i have started a thread at amd dev forum - opencl section about the 100% cpu usage bug with >1 gpu installed. feel free to jump over there and post your support for this issue being addressed...
http://forums.amd.com/devforum/messageview.cfm?catid=390&threadid=153211&enterthread=y
Topic: CPU usage 100%, any way to diable it at all? (Read 3479 times)
You didn't mention which miner you were using.
I saw this issue with poclbm, and so I switched to phoenix/phatk. No more 100% CPU usage.
I saw this issue with poclbm, and so I switched to phoenix/phatk. No more 100% CPU usage.
No waiiii
GUIminer2011-07-01-All Kernals tried
You've gotta be on linux
Meng.
In windows, no. Multi-GPU/CPU mining will load at least 1 core to 100%
In Linux, yes. OCL doesn't have the cpu usage bug in linux. I get higher oc's in Linux as well. YMMV.
In Linux, yes. OCL doesn't have the cpu usage bug in linux. I get higher oc's in Linux as well. YMMV.
so ive read alot and saw the new settings of Affinity to cores.
But my question is, is there a way to get it down to these below 10% cpu loads?
But my question is, is there a way to get it down to these below 10% cpu loads?
if your talking about gpu mining, and cpu usage during that try -f10 or number higher number till it reduces cpu usage to needs.
if i run less than 10 i get strong usage an core 0 making temps higher and fans loader, fine for garage, but not for my caravan.
i use guiminer poclbm and get next to no usage on any core
You didn't mention which miner you were using.
I saw this issue with poclbm, and so I switched to phoenix/phatk. No more 100% CPU usage.
I saw this issue with poclbm, and so I switched to phoenix/phatk. No more 100% CPU usage.
100% for me with any miner. with/out phatk as well.
You didn't mention which miner you were using.
I saw this issue with poclbm, and so I switched to phoenix/phatk. No more 100% CPU usage.
I saw this issue with poclbm, and so I switched to phoenix/phatk. No more 100% CPU usage.
on my system regardless of driver/opencl verzion, i get 100% cpu usage when using >1 card. right now im running openCL 2.1, 4cards and 100% cpu.
this is the strangest shits ever to be honest. in windows you can manually throttle the CPU under the power settings. this does seem to lower my mhash, even with openCL2.1. does use less electric though.
this is the strangest shits ever to be honest. in windows you can manually throttle the CPU under the power settings. this does seem to lower my mhash, even with openCL2.1. does use less electric though.
That is normal.
The bug is in OpenCL, Once you add the 2nd gpu your cpu is gonna peg at 100% for whatever fucked up reason.
If you have a single core proc then your SOL and i Harshly do not reccomend using Windows as your Limiter software.
If you've gotta Dual then loadup AOD and Nerf the second core down to like 1ghz and put all the miners on it, You will not notice a speed decrese
If your running Intel then i've got nothing to say, I dont use intels.
3.4ghz quad Nerfed to 2.0x3 and 1ghz x1, Threw all gpu's on the 1ghz 4th,No problems.
Anyone tried older drivers (2010) and stream 2.1?
on my system regardless of driver/opencl verzion, i get 100% cpu usage when using >1 card. right now im running openCL 2.1, 4cards and 100% cpu.
this is the strangest shits ever to be honest. in windows you can manually throttle the CPU under the power settings. this does seem to lower my mhash, even with openCL2.1. does use less electric though.
this is the strangest shits ever to be honest. in windows you can manually throttle the CPU under the power settings. this does seem to lower my mhash, even with openCL2.1. does use less electric though.
Ah yeah, Forgot to mention that the miners are all running on the 6th core. Whups
(this works great for gaming too considering most games dont have 6core support)
(this works great for gaming too considering most games dont have 6core support)
Well, Some people (if heat is your only issue) reduce thier cpu's core. One core to be specific.
Heres what i've done.
On my hexacore i have
4.0ghz
4.0ghz
4.0ghz
4.0ghz
2.0ghz
2.0ghz
And i tell All my gpu's to run on one of the weaker cores. This saves me alot of heat
But then again. Im not running a 'simple' computer
So what SOME people try is using a from of "energy saving" mode designed to reduce heat among other things. However when doing this it usually slows your pci-e cards down Alot.
another idea is to underclock the Entire cpu. Heat problems fixed. But with Obvious deficits
Heres what i've done.
On my hexacore i have
4.0ghz
4.0ghz
4.0ghz
4.0ghz
2.0ghz
2.0ghz
And i tell All my gpu's to run on one of the weaker cores. This saves me alot of heat
But then again. Im not running a 'simple' computer
So what SOME people try is using a from of "energy saving" mode designed to reduce heat among other things. However when doing this it usually slows your pci-e cards down Alot.
another idea is to underclock the Entire cpu. Heat problems fixed. But with Obvious deficits
On my hexacore we use affinity so that the mining program only uses one core (number 6 @100%) then the rest run at 0% load.
If you used dummy plugs, it can cause cpu usage to sit at 50%.
Wasn't there some kind of Catalyst / SDK combination that worked well without CPU usage or was I getting something wrong? I heard of 10.12 + 2.1 somewhere, anyone using this?
Well, Some people (if heat is your only issue) reduce thier cpu's core. One core to be specific.
Heres what i've done.
On my hexacore i have
4.0ghz
4.0ghz
4.0ghz
4.0ghz
2.0ghz
2.0ghz
And i tell All my gpu's to run on one of the weaker cores. This saves me alot of heat
But then again. Im not running a 'simple' computer
So what SOME people try is using a from of "energy saving" mode designed to reduce heat among other things. However when doing this it usually slows your pci-e cards down Alot.
another idea is to underclock the Entire cpu. Heat problems fixed. But with Obvious deficits
Heres what i've done.
On my hexacore i have
4.0ghz
4.0ghz
4.0ghz
4.0ghz
2.0ghz
2.0ghz
And i tell All my gpu's to run on one of the weaker cores. This saves me alot of heat
But then again. Im not running a 'simple' computer
So what SOME people try is using a from of "energy saving" mode designed to reduce heat among other things. However when doing this it usually slows your pci-e cards down Alot.
another idea is to underclock the Entire cpu. Heat problems fixed. But with Obvious deficits
This was my problem entirely, If I could limit to 60% or so I could get good temps and fold on my laptop CPU and GPU, and get like 2x the Mhash I get now, but alas, it doesnt work
so ive read alot and saw the new settings of Affinity to cores.
But my question is, is there a way to get it down to these below 10% cpu loads?
But my question is, is there a way to get it down to these below 10% cpu loads?
No, the problem is a combination of sdk/drivers that leads to 1 core 100% usage for each miner instance.
Atm no solution has been found.
so ive read alot and saw the new settings of Affinity to cores.
But my question is, is there a way to get it down to these below 10% cpu loads?
But my question is, is there a way to get it down to these below 10% cpu loads?
Jump to: