BIOS modding for GPU OC fun and profit!!

Under-clocking worked great. So far, no more throttling while playing over an hour of games (which used to throttle every 10 minutes or so). It even kept CPU temps below 85 (previously were over 95 quite often). VID is now 1.05, stable and no impact on performance.

Thanks for the good direction, Seeker.

 

correction: "undervolting" worked great...

 

Excellent! Would you recommend increments of 50MHz for testing stability/core temps? Lemme know what you think!

Would it be foolish to do a 650/1100 OC? Would the 650 preserve some battery under light load versus 750MHz?

 

i m running on 800/1000 for about 3 months and the battery life averages around 2 hrs for moderate use(wifi+2/3 applications+music+50-60% screen brightness)

 

hey seeker did u try disabling hpet? bcause many forums suggested that disabling hpet reduced the micro shutter in many games?

 

I haven't had a problem with micro stutter, and I'm hesitant to change something without reason. If you're having trouble, you could try the new BIOS, disable the hpet, and let us know if it worked.

 

50MHz increments are a start, but once you get close to max, you need to start trying increments of 10 or even 5. It just depends on how much time you're willing to spend, and how important getting absolute max is to you.

Also, start with your mem clocks, and get them as high as they go without artifact errors in OCCT, then worry about your core clocks afterwards.

I don't think that battery life difference between 650 and 750 core would be too much, as you probably won't be gaming on battery too often, and as long as your power play settings are set to maximize battery life when unplugged, your BIOS won't pick the OC clocks anyways. The OC clocks should only become active when plugged in and playing a game. All other times it should remain stock.

 

^ Good information. I see you're at 813/1067. I'm curious how you came about those numbers.

I want to bump my OC up a bit more since sometimes when I'm playing games, the FPS gets in the 10 range and its impossible to play.

 

Honestly, I'm good up to around 850 core if I stay below 1050 mem, or 835 core at a max of 1105 ram (835/1105 benches higher). I just flashed a bit under max to give myself some headspace. I'm slightly OCD, so I like round numbers. I figured 650+25%=813 core and 800+33%=1067 mem make good numbers that are high enough, but still give me some wiggle room from my max. That's just me though.

 

^ Haha I see. Care to share??

 

You mean the clocks? I can't share the BIOS (you have a 47), but I could make you one with the same clocks.

 

Oh duh. Sure I'll try that out. Was thinking of going with 1100 but baby steps.

 

Yup - FirePro also unlocks the ability to output 10-bit to external displays (I don't think that the communication between the GPU and the built-in RGBLED screen is 10-bit compatible - besides, the panel is only 6-bit + FRC for pseudo 8-bit anyway)... For me, that is the main capability I want to unlock, for photo editing in photoshop...

This is eminently achievable with a soft-mod (which I currently have in place) - but this means that the card is functioning with only partial FirePro capabilities (not all the 3D stuff is properly unlocked with my soft-mod, which is a very basic version - other soft-mods described on forums elsewhere manage to unlock practically all the FirePro capabilities)...

The problem with it is that you must use unsigned drivers etc, and place your windows in "test mode" - this is all fine, and the watermark is easily removed, but then, there is the inconvenience of having to manually OK all the windows updates - just in case windows decides to update your videocard driver (in which case, the card will revert back to being a plain old mobility Radeon!)...

The theory is that BIOS-modding the DeviceID will make Windows install the FirePro drivers - and this should not be a problem, seeing that the hardware is all there and capable of it (the soft-mods and benchmarks using them prove this is true)...

Unfortunately, there seems to be a mechanism in place that prevents the GPU from booting up when the BIOS has been changed in that way... I don't know how/why this is the case... I've temporarily given up the chase, but may try a different DeviceID (i.e. the one for M5725, if I can find it) in the future... Will report back on findings... There is no laptop model that uses the M5725, though, that I can find...

 

just curious but how do u get the bios from the dell site? it's in a .exe form and i dont know how to get bios itself.

 

Try referring to page 7 of this thread, Da_G explains it very well!

cheers,
dave

 

Update re: trying to BIOS alter the DeviceID - I tried changing the DeviceID from 9488 to 9489 (Firepro M5750 - which is apparently directly based on the Mobility 4670)...

Result: Epic FAIL...

Same 6-beeps issue on startup, I still haven't gotten around to finding out what the startup beep code means, but I figure that that won't be helpful in any case...

Oh well ;p Stick with soft-modding!

cheers,
dave

 

There was a post somewhere.. just can't find it but you use Universal extractor, point it to the EXE and output it somewhere else. It creates directory/files and you choose the one which is close to 5MB.

There are better instructions somewhere.. thought it was here, but maybe I'm just blind.

 

Temps overclocked, with MX4 artic paste and zalman after 3 hours of starcraft Before i changed paste gpu was in 80c range.

 

1647_A10_unlocked_5730_813_1067.zip

 

1645_A12_unlocked_4670_a09_780_940.zip

 

Hi seeker_moc.

I'm considering OC my gt425m on my xps 14, still reading the thread trying to figure out the basic procedures, never done anything like this before.

Meanwhile, it would be nice if you can make a modded bios for xps 14 with gt425m so I can jump right ahead into it

 

Sweet! It worked... although during the required restart, the computer idled on boot up (black screen). I waited a bit but nothing happened so I manually turned it off and back on again. This normal?

 

i wouldn't quite say that it's "normal" but weird things - though harmless it seems - tend to happen during firmware changes. when i updated to another firmware, the WinPhlash tool gave me an error message saying that my BIOS was unsuccessfully updated. but it did in fact work, as a reboot showed I was now using the newer firmware.

 

Are you flashing BIOS in windows? Maybe just a glitch since you can boot up normally after that. Can you take a picture of the Advanced section in BIOS menu? I did a minor correction on the previous modified BIOS for 1647 (unlocked hidden options).

I recommend flashing BIOS in DOS mode using " phlash16.exe" tool like this:-
phlash16.exe /x /s bios-file
* change the bios-file to the actual BIOS file, e.g. 1645.WPH.

 

Thanks, appreciated.

Do you know why we're using 300/800 running at 1.2v clocks when external screen is connected? I'm having some issues with the new screen and trying to figure out where the problem is.

I wouldn't call it normal but I've had this happen to me too.. turning it off and restarting showed all ok.

 

Not sure what I'm looking for.

 

I can't see the right side. With previous modified BIOS, you'll get like this:-
(taken from my Studio 1557)

The one you currently use should be fixed already like this:-


I ask because I want to get a visual confirmation. That's all.

 

can i use the A08 from this thread because it looks like this bios is for 4670, but i have 5730.

 

Let me guess, the external screen flickering when in extended mode, yes?

 

Yes, you guessed right .. it's driving me nuts! Any ideas on this.. I've tried Displayport and VGA cables, both show the same behaviour.

 

So I was playing bulletstorm and my performance was terrible no matter how low i set my settings. I opened AMD GPU Clock Tool to check the speed and for some reason it would not go past 300/300. I was on High Performance power plan, too. I had to manually set it to 750/1000 to get it to run nicely. Is there a reason why its doing this? Its not too hot or anything. I think it may be doing this in other games but i thought it was lag.

 

Your new screen is OK. The reason why it does like that because the memory clock is overclocked. I don't know the actual reason why but that is the conclusion after trial & error. I noticed when memory clock is overclocked in BIOS, if the memory clock steady, the flickering stop but when the memory clock fluctuating, it start to flickering. You have two choices to fix this issue:-

1. Set the external screen as primary/main display. I can't remember whether it will work properly in extended mode (internal screen become secondary display) or not but I remember the flickering stop when setting the external screen as the only display in Control Panel (e.g. show desktop only on 2).
2. Only overclock GPU clock in BIOS.

​

I chose option no. 2 on my Studio 1557.

 

Ah ok. There was nothing there so I didn't include it... so I guess your fix worked!

 

Thank you! I'd tried changing Primary screen to the other, no luck with that. Just flashed back to 'standard' (with the options open but no OC) and no flickering anymore. Shame.. but right now I'm just happy it's not doing it anymore!

 

Yeah, I have the same problem but it is not as bad as it was. I really like the overclock but I would prefer to get rid of that flicker.

 

Does flickering stop when you turn off the internal display & set desktop only on the external monitor? If it does, problem solved, for now. Still can play games or watching movies on the external screen.

The other workaround I found which work for me is whenever I work with external monitor in extended mode, I turn off PowerPlay. When PowerPlay off, on my notebook the core & memory clock will stayed at 500MHz & 800MHz respectively. This way external screen won't flicker & I'm able to concentrate on my work.

However, the PowerPlay trick maybe only work on my notebook since the "Default Clock" doesn't change (still stock clock), only "GPU Clock" are overclocked.

 

I kind of use both all the time (and need to much of the time) but I will give the PowerPlay settings a shot... Thanks!

 

Yeah, that happens to me every time with the A12 BIOS. I don't know why, but it still works.

 

You could use it, but it wouldn't do anything for you if the GPU doesn't match. What clocks/BIOS are you looking for?

Are you plugged in, or running on battery? The Windows High Performance power plan should automatically set it, but make sure that in your CCC that PowerPlay is set to Maximize Performance.

 

This worked! Thanks. Anyone know how to create a shortcut to toggle PowerPlay off and on? This would help a lot because going into the Catalyst Control Center to do it every time is a pain (that is, because I usually don't have it running).

 

Quoting myself in case you missed it, seeker_moc.

 

I didn't miss it. I downloaded and unpacked your BIOS. I'm having trouble finding which module your vBIOS is in. I think I found it, but NiBiTor doesn't recognize it, and I don't know how to manually edit a Nvidia BIOS. I'll take another look at it tomorrow.

 

Alright, thanks.

 

Quick question: what's the cause of artefacting with overclocking? It's definitely not temperatures (when I tried faster O/C settings, I got artefact already with temp low 70's - on the other hand, the temp can go up to 80+ with lower O/C settings, i.e. 750/900, and no artefact turns up)...

Is it likely that the culprit is memory, or the GPU clock?

cheers,
dave

 

It's not always the same for everybody, but from my experience, artifacts are caused by too high memory clocks, and instability is caused by too high GPU clocks. Temps are affected by the GPU clocks more than the mem clocks.

 

Thanks Seeker - I'll play around with the clocks then, but not looking particularly good for my particular card - 750/900 is very stable and artefact free... But 780/940 gives me artefacts... 800/1000 gave me a bit of instability as well...

I'll play around, then come up with some sort of working combo! Looks like mine's the least o'c-able of the 4670's out there, at present... (I'm still waiting on the 130W adapter though, which is on the way - previous owner of the laptop I don't think was greatly into gaming, and has a desktop anyway, so they didn't bother with it... It's on the way now, though... And, perhaps, that will enable more o'c than with the 90W...)

cheers,
dave

 

No issues/artifacts so far. Getting temps around 90C at max load for a couple straight hours.

 

Don't change both clocks at once, that will make it hard to determine what is causing your problems. What I'd do is set GPU clocks at 750, then test out mem clocks. Keep slowly turning up mem clocks until you start getting artifacts under OCCT, then dial them back about 5% from max. Then slowly turn up GPU clocks until you start crashing while trying to run 3DMark06. Once again, turn back down about 5% from max. Then run Prime95 and OCCT together to max out heat from all components. As long as you can run 30min without artifacts, or without GPU temps exceeding 90C, then you're good.

 

90C is ok, but if it starts going higher, you might want to dial back a bit, or try some new cooling solution if you haven't done so already.

 

^ Yeah, I'm planning on swapping my odd with a hdd tray so I'll upgrade my thermal paste while I have it apart.