G50VT 9800M GS Overclocking?

Anyone doing any overclocking with the 9800M GS? I know the G50v peeps with the 9700M are doing some.

Just didnt see much about the 9800... Yea its a fast card but no such thing as too fast now, haha

 

I was just about to start a thread on this, as I too have heard nothing about overclocking this beast!
Any overclockers with a G50vt feel free to work your magic.

 

Assuming there's a way to, how much could we expect before heat becomes a problem? This is my first laptop, and I'm afraid to push it too much. It already gets pretty warm in there according to Everest.

 

I think these have good fans and setup and can stay cool.


My problem is I just overclocked my CPU for the first time and started to play my game, and boom. It was giving me flickering screen saying video driver stopped responding.

Which sucks, all I did was overclock with SetFSB to the same the DirectConsole would do. I had it at the 2.7Ghz that the max in DirectConsole does and it started doing this. I did have DirectConsole closed when I set it also so its not them fighting.

So my guess is my memory cant handle it... Wish they did it so that the memory didn't have to overclock with the fsb. Unless the PCI-E is overclocking also but those clocks dont visually change so I wouldnt think it is. Just sucks that I cant even overclock the CPU to the manufactures high end ability the give us.

 

Yea, I notice that even when gaming the fan doesn't run that loud/fast on my X1, so it must not be stressing so far. I'm not a power overclocker, but maybe getting to GTS speeds would be nice.

 

ye would be nice to know.. getting my asus g50vt a1 tomorrow i guess

 

hi

tried to overclock mine.....but i really dont remember the clocks right now(im using another pc)

anyway i also oced the cpu

3dmarko6 at 1024x7--something (not standard i know)

no OCing(stock drivers): 9450

no OCing(nvidia latest): 96--somthing

with OCing(nvidia latest): 9814

in my opinion it doesnt deserve it...keep it as it is!!

i f i have the time i will post the screens.
ive seen in another site on xp it gets 10000+ at same res


peace

 

95% sure it will be the same process as overclocking the 9700.

1) 169.04 or 179.28 + ntune to find your cards max (temp- or clockwise)
2) extract the vbios from the system bios
3) nibitor won't see the extra clocks, find the extra level clock speeds
4) replace vbios and flash

if any g50vt owners want to help, post the max clocks here and ill see what i can do.

as for oc'ing the cpu; its not oc'ing the cpu but rather the fsb. since pci-e gets its timings from the fsb, bad timings = random things can happen.
and unless you enjoy benching it/compressing/encrypting things, the speed difference wont be noticeable.



a quick edit: are the clocks 625/1563/800? the g50vt bios and g50v are identical according to my tools... and theres no way that the clocks can be that high.
(i made sure i downloaded them from http://support.asus.com/download/download.aspx?model=G50Vt&SLanguage=en-us and http://support.asus.com/download/download.aspx?model=G50V&SLanguage=en-us respectively)
if anyone wants to make a backup of the bios on your machine (EZflash might have the option), a megaupload/zshare/mediafire (or whatever you like) of it would be appreciated.

 

Im not at home to post them but I was wondering how this would work because the two laptops use the same bios file some how...

Anyway are you sure that overclocking the FSB does the PCI-E? As this is not true on my desktop anymore and also SetFSB does not show the PCI-E clocks change when I overclock the CPU. Which is odd because when I run on Turbo Extreme in direct console or even match that setting in SetFSB my video is unstable and the driver likes to start to fail...

Sucks that I cant even run the stock overclocking options without having a problem. It would make me think that yes the PCI-E is getting changed, but yet it does not show that it changed and people have tested and the scores dont change at all...

 

it's why better mobo's have a pci-e lock tried many times, both with xp32 and xp64. on 64 i get instability, on 32 the lan card couldnt connect with even the smallest changes.

i don't know how turbo extreme does it, but i have a feeling its more than just fsb. my guess is it reduces pci-e to negate the effect (and keep it at 99 MHz). there was a guide on the net somewhere about it, but i dont remember it off the top of my head so google is your friend.

ive downloaded the G50V t bios from their .com.tw, .cn, and .de support site, and their all the same. only one nvidia (10DE) bios module with a 9700M device ID (064A). i just dont get it...

edit: it _doesn't_ work with turbo extreme?
i've gone through the clocks section of the 9700m vbios, and the 0xD2__ offsets which looked similar, found nothing.

 

Maybe the 9800M uses its own onboard bios and does not flash with the system bios. So the older notebooks this mod can be done on but the newer ones because its using the flash on the video card and not pulling it from the bios we are SOL?

I know they all have a bios chip on their own but like the 9700 pulls the bios from the system and flashs/updates its onboard bios if it was changed... Maybe the 9800 just uses whats on it and does not ever update or anything else.

 

just to make sure im not seeing things, if a vt owner can follow this guide ( http://support.asus.com/technicaldocuments/technicaldocuments_content.aspx?no=541&SLanguage=en-us) and rip the bios thats currently on their system using dos, it would be appreciated.

and while in dos, see if you can use nvflash to get the vbios (5.72)

 

I haven't checked out the vbios, but ntune is effective in overclocking the G50vt-X1. I've stably run 3dmark06 at 586-883-1466 which is a 10.5% overclock on all clocks. I can run COD4 and N4S:Undercover at GTS timings (600-800-1500), but 3dmark06 crashes after a couple minutes at those speeds.

 

You know what I find odd now... If I overclock my CPU with direct console to turbo extreme I am fine, but if I dont use that and use SetFSB I start to get video driver has stopped working errors like the video card is failing... I set SetFSB to the same speed as direct console even... What gives? lol (And yes I had direct console closed while using the other, hehe)

But for the overclocking I wonder if the fan in these notebooks is controlled by the CPU speed, GPU or both? Just dont want to fry anything from to much heat. I would like to OC everything to the max I can on stock voltage and than step back a tad and leave it like that but being a laptop if a part fails its much more a pain to get fixed than my desktop... I dont want to have to send my laptop to asus anymore, lol

How are the temps on your GPU?

 

Here is the Vbios from the VT just rename to .rom it's not zipted just rename.

I did not get a chanse to play with nbitor yet with it.

Chris.

 

extra step has no identifier (like the 9700).
core -> 0xCCE8, shader -> 0xCCEA, mem -> 0xCCEC.

but if overclocking works with ntune, then theres no reason for a bios mod (other than to save yourself 50MB of hdd sapce).

 

is ntunes really working for some folks?

 

I think it only works with like 2 driver versions... One being the current notebook beta drivers @ nvidia.com

I just dont like using programs to OC because thats something else to have installed. and something else to cause an issue... But works to find your limits.

 

169.04 and 179.28.

@chrous25: was it an nvflash dump or from system bios? either way, submit it to the nibitor team.

 

169.04 and 179.28 dont work from me, at least the ones from laptops2go, i blue screen as soon as i fire upntune.

 

one off nvidia site also doesnt work, does ntunes need to be reinsalled?

 

if it doesn't a bios mod is the only way.

 

is there a chance of bricking the machine with the bios mod?

 

And is there a posting for this other than the 9700 job?

 

theres always a chance to brick. if you know what you changed, then the chance is very small (power cuts are more likely).

my guess is chris used nvflash (so the 9800 has its own bios chip, separate from the system). change the addresses (if you dont know hex nibbles/words, i would recommend against doing it yourself) i listed a few posts back and reflash with nvflash.

 

Has anyone flashed the 9800 yet w/o issues?

 

probably not, but someone always has to be first

 

not first!

 

Here's proof from my X1. I'm using ntune 6.03 and driver 180.84 (from laptopvideo2go). Link to ntune: http://www.nvidia.com/object/nvidia_system_tools_6.03.html

I prefer to software overclock so my card isn't running at extreme speeds all the time, otherwise I'd just flash the vbios.

 

I've been playing NFS:Undercover recently, so I used that to test temps though it isn't really challenging for the 9800M.
At stock cpu/gpu I got 79C on the card,
at 10% overclocked gpu I got the same temperature,
when I OCed the cpu with setfsb to 2.5 with no gpu OC, I got 81, but no errors,
I'll try overclocking with setfsb and the gpu next and let you know...

All this time the P8400 has run cool as a cucumber.

 

Well what is odd is that I have not done any GPU OC yet... I wanted to but now that this happened I am a bit worried and dont want to overwork something and kill it... I know most of the time you cant kill hardware overclocking unless you go high on voltage or dont cool well, but you can stress the chips and they fail still.

What is crazy is being I did NOT overclock the GPU I got video driver errors when using setFSB to OC cpu from 2.53 to the 2.8..... Next day I did the same thing but used DirectConsole instead and it was fine... So its like direct console does not overclock the PCI-E maybe and setfsb does even though it shows the mhz staying the same? or something else I dont know, but it makes me trust setFSB less... I know a lot of people use it with no issues, but also some have issues and I guess I am one of them... Some peoples hardware can overclock better than others and I may have one that does not, but whatever caused my problem was not memory and was not CPU, so that leaves the GPU or PCI-E buss as the problems happened only with video.

But do share your findings, and if any others care to do some testing would be great. Maybe someone do some bench compares using setFSB and than useing direct console? Just use the same speeds in setFSB that direct console offers so we can get an idea.

I took setFSB off am are worried to use it again... I may just hold out and wait till I get my new CPU to replace my T9400 and wont care to overclock if I got a faster CPU... Would only care to OC GPU than...

 

Well, I think I figured the problem out. I'm afraid I can't compare benches between DirectConsole @2.5 and SetFSB @2.5 because I uninstalled direct console and now the overclock function won't work properly.

Anyway, it seems that the fsb increase makes the core and shader clocks on the gpu increase too, though GPU-Z doesn't indicate this (it doesn't seem to affect the GDDR however).
I came to this conclusion because as Bryanu says display driver errors show up when the cpu is OCed, and I hadn't been able to run any game or bench at 2.9ghz.

I experimented with the gpu timings with the CPU (P8400-2.26ghz) OCed to 2.9, and found the highest settings I could run 3dmark at are 456-915-1140. This is the same level I can overclock the memory to when at stock cpu, but the core/shaders are underclocked by 14%.
If you figure a proportional increase in core/shader speeds as FSB increases, then this -14% becomes a +10.4% on those clocks which was the highest level I could OC them to at the stock cpu speed (2.26)!

So there's the answer I think: as FSB increases, core and shader clocks do so in the same ratio, but VRAM does not. So, if you overclock the cpu 20%, underclock the core and shaders that amount before overclocking them by whatever percent you found safe before.
Let me know if this works for other people...

With everything maxed out I got 9027 on 3dmark06--not shabby for a 9800M GS and P8400. (Used an external monitor to get full resolution.)

 

I dont know, I just dont think setFSB can OC the GPU, it would need something else to be able to talk to it... I think all it can really do is play with the buss and maybe voltages on the systemboard itself. And being our NB not supported it could be playing with ones it shouldnt thus causing the problem

It is indeed odd though. Sucks because I want to use DirectConsole to OC as I am happy with how high it goes, but I want to use LCDHype for my OLED screen... Arg... Just cant win....

Someone needs to either have DirectConsole so we can change OLED, or combine LCDHype with a new working OC tool, haha

 

why does it say 9800 gt?

 

Old driver. Update to the latest official driver through nVidia's site (179.28), and it'll show the correct name.

 

the pm45 board does NOT have a pci-e lock. by raising your fsb, your also raising the pci-e bus. this doesn't increase your gpu core or shader, those run on their own crystal. a pci-e clock > 100MHz _could_ damage your card.

asm looked like decorated names,

Code:

Undecoration of :- "??0COverClock@@QAE@PAUHWND__@@I@Z"
is :- "public: __thiscall COverClock::COverClock(struct HWND__ *,unsigned int)"

Code:

Undecoration of :- "??1COverClock@@QAE@XZ"
is :- "public: __thiscall COverClock::~COverClock(void)"

Code:

Undecoration of :- "??4COverClock@@QAEAAV0@ABV0@@Z"
is :- "public: class COverClock & __thiscall COverClock::operator=(class COverClock const &)"

Code:

Undecoration of :- "?GetStepCount@COverClock@@QAEKXZ"
is :- "public: unsigned long __thiscall COverClock::GetStepCount(void)"

Code:

Undecoration of :- "?SetStep@COverClock@@QAEKKH@Z"
is :- "public: unsigned long __thiscall COverClock::SetStep(unsigned long,int)"

 

Not sure what all that is you posted but how can you be sure its not PCI-E locked?

Asus customizes a lot of things on a chipset and being they made this notebook to overclock with their own program even I dont think they would make it so you can damage the unit as than they are responsible for it... Also if you note, other users have already tested and it seems that the PCI-E is locked. For starters any programs that are able to pull it still list it at 99MHz including setFSB which I know for one I dont listen too always, but also benchmarks have shown no change in preformance on that buss (which does not have to mean anything itself) but the point here is that setFSB DOES modify or overclock something else that DirectConsole does not do. When you clock them to the same FSB in each program the results are diffrent for setFSB. Some users get crashing or issues in setFSB that they dont get using the same speed in DirectConsole.

Also most if not all newer motherboards now have the PCI-E buss seperate from the main one. Meaning they may or may not be locked but they are not linked. So even if its indeed not locked, I dont think its linked... Maybe thats the problem with setFSB, it by design changes PCI-E buss on the units it does work on but because this MB is diffrent it does something else un-normal and causes an issue.

I know from my experiance all my recent Asus mobo, low end, high end, and corporate & server level have all had the PCI-E buss unlinked/seperated.

Not trying to say your wrong or anything because I have no way to 100% know myself but just seems to be diffrent these days.

 

the realtek lan card times its packets based on the pci-e bus speed. changed speed = bad timing = packets discarded by router. i can only raise it a few points in setfsb before i cant connect. intel boards, afaik, don't and probably won't have a pci-e lock. the 965 didn't, and im fairly sure this one doesn't either.

what i posted above was a quick reference (more for myself than others) of the exported functions in OvrClk.dll. the first step in implementing the same overclock method directconsole uses.

the 2 functions of interest are GetStepCount and SetStep. it's possible we could do what directconsole does and possibly more. i've written a loader for it, hopefully i wont kill my system.

 

Thanks for that. I had issues keeping my network connection up while running setFSB. Now I know why

 

it happens with wireless as well only slightly higher tolerance.


--- offtopic directconsole stuff ---
reference hacks didn't work, so i removed the references altogether.
original OvrClk.dll and modified asm: http://www.zshare.net/download/535804312da78d35/
i don't have my machine to test it on atm, but if anyone with vista wants to compile + replace the OvrClk.dll in their directconsole and post the results, it would be appreciated.

and like usual, i take no responsibility for anything related to use of the above.
--- /offtopic ---

 

I push my CPU and GPU to look my signature. I'm using PLL ics9lpr604aglf on setfsb with this pll you don't have problem with lan and wireless and this pll don't OC GPU

 

what about your GPU, how did you oc that..what clocks and what driver are you using?

 

Thanks for the info, I got the P8400 to 2.76GHz easily with that PLL, might go higher later.

 

just out of curiosity, why are you guys oc'ing your cpu so high? I have enough with the turbo extreme at 2.49 or wtvr it is to run any game out there...

What r u guys doing with your laptop lol? ..

anyways ..still intersted on oc'ing the gpu to stable clocks...so any info would be appreciated.

 

****s and giggles, to be honest =P

Update: P8400 running @ 2.84GHz with core temp of 42C. Gonna leave it there for a bit and see if everything stays completely stable.

 

my max OC core temp is 56C on stress

 

My particular P8400 didn't like being at 2.84 (threw an error and memory dump), so it's back to 2.76.

 

Has there been any progress on this yet?

After following the steps outlined by Thalanx here: http://forum.notebookreview.com/showpost.php?p=4265412&postcount=111 I was able to pull the clocks out of the BIOS and modify them, save and replace them. I have not flashed it yet.

Everything looks great and didn't get any errors so there's no reason in theory that it should not work. I also implemented the 95 threshold and the 200 mem fix. I can upload it it anyone wants.

Based on the G50Vt-X1 209 BIOS

 

Electronicrice, are you talking about OCing the cpu or the gpu? I don't know why anyone wouldn't just raise gpu clocks with ntune.

Also, no matter which pll I use with setfsb my ram clocks increase (guess that's what happens when you raise the fsb--surprise!). I'm pretty sure the bottleneck most folks with the P8400, myself incuded, are running into is RAM related, because I get errors when I run memtest with the cpu over 2.7ghz.

I actually relaxed some of the RAM timings with a program called memset (not very reliable) and got my cpu up to 3.05 (previous best was consistently 2.95), but if anyone knows a program that will allow a reduction on the actual RAM speed I would appreciate hearing about it.

Seeing as DDR6400 is the fastest DDR2 notebook RAM I can't just go out and get faster memory, though I might try upgrading to some "performance" level PC6400 to see if that will run at better speeds than the stock stuff.

 

For the GPU. I want to modify the BIOS because I don't want to have programs running in the background. I just like having it all done by the hardware =)