Modified overvolting vbios of GTX460m on G73SW.

Im starting to make a speculation here, for those of you who are on the 1v bios mod and running high clocks, you wouldn't happen to be running that Targus 180W power brick are you?

 

I'm using the 180W power brick yea.

 

It's the g73sw 1v test vbios Saitus posted, and I was on a stock 150 brick

 

hmm. im wondering why my laptop keeps shutting off... im on the .96v bios, and the stock 150 brick.. i watched via kill-a-watt drawing a max of 138 but a constant 130+ in crysis. as soon as it powered off i glanced, it read 127

Im not sure if the brick is failing or something on the board??? brick doesn't fell extremely hot, but it is hot. I'd say somewhere in the 100 degree range.

I tried the 1V bios test as well posted. I can't get my memory to anything above 1480, and the gpu will clock like crazy. i can bench at msi afterburner 880 max setting for a round with 1760 on the shaders...

 

Okay i found something extremely weird. I clocked to the point where the driver crashes and recovers, and back it back down to the same clocks i ran with cyrsis 2 powering down in 20 min game, i noticed an decrease of 10watts on the brick, and it held gameplay without shutting off.. WTH????

i verified clocks to make sure they were set correctly an indeed 870/1740/1480

what the heck would cause this phenomenon? still at .96v

 

Thanks Saltius, will test it tonight.

@Chastity

Your results @ 900/1670 are not really showing a real improvement compared to other users of 2630qm+GTX460m o/c (some get often 16500+ or even 17000 with clocks like 840/1600+)

Anyway, on your 3dmark2006 page, the clocks are shown @stocks (675/1250), that is weird.

What guys do you think is the consumption (TDP) of this 460m@stocks and 460m@830/1660 for example ? Will you think a 460m o/c along with a 2920xm 55W (stock (55W) or o/c (65W+)) will be supported by a mere 150W adapter ?

MSI Barebone 16F2/Erazer X6813 - HM67 - i5-2410m - GTX460m@830/1660 - Crucial M4 128go - 2x2go@1333 Hyundai

 

Just FYI 2920xm uses a tdp of 55w, where 2630qm and 2820qm (and other CPUs in the middle) uses a tdp of 45w.
If you're going to the extreme route youll probably need a 180w power brick I think, althought the 150w should do fine..

 

I knew that already ! The 2920XM ES they are selling on eBay is a bit weird : TDP 45W and 2.4~3.4ghz instead of 2.5~3.5ghz that's why the TDP is lower I guess.

My plan is to get this one and no overclocking it for the moment but later on, I might try ! Cost only £235 without negotiating yet.

It's just when you try to overclock it, you have to increase the TDP on throttlestop and that's why I'm concerned with a mere 150W adapter ! (TDP 45W -> 65W+)

 

http://3dmark.com/3dm11/1912166 3DMark11 P2595

For some reason, 3dmark11 just doesn't like that 900, tho 898 works just fine. Go figure. It's Test 3 that keeps crashing it.

 

Oh it's probably the board that's doing it, Asus may have set a supply limitation on the MXM slot. Most G73/53 users face this problem when they OC the GTX460M.
And weirdly, my laptop even shuts down at 700/1250 at v1.0 after a few hours of Battlefield 3 but this only happens on Battlefield 3 1080p mode, temp was around 71C, could run at less 795+/1600 on other games.

 

im messing around with drivers a little bit, i had 270.xx drivers that would shut off completly in 20 min of crysis 2.

i updated to a beta 285.86 i think it is, game lasted longer then 20 min ended up downclocking to lowest possible and i had to restart computer to release clocks. thats a known bug with the beta... resolution i guess would having set adaptive power management in display driver to maximum performance, and setting pci power management in power options to off.

I've set those but didn't have enough time for testing last night and went to sleep. I'll have some results tonight.

Also a thought asus has a utility that detects what mode your going to be in for games, movies etc... all which have different power options, also a thought.

 

Works great ! Got now 1V, confirmed by GPUZ and nvinspector !

So far, I have raised the core to 840mhz and got +1 degree after few hours of MW2 ! WIll do a bit higher later on !

Thanks buddy!

edit: so far 845/1690 is good under BF3 and 3DMARK

MSI Barebone 16F2/Erazer X6813 - HM67 - i5-2410m - GTX460m@830/1660 - Crucial M4 128go - 2x2go@1333 Hyundai

 

System power down solved. been gaming all night on 1V 870/1740/1470

problem is driver recovery when clocked that high has a bug in older drivers.

I had driver 270 and 280, both acted the same way... system would shut down. The problem relys on clocking the gpu back to stock clocks during a recovery while still in 3d mode.

I noticed on a fluke crash back on driver 270. It would clocked once way down to practically nothing. possibly the lowest clock I ever seen on this machine. something like 53mhz weird.

so i thought ok, something is telling it to clock down. must be driver no responding issue, so i upgraded to 280 still had system shutdown. and another fluke crash where the sound cut out.. HDMI external Monitor Digital audio via hdmi.. so i thought again, alright i'll give the beta 285.38. I don't normally choose a beta and tend to stick whql. I also noticed it on chas's benchs and he doesn't seem to have the problem plus a crazy clocking gpu and ram. and bingo, ive been gaming all night on 1 volt bios no crashing @ 870/1470

So i'd venture to guess the others with the SW's and JH's system shutdown's are related to driver recovery.

 

btw by all night meaning 2-3hrs of crysis 2. i started playing around 9:30ish its 12:33 now, time for bed.

 

Good ! How high the temperature was rising ? Got mine up to 77degrees at the maximum after mw2 and 75 after bf3 beta so still fine !

tonight try 850/1700 but think I can go even further thanks to this 1V modded bios !

MSI Barebone 16F2/Erazer X6813 - HM67 - i5-2410m - GTX460m@830/1660 - Crucial M4 128go - 2x2go@1333 Hyundai

 

hello friend, How are you?
I'd like to say that step by the same problem today!
I even open a specific topic for this, You can take a look
http://forum.notebookreview.com/asu...safety-shutdown-forced-possible-solution.html

If possible, perform the same test I did with the game Mafia 2.

I would like to confirm that your model is also G73JW? Am I right?

Let me know if you can solve this problem of shutdown. this is working!

Check with the Mafia 2, will shutdown occurs.

I've even lost hope, until I was already thinking of selling my notebook.

If I go from 780Mhz, the system shuts down. Especially in Game 2 Maffia.

In GTA 4 when the shutdown occurs from 830Mhz.

I hope very much that you really can help me to end this!

Thank you!

 

Well I don't have Mafia 2 but more importantly BF3 is making my rig shutdown at 700MHz after about 5hours BF3 with the v1.0 vbios.

 

I do not believe it! I think I managed to solve the problem of the shutdowns!
The whole thing is a setting in the Clock VBIOS.

Funny that the SW model does not seem to be affected by the same problem.

More on last, after hours of testing here Configuring VBIOS, get to a perfect setup.

This GPU at 810 MHz and 1550 MHz Memory. Running at 0.912 Volts. Extremely Stable!

Test it on your JW, and then post the results.

Here it works like a charm! This perfect!

stopped the shutdowns of the system here!
before setting up the V-bios, disconnections occur at any value above 720MHz.

I'm trying to improve even more this combination in the VBIOS, I want that is stable at 840 MHz GPU to 0.962 Volts.

 

No sorry i do not have mafia 2, maybe i'll try it sooner or later... I think 3 hrs of crysis 2 at 870mhz warrents that mafia 2 should as well work. considering everyone is getting shutdowns on similiar games. even at 700mhz. I will be throughly testing this machine

MAX TEMP: 86c id say an average was around 82 for the most part. after gaming for an hour 1/2.

My Machine is an G73sw-xt1

 

This does infact impact both SW and JW. I think with the SW it was a little more revealing with the latest drivers of whats going on, causing the crash.. Or maybe i'm wrong, the latest drivers could also reveal on an JW, either way I'm glad were on the right track....

 

Ok so basically MSI Afterburner is a No go. And Nibitor is the way to go! With this way (Shown in the attached) of editing I was able to play BF3 at 860/1600 for 3 hours without any single shutdowns at v1.0 compared to the previous 5 hours shutdown at 700MHz. Interesting but it's the fact Thanks to maykon_helver! +1 Rep!

 

Hmm. Interesting, so what exactly have we found here.. My brain is scrambled from work... I work IT and solve everyone elses problems and go numb on my own hahahaa.

Anyway, so your saying nibitor

So did you manually set clocks via bios and leave afterburner alone? I remember back in the day that helps my sli 285gtx with water cooling, problems occured during clock change. but not if set via bios.

so if i edit my g73sw 1 volt in nibitor and look for clocks.. i haven't started diggin in the bios yet, i hope it will be pretty obvious

 

Yes I set the clocks via vbios.

 

Ok, just moded, clocks are stock at 870/1740

I started playing with the mem to see if i'm still stuck at 1470, i was able to bring it to 1800 before driver crash now, while i was only able to get 1490 previously. HOLY CRAP Theres Ghosts in the box

 

i used nibitor 6.03 btw.. changed clock domain 15 everything that had stock core of 1350 to 1740, and the memory clock from 1250 to 1470. im not sure if it adjusts timings as well but it just unleashed another 300mhz on ram that i could not even think about touching before. I'm using both msi afterburner and that nibitor process you just explained.

I solved crashing issue tweaking with the drivers, but i'm sure this will add a whole nother level of stability

 

OK this is the table found in the vbios maykon_helver provided.

 

Yes that is the same table i just edited.

changing domain 15:0,1,3,4 and 11 to 1740 and memory clock domain 15:5 to 1470mhz

 

Why change 15:6,7 to 740??

Stock reads 540 on 6 and 405 on 7?

 

Perhaps you need to change 15:2, 15:6 and 15:7 to a higher value like 740

 

i haven't touched those cells yet, let me do some testing and see. so far im liking this headroom on the memory.. i may leave those depending

 

Yes stock vbios had 540 on 6 and 405 on 7, not sure what are these values for but that's what he did.

 

Ok, seems pretty solid, except when pushed, but pretty much inline with other overclocks... heres latest 3dmark06

870/1740/1740

dont ask why i clocked the memory the same as the core. pretty much just for fun. can go up higher, but seems like its not the bottleneck

16460

NVIDIA GeForce GTX 460M video card benchmark result - Intel Core i7-2630QM Processor,ASUSTeK Computer Inc. G73Sw score: 16460 3DMarks

 

Hello friend!
I wonder if you could post a test result with 3Dmark06

I was looking to buy a processor i7-940XM, you think it worth it? or would you rather wait and switch to the SW line?

I would like to see the results!

Nha and if possible, when you can! Fassa that test with the Mafia 2, =)

 

There you go, 16571.

And yes XM is worth getting, but I suggest you getting a 920XM, much cheaper than a 940XM, the differences between those is only 1 multiplier and it can be achieved through ThrottleStop.

The reason we get an XM is not for it's stock speed but for its ability to fully unlock the multiplier and power supply. XM's multipliers can be changed through software like ThrottleStop, E-Leet Tunning Utility etc. But we XM people prefer changing it using ThrottleStop because you can control the power supply through it and also set profiles for different situations.

In this test I was using 27/26/25/24 which means My CPU was running at 3192 MHz when 4 cores are being used, this is like running a desktop CPU in a laptop.

But honestly I prefer a SW or SX lol.

 

You took issue with artifacts appearing on screen during testing with 3dMark06?


You took issue with artifacts appearing on screen during testing with 3dMark06?

You have an incredible speed!

could provide its Current VBIOS here to Download? I know I already have a 1V VBIOS here, most wanted to test his.

 

Had some artifacts during the Canyon Flight but driver didn't crash. This is the vbios I used for the test, at v1.0.

 

Absolutly no artifacting on my run... these are some close numbers.

I'm on a SW stock cpu 2.0ghz 2.7turbo.

Runs solid at 870/1740/1740

any idea on overclocking the CPU. thats my next venture

2630qm G73SW-XT1

thats insane on a laptop though 3.5ghz quad i7? or is that turbo?

 

Wait a second mate...what you discovered here exactly?
Using fermi voltage editor you can raise memory clock speed to gain further stability, is that right?

What I'm not getting is: doing a software overclock shouldn't bring you to the same performance level, or by modifying the clock values in bios you lower memory timings thus gaining stability?

What is happening "technically"? Because I'm still stable at 835/1690/1600 but anything above 840 is a gpu crash in a matter of minutes.

So if I edit the memory speed, let's say, from standard 1250 to 1500 I should be able to push my oc further? Is that right or I'm missing something?

I would assume that you're OCing only the memory but not the VRAM right, are you leaving the VRAM untouched?

What you think if I start from gtx560m clocks?

 

That's the turbo speed reported by 3DMark 06 since it reports the highest CPU speed detected but when 4 cores are active each core runs at 3192MHz. Not so sure for OCing the 2nd Gen i7 through setFSB tho but on mine my FSB stays at stock because I can achieve more by just increasing the multipliers.

I thought that OCing the memory means OCing the VRAM? I could achieve the same clocks through Software clocking, just my laptop will start shutting down if I go higher than 700MHz in BF3 so I started to try VBIOS OCing like maykon_helver suggested. I think if you look at the OC editing from the view of the Fermi Clock Domain Editor, the software OC only changes the value of 15:3, 15:4, 15:11 for shader clock and 15:5 for memory clock. What I did here also raised the value of 15:0 and 15:1 to the same OC value as 15:3 and then raise 15:2, 15:6, 15:7 to the core clock value. It did give me some solid runs on games like BF3 and Crysis 2 after flashing the vbios to 860/1600.

 

So basically all I need to do is mod by nibitor 6.03 these values from stock:
15:0,1,4,11 to my stable core/shaders oc 1660 right?
Then I have to mod 15:5 to a stable memory oc like 1500 or 1600?
Also I have to mod 15:2,6,7 to core clock value that is 830 by my end.
Is that correct?

What I don't get is whats the reason to mod 2,6,7? If you overclock the shaders domain shouldn't the core clock raise accordingly to it like it happens when you soft oc?
How are related those clocks to shaders domain?

I'm basing my suppositions by the screenshot you posted earlier which obviously isn't updated to your current vbios.

May I ask for an updated screenshot in order to have things clear before I attempt to mod it?
Also last time I used nibitor I wasn't able to browse the clocks domain editor, it was just all grey boxes. Did mavke updated it?
Thanks again

 

Not sure what's 2,6,7 for but I'm running pretty stable after changing it. And yes please use Nibitor 6.02 or above.

This is a screenshot of my current Fermi table.

 

Im not sure if its nessisary to mod 2,6,7... There has to be a 3d clock for while your on the Battery pack, otherwise any thing like a screensaver, or plugin, etc will crash immediatly wouldn't you think.. I know its supposed to run at half while on battery, but which cell , cells does it get its info

 

Well it was worth a shot. I've modded the clocks with my higher stable oc and flashed it, without having luck at all.
Gpu still crashes the driver in the same manner and time span if I stick to clocks above 840 and greater.
The 0,962v bios wasn't giving me stability at 830/1660/1600, it would artifact and freeze.
1,0v bios gave me stability at 835/1670/1600, and still that magic combination is the most affidable.

I'm core or shader limited. I can use 850/840 for benchmarks but not in games. Probably it's a sign of a binned gpu or something like that, thing that I already supposed before, now it's more than a confirmation.

Flashed back stock clocks at 1,0v and benched 6 mixed runs with Crysis 2 adrenaline and still I'm stable at 835/1670/1600.
Even lowering VRAM clocks to 1400-1450 doesn't keep core at 840-850, so that's my wall.

Oh well, guess I'll be fine with this

 

835 is still an impressive overclock considering your only 35mhz away from the rest of the bunch... did you try re-pasting and sinking? 835vs 870 your only talking 2-3fps

 

3.9fps to be exact... benchmarking resident evil 5 on directx 10 fixed

mem clock plays a role as well

this was 870/1740/1740 vs 835/1670/1740

 

I may do a test anyway.
Can you dump your bios at stock clocks v1 modded mate?
Is it true that I was getting only VGA output with a JW bios, but afaik the g73sw shares the same board as my g53.
Maybe I can pull something out from yours.
Thank you in advance!

 

Yep here is the stock Bios... best i could do is around 840ish and only 1470 on memory...

and the 870mhz is a 1V 870mhz core and 1740mhz memory

 

Thank you. Today i tested the one provided by Saltius, the same one used by Chastity, without luck, it acts the same way as mine.
anyway ill test yours to see what changes, thanks again!

 

so i've made a conclusion,

I've decided to go back to .91v stock asus g73 bios, with modified clocks.

I'm at 840/1680/1600 on .91v

Im not sure why the memory will clock so high now on stock bios, only thing i can think of is from beating the hell out of the VRM's trying to get a stable system without shutting off.

The gpu's temp was fine, the cpu temp was fine. System still shut off. Once vrms get hot thier resistance increases, which also feeds lower voltage. theres no way telling really whats causing this system power down that i can think of shortly of Jtagging the card and running tests.

Not worth a few fps imo, but then again stabalizing the culprit could net me another 60mhz on the core... I'm not sure if i want to go through all that hassle to find the HOT part when its now working at stock voltage.

I don't think its some MXM overcurrent protection since i've seen it power off drawing 115W from the wall and a consistant 138W working fine.

I don't think its the power adaptor either, i had the thing in the freezer running the laptop max load just fine still shuts down.

It has got to be a VRM on the board going into thermal. Which one is the question.. There are a few that arn't heat sinked.

If anyone knows how sensitive the gtx590's are to overvolting you'll see that even under a heat sink they'll blow. Runaway is a disaster.. Happened to a few motherboards of mine years ago. I thought my case was on fire, they glow bright. Computer was still running.

anyway, I haven't confirmed with asus if there is such a thing as MXM slot overcurrent protection. And if there was, wouldn't you think it would trip the second you went into 3D mode running 880 @ 1v

It crashes practically in the same amount of time wether at 880 1V or 850 .96v both of which draw a specific current. It has got to be thermal protection on the boards components. Excluding the gpu

I have solved most of the problem with clocks switching in 3dmode cause a crash and powerdown, but the powerdown still exhists. Just further down the line.

I've also read that disabling the Nvidia HDAudio in the device manager also helps with the cards random downclocking.

We made a fix around that by changing certain clockrates in the Fermi Clock domains, but it still powersdown durring gameplay. One night i thought i figured it out, and then traveled south to game again on Vacation and system powered down same way as previously.

 

Well here i am back.
I took your 870mhz vbios and modified it with my higher OC, which is 835/1680/1600 and once windows booted, i manually raised the core clock to 840/1680/1600 and benched 6 times Crysis 2 extreme DX11 with crysis 2 benchmark tool.

Well guess what, is stable. lol. Dunno why, but it is. In total i made around 12 runs with those clocks mixing maps and DX11/9 and it hasnt crashed. It crashed, however, when i upped the core to 850.

I didnt tested 845 tho, maybe it can work? 5mhz increase, could be 10 the same?

Well for sure, your bios has something special, at least is different from mine. And btw i totally understand you about the VRMS as i supposed this already in the past.
But beside thermal, i was talking about quality and timings of the ram chips themselves. That should explain why some cards go up to 880 and some others die at 830.