An Optimization Guide for the Asus G51

@autobahnen

I'm gonna look into the undervolting thing and try to work or figure something out. It does stay at my undervolting voltage when I was doing Orthos stress tests though....what method did you use to undervolt?

@rope-line

Technically, stability just means that you won't get a BSOD and the CPU will still properly function after periods of extended use. 30 minutes of running Orthos without errors basically confirms that your overclock is relatively stable.

Stability, however, does not include temperature. For me, your temps at a 2.5Ghz OC seem a bit too high for comfort (I get high 60s under load) so I'd probably stick to 2.3 or 2.4

I have observed some fluctuations but I have never see the temps go that high and it is definitely not normal. What programs do you have running in the background while idling?

 

Tevs,

How long should i run Orthos to check if my OC's are stable? It seems like i cant bump mine to 2.50ghz. but finally got my 10k mark on 3dmark06.

thanks,

Edit: Oh well, previous post answers my question...

 

Hi! I have a few questions, and though I've read alot of the thread- I might have missed these answers (seen some things that were similar but didn't really make sense to me) so I apologize if I'm being a broken record here. Love the thread btw, definitely a must read imo, for anyone purchasing an asus for gaming. My questions:

1. Would it be possible to put something in there about what will/won't void your warranty? I'm all for having fun with the system- but I'd like to know what will and won't be costing me later on if something goes awry and I need to RMA it.

2. Is it possible to undervolt, and overclock with the stock asus overclocking utility. If not, is it possible to do it with SetFSB.

 

I am using Multiplier Management. I found that SpeedStep has limited or no results for me so I gave up on it after a while.

May I ask what kind of Orthos test you are doing? My undervolt stays stable when I do the Blend test just fine. But when I do the Stress CPU test (the fifth choice in the drop down list), I get the same thing where my voltage reverts back to 1.250V after a few minutes.

 

Considering I haven't under-volted either of latter two options, and that orthos is on "blend" for the test, would any of that change the main opinion of my situation?

As for what programs I have in the background while idling: Skype, MSN Live, and NOD32. Any suggestions in general as a course of action to take?

EDIT: I went back and decided to test out the 2.3GHz state and came up with these results, I'd like to say they match what you mentioned about your own TevashSzat. The undervolting was only stable at 1.163V with my settings.

[Without Under-volting]

[With Under-volting]

The temperature drops are roughly 2C on the max value, and 3-4C on the current value.
Would this be considered a good/proper overclock?

 

Software CPU or GPU overclocking will not void your warranty. Asus has no way of proving that you overclocked. If you overclock by flashing the GPU, it will also not void your warranty because you can always flash your original BIOS back and Asus would still have no way of proving that you overclocked.

 

Awesome thanks.

 

Technically it does void your warranty if you tell Asus you overclocked the GPU or overclocked the CPU outside of their own utility, or if you send your RMA'ed system to them with say the modified GPU BIOS. But I'm sure we're not stupid enough to do make such a novice mistake like that, haha.

 

definitely I doubt I'll be messing with GPU clocks, but I'll definitely be undervolting the CPU as far as I can stability wise while OCing a bit. have the X1A so I've got 2.53ghz stock, am hoping to hit 2.7ghz or so with a stable undervolt to OC ratio.

 

Cool. That's a good start on lowering temps and getting a bit more horsepower out of your CPU.

 

Hello to everybody, I'm new to this forum and to notebooks, the G51 BB is my first one. I tried to adjust the fan speed using RivaTuner but that tab was disabled, also I ran a system check with the same program and it said that my fan was already at 100%. Although I don't plan to overclock any time soon and I-m using a cooler pad I would like nonetheless to lower my temp while playing (My CPU usually reaches 71C and my GPU 86-90C). Thanks in advance and great forum.

 

So now I just got off work, I opened up CPU-Z and CrystalCPUID and now they are showing conflicting voltages and clock speeds...Which is weird because it wasn't doing that before. I changed the voltage and multiplier back this morning from the undervolt it was sitting at standard for some reason. The change didn't do anything this morning, but when I came back tonight it suddenly showed up as changed on one program but not the other.



I am going to run 3dMark06 and find out if one of them is not showing the right number, but I would still really like to find out what caused to multiplier and voltage issue when I hadn't even had a chance to even download CrystalCPUID to even make the change manually...


EDIT: Sweet, so apparently the clock back up brought my performance back:



I still am really wondering what caused my multiplier to get cut before, though...It had to have been an issue with Windows 7, but for now I'm just going to have to manually keep it up to stock voltages...

 

Im also having some issues with CPU-Z. Without OCing sometimes at idle it shows a core speed of 1.6 GHz and a Multiplier of 6.0...
If I do anything, like maximize/minimize a window, run Orthos or even click on the Start button CPU-Z updates itself and shows normal values, but changes again to the lower values when doing nothing.

May its just a problem of CPU-Z, I guess, or the CPU underclocks itself during idle... strange.

 

Yeah, mine pops back and forth from 6.0 to 7.5 every once in a while...I was definitely having issues with my CPU, but even when my voltage is good, it still pops back and forth. I don't think its just an idle thing, because during Crysis, and during 3dMark06 tests my CPU was getting raped, until I had to manually clock my voltage back up and set the multiplier.


Are you running Vista or 7?

 

Vista. It was OK while playing some Splinter Cell, but then when I closed it, it went back to 6.0 and 1.6 GHz. I'm fine with this because its not affecting my performance.

 

Do you switch power4gear to performance mode?

 

Yes, it actually was on Extreme Turbo High Performance Mode

 

You GPU and CPU temps are both fine; what games are you playing?

CPU temps can be lowered through undervolting. On my P7350 each .025v decrease in voltage led to a 2C decrease in temperatures.

 

Your multipliers automatically decrease when your CPU isn't under load. Run something like Orthos and you'll see the multiplier increase.

Are you sure the "Enable Multiplier Management" setting is checked in Crystal? You ought to be able to select your voltages in Crystal, but your voltage lists are grayed out.

 

I think the P4G settings will negate the SetFSB/CrystalCPUID settings. I tried to see what would happen if I tried to enable them both (I naively thought that my overclocks would be... overclocked), but P4G seems to override any other settings.

So set P4G to normal and then do your overclocks/undervolts.

 

Then there must have been a problem before because the multiplier wasn't increasing while in-game. I noticed a decreased performance in any game before volting back up. I'm thinking it was just a driver issue or something...

 

That definitely is odd. Maybe the reading in CPUz were wrong? I'd wait for Tev to chime in, he knows more than I do.

 

Well, got my package of MX-3 today. Cleaned off the CPU and GPU, cleansed, and applied. Smeared a bit over the the non-die part of the GPU. Remembered it doesn't conduct electricity. Cleaned off as best I could. Noticed that it's a real female dog to work with. Apply a tiny bit of pressure and a lot comes out. However, method is not the same as AS5. No spreading is required. Just dab and smoosh.

Cold booted. Ran 3DMark06 for halfway through the first trial, noticed lower than normal frames. Panicked. Exited out, noticed temperatures weren't bad at all. Noticed that Windows was updating. Shut it down.

Ran 3DMark again, at 1280x768. X9100 CPU was 3.06 GHz, multiplier of 11.5, voltage of 1.350. GPU was stock underclocks, standard voltages.

Initial CPU temperature was 46/45. Initial GPU temperature was 65.

AND HOLY #### THAT #### REALLY #### WORKS


I had EVGA real-time GPU temperature readout running. Noticed some things. On the first test (Return to Proxycon), I maxed out at 87C, at about where the heavy chaingun guy waxes the first bunch of defenders. However, I only maxed there for... two seconds? As soon as the fans kicked in the temperature dropped to 86C for the rest of the trial. I maxed 88C on the next test, Firefly Forest, but, again, for about two seconds before fans dropped temperature down to 87C. The two CPU tests had the GPU at something like 72C for the majority of both trials. The Canyon Crossing maxed me at 88C, but, again, only for a few seconds before stabilizing at 87C. Deep Freeze GPU test maxed at 87C, but, you guessed it, only for a few seconds. Stayed 86C for the last third of the test.

Damn, I'm a happy customer.



EDIT: With CPU underclocked/undervolted to 2.793 GHz, 10.5x, and 1.174 voltages, the CPU is at 37C/38C while the GPU is chilling at 50C.

Nice to be appreciated.

The thing with the speedstep shortcut is that it forces your CPU to that voltage and multiplier, 24/7. It works fine for me, so there must be something finicky about your harware. The thing that I've done is make multiple shortcuts. For example, I have a 10.5x @ 1.175, 11.5 @ 1.250, 11.5 @ 1.300, and 11.5 @ 1.350. I just doubleclick depending on what I want to do. For example, while writing this post, my settings are 10.5x @1.175. I could probably kick it lower and not notice anything.

For the most part fan speed cannot be adjusted, and certainly not in windows. You might be able to do it in the vBIOS, but I haven't experimented yet.

And congrats on your purchase.

 

Mission accomplished, ladies and gentlemen.

 

Well, I didn't want to stay in the "stone age".

If we could just get a stable undervolt for the GPU I'll be able to refrigerate stuff inside the GPU.

Okay, okay, your Sager still stomps me, but I'm happy. Very happy.

 

I don't have a Sager. Glad to hear the MX-3 has brought your temps below 90*C. I'd be jumping for joy too if I had a dramatic change like that from switching to a new paste.

I was playing CS:S a while ago to test my new heatsink revision. After 15 minutes of playing in a cs_office pub, my GPU never exceeded 48*C.

 

I saw that your copper mods were posted in a Sager/Clevo thread. Automatically assumed a Sager.

And that was without my Zalman NC2000 on. I can get a nice happy overclock and keep sub-90 with it full blast. Probably even better if a [v]BIOS mod could get me lower trigger temperatures for the fan. Or a good undervolt.

Now that I think about it, a fanspeed mod would be better than a simple undervolt. I can keep full voltages, and the fan would keep the whole thing cooler.

I might even try a copper mod. Why not. I finally killed my warranty sticker when my fingers spazzed when reinstalling the heatsink.

And 1.6 is where it's at. Hah. I'm still rocking that Half-Life mod.

 

I have a Clevo, but it was never touched by Sager. It was imported directly from Clevo.

Tinkering with the fan speeds in NiBiTor is impossible for mobile GPU's, or atleast with the G92's. Our GPU's use some proprietary settings that keeps resetting itself after you save the BIOS. This is because NiBiTor is made to edit desktop GPU fans, not mobile GPU fans. I suppose I can hex it, but that's expert stuff even I wouldn't touch.

I'm still trying to come up with a theory as to why my unstable BIOS failed in your G51. Part of me is saying that 0.92v cannot be regulated by the MOSFET's, and that they can only power the card in increments of 0.05v. But then again I have seen G92's use off-increment voltages before, so I'm not completely sure.

 

Take your time. My MX-3 goop is keeping me happy.

If you could try a .93 voltage (and if you think it would work) I'll test.

 

Just finished adding the vBIOS editing and flashing portion. It is basically done now but I'm gonna add some more things later.

 

Basically done with the vBIOS things now. Tell me if I've made an error anywhere and such....

Also, I'd appreciate it if you guys who have undervolted the GPU via flashing to post your results. Tell me any temperature drops you may have had as well as your new maximum overclock via GPU.

Actually, I'm not too sure. I'll check tonight.

The underclock seems to be good.

As for the temperature fluctuation issue, the programs you have in the background shouldn't really cause that. Has it occurred repeatedly or was it just a one time thing. (the GPU up to 87C that is)

You can't change fan settings with any software as of right now.

The best options would to either somehow edit the vBIOS via hex editor or edit the BIOS itself or softmod and change the dsdt tables for acpi settings.

None of those are easy. I've been looking into changing the fan settings for a bit now and nothing seems to be overly promising.

The CPU will underclock itself in idle to reduce temperature. That is normal. Thus, that is why you have to put the CPU under stress via Orthos to confirm that you have successfully overclocked with SetFSB.

@EIment

You should be able to overclock the CPU with the Asus utility but I believe it forces the GPU to stock clocks and overrides any settings there. Thus, you would be unable to overclock the GPU.

 

Sweet, I was right.

 

Bah....missed it earlier.

Well, the good thing is that we know what the problem is. The bad thing is that I have no idea what would cause your CPU to underclock and undervolt itself so badly normally.....

You wouldn't by any chance have done some benchmarks under Vista?

There can't be a driver issue since the CPU doesn't have drivers either.......

Anyways, at least you can automate the CPU forced voltage and multipliers and it'll perform normally.

 

I notice my fan kicks in around 150F this true for anyone else?

 

That sounds about right as the point at which the "low" fans kick in.

Medium fans kick in around 75C I believe and the max fans kick in at 89C.

Max fans go down to medium speed at ~79C

 

I noticed it only occurs while I have CrystalCPUID running (@undervolt of 1.163V). It seems in addition, to also occur if there isn't enough load on the processor that would increase its speed to the OC of 2.3GHz.

Meaning: While I'm gaming or web surfing, or watching videos, etc... the problem can't be replicated. But if I let it idle, it will climb from idle temp of 59C to a possible max of 87C (can be lower) and then drops back to 59C. The fan doesn't go off, nor does the exhaust air seem any warmer than usual.

I was wondering, in CPUID when I go to multiplier management settings, should all the multipliers be 7.5? I've been reading throughout this thread and that was suggested to someone who had problems with undervolting.

 

Well, it really shouldn't matter in regards to the multiplier.

Okay, can you get GPU-Z, go to the second tab when it starts, and then log the temperatures and check the box that refreshes when in the systray.

After you've been through one of your temperature spikes, attach the log to a new post.

To do so, look at additional options within the post editor.

 

Thanks for the reply guys, I'll be checking out your posts if any of you finds a way of keeping lower temps. Great forum.

 

Hey, I get this problem too. GPU temps spike randomly when I'm not doing anything intensive. Just earlier I got a 77C spike (laptop on NC2000 cooler) from normal temps of low 60s.

 

Aside from applying new thermal paste, using a notebook cooler, undervolting, and underclocking, the last option a G51 user has to lower temperatures is to perform a heatsink mod. It will be very effective on a G51 as there is no intake vent directly over the fan, rather there are smaller multiple intake vents over key components such as the CPU and GPU. Adding more surface area to those components will allow better cooling as cool air passes over them.

 

Voltages on my G71:

High Performance mode: 1.225V
Asus Standard Mode & Power Saver Modes: 1.113V

Wouldn't putting the computer in either Asus Standard Mode or Power Saver mode have the same underclocking results as using CPUID?

 

Has anyone tried the Resident Evil 5 DX10 Benchmark on the G51 yet?Would be interesting to see temps and FPS... Just tried it and my OC'ed 4870,which until then breezed past every game at 19x12,struggled in some parts with all settings maxed out... Hope it's just a case of it just being some a benchmark tool and not the finished product .

 

I did this test today, maxed out ( Max resolution was 1360x768, weird )...
Sorry but no screenshots, if i press Print Screen, then past it in Paint, all i get is a Full Black Screen =/

Anyways, on variable test, i did 35.1 FPS, didnt notice any lag during the test,
GPU temps top at 90C ( that alot for my system ), and CPU 57C and 61C...

 

If you don't mind, could you download/install EVGA Precision? You are able to run a real-time temperature readout while you game/benchmark.

Quite helpful, since you can get an idea of your average temperature versus your max temperature.

Great item
(Thank Soviet Sunrise for pointing this out to me)


Eh, might as well say it myself

THANK YOU SOVIET SUNRISE FOR POINTING THIS OUT TO ME. IT IS AWESOME.

Heck, I might try a copper mod next. I'll take a ruler into the G51 guts and see what I need.

 

Yeah, thank Soviet Sun... wait a minute. Oh sh...!

 

Thanking yourself = /0 ?

 

Haha....I should probably do that. Right now, I'm doing it the old fashion way which is to get it as a log file through GPU-Z and then importing it into Excel to do graphs.

Anyways, I've done some retests with Orthos and I still get my 1.113v undervolt. It'll jump to 1.250v for a fraction of a second at first but then it goes down to 1.113v and stays there.

Weirdly, I am now getting 1.038v at 6.0 multiplier even though I have it set as 1.113v in CPUID......

 

Wow, that's how my dad used to do it.

 

Used to? How does he do it now?

 

You're doing "Blend - stress CPU and RAM" test option in Orthos? It works for me on that too. I've gotten over 40 minutes stable on 1.113V using that test (didn't test any further). Problem is when I try the other Orthos test options, other benchmark tests like Street Fighter IV, or game demos like the Crysis demo.