Open GoogleCodeExporter opened 8 years ago
I'll handle this one.
Original comment by sven.wit...@gmail.com
on 13 Oct 2011 at 7:29
Hi,
thanks for the feedback and the screenshot. That is always helpful. You can
also click on "Log now" and cut and paste from the opened text, if you like
that more than uploading a picture. Could you just do that and paste the output
here. It is possible, that there are more than 3 PStates in newer APU's, which
I just don't know yet.
I've got two questions:
1. What APU is built into your system?
2. Did you use a previous version of BrazosTweaker, where this didn't happen?
Regards,
Sven
Original comment by sven.wit...@gmail.com
on 13 Oct 2011 at 7:37
Hi Sven,
Thanks for making this program. I've got the AMD C-60 APU. I never tried a
previous version with this APU, but I'll download one and let you know.
While trying to get the Log Now output, I realized this only occurs if I am
running with the "Power Saver" profile. Hope that helps. Here's the output
after making a small change to P2 voltage:
Family: 14h
Bit numbering 63 59 55 51 47 43 39 35 31 27 23 19 15
11 7 3 0
COFVID 0071 0000 0000 0011 0000 0000 0001 0001 0010 0110 1100 0000 0011 0101
0110 0101 0010
P-State0 0064 1000 0000 0000 0000 0000 0001 0001 1100 0000 0000 0000 0000 0100
0100 0010 0000
P-State0 0065 1000 0000 0000 0000 0000 0010 1011 1001 0000 0000 0000 0000 0101
0000 0011 0000
P-State0 0066 1000 0000 0000 0000 0000 0010 1001 1000 0000 0000 0000 0000 0101
1000 0100 0000
P-State0 0067 0000 0000 0000 0000 0000 0010 1000 0111 0000 0000 0000 0000 0101
0110 0101 0010
P-State0 0068 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 0069 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006A 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006B 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
Bit numbering 31 27 23 19 15 11 7 3 0
NB P-State0 D18F3xDC 1001 1010 0100 1011 0110 0011 0010 1011
NB P-State1 D18F6x90 1000 0000 0000 0000 0011 1000 0011 0000
ClockTiming D18F3xD4 0000 0000 0000 0010 0100 1111 0101 1000
BIOSClock D0F0xE4_x0130_80F1 0000 0000 0000 0000 0000 0000 0110 0100
D18F3x15C VidL0 0.85V VidL1 0.85V VidL2 0.875V VidL3 0.9V
D0 00 D1F0 90 SMBus A0 D18 C0
1510 1022 h
4397 1002 h
4385 1002 h
1700 1022 h
MSRC001_0061 P-StateMaxPState: 2 MinPState: 0
BIOS vendor BIOS version MoBo vendor MoBo name
Acer
V1.04
Unknown
JE10-BZ
Original comment by jdub2...@gmail.com
on 16 Oct 2011 at 5:04
Disregard what I said about it only happening in the Power Saver profile.
Happens with any profile, as long as the max frequency is below 100%
Original comment by jdub2...@gmail.com
on 17 Oct 2011 at 12:23
Thanks for the input. Can you please do the following:
1. Uncheck the "make permanent" box in the service and switch off the system
(don't restart).
2. Switch on the laptop
3. Start BrazosTweaker
4. Do another "Log Now" and paste it here. Save the log somewhere, since
another click is going to overwrite it.
5. Do some tuning and click "Log now" again (after having copied the first
one), if the P3 settings are coming from BrazosTweaker.
I'd like to know, where the entries from Pstate P3 are coming from. That looks
strange to me. With your help, I might find out and can help.
Thanks,
Sven
Original comment by sven.wit...@gmail.com
on 17 Oct 2011 at 1:34
[deleted comment]
Log after clean boot, with service disabled:
Family: 14h
Bit numbering 63 59 55 51 47 43 39 35 31 27 23 19 15
11 7 3 0
COFVID 0071 0000 0000 0011 0000 0000 0001 0001 0010 0111 0000 0000 0010 0101
0110 0100 0000
P-State0 0064 1000 0000 0000 0000 0000 0001 0001 1100 0000 0000 0000 0000 0100
0100 0010 0000
P-State0 0065 1000 0000 0000 0000 0000 0010 1011 1001 0000 0000 0000 0000 0101
0000 0011 0000
P-State0 0066 1000 0000 0000 0000 0000 0010 1001 1000 0000 0000 0000 0000 0101
0110 0100 0000
P-State0 0067 0000 0000 0000 0000 0000 0010 1000 0111 0000 0000 0000 0000 0101
0110 0101 0010
P-State0 0068 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 0069 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006A 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006B 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
Bit numbering 31 27 23 19 15 11 7 3 0
NB P-State0 D18F3xDC 1001 1010 0100 1011 0110 0011 0010 1011
NB P-State1 D18F6x90 1000 0000 0000 0000 0011 1000 0011 0000
ClockTiming D18F3xD4 0000 0000 0000 0010 0100 1111 0101 1000
BIOSClock D0F0xE4_x0130_80F1 0000 0000 0000 0000 0000 0000 0110 0100
D18F3x15C VidL0 0.85V VidL1 0.85V VidL2 0.875V VidL3 0.9V
D0 00 D1F0 90 SMBus A0 D18 C0
1510 1022 h
4397 1002 h
4385 1002 h
1700 1022 h
MSRC001_0061 P-StateMaxPState: 2 MinPState: 0
BIOS vendor BIOS version MoBo vendor MoBo name
Acer
V1.04
Unknown
JE10-BZ
Original comment by jdub2...@gmail.com
on 18 Oct 2011 at 2:30
After making a change (P2 voltage from 1.0125 default to 1.0), and dropping to
P3 444MHZ:
Family: 14h
Bit numbering 63 59 55 51 47 43 39 35 31 27 23 19 15
11 7 3 0
COFVID 0071 0000 0000 0011 0000 0000 0001 0001 0010 0111 0000 0000 0011 0101
0110 0101 0010
P-State0 0064 1000 0000 0000 0000 0000 0001 0001 1100 0000 0000 0000 0000 0100
0100 0010 0000
P-State0 0065 1000 0000 0000 0000 0000 0010 1011 1001 0000 0000 0000 0000 0101
0000 0011 0000
P-State0 0066 1000 0000 0000 0000 0000 0010 1001 1000 0000 0000 0000 0000 0101
1000 0100 0000
P-State0 0067 0000 0000 0000 0000 0000 0010 1000 0111 0000 0000 0000 0000 0101
0110 0101 0010
P-State0 0068 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 0069 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006A 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006B 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
Bit numbering 31 27 23 19 15 11 7 3 0
NB P-State0 D18F3xDC 1001 1010 0100 1011 0110 0011 0010 1011
NB P-State1 D18F6x90 1000 0000 0000 0000 0011 1000 0011 0000
ClockTiming D18F3xD4 0000 0000 0000 0010 0100 1111 0101 1000
BIOSClock D0F0xE4_x0130_80F1 0000 0000 0000 0000 0000 0000 0110 0100
D18F3x15C VidL0 0.85V VidL1 0.85V VidL2 0.875V VidL3 0.9V
D0 00 D1F0 90 SMBus A0 D18 C0
1510 1022 h
4397 1002 h
4385 1002 h
1700 1022 h
MSRC001_0061 P-StateMaxPState: 2 MinPState: 0
BIOS vendor BIOS version MoBo vendor MoBo name
Acer
V1.04
Unknown
JE10-BZ
Original comment by jdub2...@gmail.com
on 18 Oct 2011 at 2:32
It looks to me, as if the settings for P3 are coming from the chip itself, but
P3 is actually not activated. Btw, P3 is actually set to run at 615MHz, which
is little different than what the tool reports (444MHz). I'm wondering, why the
APU goes to that PState, because it is actually not supposed to do that by
"P-StateMaxPState: 2".
Many, many thanks for your time. Please uncheck all options in the service tab,
if you think, it doesn't work as needed. I'm not considering this as any risk
for your system, but it might influence the time you can run on battery.
I will to fix that problem as soon as possible, but due to not having this
piece of hardware you have, it can be hard to find the root cause.
-Sven
Original comment by sven.wit...@gmail.com
on 18 Oct 2011 at 7:37
[deleted comment]
In addition to encountering the same issue above (dropping into P3), I
additionally found that modifying the voltages for P1 (P2 at default and P0
changes seem to have no bug) causes the C-60 to lock into P2. This might be
some kind of (thermal?) failsafe since stopping the workers on prime95 allowed
everything to return to normal operation. I think thermal might be the culprit
since this condition is not consistently triggered at the same voltage drops
and restarting the p95 workers doesn't necessarily cause it to return to the
P3/P2 locked state.
As far as the bouncing between P0 and P1 as noted in the OP, I don't think
there's anything we can do to control this since entering the "turbo" P0 state
is (in theory) controlled at the apu level based on some predefined thermal
envelope. Effectively, windows treats P1 and P0 as the same 100% state and the
apu decides for itself which cores have the capacity to run at P0...
Original comment by tws...@gmail.com
on 19 Oct 2011 at 11:19
Hi,
Since you are talking about thermals. What are the temperatures, while seeing
the CPU being locked in P2?
The next thing, I'm wondering is, why there are P3 settings available, but
actually they are not supposed to be used, since P3 is not activated (see Bit
63 P-State0 0067).
I'm curious, if the CPU goes into P3, without having the BrazosTweaker service
running (uncheck everything on the service tab and power off the system - power
on). I would like to know, how the CPU Pstates behave, if BrazosTweaker is just
monitoring (CPU-Z can be used too).
As I said, I'm not owning a C-60 or any APU with turbo, so I can't test. What I
know, is that the turbo works completely different than in Phenoms or Llano.
There are no Boost states.
-Sven
Original comment by sven.wit...@gmail.com
on 20 Oct 2011 at 7:47
I have (most likely) the same Acer 722 with a C-60 as original poster and am
not experiencing these problems, even after running prime-95 for 10+ minutes
and thermals not seeming to go above 65C. This is with combinations of CPU
cooling set on either active or passive, and CPU frequency at either 90% or
100%.
I do recall something like this in a previous version of BrazosTweaker that I
tried to use (1.04 I believe) and ended up not using it because of this. Fast
forward a bit and I tried using 1.06 and didn't have this issue. Upgrading just
now to 1.07 and testing more, I still can't replicate this. Might it be AMD
driver (Catalyst Control Center) related in any way, as I now have this turned
off. I can plug the original hard drive back in and run it on an OEM system if
that would be of any help to you.
Much appreciated for the work you put into this, hope I can be of some help.
-Alex
Windows 7x64
C-60
Acer Aspire One-722
Crucial M4 SSD
Catalyst Software Suite 11.10 w/ no CCC
Family: 14h
Bit numbering 63 59 55 51 47 43 39 35 31 27 23 19 15
11 7 3 0
COFVID 0071 0000 0000 0011 0000 0000 0001 0000 1010 0110 1110 0000 0010 0111
0100 0100 0000
P-State0 0064 1000 0000 0000 0000 0000 0001 0001 1100 0000 0000 0000 0000 0101
0110 0010 0000
P-State0 0065 1000 0000 0000 0000 0000 0010 1011 1001 0000 0000 0000 0000 0110
0110 0011 0000
P-State0 0066 1000 0000 0000 0000 0000 0010 1001 1000 0000 0000 0000 0000 0111
0100 0100 0000
P-State0 0067 0000 0000 0000 0000 0000 0010 1000 0111 0000 0000 0000 0000 0101
0100 0101 0010
P-State0 0068 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 0069 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006A 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
P-State0 006B 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000
Bit numbering 31 27 23 19 15 11 7 3 0
NB P-State0 D18F3xDC 1001 1010 0100 1011 0110 0010 0010 1010
NB P-State1 D18F6x90 1000 0000 0000 0000 0011 0111 0011 0000
ClockTiming D18F3xD4 0000 0000 0000 0010 0100 1111 0101 1000
BIOSClock D0F0xE4_x0130_80F1 0000 0000 0000 0000 0000 0000 0110 0100
D18F3x15C VidL0 0.8625V VidL1 0.875V VidL2 0.875V VidL3 0.9V
D0 00 D1F0 90 SMBus A0 D18 C0
1510 1022 h
4397 1002 h
4385 1002 h
1700 1022 h
MSRC001_0061 P-StateMaxPState: 1 MinPState: 0
BIOS vendor BIOS version MoBo vendor MoBo name
Acer
V1.04
Unknown
JE10-BZ
Original comment by vaint...@gmail.com
on 7 Nov 2011 at 7:10
Another thing that I noticed is that in the main program window, there is P0-P2
and NBP0, NBP1 (pic). When I click the service tab, I see P0-P4 and no NB
values (other pic), and one of these values is set at 444, the other being 333
as in the first posters image of the NB frequency. The voltage values of NB P0
and P1 are the same as P3 and P4 in the service tab, and change when I change
them in the main window and update. Might this be related?
Might be off topic but after playing with CCC and AMD drivers, my NB seems to
be stuck in P0 state at max (first pic), and my temps have risen 10C+. This is
happening independent of BrazosTweaker service running or not.
-Alex
Original comment by vaint...@gmail.com
on 7 Nov 2011 at 8:25
Attachments:
Hi Alex,
I noticed you are using version 11.10 catalyst drivers. I'm still using the
original drivers that came with the computer (8 something). Maybe that is part
of the problem. How did you get 11.10 to work with the C-60? I searched AMD's
website, but they don't list any drivers for the C-60.
Thanks,
Jerry
Original comment by jdub2...@gmail.com
on 8 Nov 2011 at 2:18
Hi Alex,
thanks for the feedback. Always appreciated.
To your question about P3 and P4 looking like NB P0 and NB P1. That is actually
right. On the service, the NB Pstates get handled internally like the CPU
Pstates (has to do with the programming), thus the numbering is just increasing
(could be cleaner, I know ..).
Question about NB P0: Take a look into the advanced power options under
Powerplay. You will find the settings for battery and plugged-in. Play around a
bit there and keep in mind, how the NB Pstate switching is doing with the
different settings.
Furthermore, it is very possible, that CCC is playing a role regarding the
reported behavior, which that issue was opened on. Please update to the newest
version.
Regards,
Sven
Original comment by sven.wit...@gmail.com
on 8 Nov 2011 at 7:32
What I did:
Insert OEM harddrive, system restore to OEM state. -> Had P3
Bug
Uninstall all bloatware and any non microsoft services. -> Had P3
Bug
Install Catalyst 11.10 Drivers and turn off CCC -> Had P3
Bug
Change power options of power saver to match balanced. -> Had P3
Bug
Open BT-> Service-> Make permanent-> Update-> Restart Comp->
Open BT again and attempt to make changes -> No P3 Bug
Removed OEM Drive and install SSD (one with no bug -> No P3
Bug
Open BT-> Service off-> Restart comp-> Power saver-> Change values -> Had P3
Bug
Open BT-> Service on-> Restart comp-> Change values -> No P3
Bug
So apparently the driver versions I had made no difference, nor did
uninstalling CCC. I simply always had the service running. When you start
BrazosTweaker without the service first starting, and being in either 99% CPU
max limit in any power state, or in power saver mode (with oddly enough 100%
CPU max), it bugs out. When you have the service running already at windows
start, it seems to do fine, and can make changes without it sticking in P3. It
does however jump from time to time into P3.
Thermals don't appear to matter, as it happens even after CPU idle for a while,
hardware monitor showing sub 60C (52C tested).
The stick at 444Mh NB P0 state happens whenever I choose Maximize Performance
as the ATI graphics power setting, and drops to NB P1 when I choose Maximize
Battery Life. Running furmark does not seem to make it jump to P0 if in Max
Battery mode. Not sure if this is useful info to you but just noting that it
doesn't work like a turbo core.
jbud2: The drivers are simply the AMD mobility drivers. If you go to the amd
website, and click "support and drivers" just above "find a driver", then on
the next page there is a link available for AMD Mobility Radeon / 64 Bit
Windows, and click download on the next screen. It will download an installer.
I believe that's the one I used that recognized the card.
Direct HTML:
http://www2.ati.com/drivers/mobile/catalyst_mobility_64-bit_util.exe
-Alex
Original comment by vaint...@gmail.com
on 9 Nov 2011 at 5:14
Hi Alex,
thanks for trying all this.
At which point did you install BT on your OEM harddrive? Right in the
beginning?
Just for information: Before activating the service ("Make permanent") or
changing a setting, BT is really just monitoring and doesn't write anything to
the registers. If the P3 bug was already apparent before enabling the service
or changing any voltage, the issue can not be caused by BT. Btw, I'd like to
know, what CPU-Z is reporting as frequency, while seeing the CPU in P3.
Let me know your thoughts,
Sven
Original comment by sven.wit...@gmail.com
on 10 Nov 2011 at 12:20
I installed BT right away on the OEM drive, but did not have the "make custom P
states permanent" checked. To test each variable I simply changed the voltage
and clicked apply on the main window. It doesn't jump into supposed P3 just
from changing the number for the voltage, but from hitting apply.
CPU-Z shows it dropping from an 8x to a 6.15x multiplier when clicking "apply"
on the main window in BT, roughly 624 Mhz (picture attached).
Original comment by vaint...@gmail.com
on 10 Nov 2011 at 3:30
Attachments:
Once you hit "Apply" BT will write to the registers, which can possibly lead to
having a fourth (P3) pstate being set up (not sure how, but maybe). The thing,
I like to know is, if there is a P3 pre-defined before hitting "Apply" or
enabling the service. I believe, this can be checked by disabling the service
and switch off/on the system. On restart start BT and do a "Log now", without
anything else. If I know, that BT is creating the entries for P3 and I can
debug the code, but until now I'm not convinced, that BT is the culprit.
Currently I have no clue, how this could happen.
Ok, at least I know now, where the 444 comes from. It's equal to the NB
frequency from NB P0. This is because, I'm not considering getting a fourth CPU
Pstate and pointing in the frequency array on the value for NB P0.
-Sven
Original comment by sven.wit...@gmail.com
on 10 Nov 2011 at 3:47
To be clear, I'm not actually disabling the service in windows
msconfig/services/registry, I'm simply checking/unchecking make P states
permanent.
Also, on a fresh restart with the service off, even if i move P2 voltage from
say 1.0250 down to 1.0125 then move it back up to 1.0250 and only after hitting
apply, it still does it. So I didn't really change the values at all.
So I did these things in order...
Unchecked make custom P states permanent and hit apply.
Made sure power plan was set to power saver.
Close program and restart computer.
Opened CPUZ and saved log [noservice-cpuzlog]. (dunno if you may need it)
Opened BrazosTweaker and saved log [noservice-BTlog].
Changed value of P2 voltage up one notch, then down one notch, then clicked
apply (no theoretical voltage change).
BT now showing 444Mhz and CPUZ roughly 630Mhz.
Saved logs of CPUZ and BT [apply-log].
Let me know if you need anything else.
-Alex
Original comment by vaint...@gmail.com
on 10 Nov 2011 at 5:05
Attachments:
Hi Alex,
Thanks! The logs you saved from CPU-Z and BT match in regards of the MSR
0xC0010067 register. This one stores the values for P3. This PState seems to be
pre-defined within the C-60 APU, but not activated. It tells me, that BT is not
writing to this register as you haven't done anything with BT before logging
with CPU-Z.
What I really need to do now, is trying to figure out, why after hitting apply,
P3 is activated. At least, I know now, that the register writes (P3) are not
from BT.
Btw, the service gets activated/deactivated by checking/unchecking "Make
Pstates permanent".
Regards,
Sven
Original comment by sven.wit...@gmail.com
on 11 Nov 2011 at 8:50
Enabling the service allowed me to stay out of P3. Thanks Alex and Sven.
Original comment by jdub2...@gmail.com
on 16 Nov 2011 at 2:22
Enabling the service wasn't helpful in my case.
Original comment by emide...@gmail.com
on 17 Dec 2011 at 5:13
This also happens to me: "When I hit apply after adjusting a voltage, the
reported PState on both cores drops to "P3" and 444MHz".
But only when using Cool 'n Quiet in the BIOS. After I disable it, I more or
less don't have this bug.
Thing is, I really-really thing there's a HUGE NEED for a forum for this
project. There are too many variables and different hardware/vendors, so I see
a need to open up a forum and/or mailing list, in order for us to communicate
and help us each-other.
Please Sven, do open up a forum and a mailing list. Thanks in advance!
Original comment by henrique...@gmail.com
on 29 Jun 2012 at 4:07
Hi henrique,
I totally agree with you. We are DESPERATELY in need for a FORUM for this
project in order to help may of our users. Hope Sven can consider about this.
MyBoy
Original comment by myboy.my...@gmail.com
on 14 Apr 2013 at 5:34
I found a solution for this problem. When the 444 MHz situation appears, just
go to "Super Hybrid Engine" and choose "High Performance". AFAIR, The P-State
comes back to P1 and the Mhz goes to 1333. Then you can try changing back the
frequencies and choosing "Power Saver".
Original comment by henrique...@gmail.com
on 18 Apr 2013 at 8:12
Hi myboy.myboy,
Just start a discussion for instance on http://forum.notebookreview.com and put
a reference in here. I can set a link on the homepage to it later on.
-Sven
Original comment by sven.wit...@gmail.com
on 19 Apr 2013 at 6:20
Hi guys, i need help, rigth now, i now has been 2 year and i dont now anything
about this problem, please, some one can tell me if there is a solution for
this or my pc is dead???
MuSIC_Pro
Original comment by pipeme...@gmail.com
on 17 Apr 2015 at 10:18
Original issue reported on code.google.com by
jdub2...@gmail.com
on 13 Oct 2011 at 2:20Attachments: