I'm running 3&3 3 MWay's & 3 Regular Wu's and getting 85,000-90,000 Per Day from 1 Card (4870 512mb with no Overclocking) ...

That's crazy. :P It puts my 1 pc to shame. In 2 days you can put out more than my overall total, I did ~50k all of last year. I need a Gpu...(sigh).

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

PS:
I still don't get why people raise the number of concurrent WUs. Two are really enough for maximum efficiency. So even taking the start/finishing of WUs into account, you will never need more than 3 WUs (default value) for maximum efficiency of the GPU. It would be better for the throughput, if you lower the wait factor closer to 1. Or are you just trying to avoid the eviction of WUs by the boinc manager?

I don't understand the settings in the BOINC parameters, something to get into at some time, but I may have a tip that may work for others. I'm trying to get MW to run with another task, eg, CPDN, but usually 4 MW are running on their own. Not efficient as you pointed out Cluster Physik.

So what I do is put MW on suspend, two CPDN tasks run (I'm hyper-threading). Then I resume the CPDN project. 4 MW tasks then 'freeze'. I then suspend one of the CPDN tasks and have two MW and one CPDN crunching away nicely.

When things don't look right, exit the BOINC manager and restart it - that usually sorts it out for me.

---

I'm running 3&3 3 MWay's & 3 Regular Wu's and getting 85,000-90,000 Per Day from 1 Card (4870 512mb with no Overclocking) ...

That's crazy. :P It puts my 1 pc to shame. In 2 days you can put out more than my overall total, I did ~50k all of last year. I need a Gpu...(sigh).

It's still about contributing what you can/want. Even with GPU, both ATI and nVidia, I'm still crunching the old fashioned way with actual computers and laptop ;P

---

This behaviour should go away, if you disable the "leave aplications in memory" option. That way, the paused WUs are really stopped and not just suspended.

I set this option to disable, update project and try to suspend.
Then i suspend abc, application remove from memory.
But if suspend MW, all aplication leave in memory and not remove.

Brickhead, what your saying is that your only running on one side of the x2 Video Card, but how many WU's are you running @ once with it by it's self or running another Project with it ???

Doesn't matter. I've tried anything up to four WUs at once. GPU0 gets all the load it needs, and GPU1 remains idle. No surprise, given that multi-GPU functionality isn't built into 0.19 or 0.19b.

I am however testing a new development version which supports multiple GPUs. Not quite the doctor's order, as the combined load still doesn't exceed what would fit in one GPU. So no practical improvement for X2 cards or crossfire yet AFAIK, but still an important step forward. I'm confident that our mastermind Cluster Physik will eventually be victorious in the battle against cold GPUs :)

I do run four other projects in addition to MW on this quad-core dual-GPU machine, and by assigning resources 400 (MW) : 4 x 100 (others), I've managed to get 4 MW WUs running alongside 4 from the other projects. I had to set avg_ncpus very low (value * number_of_GPU_WUs_at_once < 1) to tell the BOINC CC that there is no need to kick any of the 4 "foreign" WUs to accomodate 4 MW WUs at the same time, in order to have all eight *running* simultaneously.

---

Okay, was just curious, so far I've just stuck with the single core cards but when dual core is enabled I may get one ... Thanks

I tried to get ATI Card to run the Milkyway WU's on the Linux side of a Dual Boot Windows/Linux Box but so far no luck. I installed the Linux 8.12 Drivers or at least I thought I did because everything seemed to go alright & the ATI Counsel showed up. I did it from the Terminal because when I tried to install them from a Counsel I got a error message saying something about a wrong ID.

Anyway the Milkyway WU's run but Dog Slow at best, about 30 min's Per WU so I'm assuming something isn't right & their not running on the ATI Card but the CPU side instead ... So I booted back into Windows until I can get it figured out, any ideas on whats wrong would be helpful ...

I tried to get ATI Card to run the Milkyway WU's on the Linux side of a Dual Boot Windows/Linux Box but so far no luck. I installed the Linux 8.12 Drivers or at least I thought I did because everything seemed to go alright & the ATI Counsel showed up. I did it from the Terminal because when I tried to install them from a Counsel I got a error message saying something about a wrong ID.

Anyway the Milkyway WU's run but Dog Slow at best, about 30 min's Per WU so I'm assuming something isn't right & their not running on the ATI Card but the CPU side instead ... So I booted back into Windows until I can get it figured out, any ideas on whats wrong would be helpful ...

If you take a look at zslip you will see that the ATI cards and the optimization files made available for MilkyWay are only available to Windows 32 and 64 bit. No mention of it working for Linux, and none that I have heard about.

---

I could have sworn I seen a post by somebody saying they got some app to work in Linux but could be mistaken, anywho I didn't pay attention when I downloaded the Linux App, just thought it was for running the ATI Cards ... :)

I could have sworn I seen a post by somebody saying they got some app to work in Linux but could be mistaken, anywho I didn't pay attention when I downloaded the Linux App, just thought it was for running the ATI Cards ... :)

There are certainly optimized apps for Linux by speedimic, but I don't think he's gotten round to ATI GPU, but I don't have Linux myself and I really don't know.

---

I could have sworn I seen a post by somebody saying they got some app to work in Linux but could be mistaken, anywho I didn't pay attention when I downloaded the Linux App, just thought it was for running the ATI Cards ... :)

There are certainly optimized apps for Linux by speedimic, but I don't think he's gotten round to ATI GPU, but I don't have Linux myself and I really don't know.

LOL, haven't even gotten round getting me a ati/nvidia card... only low-budget-onboard stuff here.

---

mic.

LOL, haven't even gotten round getting me a ati/nvidia card...

The development of apps for GPU is really quite recent. If I recall, it is less than 6 weeks or so when there was talk about ATI being faster than nvidia, and someone ought to develop an app based on the ATI cards...

Come on speedimic, you're missing the fun :P

---

Come on speedimic, you're missing the fun :P

ATM I'm missing the money...
Just got my '83 BMW M3 out of winter sleep - gotta invest in fuel now. ;-)

---

mic.

Come on speedimic, you're missing the fun :P

ATM I'm missing the money...
Just got my '83 BMW M3 out of winter sleep - gotta invest in fuel now. ;-)

Awww, and there's me forgoing the lens I've been lusting for to buy plastic bricks that can't even autofocus :~p

---

Come on speedimic, you're missing the fun :P

ATM I'm missing the money...
Just got my '83 BMW M3 out of winter sleep - gotta invest in fuel now. ;-)

I know what you mean...fuel (& fun) or everything else.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

I have been running v .17 is there any benefit in spending the time to upgrade to 19 or 19b -

I also do not see a download on zslip for anything more recent that 19b... isn't there a 19d ?

---

.

I have been running v .17 is there any benefit in spending the time to upgrade to 19 or 19b -

I also do not see a download on zslip for anything more recent that 19b... isn't there a 19d ?

0.19d is still under develpment/test for multi-GPUs. I can't say about 0.17 v 0.19/b

---

I have been running v .17 is there any benefit in spending the time to upgrade to 19 or 19b -

I also do not see a download on zslip for anything more recent that 19b... isn't there a 19d ?

The changes to 0.17 is only the CPU (0.19) and GPU detection (0.19b) output to stderr.txt visible under the task details. The calculation stuff is unchanged.

Version 0.19d does not use a full core anymore (down to 10% or even less) and was supposed to add some experimental multi GPU support. But the latter isn't really working, yet.

Version 0.19d does not use a full core anymore (down to 10% or even less) and was supposed to add some experimental multi GPU support. But the latter isn't really working, yet.

In that seems to be taking a little longer than expected ... any chance of a 19d without it so we can get the lower CPU use?

Version 0.19d does not use a full core anymore (down to 10% or even less) and was supposed to add some experimental multi GPU support. But the latter isn't really working, yet.

In that seems to be taking a little longer than expected ... any chance of a 19d without it so we can get the lower CPU use?

I'm on a conference this week. I can't do much here. But 0.19d is floating around already (it is usable, even on a multi GPU setup, it just doesn't use multiple GPUs to the full extent). So if Ice wants he can put it on zslip. He has it already.

Alternatively, you can use this link. Be sure to read the included readme.txt!