NotebookForums.com › Forums › Notebook Manufacturers › Sager & Clevo Notebook Forums › Sager & Clevo Notebooks › 3790 Speed Stepping Issue & Fix
New Posts  All Forums:Forum Nav:

3790 Speed Stepping Issue & Fix - Page 2

post #21 of 122
load is just how much work the CPU is doing.
speed is how much work it is capable of doing MAX.
what you would want to do is run a program that causes the CPU to stay at 600MHZ... then run something like Doom3 for about 10 minutes, and then exit, and check the graph history and see if the CPU speed jumped from 30-40% to 90-100%
post #22 of 122
2nd Update Using SpeedstepXP: Loaded up Doom3 and played for a couple of minutes. Pegged out my cpu load, but CPU speed remained exactly the same. What the hell does that mean?
I will mention that, for the first time, I was able to play Doom3 without any choppy framerates (previously, even at the very lowest settings, I could not play Doom3 well do to poor framerates while on DC power). But, I would expect this because I've maxed out CPU power.

EDIT: Okay, so, agentsim, I should interpret the above as my cpu was not, at least according to speedswitch, increasing its speed to accomodate increased load? How can that be possible when I was able to play it as smoothly as I can play it when on AC power? Previously, before changing my BIOS settings, I couldn't play Doom3 on battery. So, this must mean that Speedswitch is not able to report increased cpu speed? Obviously, it was working faster than ever before . . . but why wasn't this represented on the graph?
post #23 of 122
hmm.
100% load is what you would expect.
no change in speed is odd.
Let's figure out what's going on here:

1)When on battery, at max battery setting (lock at 600mhz) Doom3 is choppy at any setting?
2)And now following Luke's advice, when on battery Doom3 is noticably less choppy?
3)When on max power (lock at 2ghz) is Doom3 noticably faster than in case #2?
post #24 of 122
ok... you edited and answered my questions

this means that your CPU IS scaling as it should
it would appear that CPU-Z is unable to detect this.

the problem is resolved!
post #25 of 122
actually I guess there is one other test to do before we can be sure that everything is fine.
you need to confirm that it was the CPU scaling that made Doom3 faster, not GPU scaling.

to do this, set BIOS to max battery (lock at 600mhz), turn off power savings on the GPU, make sure that GPU is clocked as you normally have it, run Doom3, ensure that it IS choppy.

If it isn't choppy, that means the 3790 is able to play Doom3 at 600mhz smoothly... if that is the case, why complain at all?
post #26 of 122
Quote:
Originally Posted by obimattkenobi
G,
You would be able to answer this for me. Will having the BIOS settings to Maximum Performance Always (ie, locking the CPU frequency at maximum settings of, in this case, 2.0ghz) take its toll on my cpu? Will it place it under undue strain? Will it shorten its life?
Thanks!
If the configuration was designed and tested properly to support that processor, you should not encounter any problems. It's not like it is "overclocked" or anything like that. The battery life, while at 2Ghz may be another issue because of the possibility of a "quick discharge" and the heat that could be associated with it. (The 8790's 8 Cell batteries bur up quickly under "max load").
post #27 of 122
Thread Starter 
Quote:
Originally Posted by agentsim
ok... you edited and answered my questions

this means that your CPU IS scaling as it should
it would appear that CPU-Z is unable to detect this.

the problem is resolved!

No, it means his CPU is just locked at max since he is running max performance in the bios.
post #28 of 122
gaaaa.... I didn't consider that
post #29 of 122
can you check the frequency under the "My Computer"->properties in WinXP and confirm that while on battery, and idling, the frequency is 600mhz? Or is this feature not reliable?
post #30 of 122
To answer posed questions:

#1) Yes. My previous experiences have been that, while on AC power, Doom3 flies. But, until I changed my bios, I could set the game's options to the lowest resolution, lowest detail and still would be choppy while on DC. That was, of course, when my BIOS was previously set to a cpu speed ceiling of 600mhz.
#2). Yes. After modifying my BIOS settings to make the computer run at maximum performance while on battery, I am now able to play Doom3 as smoothly and with as high as settings as I have previously been using while playing on AC power. FRAPS indicates framerates are comparable to AC power framerates now.
#3). Some confusion here. According to Luke's post, setting the BIOS to max DC power and setting the OS power settings to laptop/notebook scheme should basically allow the computer cpu to run at around 600mhz, but to tap into the extra 1400mhz when needed to help run demanding games/software. My settings correspond to this.
What is confusing is what these third party cpu-monitoring programs are reading to me. Mobmeter reads cpu speed locked at 2ghz, even though, according to Luke's post, it should be throttling down when using less-demanding applications. This has been discussed previously -- drawmonster stated mobmeter will not effectively read/report cpu speed once the BIOS is changed.
So, can we assume Mobmeter is inaccurate? Well, if you look back at Luke's post, the Sager tech guy stated Mobmeter should actually reflect the changes in cpu speed.
So, is Mobmeter accurate or what? If it is inaccurate, then I very well could be throttling up and down as in Luke's post, but the cpu speed is being erroneously reported at "locked-on" 2ghz. If it is accurate, then, despite the advice of Luke's post, there is no adjusting of cpu speed going-on to match the demands of the applications I am using.

So, let's forget about Mobmeter for a bit. What about SpeedswitchXP? This is another third party program that has perplexing results. The realtime graph function of this program seems to accurately reflect the demands certain apps place on the cpu load. Load stays low while running WMP, IE, Word, etc, but then hits the ceiling when playing Doom3. This makes sense -- I would expect the load to fluctuate with the apps I'm using.
Also, on SpeedswitchXP's graph, is a line plot for cpu speed. This is where things get interesting. The speed hovers at approx 30% of max (translate to 600mhz) no matter what app I run, including Doom3. So, what does this mean? Does it mean I am really stuck back at 600mhz for my cpu speed ceiling again? I don't think so, because Doom3 has been running superbly as if I'm on AC power. Again, that didn't happen until I modified my BIOS power settings to "Max Performance" this afternoon. So, despite what SpeedswitchXP graphs, I believe I'm running at a higher frequency than 600mhz.
Could it the lack of graphed fluctuation in cpu speed mean that SpeedswitchXP, like Mobilemeter, is unable to accurately report that value? That would seem odd, b/c it can report increased cpu load. So, why can't it report increased cpu speed?
So, to sum it all up . . . . I've followed Luke's advice. Doom3 runs great. None of the 3rd party software programs are reflecting any change in my cpu clockspeed, so I am unable to verify whether or not, as the Sager guy in Luke's post suggested, my cpu is able to throttle up and down to meet application demands while maximizing battery life as well. And that's what we really need to do here. This is the crux of this problem. Does making the recommended BIOS change fix this troublesome issue? Does it provide an alternative to the two crappy options of either: A) having cpu speed max out at 600mhz while on DC power to maximize battery life at the cost of performance; or, B) locking cpu speed at max all the time to maximize performance at the cost of battery life? Unfortunately, I don't think either Mobilemeter or SpeedswitchXP provide an accurate means of measuring this. Either that or, both are indeed accurate and the BIOS switch recommended in Luke's post does absolutely nothing.
post #31 of 122
ok. if we can't trust any program's output (including WinXPs speed indicator) then the only thing I can think of is comparison of battery life.

if you set the machine to lock at 600mhz and unplug, measure your life.

then set it to what should be variable, and unplug, do nothing (idle the machine) and theoretically WinXP should clock down to 600mhz, and the battery life should be very similar.

finally, and optionally, set the machine to variable mode, and run Seti@Home until the battery dies, if speed switching is working this should confirm it with significantly lower battery time.
post #32 of 122
post #33 of 122
Quote:
Originally Posted by agentsim
actually I guess there is one other test to do before we can be sure that everything is fine.
you need to confirm that it was the CPU scaling that made Doom3 faster, not GPU scaling.

to do this, set BIOS to max battery (lock at 600mhz), turn off power savings on the GPU, make sure that GPU is clocked as you normally have it, run Doom3, ensure that it IS choppy.
Did as you recommended. Choppiness abounds!
post #34 of 122
Quote:
Originally Posted by Andy
basically throttling down the CPU should save battery life. so if we want to check that the CPU is throttled under battery we compare life.
if we also want to confirm that it can run at max power on battery, we can check that by using CPU to its max, and confirming that battery life was worse.
post #35 of 122
Quote:
Originally Posted by obimattkenobi
Did as you recommended. Choppiness abounds!
so now we know that the recommended mode is at least able to provide 2ghz under battery.
now you need to do the batt. life test to confirm that it is able to do only 600mhz under the recommended mode.
post #36 of 122
Interesting:

Just moved my arrow over the SpeedswitchXP icon on the taskbar. It indicates my cpu speed -- simply while idling -- is at 2ghz. Yet, it's real-time graph indicates my cpu speed is idling at 30% of max.
What a crock of sh*t. Clear as mud, clear as mud.
I tend to think you are right, Agent. The only way to truly test this is to discharge the battery at various settings check and see if the cpu is truly throttling down as its supposed to (according to Sager). What a pain in the ass.
Oh well, better start charging the battery. It's going to be a long night.
post #37 of 122
the hidden plus of not being the first to order...
I get to suggest tests, and don't have to do 'em
post #38 of 122
Karnac: Hmmmm, may your 3790 be in the cargo hold of a capsized freighter resting on the bottom of a Taiwanese (sp?) harbor!
post #39 of 122
Quote:
Originally Posted by obimattkenobi
Karnac: Hmmmm, may your 3790 be in the cargo hold of a capsized freighter resting on the bottom of a Taiwanese (sp?) harbor!
can you feel the love?
post #40 of 122
couldnt you use CPU-Z to detect the speed? First run CPU-Z by itself and confirm that it's 600mhz, then run UT2004 in a window or something and launch CPU-Z again. It should be 2000mhz.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sager & Clevo Notebooks
NotebookForums.com › Forums › Notebook Manufacturers › Sager & Clevo Notebook Forums › Sager & Clevo Notebooks › 3790 Speed Stepping Issue & Fix