Looking for Thuban user running Linux/Windows (a bit of TPC testing)

Thanks for the test. I'm fairly certain it's not a BIOS bug but TPC bug... == don't think we need to revert to older BIOS.

Code:
C0: 0 1 15 20 542 C1: 0 1 16 20 541
C2: 0 1 16 20 541 C3: 0 1 16 20 541
C4: 0 1 16 19 542 C5: 0 1 16 19 542
It almost seems like it's replicating single core data to other cores... per-core statistics are too similar:
0/1/16/20/542
0/1/16/20/541
0/1/16/20/541
...


Ok, back to the code... I'll let you know as soon as I've got something.
 
Found it. I should have something done in 30 minutes or so...
Sorry about that.
 
No problem tear. I have to take off anyways for a couple of hours. I will be back around 9 pm est. it is 6:17 right now.
 
I am sorry about the delay in doing the tests. Here is the results you wanted.


computer idling with just IE, Overdrive, and Command prompt window open on desktop

Ts:12119717 - ed.
Node 0 c0:ps4 - c1:ps4 - c2:ps4 - c3:ps4 - c4:ps4 - c5:ps4 - Tctl: 20
Node0ndows\system32>d:
C0: 0 1 15 6 456 C1: 0 2 15 6 455
C2: 0 1 15 6 456 C3: 0 2 15 6 455
C4: 0 2 15 6 455 C5: 0 2 15 6 455
MinTctl:19 MaxTctl:201F-1871


F@H running on 1 core

Ts:12282847 - ed.
Node 0 c0:ps1 - c1:ps1 - c2:ps1 - c3:ps1 - c4:ps1 - c5:ps1 - Tctl: 22
Node0ndows\system32>d:
C0: 0 469 0 0 7 C1: 2 473 0 0 1
C2: 0 476 0 0 0 C3: 1 475 0 0 0
C4: 1 473 0 0 2 C5: 0 476 0 0 0
MinTctl:22 MaxTctl:221F-1871
 
Last edited:
Kind thanks :)

Humm. Somewhat different. $me ponders...
 
I'll run some more testing locally (non-turbo CPU) but I may possibly need to switch to Linux for a few tests.

That is, If your patience hasn't run out yet :)
 
It is fine with me. I am getting ready to head to bed in a few minutes. I will run what ever tests you want tomorrow night. Not very familiar with ubuntu and linux yet but willing to run it on ubuntu.
 
It is fine with me. I am getting ready to head to bed in a few minutes. I will run what ever tests you want tomorrow night. Not very familiar with ubuntu and linux yet but willing to run it on ubuntu.
With sc0tty's help we were able to squish few more bugs using Linux box (and it seems
there are no outstanding issues there).

So... little change of plans. I'd like to stick with Windows for little longer and figure out
why CM stats are off... I've got another test idea (that should provide us answers we
need) -- I'll write it up...
 
Well I am willing to get back at helping you when ever you are are ready.
 
All righty... let's try something else :)

1. Disable PowerNow and Turbo/CPB in the BIOS

2. Discard all previous TPC copies and download this one: http://darkswarm.org/tpc/tpc-svn111-windows.zip

3. Unzip

4. Check "Run as administrator" on the x64 executable

5. Open CMD as administrator

6. Run:
Code:
TurionPowerControlx64.exe -core 5 -fo 2

7. Run:
Code:
TurionPowerControlx64.exe -CM
for usual 30 seconds (until summary pops up), then interrupt (Ctrl+C) and paste results please.


Thanks again for your persistence :)
 
Ok I had a problem right from the get go. Here is a copy of the Command window
I turned off Power Now and Turbo core boost in the bios.

Microsoft Windows [Version 6.1.7601]
Copyright (c) 2009 Microsoft Corporation. All rights reserved.

C:\Windows\system32>d:

D:\>cd\tpc-svn111-windows

D:\tpc-svn111-windows>TurionPowerControlx64.exe -core 5 -fo 2
TurionPowerControl trunk-r111 (Turion Power States Optimization and Control - by
blackshard)
Wrong node. Allowed range: 0--1

Done.

D:\tpc-svn111-windows>
 
And the excitement never stops! (yay for more bugs)
Stand by...
 
Well I did as you asked and I must say I have something strange going on. I even tried rebooting to end up with the same effect.

Here is a picture

corestrange_zpsdbbf4c69.jpg


And here is the results you wanted


D:\tpc-svn111M-windows>TurionPowerControlx64.exe -core 5 -fo 2
TurionPowerControl trunk-r111M (Turion Power States Optimization and Control - b
y blackshard)

Ts:5029347 -
Node 0 c0:ps1 - c1:ps1 - c2:ps1 - c3:ps1 - c4:ps1 - c5:ps3 - Tctl: 22
Node0
C0: 0 586 0 0 0 C1: 0 586 0 0 0
C2: 0 586 0 0 0 C3: 0 586 0 0 0
C4: 0 586 0 0 0 C5: 0 0 0 586 0
MinTctl:21 MaxTctl:22


*edit* - I went back into bios and turned on PowerNow and Core Performance Boost and still ended up with the same thing. Only the one core reduced speed.

*edit2* - The Smart Multicore option in the boot section of the bios didn't do a thing either. I didn't see anything else in the bios that looked like it might have to do with the cpu cores.
 
Last edited:
That's exactly the desired effect.

We wanted to make sure that -CM accounting wasn't botched
so we forced single core (core 5) to a lower-performance P-state
to see if summary would correctly reflect it....

and it did :)

C0: 0 586 0 0 0 -- P-state 1 -- 3000 MHz
C5: 0 0 0 586 0 -- P-state 3 -- 1600 MHz

I'd say we're done. Feel free to reset BIOS settings to originals (enable PowerNow and
Turbo/CPB) any time.

Thanks again for your extraordinary patience and time, dcds1 -- we're much closer to
TPC release now :)

EDIT: I don't know why you get same results w/PowerNow; you shouldn't... unless
Windows power profile's CPU speed range is different than usual 5%-100%
 
Last edited:
Well I first loaded all the cores with Boinc and then I actuated the command in the command prompt. With the cores loaded they stayed at 3000 MHz except core 5. I was just trying to make sure the bios wasn't messing with your program. I was trying different bios settings to see if I could get all the cores to drop. I didn't know you were only wanting to lower core 5.

But I am glad I was able to help you out and get you closer to your goal. If you ever need any help again I willing help you out.
 
With PowerNow disabled, OS does not have the ability to adjust speeds
so that's why you saw no difference between "no-load" and "load" cases...
== we controlled the speeds via TPC (lowered core 5 speed to 1600).

What I initially found strange was that your Turbo (back when we had PowerNow
and Turbo/CPB enabled) wasn't behaving the way I expected. I was hoping to see
(with single-core load) one core in P-state 0 (full-turbo) and other cores in much
lower P-states (4). I blame it on Windows :)
 
Back
Top