Pages:
Author

Topic: Marlin - SiaCoin OpenCL/CUDA Miner - CPU/GPU/iGPU - page 2. (Read 12012 times)

full member
Activity: 168
Merit: 100
You're definitely doing better than me, bathrobehero.

Which specific 970 do you have? Is it overclocked? Mine is all factory settings.

(Edited my post with more results)



No fees apply? That's rather honorable! Especially in these times where everyone is pickpocketing like their lives depend on it

So no fees at all for marlin itself?

What are the fees for the pool - 3%? Does xurious get something for that (I hope so)?
legendary
Activity: 2002
Merit: 1051
ICO? Not even once.
I've been running this on my EVGA GTX 970 with Intensity = 25 and getting a pretty consistent 838.8 MH/s.

Is there some way to find the optimal intensity, or is it only trial and error? Is there any dynamic option so that marlin checks itself and settles on the intensity that gets the best results?


EDIT:
So I ran with --benchmark and it appears that the m2p4 kernel is best, at intensity 27?

Running for a few minutes now and it's up to 894.7MH/s even though the benchmark returned 1016.7. Still a decent improvement. Smiley

I get 1090 Mh/s per 970 with -I 29 or 30 and with -K m2p4.
full member
Activity: 243
Merit: 100
No fees apply? That's rather honorable! Especially in these times where everyone is pickpocketing like their lives depend on it
full member
Activity: 168
Merit: 100
I've been running this on my EVGA GTX 970 with Intensity = 25 and getting a pretty consistent 838.8 MH/s.

Is there some way to find the optimal intensity, or is it only trial and error? Is there any dynamic option so that marlin checks itself and settles on the intensity that gets the best results?


EDIT:
So I ran with --benchmark and it appears that the m2p4 kernel is best for me, at intensity 27.

Running for a few minutes now and it's up to 894.7MH/s even though the benchmark returned 1016.7. Still a decent improvement. Smiley


Various results after a few minutes:
default kernel at intensity 25: 838.8 MH/s

m2p2 kernel at intensity 27: 888.8 MH/s (most auspicious if you are Chinese!)
m2p2 kernel at intensity 24: 883.0 MH/s

m2p4 kernel at intensity 30:897.0MH/s (GPU temp at 82 and desktop basically unusable)
m2p4 kernel at intensity 27:894.7MH/s
m2p4 kernel at intensity 24:883.0MH/s

x1p1 kernel at intensity 27: 888.8MH/s
x1p1 kernel at intensity 25: 883.0MH/s

So all things considered, it looks like m2p4 it is.

Would love to hear some recommendations on how to determine the best intensity. Thanks.
full member
Activity: 349
Merit: 102
Gigabyte 1070 Mini ITX OC +255 core ~ 1815MH/s (using m2p2 kernel)
iom
newbie
Activity: 31
Merit: 0
Gigabyte 1070 G1 +150 core ~ 1750mh/s
legendary
Activity: 2002
Merit: 1051
ICO? Not even once.
Anyone knows what is the expected hashrate for 1070 OC?

I'm curious as well.
sr. member
Activity: 463
Merit: 250
Anyone knows what is the expected hashrate for 1070 OC?
sr. member
Activity: 291
Merit: 250
just started trying it .... will give it a shot for the day, but so far in my 10 or days mining limited experience ... as simple as the Minergate windows app ... but on linux ... Smiley

Code:
2016/09/23 20:09:12 Starting marlin 1.0.0
2016/09/23 20:09:12 OpenCL: AMD Accelerated Parallel Processing
2016/09/23 20:09:12  [0] GPU: Spectre
2016/09/23 20:09:12  [1] GPU: Cedar
2016/09/23 20:09:12  [2] CPU: AMD PRO A8-8650B R7, 10 Compute Cores 4C+6G
2016/09/23 20:09:12 Connecting to siamining.com:3333...
2016/09/23 20:09:12 [0] Initializing Spectre
2016/09/23 20:09:12 [1] Initializing Cedar
2016/09/23 20:09:13 Difficulty set to 4.2G
2016/09/23 20:09:13 New block ...3e4f76f2 detected, difficulty 13P
2016/09/23 20:09:13 Authentication successful
2016/09/23 20:09:15 [1] Initialized, work size 16777216
2016/09/23 20:09:17 [0] Initialized, work size 16777216
2016/09/23 20:09:49 Difficulty set to 536M
2016/09/23 20:09:51 [0] Accepted 6088850a D:611M/536M  121.4 MH/s
2016/09/23 20:09:56 [1] Accepted 95f2ed04 D:564M/536M   13.4 MH/s
2016/09/23 20:10:12 A:2 R:0 E:0 0:124.5 1:13.4 138.0/137.2 MH/s
2016/09/23 20:10:15 [0] Accepted 6132453c D:1.0G/536M  121.3 MH/s
2016/09/23 20:10:15 [0] Accepted ccffa33e D:1.8G/536M  124.5 MH/s
2016/09/23 20:10:15 [0] Accepted 45aedb3e D:897M/536M  124.5 MH/s
2016/09/23 20:10:16 [0] Accepted 91193248 D:653M/536M  124.4 MH/s
2016/09/23 20:10:23 [0] Accepted f4fd2c79 D:1.2G/536M  124.3 MH/s
2016/09/23 20:10:23 [0] Accepted 99033a79 D:679M/536M  124.3 MH/s
hero member
Activity: 952
Merit: 508
I can report 55mh using the iGPU on an Intel i7-5775c.

Not much I know but it's better than nothing when the GPU's in the system and the processor are mining other coins Cheesy
sr. member
Activity: 413
Merit: 250
it doesn't work for cuda 2.1 card,

http://prntscr.com/ckyrjx

Could you provide more specs? I've verified it works on an NVS 4200m. While not an amazing chip, it's certainly cuda 2.1, equivalent to a "GT 520m."

r9 290 gets 1147MH/s with 1050/1250

Most however just dual mine this coin with Claymore since its more profitable.

For the rest, Marlin is a great miner that offers significant speed advantages over some of the others.

after it i tested with pool and it worked,that was benchmark.and card was geforce gt 720m

Thanks, I'll look into the benchmark on the nvs I have, and verify on a 740.
sr. member
Activity: 462
Merit: 250
Arianee:Smart-link Connecting Owners,Assets,Brands
it doesn't work for cuda 2.1 card,

http://prntscr.com/ckyrjx

Could you provide more specs? I've verified it works on an NVS 4200m. While not an amazing chip, it's certainly cuda 2.1, equivalent to a "GT 520m."

r9 290 gets 1147MH/s with 1050/1250

Most however just dual mine this coin with Claymore since its more profitable.

For the rest, Marlin is a great miner that offers significant speed advantages over some of the others.

after it i tested with pool and it worked,that was benchmark.and card was geforce gt 720m
sr. member
Activity: 413
Merit: 250
it doesn't work for cuda 2.1 card,

http://prntscr.com/ckyrjx

Could you provide more specs? I've verified it works on an NVS 4200m. While not an amazing chip, it's certainly cuda 2.1, equivalent to a "GT 520m."

r9 290 gets 1147MH/s with 1050/1250

Most however just dual mine this coin with Claymore since its more profitable.

For the rest, Marlin is a great miner that offers significant speed advantages over some of the others.
legendary
Activity: 3808
Merit: 1723
r9 290 gets 1147MH/s with 1050/1250

Most however just dual mine this coin with Claymore since its more profitable.
sr. member
Activity: 462
Merit: 250
Arianee:Smart-link Connecting Owners,Assets,Brands
it doesn't work for cuda 2.1 card,

http://prntscr.com/ckyrjx
sr. member
Activity: 413
Merit: 250
Lightning fast with both AMD and Nvidia GPUs.

Can we get some actual screenshots or examples how fast it is with a couple of cards?

I added the one card I have on hand (rx 480 while at work). I'll add more as time permits the testing again over the next few days.
legendary
Activity: 2002
Merit: 1051
ICO? Not even once.
Lightning fast with both AMD and Nvidia GPUs.

Can we get some actual screenshots or examples how fast it is with a couple of cards?
sr. member
Activity: 413
Merit: 250
Marlin - OpenCL/CUDA miner for SiaMining.com
==============================================

Up to date info and download links: https://siamining.com/marlin

Info:

Marlin is a new Stratum miner for OpenCL and CUDA devices developed by SiaMining. It is designed to be fast, lightweight and easy to use.

Notable features:

  • Native Stratum support with SiaMining.com.
  • Lightning fast with both AMD and Nvidia GPUs.
  • Multiple built-in kernels.
  • Kernel and intensity can be set separately for each device.
  • Detection of hardware errors.
  • Detailed share reporting, including difficulties and reject reasons.
  • Throttling support

Usage:

Code:
marlin -u YourPayoutAddress.WorkerName -I 28

  • By default, the miner connects to SiaMining.com's Stratum server.
  • By default, all OpenCL GPUs are used. To use a specific subset of OpenCL devices, use option -d with a comma-separated list of device IDs. The full list of IDs is printed every time the miner is started. Note that Nvidia GPUs should be listed twice, once as CUDA and once as OpenCL devices; the OpenCL interface is disabled by default for these GPUs, as CUDA usually gives better results.
  • The default intensity is 24. You can use option -I to adjust it. You can also specify a different intensity for each device by passing a comma-separated list.
  • You can use option -K to use a kernel other than the default one; like with intensity, you can specify a different kernel for each device by passing a comma-separated list. Run a benchmark with option --benchmark to see all available kernels for a given device.

To see all supported options:

Code:
marlin -h


Example performance:

RX 480 - 1185 core, 2000 mem: Marlin v1: 945MH/s vs SGMiner: 904MH/s GoMiner mod3: 845MH/s.
GTX 960 - 1303 core, 3004 mem: Marlin v1 m2p4: 657MH/s , CCMiner: 627MH/s
GTX 760 - 1162 core, 3004 mem: Marlin v1: 384MH/s, CCminer 263MH/s

Example screenshot of the 760:


Pages:
Jump to: