06-20-2010, 06:41 AM
Then you seem to have the same problem as me! But I haven't installed ETQW to test but I guess that it's the same as in HoN.
Originally Posted by Muad'Dib
06-20-2010, 06:48 AM
Is anyone else having problems with power saving features on Catalyst 10.6? I had to revert back to 10.5 because with 10.6 the clock speeds were throttled up almost constantly - even if idling on desktop.
The difference is just a couple of degrees, but in my case it's a question of running totally passive versus active cooling, which is quite an audible difference.
06-20-2010, 08:17 AM
Originally Posted by Minigun
yeah, it's not much - but as temperatures outside (and inside) are getting higher it's quite noticable (and sometimes also annoying)
06-20-2010, 09:01 AM
We now have quite some people around here, that have problems getting their Radeon cards working on that particular nForce 430 chipset(MCP51 variant).
It's not just that they can't use some application, no, their graphical interface just fails to boot, thus they can't use their machines in any usable way.
In this thread we have
Phoen1x, j0v4nni and probabaly guilio.
You might want to add dartmouth, whose thread is here:
He tried his way using slackware and Ubuntu.
There's guria on the Ubuntu Bugzilla:
There's sling-shot who tried his luck with PCLinuxOS
Lastly there's my own bug-report at
which contains detailed Information about everything I tried
with catalysts ranging from 10.3 to 10.6 on Ubuntu 9.10(32bit/64bit) as well as Gentoo.
My bug is known in the internal AMD database by the ticketnumbers
8200313953 and 8200313951.
A few of the people I've mentioned are using Radeon 4000 and older cards and they seem to have been able to work around their problems with sticking to catalyst version 9.8. That's some sort of workaround, but of course it is not satisfactory, seeing the progress of the AMD driver feature-wise. This also indicates that AMD has introduced some kind of regression into its driver at that point of time.
As Radeon 5000 series cards are only supported from 9.10 (I believe) onwards that workaround is not applicable to users(like me) of Evergreen cards.
I think if you look through the informations we have provided you
(I tried to collect everything I know so far in my bugreport:
http://ati.cchtml.com/show_bug.cgi?id=1794) AMD should be able to locate and fix this bug.
In case, you, or the AMD developers should miss any information I will gladly provide that information, and I think most people I mentioned here will do so too.
In case you ask I have the most recent BIOS for my motherboard
and my /proc/mtrr output looks like this:
(...)@wthp094 ~ $ cat /proc/mtrr
reg00: base=0x000000000 ( 0MB), size= 1024MB, count=1: write-back
I have currently one DIMM with one GB of memory installed.
Thanks for your attention, and let's hope that 10.7 has one bug less.
06-20-2010, 09:07 AM
Well even in best case it would take 3 months for a bigger bug
06-20-2010, 09:10 AM
Well, that means I have 3 months for getting my OpenCL code working on it.
Originally Posted by Kano
Depending on that outcome we will decide whether it is cost efficient to stick with CPU's or if we have code paths that profit from GPUs.
06-20-2010, 09:14 AM
I forgot to mention doobiest and Syon from the Ubuntuforums:
Syon could get the card working in his box with Windows 7, but not on Ubuntu.
So it seems not BIOS related.
06-20-2010, 10:47 AM
You were right and it didn't solve the problem for me.
Originally Posted by phoen1x
The weirdest thing is that everything worked fine prior to catalyst 9.9. I didn't change any hardware/firmware/etc. just the fglrx driver and X just doesn't work.
I can also add that a long time ago I even completed Linux Crew Survey reporting the bug but no one contacted me with any feedback.
06-20-2010, 11:46 AM
The Catalyst Crew Survey is one-way, I'm pretty sure the form says there will be no response. Bugs should be filed at ati.cchtml.com - there are a couple there already so probably better to update one of those with current information, and probably a bit of cleanup would help.
Ticket #1794 has most of the good information but neither the title nor any of the comments indicate that the problem is specific to the NForce430; ticket #1700 explicitly mentions the NForce chipset but all the information is in a ticket on another system which has been closed.
06-20-2010, 12:18 PM
Well, you're right, the title doesn't mention the nForce4 chipset - I've fixed that now.
Originally Posted by bridgman
But I don't understand the rest of your response. I write in the second sentence of the first post of that Bug report that I have an nForce4 chipset. As there are probably more of them out there, I thought it is more convenient to provide PCI IDs. I provide all of them, which I think are relevant
(simple lspci output) in the first post.
That the cause of the hang is most likely due to the chipset can be proven by looking at the lspci output of the people around here. They share the same symptoms AND the same chipset down to the PCI ids. Therefore I think that this is the cause of all evil...
I don't think that ticket #1700 is relevant because the reporter gets explicit messages from the kernel that his BIOS is providing faulty MTRR ranges. This is, at least for me, not the case here.
So thanks for your response and your help in getting the report in shape so that we and AMD can nail that issue down and fix it.