BIOS modding for GPU OC fun and profit!!

Well, after a bit of work, headaches, and testing, I'm going to start producing a new "A13" BIOS for the 1645. All A13 BIOSes will include:

Unlocked BIOS options (if you don't need them, then don't change anything, then it will be the same as stock)

Updated i7 xxxQM microcode (v3 upgraded to v4)

All 4670 vBIOS is ver A09

Fixed a bunch of DSDT problems:

Syntax errors including:
Misplaced IF (CondRefOf(FPED)) statement moved into a method
Unallocated memory IO addresses fixed
_PLD (physical location description) fixed for USB ports

Linux improvements including:
Fixed the _OSI (Operating System Identification) method to treat Linux as 'XP or newer Windows' instead of the default 'pre-XP Windows' which fixes some things like correctly enabling HPET in Linux

Fixed the _OSC (operating system capabilities) method to improve PCIe support in Linux, fixing a bunch of dmesg errors, though this has the minor side-effect of adding a bunch of devices to the Windows 'Safely Remove Hardware and Eject Media' box. If Windows-only users find this annoying, I can re-break it for you.

Things I couldn't fix:

The 2 error messages that BITS gives you regarding CPU power management. These problems are in the SSDT, and when I tried to modify it, the results were disastrous. You can implement a fix in the DSDT, but the code made the DSDT way too long to reintegrate, so I'm going to give up on this for now.

Thanks again to Kizwan for his help in my efforts.

 

1645_A13_5730_800_1000.zip
1645_A13_5730_813_1067.zip

 

Thanks. Just curious which BITS problems you see, the issue I see is the C1 auto-demotion because it's not enabled and then of course the 0x199 (or something which isn't applicable AFAIK).

You can fix the C1 auto-demotion error either via BITS by enabling it and/or better yet, use TS and set/enable both C1 and C3 auto-demotion which will speed up your SSD 4k writes (mobile platform) without having to play around with power settings (at least 1 other person has confirmed they can see speed improvement besides me, nobody else has commented).

BTW: any chance to get a 'normal' BIOS with the fixes as the 2nd monitor doesn't like OC and the BIOS options don't really do anything AFAIK?

 

Thanks works great. But as i m an only windows user can u rebreak OS capabilities please.

 

What's that mean?

 

Any chance we could get your new bios for the 4670 users, or does the 5730 version work for us?

 

So I installed a SSD and upgraded my thermal paste using AS5 polysyn. Now the max temps I get are 78C and no more throttling.

 

any news about 1640.. i can test different settings if you'd like.. it would be really good.. because on idle computer is making too much noise...

 

It's like a virus definition update but this is for processor.

 

733/1000 works great, thanks seeker

 

Anything you can do to allow for FSB/QPI control to allow some overclocking of the i7?

 

There are many unlocked options which does work. The one I noticed works for you is the C-states options which help improved your SSD performance. Several unlocked options depend on the hardware compatibility but most of them works. It is common for notebook to have FAN which can't be controlled manually. It's like a standard now. BIOS FAN settings follow ACPI specification but the FAN is controlled by a controller (thermal sensors) which override the BIOS settings. The only way to control the FAN manually is by reverse engineer the Dell Diagnostic tool. One of the test run by Dell Diagnostic tool is turn OFF/ON the FAN. So, if anyone successful reverse engineer it will be able to wrote simple program to control the FAN manually. The "normal" & modified BIOS are the same. The only different is the hidden options are now visible with the modified BIOS. With original BIOS, the unlocked options are there but hidden. All of the unlocked options settings always loaded into NVRAM/CMOS whenever you start the computer, whether you're using "normal" or modified BIOS.

 

First, need to identify which BIOS modules hold the PLL registers & routine. It's difficult to identified the responsible module in Phoenix BIOS. So far no success yet.

 

Yeap, good point it does provide options which do work. BTW: on the fan I noted earlier that with the High Performance plan the fan (for me) starts when it hits 55 degrees but with the Power Saver plan it starts at 52 degrees.

This happens everytime on my laptop so I doubt BIOS itself is the answer here but it should (in theory at least) be possible to find what changes that and play with the value. All other fan values remained the same (even the off value).

 

I think it is good idea if seeker_moc can provide a non-overclocked BIOS with below fix integrated in the BIOS for 1645.

 

The 5730 version will work for the 4670, sans the overclock.

I've been unable to get any 1640 mods to work. I can modify it, but both times I did the 1640 owners who tried were unable to flash them.

I'm going to release a non-OCd BIOS today, I was out all day yesterday and wasn't able to.

The temp option does do something. When you go to the Linux ACPI device /proc/acpi/thermal_zone/TZ01 and open trip_points, it normally says this:

Code:

critical (S5):           85 C
passive:                 95 C: tc1=0 tc2=10 tsp=2 devices=CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 CPU6 CPU7 
active[0]:               71 C: devices=FAN1 
active[1]:               55 C: devices=FAN0

When changing the options in the BIOS, it changes the trip_points in Linux accordingly. It's possible that the CPU itself, or some thermal sensor that the BIOS doesn't control, overrides the values set in the BIOS.

As far as overclocking and/or undervolting the CPU, as I understand it, you'd need to modify the same _PSS method that's responsible for the C-states and power management. Like I mentioned earlier, that is embedded in the SSDT, not the DSDT. I have been unable to find where exactly the stock values are drawn from to modify, trying to replace the entire SSDT didn't work, and putting override values in the DSDT makes it too big.

Or, like Kizwan mentioned, you can overclock (but not undervolt) by trying to change the values of the PLL generator (the 'clock' itself) either in the BIOS or with software like SetFSB, though SetFSB AFAIK does not currently support the 1645 (you'd have to take apart your entire laptop to find the model# of the PLL chip, just for the chance that SetFSB might be made to work for you). I don't think there's too much demand for OC of the CPU, considering the heat/throttling problems that many are having with the quad cores.

 

Stock clocks A13:

1645_A13.zip

Also, I did manage to fix the _BQC Linux error messages. I am still unable to make (PCIe / _OSC) work error free in Linux without adding extra devices to the Windows "Safely Remove Hardware" menu, though I have been trying.

 

seeker, will u be releasing more revisions of A13? should windows 7 users get A13?

 

Hey seeker,

Still waiting for a 800/1000 OC for the 1647s.

Thanks!

 

@ seeker man the 813/1067 clocks are running hot..! the system is throttling in 10-15 mins... can i have a 800/1050 without the os capablities thing?? thanks..!

 

My bad, the 800/1000 I made for you was a 1645, I must have mis-read I'll make you one tonight.

 

I'm thinking of making a separate A13B for Windows only users, though there is nothing wrong with using this current A13 in Windows. The only glitch is that allowing the BIOS to pass control of PCIe to the OS causes all PCIe devices (Ricoh card reader/firewire, WiFi card, Network Card) show up as removable in Windows, even though you'd never want to actually eject them. It's not broken, just slightly annoying. The non-standard ACPI routine that Dell used to mask the devices from Windows causes a whole bunch of error messages in Linux.

 

No worries. Just figured I'd remind you. My laptop runs fine with the 813/1067 OC but I want to see if the benefits outweigh the higher temps.

 

What improvements would we (windows 7 x64) users see from A13?

 

Honestly? Probably not much, except for the knowledge that some minor bugs were fixed that you probably would never notice anyways .

 

1647_A10_unlocked_5730_800_1000.zip

 

Hi, really amazing work you've done here

Do you think that fan voltages can be adjusted somehow to make it spin all the time, just slower?

Also do you think it is a good idea to try to change memory frequency to say 1600MHz although they are rated 1333MHz? On desktop most memories are capable of running one step faster...

Thx

 

1. Nothing I've seen so far will allow this.

2. I tried it, but it didn't do anything, just stayed at 1333. Though it does work when you turn it lower. I think that to get the 1600 setting to work, you'd need RAM with 1600 in the SPD, though even that may not work as 1333 is the spec limit of the memory controller in the i7 quads.

 

seeker_moc: I established my stable clocks to be 840MHz core and 960MHz memory. I originaly thought that I'll settle with your "Mild OC Edition", but these clocks just seem to be too much higher and thus tempting...so can I ask you to mod 1 bios for me (with all options unlocked)? I'd really appreciate it...

System specs are in signature.

Thx

 

The Dell download won't let me. Says they are the same version so it won't update. Need to put stock back on.

Thx

p.s. never mind, i found this post by seeker from a while ago. It worked for me

http://forum.notebookreview.com/del...downgrading-your-1645-bios-5.html#post6904588

 

Thank you sir.

 

something didn't go right. I tried to flash twice and got the same results.

 

Is there a way to do the same thing for my XPS studio 1340 with modifyed BIOS update?

 

It worked, or it wouldn't say 800/1000 in the default clocks block. Do you have a software OC tool, like Rivatuner, that starts automatically when Windows boots? It might be forcing your clocks back to stock. The only time that I know of where the GPU Clock and Default Clock should be different is if you're using a software OC tool.

 

No software OC tool whatsoever. Which is correct? GPU or default?

I've used your BIOS OC's before and the MHz values are always the same.

 

Go to the sensors tab, check both the blocks 'continue refreshing in background' and 'create a log file', then run a game or a benchmark. If the log doesn't show it hitting 800/1000 then something is wrong, but I don't know what.

 

Somethings wrong then...

 

PowerPlay turn OFF or ON? Did you try shutdown your notebook & turn ON again. Based on the log file, the clocks are constant which usually caused by PowerPlay off.

 

^ Strange. The PowerPlay toggle isn't where it normally is and I can't locate it.

Edit 1: Powered down and booted back in Win7... still no changes.

Edit 2: Reflashed with the 813/1067 OC. PowerPlay is back. Clocks are right in GPUZ.


Attached log file too.

 

Looks OK now. Probably just a glitch earlier.

 

I was using the 800/1000 OC seeker made for me. Maybe somethings up with that one?

 

Do you all flash from windows or dos?

 

windows..!

 

hey seeker wen will u be back on earth?? The usb bug is killing me... it wont allow me to boot with a flash drive..! when i try login thought usb hdd in bios it boots to broadcom and just exits!

 

I'll make you a Windows only BIOS today. However, your problem sounds more like you have USB Emulation disabled in the BIOS. That prevents USB booting, and should be something you can change in the BIOS menu.

 

1645_A13W_5730_800_1050

1647_A10_unlocked_4670_840_960

1647_A10_unlocked_5730_800_1000

 

^ Thanks. Works great

 

You're welcome. Not like I did anything different than I did last time.... Must have been the magic computer bug faeries.

 

Haha strange.

 

I can't find your other posts so I'm not sure what you've already tried etc, but I get this same thing when I have a USB drive inserted with no OS there to boot from but when I put Ubuntu USB it works just fine (USB being the first device). So.. are you sure the OS is actually installed right?