An Optimization Guide for the Asus G51

So there's a chance that the flash will work, I boot up into Vista, and five minutes into a game/benchmark it'll brick?

Or will I be able to visually go and reflash?

EDIT: By the way, when I looked at the .rom in NiBiTor, there was no 2D voltage. It was blank. I had to pick either .92 or .8. Would it be safe to save it from there?

 

It depends if you get past POST or not. If you do, then you can flash back normally if SHTF.

Off topic: Sorry. I'm kind of keeping my eye on my flashlight right now. The current bidders are playing the waiting game right now and are ready to out snipe each other at the last few seconds.

 

Ew.

It's possible that it will brick before/during POST?

 

In this case, I only hexed the lowest level. Typically if the lowest level fails, the system should use the next level up, throttle, as the backup and use that as the boot clocks. I wouldn't hold my breath on that though.

 

Phew. Daunting.

Like I asked, though, should I modify the vBIOS so my 2D settings are .8, or should I leave it blank?

 

I created a new voltage table that includes three voltage entries: 0.92v, 0.8v, and 0.75v. 0.75v is hand entered into the voltage identifier, hence why the 2D box is blank. However, the BIOS still reads it as 0.75v. You may change it back to 0.8v if you want as the whole point of your request is to test 0.92v. Because NiBiTor doesn't natively support voltages below 0.8v, I had to force it in the code, which is why I'm being extra cautious right now, and to confirm your assurance that you can force flash on your own.

 

All right, then. I'll just change it to .8 for the time being. .75 can come later. Wish me luck, comrade.

EDIT: Bad news and not-so-bad news.

Bad news is that the BIOS doesn't work. I flashed successfully, but after signing onto Vista, my screen would hang up (black screen, only my pointer visible) before going to Aero/desktop.

Not-so-bad-news is that the GPU worked in while in BIOS, and I flashed back to my original .95/.85/.85/.85 voltages. I think I'll stick around stock voltages awhile until you and whoever is insterested in this can come up with a stable overclock-capable undervolt.


Much thanks. I'd rep, but apparently I have to spread it around some more first.

 

Oh well. Atleast I got to pass time while 3D'06 was downloading. Forgive me, but I'm a touch buzzed on booze and bull right now. Perhaps I made an error under the influence, haha. I had a feeling that the BIOS was going to fail anyway. I'll take another look at it later and see where I screwed up.

My card is running 500/1250/800 @ 0.9v and is rock stable at 88*C on OCCT. Time to put it to the test.

 

Hah. Well, give a Sober Soviet vBIOS whenever you can.

 

Alright, I installed Windows 7, and installed the Dox drivers per original guide. I ran 3dMark06 and got this score:



Can anybody possibly explain why I would get such scores??

 

Oh snap.

I'm assuming that otherwise your computer is running fine (like you can run Crysis-level games), but that 3DMark06 doesn't like Win7.

 

I would like to le everyone know that I have had FANTASTIC success with Tev's NEWBIOS.rom flash.

My gpu hovers at 96 degrees on the gtest in OCCT, and one day even hovered at 102 degrees.
Now it stays at a ceiling of 85 degrees. That's an amazing 11 degree drop.
I'm running at the stock 500mhz core clock, and everything seems stable so far.
About to try 3dmark06 and some Crysis gaming to see how those match up, but I'm expecting good things.

 

Crysis was running slightly choppier on Windows 7 than it was Vista. Not by a whole lot, but noticeable. Otherwise, it ran very well, nothing even close to what the scores would suggest.

During the test, the CPU tests would not even give me more than a frame per second, it would be more like a frame every 3-4 seconds. I don't know why it would be so bad...

Oddly enough also, Windows 7's "Experience" gives the P7350 a 5.5 instead of the 5.0 it gives it in Vista. I'm assuming it scales them slightly differently...

 

Alright, I've done more tests with my undervolt and overclock combo and I found something that worked finally. It appears that Orthos, when set at the StressCPU option, undoes my undervolt after a few minutes. When I switched it to the Blend option, it ran stable at 1.113V and 2.25Ghz for over 45 minutes with Orthos. Temps hovered around 68-69 degrees Celsius. Thanks for the help anyway, Forge. And of course, thanks Tev for his guide. Another success!

 

My GTX 280M is clocked at 500/1250/800 @ 0.9v, running on Windows Seven x64 build 7264, running 186.03 WHQL, and using the Edova Xpad. Ambient temperature is 73*F.

I ran two back to back sessions on 3DMark'06. I began the benchmark when the GPU core temperature rose to 35*C minutes after the GPU fan turned off at 34*C. For the duration of the benchmark, the fans were in normal mode, not at maximum or silent mode. At the end of the two benchmarks, the GPU never exceeded 53*C. The CPU never exceeded 58*C on the hottest core.

My scores are classified.

 

All hail Sager cooling solutions.

Now, all I need is that MX-3, a stable undervolt, and I might get only thirty degrees more than you.

 

So yeah, I can't find anything online that would explain the scores...anyone have any insight as to what could cause such a thing?

EDIT: Yeah, with Windows 7 and dox drivers, my fps in Crysis have dropped to about 10-15...i'm going to try going back to the original drivers and see what happens, I probably messed up with the driver install.

This still wouldn't explain the horrible 3dmark ratings, though...

EDIT2: Back to original drivers and my fps was cut in half yet again! I was only getting about 5-10 fps average, it was ridiculous...I have no idea whats up, but i'll get back to it tomorrow...hopefully some people provide some possible insight before I wake up...

 

I think I have a theory as to why the BIOS failed. My head is starting to spin right now so now isn't the best time to explain it. Maybe in the morning.

 

Wow, some interesting stuff happened to me in the past few hours.

I tried making the SetFSB shortcut link on my desktop as per the guide's directions, and it was fine until I actually clicked it. My computer crashed without a BSOD. I try clicking it again, same thing happens.

I recheck the guide and figured out that I made a mistake and forgot to change the ### to actual values. So I fixed it and tried clicking it again. Now this time I actually get a BSOD. At this point I'm pretty freaked out. I reboot and check out my computer's numbers to find my CPU was downclocked to around 800Mhz.

I tried restoring stock clocks and testing again, and they seemed to work. After a few shut downs and reboots just to be safe, I start CPU-Z to find something strange. My computer starts out at boot with 1.113V on the CPU without me even clicking the CrystalCPUID shortcut.

I tried to revert my voltage back to stock 1.250V manually and tried the StressCPU option on Orthos (the one that reverted me a bunch of times earlier) and I see that the voltage reverts back down to 1.113V after a few minutes, as if 1.113V was my stock voltage. This is pretty cool and all, but I'm kind of worried that something might have happened to the BIOS settings or something as a result of the BSOD I got earlier.

I really don't want to make it obvious I voided the warranty so I'm hoping there's some way to fix this. Can I go back to 1.250V stock voltage without manually configuring it?

 

Do you have CPUID on autostartup?

 

Nope. I opted out of automation because I was afraid I would forget about it after a few weeks. Thus, I have nothing set up.

 

Hmm. And I'm sure you don't have the shortcut as a startup, either.

Make a new shortcut at stock multipliers and voltages. Click it, and reboot. Check voltages then.

 

Sweet, I tried this and it works perfectly. Thanks a bunch.

It would seem that the simple existence of the shortcut affects the voltage. Am I the only who has experienced this so far?

Come to think, using that logic, it makes sense now why my computer crashed when I clicked the SetFSB shortcut that I made earlier. I should not have clicked it and instead just let it sit on the desktop, as it works its magic without the need for me to touch it at all.

Edit:
Okay, I actually tested that out with a SetFSB shortcut. Seems like it doesn't work that way with SetFSB. I still get 2.0Ghz when I boot up. I don't dare click it this time though lol. The mere sight of it scares the hell out of me.

Aww man, I tried remaking the shortcut for CrystalCPUID and getting the static 1.113V at boot again, but it won't do it. I guess that was a one time problem thing related to the BSOD.

 

Wow.....tons of posts since I went to sleep.

Regarding the undervolting thing: CPUID has somewhat of a tendency to not always keep the voltage at that level. At times it'll go back to the default voltage briefly. I'm still looking into it and if not too lazy, I'll try to find a more stable undervolt.

@MooCwzRck

What are your component scores in 3DMark06? What temps are you getting?

@autobahen

That instant BSOD was due to too high of an overclock. The voltage thing I haven't seen....

Regarding the vBIOS flashing, I'm still working on getting that 10K 3DMark06 score right now..... I just got 9999 last run.....and I'll go work on the 3D, Throttle, 2D voltages in a couple of days......

Edit:

Thats it....I'm giving up.....just going to stick with 9999 right now.....

 

Haha, 9999 points. That's a keeper.

 

I get 10468 on 3dmark06 with an undervolted 260M. I have the Q9000 which is probably pushing me up though ^_^

 

Yeah....upping the CPU clocks or performance gives you a huge boost in 3DMark06 while OCing the GPU doesn't really give you too much.

 

It's funny sometimes how small things are the source of bigger ones... All this G51 GPU undervolting bizzo started out from the other thread about battery life...

No offense to the other thread's starter,but it generated something MUCH more interesting,at least for me.I'll be getting my BB G51 next week,U can count on me flashing the vBIOS 1st thing out of the box.For me components security and longevity are more important than 3DMark scores at the moment,as RMAs would be a complete pain given where I live right now(I remember how I had to return my then X1900XT to ExcaliberPC because it was under a US-only warranty...It wasn't fun).So cooler temps all the way for me,even if it has to be at the cost of some FPS.

I'd like to thank Soviet Sunrise,who is a well of knowledge for which my thirst will never be quenched,and our 2 G51 mod crusaders,Tev and Forge.All your efforts are much appreciated out there,guys.It's been a llooooonnnggg time since I last hex-edited,perhaps I'll take a refresh course with U,SS.

After this convoluted preamble,I would like to know if we could gain some more degrees depending on the driver used(DOX,stock or others from LV2G.com).

 

I remember reading some people reporting 2-3 degrees cooler with DOX, don't quote me though.

 

@Y_Double

Well technically, the Dox drivers are supposed to make your GPU slightly warmer but for most G51 owners, there have been no temp difference and some have seen it run a little bit cooler.

As for further reducing temps, I'm gonna go find the lowest voltages for 3D and Throttle which will reduce heat from non-gaming like web surfing or watching videos. As for reducing temps under load, the only things I could think of is to get a cooler or do a fan mod, ie install more or more powerful fans in the G51.

 

These DOX drivers seem to be a combination of Jeff Wilson,Glen Osbourne and Christian Cullen all-in-one .Offering the best performances at reduced temps?Me likey.

 

Goddamn I got a pounding headache right now. I actually did the most bizarre thing a few hours ago while I was toasted and dismantled my entire GPU piece by piece. What's even more insane is that I stuffed it with more thermal pads and heatsink blocks inside, yes inside, of the main heatsink itself. I'll link you guys to the Clevo forum after I upload the pics and find out what the hell happened to me after I had that Red Bull Cocktail.

 

Umm.....and thats why you don't disassemble computers while inebriated

 

I seriously can't remember what I did after I turned off my computer, went to the little boy's room, came back and tried to go to sleep. Could it be the adverse effects of a combination of sleep deprivation, alcohol, and caffeine?

 

Maybe next time try mixing czech absinthe with vodka and mana potions. Who knows, you might wake up with a phase-change cooling mod on your 260M ^_^

 

Whatever. I'm going to try and go back to sleep. Here's the link as I promised of my even further modded GTX 280M.

http://forum.notebookreview.com/showpost.php?p=5190191&postcount=645

 

I just updated the guide a bit by rewriting some passages that I though may have been unclear and then added some more information.

It still doesn't have vBIOS flashing instructions yet. Probably will do those tonight when I get home.

 

tevs, how can i save changes using setfsb? everytime i shutdown and boot up it reverts to stock. thanks

 

Go to the Task Scheduler section of the guide and read.

 

and i'll post my temps in 1 minute...

EDIT:


the temperature was taken just after the 1st CPU test, where I was only getting like a frame every couple seconds. It doesn't seem like its the CPU or GPU itself thats the problem, it seems like a driver problem, or a compatability with Windows 7 problem. My performance in Crysis dropped noticeably when I installed a clean version of Windows 7 on there, and my temps in game are generally very good. Crysis tops it out to about that 91 degree ceiling, then the fan kicks in and it stays consistently below 86 degrees, and I've only seen it stay around there during firefights and areas with a lot of units.

What do you think?

 

Okay....both your GPU and CPU scores are way too low.

After you post the temps, would you mind posting a screenshot of CPU-Z while Orthos is running?

Edit:

Your CPU is obviously being throttled. I'll know for sure when you do the CPU-Z with Orthos running.

3DMark06 GPU subscores are too low too which suggests that it is being throttled somehow.....

Post a GPU-Z screen shot while you are at it too.

 

how long do you recommend having it run before the screenshot?

 

Not long, two minutes would be fine. To save space, you can just put the CPU-Z and GPU-Z screenshot together.

 

Okay, how did you undervolt your CPU?

You have locked it at the 6.0 multiplier when it is supposed to have a maximum multiplier of 7.5 which is essentially underclocking it.

If you followed Forge's shortcut option, make sure that when you did it, you are doing it for the 7.5 multiplier.

 

I didnt undervolt anything, I never even got a chance to get to that process. The only changes I have ever done to my laptop so far, was the installation of Windows 7, the installation of Oblivion and the Crysis demo and the temp monitor and Chrome, resizing the desktop and icons and other features, installation of DOX drivers per Forge's guide, 3dMark06, then a reinstallation of the nVidia drivers to see if that was the issue, and most currently, the installation of CPU-Z and GPU-Z. That is literally every single thing I have done in this computer. I havent made any other adjustments to anything.

So something else clocked it to the 6.0 multiplier, because I never messed with anything as far as voltage goes. I'm going to go ahead and go through the guide and see if I can just have it go to stock voltage...but I'd like to know what could have caused this in the first place...

 

Well.....then that is very very weird indeed. Something is underclocking your CPU.

Okay....download Crystal CPUID, Go to Function, then "Intel Enhanced SpeedStep Control"

Set the multiplier to 7.5 and voltage to 1.250v and then apply it.

Do not close CPUID, keep CPU-Z and HWMonitor open, and then run 3DMark06 again and give me your new score (as well as subscores).

Edit:

There is a link to CPUID in the guide. Also, get CPU-Z 1.51 instead of 1.52. There may be problems with 1.52 and CPUID.

 

Damn, I've got to go to work...I'll mess with it more tonight, but thanks a lot dude, you've been an amazing help so far!

 

After more tests this morning, I've found that the undervolt doesn't stick when I play games. I tried the Crysis demo and the Street Fighter IV benchmark and both times, I noticed the voltage shoot back up to 1.250V.

Also, it seems that the voltage stays stable at my set voltage during the Orthos test if and only if the Core VID is below the voltage that I set (usually 1.113V or 1.150V), or else it defaults back to 1.250V when the test starts. It appears the only time that the voltage stays stable for extended periods of time is when the Orthos test is kept running. Otherwise, it keeps popping back to stock when I try other things that stress the CPU, like games or 3-D Mark. I've tried the automation via task manager thing as per the guide and I get the same results as I explained in this post.

Edit:
Oh yeah, and this is while running stock CPU 2.0Ghz.

One more thing I noticed is that the mouse-over message on the task bar icon for CrystalCPUID says "Multiplier=7.5x, Voltage=1.250V", despite my settings being 1.113V in Multiplier Management at the moment. This is during the Street Fighter benchmark.

 

Hello Gents,

I'm new to the board, but I've been reading up on a variety of your topics regarding this notebook PC. I'm an experienced user, but I've never attempted OC'ing or under-volting, until buying the G51 that is.

I kind of wanted an opinion, and maybe a clarification? If possible.
What I'm looking at currently is:

a) I successfully OC'ed to 2.3GHz a few day ago, ran fine for a while, but I felt like the temps should be a bit lower, so I decided to undervolt and I was able only to go as low as 1.163V. This produced temps of around CPU: 68C and GPU: 64C, running orthos for 30min (Orthos errored if I went any lower than that voltage).

I decided at that point to reset the voltage back to the original setting of 1.250V and restart, and attempt to OC to a higher value. And yes, I do realise higher OC produces more heat.

b) I successfully OC'ed to 2.4GHz this afternoon, but I'm not sure that the temperatures are stable? Here is a description of the controls: 80F room ambient and a Zalman NC1000 cooler on medium speed setting, the desk is wooden, vent not obstructed. No under-volting has been performed on this setting yet.

(image here)

c) After achieving those results, I pushed my luck a second time and I think I successfully achieved an OC of 2.5GHz. But this is where I pose my question: "What exactly is 'stable' in terms of OC'ing?" Is it based on not BSOD'ing and/or heat.. or? Same temperature controls as in Part B, and again no under-volting.

(image here)

Being even more daring, I decided to push past 2.5GHz and ended up BSOD'ing only after adding around 20-30Mhz to the clock. Based on what I've shown above, does anyone have any clue which build would be most stable to use? I game occasionally (MMOs or FPS) so it would be in load for long periods of time, and it will also be used for University.

My last question might be a bit random but: Has anyone had issues with temperature fluctuations when the screen is off along with the HDD and idling? I get GPU temperatures of around 59-62C idling w/screen and HDD on, but it can sometimes leap to 87 when I'm afk and the fans kick on (can be heard) it cools back to around 68C and it might repeat the cycle, depending on its mood. Any suggestions?

Thanks for any help.

Cheers