Bitcoin Forum

Ok. So I had a 5770 running in my PC running poclbm.exe using -d0 (Juniper). Was getting 150000 khash/s. CPU was naturally at 0% as I was using the GPU.

I've added a second 5770 to the same motherboard. I now have 2 x 5770 -

[ 0 ] Juniper
[ 1 ] Juniper
[ 2 ] AMD Athlon

The problem is that whenever I run poclbm.exe against either -d0 or -d1, my CPU is taxed 100% by poclbm ? I still get the 150000 khash/s on each card, but my CPU is now running at 100%. What's up with that?

Thanks.

Please give some more details about your OS, driver version, ATI Stream (APP) SDK version... Do you have crossfire?

I am also having this problem. 2x 5570's, non crossfire, amd system too

lastest version of the miner with 2.3 sdk

Hmm same here...haha looks like i posted on the main thread. But im running 5970 + 5850, 11.2 ccc and 2.2 stream with 25% load per poclbm instance. Win 7 64 bit here.

I am using streamsdk_2-3_win764

ATI Radeon HD 5700 Series. Driver Version: 8.821.0.0. Driver Date: 26/01/2011.

11-2_vista64_win7_64_dd_ccc_ocl.exe

W7 64.

Very strange.

If you have more than 1 video card detected, Crossfire or not, in windows you will have high cpu usage. Try to start it with only 1 card plugged.

I got this same problem after i added second HD 5850.

I have quad core i5 processor so when i start first instance CPU goes up to 25% and after starting second one it goes up to 50%
MH/s rates are always normal nevertheless.

Specs:

Win 7 64bit
Cats 11.2
AMD Stream 2.3
Dual 5850s
Miner: poclbm_py2exe_20110311

You all should probably read through a few of the other threads on this forum.

To get better performance, you should:

-Use Stream SDK 2.1.
-Use the display drivers from Catalyst 10-11.
-Turn off crossfire if possible.
-Use Linux. I've heard that people had problems running DiabloMiner in Ubuntu, so I might suggest Debian instead. There are guides for getting many miners up and running on different linux platforms if you're unfamiliar with how to use it. Look around on the forums to find them. Get ready to use the command line.

whats your gpu usage at?

Kinda hard to do that with Cayman gpu supporting only 11.1 and later. Under linux there isnt voltage controll to reduce power consumption.

Just thought id add im win7 64bit too.

Is the dud combination multiple ati 5xxx gpu's+win7 64bit?

Any one with 2 cards & windows 7 32 also have this problem?

I have HD6870 & getting around 270Mhash/s after OC of core to 1038 & mem to 375, lowering mem clk reduced heat from 74C to 66C.
I do have the poclbm taking some cpu percentage & most time i see 20% in task manager.
I have 11.2 & 2.3 SDK.
However yesterday i find the preview of ati next driver version 11.4(not 11.3) with SDK 2.4 & installed it.
I see 2 difference mainly.
1. Before its always 99% GPU in MSI afterburner(only a straight line), now it floats from 92 to 99.
2.poclbm uses now 0 cpu, even though i am running with/without  -f1.

So, i think its not bad to try new version even though u have 5XXX series card.
Also, i still dont understand why all saying to use -w128 , as 6870 actually max work is 256 &  placing -w256 increases 2-4Mhash/s.

U can download the drive from here.
http://support.amd.com/us/kbarticles/Pages/Catalyst114earlypreview.aspx

Also, instead of express, go for custom & select only APP, driver, catalyst..only what u want, many r useless to mining, hydra...
Today, itself without a single reboot , i upgraded, downgraded, upgraded, downgraded.....from 11.2 to 11.4 & from 2.4-2.3-2.2-2.4....
Just go fro custom & don't install useless. U can verify ur SDK & ATI version using gpu caps viewer.
Hope this helps.

thanks for that,
installed the new drivers & SDK and no change. Each gpu is still maxxing out a core when mining :(

I'm also on Win7 64bit and seeing the same issue. Newest Catalyst drivers and Stream SDK.

Must have something to do with the 64bit windows. IIRC, I tried this on 32bit win7 and didn't have such problems. Maybe someone else could confirm if they're not having this issue on 32bit win7.

Luckily I have 4 cores.. But the rig is still consuming almost 100w more than it should :--/

I am on windows 7 32 bit with Intel Quad core 2.66GHz Q8400.

I ran into the same problem, though it came to my attention in a different context (http://bitcointalk.org/index.php?topic=4648.msg68830#msg68830).  My system is now:

Ubuntu 10.10
Radeon HD 5870
AMD Sempron 140

I just  spent 2 hours trying to solve the issue by installing different versions of both the Catalyst drivers and the stream SDK- nothing helped.

Combinations I tried:
10.09 + 2.1 SDK
10.10 + the included OpenCL runtime
10.10 + 2.1 SDK
11.2 + 2.1 SDK
11.2 + 2.3 SDK
11.4rc2 + 2.3 SDK

.. and maybe some others. Every combination still used 100% of a CPU core for each GPU core.

Hope this saves some time from others. Any ideas on how to fix the issue are also welcome :)

Uninstall every ati driver & sdk using some uinstallers like revo uninstaller.
Also never install sdk from stand alone adk package, instead install from driver package.
If you have 5000 series card, install 10.12 & 2.1 sdk.
If you have 6000 series card, install 11.2 & 2.3 APP.

You haven't mentioned what card you have, what OS & how many Mhash/s you getting.
Instead of going express install, go for custom install & install only these 4.
ATI display driver, ATI catalyst install manager, APP 2.3 or SDK 2.1 & VC++

Thank you for the response. I tried this, but still no luck. I'm going to try some wild stuff, like removing some AMD specific drivers etc next.

Hardware specs:

AMD 880G (3x PCIE slots for GPUs)
AMD Phenom II x4 BE550
1800Mhz DDR III
1x Radeon HD 5970 (about to get another one soon)

and I'm running win7 64bit.

I'll post if I figure out something. I need to fix this before I get the other card or this will be 100% of all my CPU core :)

exactly the same issue here!

rig:
intel 430 (one core)
X38 + 1GB DDR3
5970 + 6970 (3 GPU cores total)
drivers 11.2 with SPP and 11.4 beta, no extra SDK drivers are loaded 
win7 x86

when i am running one poclbm for the first GPU (whatever that be) cpu is at 1-4%
with the second poclbm, cpu goes 99%!!!!
with the third poclbm, cpu is devided for the 2 last poclbms (48%+48%)
i think, perhaps a bug with poclbm program?

Are you guys all using -f 1 or some absurdly low -f?
Set -f to something higher and your cpu util will go down (though hashrate will as well)

No, that doesn't help yrral.

This isn't the issue.

Using -f1 with a single card will require very little cpu usage, 0-3%.

Adding the second card, regardless of what you put after -f, will result in much, much more cpu usage for each gpu mining. I get 25% per gpu on a quad.

i was testing the RPC miner and i see the same behavior. I wonder if the drivers + windows have an issue rather than the software!
anyone else trying other miners?
i was said that in linux the cpu goes down to 20-30%

This is most likely a driver side issue. From what I gather, bridging the cards into crossfire (or having a dual gpu card) will prevent cpu load, even though you are running several instances of your miner.

It's working exactly the other way.

I'm a bit confused.

In which situation does the extra cpu load occur?

a. Two single gpu cards - crossfire enabled (causes too much hashrate fluctuation to be worthwhile)
b. Two single gpu cards - crossfire disabled
c. One dual gpu card - crossfire enabled (can you even run a dual gpu card without crossfire?)

Or does it occur in all of these?

I haven't seen anyone complain about high cpu usage with a single, dual gpu card, but I could have missed it. If it doesn't, why?

I've experienced b. and I could test a. but that would require shutting down my miner and I'm reluctant to do that!  :D

It depends on the model. Any combination of HD 6900 will cause the 100% cpu usage, no matter about crossfire or not.
The only woraround is to disable crossfire and have only 1 card plugged to a monitor, start a miner, plug that monitor to the other card, start the miner and so on. Yep, it sucks big time.

Indeed, that is the work around for making the card active in windows. Although, I'm not sure what it has to do with the extra cpu load. Or are you suggesting that this will get rid of the extra load?   ???

I don't mind taking extra steps for windows to make a card active. I do mind the extra cpu load.

Also, does a 6990 cause the extra cpu load?

I'm using a 5970 in two systems, an AMD Phenom II X2 555 and an Intel i7 920.  Both running Stream SDK 2.2 and the intel avgs about 30% CPU load, the AMD is 100%.  This hasn't been solved yet?  Weird.

-EP

Dual core vs quad core with hyperthreading, each instance of poclbm should peg an entire core, so that sounds about right.

I've got the same problem
AMD + one 5850 consume 100% of CPU
Windows XP
 :(

with my i7 920, I get 38% load, went into task manager and there was an svchost that was taking 13% that I close, and I put affinity to CPU0 to both my miners and they share the 12/13% 6/6 or 7/5 or 7/6

So this hasn't been resolved in the slightest ? 

All my PC do that.

Some one please please please please try granola & see at least it reduces your CPU.
Please please try.

http://bitcointalk.org/index.php?topic=6188.0

please try

I only do things for 8 pleases. 7 pleases is far too few.

Tried it, no change

damn

I just want to chime in to report that I'm also getting 100% CPU usage with poclbm-gui with my 6990.  I'm using Catalyst 11.3 with SDK 2.4, and my CPU is an Athlon 64x2 4200+.  This is a pretty big problem, since 100% CPU is essentially just an additional cost to mining.

At least on linux you can try to export GPU_USE_SYNC_OBJECTS=1 environment variable.

That doesn't help for the problem with 100% CPU on multi GPU setups.

I know this is an ATI focused thread, but I'm seeing the same behavior on my nVidia cards. 

Two seperate systems.

1 - quad core 2.66ghz core2, GTX 260, drivers 270.61. OS Win7 64.
2 - Dual core 2.93ghz core2, GTX 460, drivers 270.61. OS WinXP 32.

On system 2, poclbm.exe hits 50% cpu (all of 1 core) and causes the temperatures to skyrocket when running only on GPU.
On system 1, poclbm.exe hits 25% cpu (again 1 core). 

Since I do not have multiple GPUs in each of these systems, I can't confirm if its the same problem, but considering the other symptoms marked here...

So, it seems that each instance of the miner is using the GPU, but also maxing out 1 core of the CPU per GPU client loaded.  This has been stated many times by m0mchil to not be proper behavior.  Seems it's not relegated just to the ATI cards any more.

I'm lucky to have an i7 920 with HT and 2 cards, so both together do 13% after affinity adjustment....

Has anyone found the fix yet?

im running 4 identical rigs. and on 2 of them im having this issue.

what gives.?

Same issue here, tried a bunch of stuff , nothing worked so far. Only thing I haven't tried are different miners.

Anyone else try any?

I guess that is what I will do, I'll come back with the results.

Apparently the miner doesn't make any difference. I have the same issue with both guiminer (essentially poclbm with a pretty gui on top) and phoenix.

GPU_USE_SYNC_OBJECTS works only for SDK2.3 and above. With older SDKs you are out of luck.

Any way to turn this variable on in Windows? Modify poclbm?

I'm having the same issue here.

Running:
ATI 4870 - Single Card w/ 1gig of ram.
Win XP 64bit
8 Gigs RAM
AMD Phenom II x4 940 CPU

Same problem here,
Got two 6990's so 4 gpu's and the each rape a core without me slapping affinity

my problematic situation:
I have on ATI 4670 11.3 & 2.3 SDK. But CPU100% (and GUP with guiminer on always 70°C, too much?).
i seen other posts about cpus but 100% still remain..  :-\
any solution?

thanks ;)

So far no matter what I tried in windows I cannot fix this problem. I will try linux mining over the weekend to see if that fixes it.

Guys give this a try:

https://www.bitcoin.org/smf/index.php?topic=7374.0

not saying that it fixes it ...just to try it.

there are ALOT of people in the world that dont know JACK about linux
Very honestly here is my Entire knowledge of Linux

Good for wireless cracking
almost completely open source
If your a noob GTFO
Uses some differnt HDD encryption thingy
and uhhh, Has a penguin, no windows key.
Lots of weird "Buntu" things

Literally that is Everything i know about Linux

I lol'd

I facepalmed :/

rofl!!

But seriously Linux and other open source operating systems (OpenBSD, FreeBSD, NetBSD) kick ass. I used to be all hardcore into linux from back in the days of the 2.0 kernel. I was a big slackware fan and was so into compiling all my programs. Along the way the evil M$ empire got me and I installed Windows XP on my main machine (I wanted to play games) and the rest as they say is history. I still use Linux as my server operating system and I refuse to use Windows for that purpose.

You gotta admit that Linux has come a long way - a VERY long way. I don't think M$ is shaking in their boots yet though.

For mining Linux is perfect. It uses minimal resources (can be trimmed to run with very small amounts of ram) and if you use a more user friendly distro like Ubuntu it can be very well suited to quickly putting together a mining system.

Long live Linux!!!

Hi!

About CPU usage - i just mounted 1 GPU on core 2 duo and it was 0% on poclbm. Installed another - both uses 1 core at 100% (consuming whole processor at 100%).

Is here any way to stop this? got 2 comps running win 7 (one 64 bit another 32 bit) and both have same problem with multiple GPU's

I would also like to share the same issue: 2 5850s Crossfire 16x and 4x. Two threads of poclbm uses around 25% of CPU each making the whole cpu 50% utilized. On my other mining rig a single 6850 and CPU is @ 12% utilization. I hope this get's fixed.

I heard changing GPU_USE_SYNC_OBJECTS  variable to 1 puts cpu's on 0-5% state while mining. Any idea how to change it? in ANY sdk, can be 2.3 or 2.4 even.

its an environment variable... I tried in Windows didn't help... seems to be only Linux

Same problem under Ubuntu 11.04.
AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
Two ATI cards.
GPU_USE_SYNC_OBJECTS=1
Latest poclbm

1 miner = 0-5% CPU
2 miners = 97-100% CPU

export GPU_USE_SYNC_OBJECTS=1

don't forget the export thing...otherwise it is useless.

How do you use the GPU_USE_SYNC_OBJECTS environment variables in windows, seems to be the key to the problem. This post http://forum.beyond3d.com/showpost.php?p=1506556&postcount=9 (http://forum.beyond3d.com/showpost.php?p=1506556&postcount=9) leads to this one http://forums.amd.com/forum/messageview.cfm?catid=390&threadid=143851 (http://forums.amd.com/forum/messageview.cfm?catid=390&threadid=143851) hinting that it fixes the 100% CPU problem. Anyone knows how to apply this in windows?

I was able to update my video drivers and it detected my video card. Now I'm cookin' with GPU.

Not to be an ass, but how is that relevant to the thread?

Because I was having the same issue and I updated my video drivers and it went away.

Umm, are you mining with more than one video card?

And to what ATI driver version did you upgrade to?

Can you show proof that it went away.

I'm experiencing this problem as well with the GUIMiner whenever I have more than 1 card on a system. Windows 7 64 bit in both cases.

Rig 1: Two 5870's - each miner instance takes up a CPU core. So if I try to set 2 miners on each GPU (one for a backup in case my main pool goes down) and have 4 total miners, I use up all 4 cores on my i5-2500k.

Rig 2: Same deal, but with 1 5870 and 1 5850.

Seems like it's a problem with the number of miners you have rather than the number of GPUs. I'm sure if I had a hexacore system and tried to make 6 total miners, I'd experience the same problem even though only 2 miners are actually "working"

So it's been 3 month and nobody's found a solution yet ?

With a single GPU poclbm use 30% on a C2D 2.4ghz,

I'm starting to think that there is no solution in Windows..at least nothing super easy.

In Linux use this:

export GPU_USE_SYNC_OBJECTS=1

The reason why I don't use linux is that I want the PC to be usable for something else.

Not that linux is unusable, just "too different" for people's taste.

Yep, I understand your reason. But unfortunately you'll have to put up with the CPU usage. Just set CPU affinity for each miner to one core.

This issue led me to switch to the Phoenix miner, which is getting me more hashes anyway, and 1% CPU usage.

huh? It doesn't matter what miner you use. For me Phoenix does the same thing! Am I the only one??

I am beginning to think this happen to every windows GPU setup,

There is definitely people who don't have this problem, isn't ?
Otherwise this page would have 30 page.

What is it you're doing ?

I have PCs running many GPUs with the crappiest 1.6ghz single core you can find and it does not affect hashing at all.

These cpu cycles are definitely doing nothing useful at all.

I have the same problem:

6990
AMD Sampron (one or dual cores - not sure), running at 100% all the time.

I think i have found a solution to this problem atleast on Win7, if you go into the advanced power settings in Win7 you can set the Maximum Processor State to say 40-50% and with my system
5770
5850
AMD Athlon II X3

Causes the CPU to remain at only around 40watts rather than 110watts as shown by HWmonitor program when mining with dual cards. I had to do some testing to see what was the optimal setting for me by adjusting the percentage until GUIminer was getting the highest MH/s but also before the CPU started to go above 40watts.

Hope that helps.

But then U are limiting something from doing it's job, no?

I mean -there is a reason the CPU is on a 100%, no? Or actually the fault is that it's doing NOTHING while running on 100%???

I am also seeing this.  I have 3 ATI 6950s and a 5970 with an i3 3.2Ghz CPU and each python instance maxes out a core at 100% (on Win 7).  [And one of my cards runs at about 2/3 max capacity].

Has anyone come up with a reliable solution for this?

My investigations shows that problem may be in driver version.

When I was using Ubuntu 10.10 and NVidia 260.19.36, there are no problems - poclbm was using only 0-5 %
Today I upgraded to Ubuntu 11.04 and fresh NVidia driver 270.41.19, and now poclbm (and phoenix too) are using up to 100% :(

When I install old NVidia driver 260.19.36 on Ubuntu 11.04, poclbm is digging and "top" shows low cpu usage (0-5%), but unfortunately X server cannot starting complaining about "ABI version mismatch". Now I am on Ubuntu 11.04 with fresh driver, and also seeking for better solution... Maybe I'll have to go back to Ubuntu 10.10.

EDIT: videocard is - nVidia Corporation GT215 [GeForce GT 240] (rev a2) ~20.000 khash/s

Hmm if that's the case does anyone know of a windows catalyst driver package version for win 7 64 bit that does not exhibit this annoying 100% cpu usage? I'm currently  using 10.11 (why you may ask, well it's cause it actually works faster than 11.5 with phoenix and the phatk kernel).

All miners can be shoved onto the same core. Through BIOS, AOD, or similar tool, you can downclock and downvolt individual cores in most modern setups. Shove all miners onto one core, downclock and downvolt that core as much as possible, and losses will be minimal. There is no effect on hash rate by doing this because (for the most part) the CPU usage does not seem related to the work the miner is doing. It does not solve the issue -- one core will still effectively be useless -- but I think this is preferable to the earlier suggestion of downclocking/volting all of the cores.

Edit: Meant to say entire CPU voltage can be lowered more after individually underclocking the core the miners use.

This.

I recently moved to Fedora 15 and upgraded drivers for my 8500GT.  poclbm was maxing at 100% CPU across two CPUs.

Threw the machine into run level 3, downloaded the nVidia driver mentioned by the user above (260.19.36), installed, reloaded the module - viola.  poclbm is taking < 1% CPU.

Very, very, very easy to recreate.  Use the versions mentioned above.

I am limiting it but it does not seem to effect my hashrate as long as its set at 57% or so even if i drop it down too something like 20% i do not see any reduction in my hashrate with my two cards, i set it at 57% as at 58% it starts to use about 70w rather than 40w when at 57% mostly so it quieter although the powersaving is also good :).
This is because i am still using my computer for search the web watching youtube etc when the two gfx cards are running and i the computer becomes sluggish if it was set real low but does not seem to effect the hashrate just the responsiveness of my system in general.

My rig has a one core CPU, so that doesn't help.

CPU at 100% from poclbm does not affect hash rate but CPU at 100% from anything else almost stop it.

I can't use the afterburner UI when poclbm use 100%

Yeah that sucks. For you the only option is to go the linux route or to get a cpu with dual cores (AMD Athlon II X2 250 may be a good one)

So this have not been solve for ATI yet huh... Man. And I'm having problems for my windows to install correctly. damnit.

It doesn't seem like it should be that hard to debug python the to find where the cpu peg is happening.  Even if it's not poclbm's fault it would still be good to know on which opencl calls this happens.  I don't know opencl or bitcoin otherwise I'd dive in.

Sounds like the perfect opportunity to learn something new, eh? ;)

I realized that if I don't overclock the GPU (meaning I disable ATI Overdrive) then poclbm works normally. Idk if this is universal, could someone verify if this is repeatable (forgot the right word for bugs). If it is we might be getting closer to solving this problem.

Umm, that sounds odd. I don't overclock at all but I do have ATI Overdrive enabled just to set the fan profile manually, yet I still have high cpu usage. I also have the memory underclocked but not via ATI overdrive but MSI afterburner. Hmm, I guess this weekend I'm going to have to try disabling overdrive and see if this works. But maybe someone can beat me to it and report back their findings here?

IMHO this problem is caused by the OpenCL implementation in the driver, I am testing the RPC miner (CUDA) (http://forum.bitcoin.org/?topic=2444.0) under Windows 7 and it works fine, with very small CPU usage (< 1%) and 80% of the poclbm efficiency (43 vs 55 mhash/s).

I have tested also the OpenCL version of the RPC miner, and the tests were inconclusive, it used 100% of one of the cores for a few minutes and repeatedly generated 0 khash/s, then crashed because of the memory allocation error.

I have GTX 275 and C2Q 9450 @ 2,66 GHz.

Well dammit why doesn't AMD fix this sh*t! Is it really so hard to at the very least allow us to turn off the cpu spin state flag off? I'm thinking a simple .ini or config file option would be all that is needed.

I can confirm this problem too, new drivers 270... makes my cpu work more for nothing! I've downgraded to 260 and now it's fine. Thank you ;) and about your ubuntu problem you should rmmod and modprobe your nvidia kernel module before restarting xorg, or reboot. It should fix your error.


I've got a GTX 285 and the same CPU and got 60 Mhs on linux with poclbm (at max without Xorg). Do you think CUDA could change dramatically the performance? Something like > 65 Mhs?
EDIT: I've tested puddinpop's miner with CUDA and I'm getting less (45-55 Mhs and it has frozen my kernel!) than poclbm OpenCL at "low rate" (-f at 150 I get 56 Mhs) which allows me to use my computer without refreshing latency at all (if I whish I can set -f0 and mine at 60 Mhs without Xorg! How can it be better?).

Check this link, It MUST solve CPU usage problem.

http://forum.bitcoin.org/index.php?topic=6188.msg168009#msg168009

I already use 10.9 on two of my dedicated miners and it does not solve the high cpu usage problem. So just running 10 series does not work on windows. I'll look into that granola program. I hope though that I can trust it and it won't be some spyware/malware scam.

does the phoenix miner exhibit this same problem ?

Yes, for at least some of us.

Guys this is not a miner issue..it's just the opencl drivers...

Guys...
Just set the CPU affinity...

This question gets asked a lot. Here's how to alleviate it: http://forum.bitcoin.org/index.php?topic=12292.0

Yes, that's what I was saying. I know how to do it...

I got mine to work after having this problem...maybe this will work for you guys as well. Start computer with main graphics card plugged in AND secondary graphics card plugged in. Launch miner with device set to 0, then unplug primary graphics card from monitor, switch to secondary graphics card, then run again with device set to 0. For me, it starts out with one of them using 25% CPU then it drops back down to 0% after I launch the second miner... I have no idea why this works, but this is how I fixed the problem.

I am running a W7 rig with 6950x2 and I'm getting the same problem. Either it's GUIMiner's poclbm sucking 100% of CPU (divided equally between all instances) or a single instance of Java.exe for 100% CPU when using DiabloMiner.

Currently, I am using DiabloMiner, because there's only one process (java.exe) for me to set the affinity to a single core, instead of multiple instances of GUIMiner's, and I guess that's the only viable solution for now - trading off 25% for saving 75% on a quad-core CPU. :/

It's the OpenCL drivers! :) NOT any miner application..

Please all e-mail AMD to fix it :)

hey just adding that I'm also experiencing full CPU usage.  I'm using guiminer with CPU affinity *off*.  I have my aggression right now set at 9.  My poor fans are going full bore :)
 
Win7 64bit.  2 HD5770s.
 
Anxiously awaiting someone to figure out something :)

Easy way to alleviate the problem, not solve it. Set an affinity for each instance of poclbm.exe you find in task manager.

Or get AMD to fix this issue lol

seems always related to 2+ video cards?

This happens on NVidia cards as well (GeForce GT 330M). I don't think AMD is the cause.

Hardware:
Via C7 1800Mhz http://www.biostar.com.tw/app/en/mb/content.php?S_ID=470
2GB RAM
ATI 4650 / 5850

Software:
Debian Squeeze ( ATI 10.9)
m0mchil's miner
SDK 2.2

I startet building this miner yesterday with the 4650. Didn't notice the high CPU-load until I hit htop, but the machine was still easily operated.

Then I borrowed a 5850 from a friend. CPU-load is still 100%, but now the commandline is getting a little sluggish.

I wouldn't bother about it, because this machine is a dedicated miner, but I think it would save some energy to get the CPU-load down...

I tested with dual 6970 and 6950 and both use full CPU .. AMD x250. so I downgraded teh processor poer to 800 mhz instead of 3200mhz. not using affinity as it sometimes hangs.


I guess it is related to AMD. And using ufasoft to mine using the CPU is yet not a good solution.


I am getting around 400 Mh/s from 6970, 350 mh/s from new 6950 with single fan . and 320 mh/s from old 6950 with dual fan.


not much of a problem here with low power prices but yet annoying!

but what about the people who have nvidia who have this issue?

That is a bandaid but not a fix. For a two core CPU having one running 100% 24/7 is quite a load on both power and resources.

The only solution at the moment seems to be switching to Linux, but that is either undesirable or impossible for many without dedicated rigs.

I have this issue as well =( its very annoying. I have tried all sorts of different miners w/ my new 6950 and it still has the one core usage on my single card!

I have tried the preview 11.7 drivers which are said to fix this issue. Is it possible to individually try different OpenCL packages w/ different driver versions till find one that does not have this issue? E.G. use the 10.x drivers OpenCL (which is said to work correctly) w/ newer drivers?

On my end, I never had this problem, until I upgraded to Catalyst 11.7. Now I have one core maxed out nonstop in Guiminer - using any kernel, poclbm, phatk, phoenix, anything. Aside from the power drain, it also hits the performance (so it's not just burning idle threads, but it's stuck in an infinite loop).

It's a bitch alright. It would be nice if switching to Linux full time wasn't the only solution.

Not a bad idea. If we all threw a quarter/half BTC in that could probably amount to a nice sum.

http://i55.tinypic.com/1p8q5v.jpg

Windows XP 64-bit / 11.6 and 2.5 / -f 10

Motherboard is a Gigabyte 890GPA-UD3H and I use the IGP as the primary display so I can game or watch videos without problems. Tried -f0 with both cards but the IGP seems to be affected, mouse lags and jumps sometimes when moving the pointer across the screen.

I haven't encountered this 100% cpu utilization problem you guys are talking about even with 11.2 and 11.4 and previous versions of poclbm.

You got XP!,,, ohhh let me try.

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

So this is a Win7 bug only?

ive read of it happening with other's though my experience is just with 7, but i have xp which i should test.

I have got alot less problems after setting cpu affinity for one core and also setting priority to low. One day and no freezes now, still stuttery though I have to be slow and gentle with opening windows.

Maybe it has to do with how vista/seven manage the cards error.

On XP you got a blue screen, on seven you get a reset/notification.
These GPU errors must be caught by the OS/CPU somehow.

Your thoughts ?

I noticed I do not get any CPU usage running poclbm when on a 512mb ram PC running a Win7 64bit

https://bitcointalk.org/index.php?topic=36134.msg445411#msg445411

Interesting. I'll try fixing the problem with that in mind later. Being a bit of a necro here - so has there been any progress on this while I've neglected this thread?

There is some hope that AMD has resolved the issue (at least somewhat) with the new drivers

https://bitcointalk.org/index.php?topic=43604.0

Noob.

Were installing different versions of the ATi STREAM sdk technology, We grab the sdk, because it has Everything were ever going to need, And the simplist way to know What version of ATi Stream tech. Is by picking one of it's SDK's

Because older catalysts doesn't have OpenCL drivers included in the installer you download from AMD. You have to download an SDK to have OpenCL support.

WELLLL if you'd still wanna know, Some """"""sdk's""""" give better hashrates than others

Now that answer is not entire correct i guess, since in the boinc community we used much older versions like 10.11, 10.12 and when you install those opencl do not work, you have to install sdk 2.1 to get stream and opencl working.
So i guess that has been passed on to all following versions even though i think from 11.2 the opencl driver should be working with the standard install.
These old version do not have the cpu bug at all, but i found cgminer not able to run with this driver.
Installing sdk 2.5 did not solve the problem, so i had to install one of the bugged ones to get the card running with cgminer again.
I have not tried any of the other miners because for me the cgminer is the best ever made.
Especially since phoenix allways seem to have aproblem connecting to the pools on my machines and all others are much slower on my cards.
And most important cgminer has so darn many nice options which can be set :D, i simply love it.
Till today non of the opencl group have tried to solve this opencl problem as far as i know.
And yes JackRabitt the fastest is still 2.1 in the boinc community but ofcourse program makers do incorporate new things from the newer sdk's so you end up with not good working programs. But i can promise you that sometimes you see fake results as well.
I had installed 11.10rev3 and i suddenly had 104 Mh/s on a 4770 which in fact does around 65 Mh/s so it simply was showing bugged numbers
To answer on other people mention sdk 2.2, 2.3 and 2.4 as being good, well lol i have bad experience with them on windows and no i do not have a clue about linux i hardly use it anymore ;)

Nope its present on every windows version no matter which, but you might wanna test win2000 ;). Anyway its a opencl bug as far as i know and only happens normally on multi gpu setups.
The reason why on some machines they have reported similar situations must be some other bug, since on one i got it fixed by setting the screen to 1 screen only, that machine has 2 monitors installed on one videocard.

I am having the 100%+ usage issue on my system.

It's different because I am running a Intel C2D E6600 with a nVidia 460GTX, I am running Win7 64bit nVidia Drivers 314.07.  Oh and I have a nVidia 680i SLI mb running the latest MB (15 something) drivers.

When GUIMiner is running using poclbm 140% of my CPU is used (1.5 cores used) even though I set it to not do CPU mining.  Even when I turn on CPU mining nothing changes.

Anybody have any clues or ideas on this?

Is nvidia mining bitcoins?
Have you able to mine & got any coins so far?

I am via Slush's Pool.

It's slow at only 70mhash/s though.