screws up the hash calculation for those threads. The wrongly calculated hashes pop up as HW errors when they meet the minimum share difficulty and are double-checked by the miner.
newbie
Activity: 29
Merit: 0
I did a brief test on my 7870, 7870xt and 7950.. The formula failed on all those.

I have tried differed TCs a bit, and currently I get 78 kH/s with 4096 TC on 7870xt, and 72 kH/s with 4096 TX on 7870. Both are clocked at 1150/1000 (GPU/Mem). These are the highest / best performing settings. Intensity is 11 on both. With higher intensity I get better hash rate, but there are more HW errors.

On my 7950 I now get 67 kH/s, I haven't been able to get better performance out of it. TC is at 7936, intensity 11.

BTW. I think it would be useful to add a percentage of HW errors of all processed blocks. That way one could see the effective hash rate when taking HW errors into account.

hero member
Activity: 693
Merit: 500
I've updated my settings on my 7850 with 1GB of RAM to the following :

yacminer --scrypt --worksize 256 --lookup-gap 3 -I 15 --thread-concurrency 18200 --gpu-engine 1100 --gpu-memclock 1250

this produces 61.15 KH/sec - higher intensities push it up slightly, but the WU/s go down.  I get no hardware errors with this configuration.  With a LUG of 2 best I could get was 76 KH/sec with a significant number of hardware errors. 

I've been tuning my thread-concurrency on each card by setting it to an outrageous number, and running cgminer.  It will then tell me what I'm trying to allocate memory wise (TRY), and what it could allocate(MAX).  I then use those two numbers to lower the TC down to (MAX/TRY) * TC and then subtract 256. 

My 5870 still gets about 51KH/sec using the same settings as N=10, just with a lower intensity.  Again, with no hardware errors.

In the code, what does the hardware error actually occur from?  is it running out of memory, hashes not adding up, or something else?
sr. member
Activity: 301
Merit: 250
Here's an alternative table how the miner scales with lookup gap:

Code:
Lookup gap | Hashrate
-----------+----------
2          | 61.01
4          | 39.24
8          | 21.61
16         | 11.97

Power consumption is likely to remain the same but I haven't measured it.

As you can see, there are at least 2 options for scaling. One option reduces both the hashrate and the power consumption. The other option provides higher hashrates at the cost of high power consumption. These options can also be combined. And this is exactly why I don't back of the envelope calculations of when CPUs are going to break even with GPUs.
Pages:
Jump to: