Jump to content

Recommended Posts

OCed 3840QM and temps under 80C? Madness!

Haven't started really stressing it yet, I'm sure it will hit 90c+ in no time. but I'm optimistic, at 40W I don't need to improve the cooling much, maybe an extra 5-10W would be all that's required.

Link to comment
Share on other sites

Ok, so here are some details about clocks and scores:

[ATTACH=CONFIG]9008[/ATTACH][ATTACH=CONFIG]9009[/ATTACH][ATTACH=CONFIG]9010[/ATTACH]

I was able to push the BCLK to 105.7 so far and got slightly over 3.6ghz. So far so good, but I hope there's a way of getting control over amps and voltages. Funny how the scores are better than those of the new haswell processors including the 4930MX lol

I'm a little confused why your CPU is running 3.4Ghz in 4-core mode (4C). XTU shows a x36 multiplier for 4C. Shouldn't that mean then that with your BCLK overclock you'd be running 3.81Ghz with 4C?

Link to comment
Share on other sites

I'm a little confused why your CPU is running 3.4Ghz in 4-core mode (4C). XTU shows a x36 multiplier for 4C. Shouldn't that mean then that with your BCLK overclock you'd be running 3.81Ghz with 4C?

It should, you are correct. The problem is in power draw, the BIOS only allows about 40W total for the CPU. It would need about 10-15 watt extra to hold 3.6 @ stock BCLK.

  • Thumbs Up 1
Link to comment
Share on other sites

I'm a little confused why your CPU is running 3.4Ghz in 4-core mode (4C). XTU shows a x36 multiplier for 4C. Shouldn't that mean then that with your BCLK overclock you'd be running 3.81Ghz with 4C?

I got strange behavior like that too with the 2920xm in the latitude. 4c turbo never went above x29 even though it was x31. When the CPU hit 70C it would drop to x28, and when it hit 80C drop to x27. x31 ran properly in the clevo with all default settings. The latitude did not seem to have short term turbo properly implemented. He isn't hitting 45W though, while I was hitting 55W, so I guess the BIOS is just being conservative.

Link to comment
Share on other sites

I got strange behavior like that too with the 2920xm in the latitude. 4c turbo never went above x29 even though it was x31. When the CPU hit 70C it would drop to x28, and when it hit 80C drop to x27. x31 ran properly in the clevo with all default settings. The latitude did not seem to have short term turbo properly implemented. He isn't hitting 45W though, while I was hitting 55W, so I guess the BIOS is just being conservative.

Yes and I wonder, since this is an Insyde BIOS, could there potentially be a way of stripping off all the custom menus and revealing the hidden stuff (if it even exists) like we used to do here?

Link to comment
Share on other sites

I got strange behavior like that too with the 2920xm in the latitude. 4c turbo never went above x29 even though it was x31. When the CPU hit 70C it would drop to x28, and when it hit 80C drop to x27. x31 ran properly in the clevo with all default settings. The latitude did not seem to have short term turbo properly implemented. He isn't hitting 45W though, while I was hitting 55W, so I guess the BIOS is just being conservative.

Any chance of unlocking power and/or multiplier control in the 2570P bios? I realize we don't have an unlocked bios to reference. Though m18x series uses Insyde bios. Could they be used as a reference. Eg: http://forum.techinferno.com/alienware-m18x-aw-18/1125-m18x-a03-bios-modifying-via-da_gs-method-hex-registers-editing.html

If that's a dead end and we are working within the stock bios power limits then it will be desirable to know which i7-quad CPU (720QM, 820QM, 840QM) and stepping has the lowest temps to get best performance. Eg: Jacobsson's has a 820QM-ES. Could his hold a x35 turbo mode AND apply similar 105.7 BCLK? If so he'd be hitting 3.7Ghz on 4-cores.

Link to comment
Share on other sites

Any chance of unlocking power and/or multiplier control in the 2570P bios? I realize we don't have an unlocked bios to reference. Though m18x series uses Insyde bios. Could they be used as a reference. Eg: http://forum.techinferno.com/alienware-m18x-aw-18/1125-m18x-a03-bios-modifying-via-da_gs-method-hex-registers-editing.html

If that's a dead end and we are working within the stock bios power limits then it will be desirable to know which i7-quad CPU (720QM, 820QM, 840QM) and stepping has the lowest temps to get best performance. Eg: Jacobsson's has a 820QM-ES. Could his hold a x35 turbo mode AND apply similar 105.7 BCLK? If so he'd be hitting 3.7Ghz on 4-cores.

I'm wondering the exact same thing, haha. Regarding the temps and quads, the 3840QM would be the most power efficient of the high end quads, IMHO. In a screenshot provided by Jacobsson the 820QM is at 3.3ghz on all cores during prime and that's the max it can hold within the 40W envelope. That's my guess.

Link to comment
Share on other sites

I'm wondering the exact same thing, haha. Regarding the temps and quads, the 3840QM would be the most power efficient of the high end quads, IMHO. In a screenshot provided by Jacobsson the 820QM is at 3.3ghz on all cores during prime and that's the max it can hold within the 40W envelope. That's my guess.

If you can a bios dump (inc ME FW) from a 8470P/W or 8570P/W then it would give us a reference to increase power limits. Some were optioned with a factory 45W i7-quad CPU. I'd imagine then getting your 840QM to hold 3.6Ghz (4C), with potential for another 200Mhz via a BCLK overclock (3.8Ghz) would be maxing out the thermals of our little system. That would represent 400Mhz more than the stock settings.

The 8470W/8570W ME FW may also enable the SATA port associated with our mSATA/WWAN slot. If it does then it's a good sign to then solder 0ohm SMT bridging resistors at pins 23, 25, 31, 33 in the WWAN thereby making it a WWAN+mSATA slot.

  • Thumbs Up 1
Link to comment
Share on other sites

If you can a bios dump (inc ME FW) from a 8470P/W or 8570P/W then it would give us a reference to increase power limits. Some were optioned with a factory 45W i7-quad CPU. I'd imagine then getting your 840QM to hold 3.6Ghz (4C), with potential for another 200Mhz via a BCLK overclock (3.8Ghz) would be pretty much maxing out the thermals of our little system. That would represent 400Mhz more than the stock settings.

Not sure if there's someone who would be willing to dump his/her bios for us but we can try... But even if we succeed it would give us very minimal gain while bumping the temps significantly. I'm still hoping there's a way of getting a granular control over settings. With my old M18x I used to be able to fine-tune the power settings in such a way that @55-60W it (840QM QS) would hold +400mhz (40-40-41-42) + extra 200mhz from BCLK (~104) resulting in almost 4.5ghz on a single core! And all that @ low temps.

Link to comment
Share on other sites

Not sure if there's someone who would be willing to dump his/her bios for us but we can try... But even if we succeed it would give us very minimal gain while bumping the temps significantly. I'm still hoping there's a way of getting a granular control over settings. With my old M18x I used to be able to fine-tune the power settings in such a way that @55-60W it (840QM QS) would hold +400mhz (40-40-41-42) + extra 200mhz from BCLK (~104) resulting in almost 4.5ghz on a single core! And all that @ low temps.

Having a 8470W/8570W ME FW dump would allow us to do a hex comparison against the 2570P one. The bytes that set the power limits might then be obvious so could alter them directly. The process would then be alter ME FW dump power limits, flash dump, test thermals with wprime, repeat until find lowest power limit for 4c active + BCLK overclock [or lower if cross thermal limits]. You'd stand to gain an additional 200Mhz and possibly a mSATA slot for your efforts.

Link to comment
Share on other sites

Having a 8470W/8570W ME FW dump would allow us to do a hex comparison against the 2570P one. The bytes that set the power limits might then be obvious so could alter them directly. The process would then be alter ME FW dump power limits, flash dump, test thermals with wprime, repeat until find lowest power limit for 4c active + BCLK overclock [or lower if cross thermal limits]. You'd stand to gain an additional 200Mhz and possibly a mSATA slot for your efforts.

It's the BIOS through setting MSRs that really determines TDP, not the ME FW. It sounds like TDP DOWN is actually flagged in BIOS, reducing your long term TDP from 45W to 35W. I may be able to find where to switch this to TDP UP for a 55W TDP, but it will take quite a bit of staring at assembly. As for 55W possibly being too high, it is much better to limit CPU speed through hard multiplier limits than a power limit. Using power limits to do it causes the CPU speed to jump around a lot, hurting performance.

  • Thumbs Up 1
Link to comment
Share on other sites

It's the BIOS through setting MSRs that really determines TDP, not the ME FW. It sounds like TDP DOWN is actually flagged in BIOS, reducing your long term TDP from 45W to 35W. I may be able to find where to switch this to TDP UP for a 55W TDP, but it will take quite a bit of staring at assembly. As for 55W possibly being too high, it is much better to limit CPU speed through hard multiplier limits than a power limit. Using power limits to do it causes the CPU speed to jump around a lot, hurting performance.

Presumably it's a locked setting. Otherwise surely Throttlestop could alter it.

Unfortunately if the BIOS sets and then locks the TDP, rather than the ME FW, then it's may be a lost cause staring at assembler. That would be the case if modifying the BIOS code was to them be the workaround. Problem there is we've found that HP RSA protect the bios (Elitebook 2530P F.20 or newer) such that flashing a modified one results in no boot.

The remaining workaround then to raise TDP is to flash a 8470P/W or 8570P/W bios dump using fpt.exe/fptw64.exe and hope for the best. Anybody interested in doing that might want to try just flashing the ME FW component first. By some stroke of luck HP may be flagging the TDP in that component.

Link to comment
Share on other sites

I've checked the wwan slot's pin, the power supply and GND pin are the same as msata slot, just pin 23,25,31,33 were cut by HP, so I connect them and test with a msata card. Unfortunately startup bios detect it is not wwan and disable it, maybe is whitelisted. And then I cover pin 45,47,49,51, it still stop at bios detect. So i think it also should remove wifi/wwan whitelist from bios.

Link to comment
Share on other sites

I've checked the wwan slot's pin, the power supply and GND pin are the same as msata slot, just pin 23,25,31,33 were cut by HP, so I connect them and test with a msata card. Unfortunately startup bios detect it is not wwan and disable it, maybe is whitelisted. And then I cover pin 45,47,49,51, it still stop at bios detect. So i think it also should remove wifi/wwan whitelist from bios.

Other thing to try is moving the HP supplied wifi card into the WWAN slot and seeing if the system boots and detects it since it's in the whitelist. If it does, then those additional lines you soldered are not SATA but rather are pci-e.

If above test is not conclusive,try isolating the mSATA pin20 (W_DISABLE) to see if you can boot with the mSATA card.

If still not conclusive AND you are quite daring, then consider obtaining a dump of a 8470W or 8570W BIOS using fptw64.exe (hit WIN+LEFT_ARROW+RIGHT_ARROW on poweron to unlock the flash descriptor to allow this to work). Either of those systems have mSATA support. Then do the same but flash the bios dump to your 2570P and hope it boots. Risky but could be rewarding not only for mSATA but full 45W i7-quad power limits support.

Link to comment
Share on other sites

Presumably it's a locked setting. Otherwise surely Throttlestop could alter it.

Unfortunately if the BIOS sets and then locks the TDP, rather than the ME FW, then it's may be a lost cause staring at assembler. That would be the case if modifying the BIOS code was to them be the workaround. Problem there is we've found that HP RSA protect the bios (Elitebook 2530P F.20 or newer) such that flashing a modified one results in no boot.

The remaining workaround then to raise TDP is to flash a 8470P/W or 8570P/W bios dump using fpt.exe/fptw64.exe and hope for the best. Anybody interested in doing that might want to try just flashing the ME FW component first. By some stroke of luck HP may be flagging the TDP in that component.

So you're saying there's some checksum that causes a laptop to intentionally brick itself if it detects a modified BIOS? That is extremely silly. And I guarantee you that the BIOS directly controls TDP so if the BIOS cannot be modified, then you're stuck at 35W for quad cores. As for crossflashing the BIOS from a bigger laptop to get 45W, it may work, but don't count on it. FPT does not flash over the EC BIOS, so even if the BIOS is compatible with the hardware, it may not be compatible with the existing EC BIOS and result in a brick due to that. If you want to try it safely there are tools you can buy that clamp onto the BIOS chip and can read and write to it without even desoldering the chip. You just need to make sure you have the correct tool for the correct chip type.

And never flash in windows with fptw64. People brick graphics cards all the time trying that with the windows version of nvflash. They brick toshibas all the time as well since they only provide a windows utility to flash. Also if you flash a BIOS from another laptop with FPT or FPTW64 and you have windows 8, you will lose your CD key.

  • Thumbs Up 1
Link to comment
Share on other sites

Ok, so here are some details about clocks and scores:

[ATTACH=CONFIG]9008[/ATTACH][ATTACH=CONFIG]9009[/ATTACH][ATTACH=CONFIG]9010[/ATTACH]

I was able to push the BCLK to 105.7 so far and got slightly over 3.6ghz. So far so good, but I hope there's a way of getting control over amps and voltages. Funny how the scores are better than those of the new haswell processors including the 4930MX lol

Just for kicks I installed windows 7 and ran the same benchmarks and got significantly better scores across the board (wPrime1.55, Superpi, CineBench R10, 11.5, etc) Dunno why but the system performs way better in W7....

Link to comment
Share on other sites

Just for kicks I installed windows 7 and ran the same benchmarks and got significantly better scores across the board (wPrime1.55, Superpi, CineBench R10, 11.5, etc) Dunno why but the system performs way better in W7....

win8 loses track of time when you overclock BCLK in it. When you overclock BCLK 5%, to windows time now happens 5% faster. win7 does not have this bug and thus will properly report a performance increase with overclocking BCLK.

Link to comment
Share on other sites

win8 loses track of time when you overclock BCLK in it. When you overclock BCLK 5%, to windows time now happens 5% faster. win7 does not have this bug and thus will properly report a performance increase with overclocking BCLK.

Buggy indeed. Here's the new wPrime155 run @104.8 :

post-85-14494996123184_thumb.jpg

Link to comment
Share on other sites

win8 loses track of time when you overclock BCLK in it. When you overclock BCLK 5%, to windows time now happens 5% faster. win7 does not have this bug and thus will properly report a performance increase with overclocking BCLK.

Affected users can alter their system to use the platform clock as described here. That is, at a DOS prompt type:

bcdedit /set useplatformclock true

You will have to exit the command window and reboot before this setting can take affect. What this does is it changes Windows so the QueryPerformanceCounter function will be based on a fixed counter that is not influenced by BCLK overclocking. That one line of code should correct this problem.

If you ever want to go back to the original buggered up timer that Windows uses then open up a command window and type in this.

bcdedit /deletevalue useplatformclock

So you're saying there's some checksum that causes a laptop to intentionally brick itself if it detects a modified BIOS? That is extremely silly. And I guarantee you that the BIOS directly controls TDP so if the BIOS cannot be modified, then you're stuck at 35W for quad cores. As for crossflashing the BIOS from a bigger laptop to get 45W, it may work, but don't count on it. FPT does not flash over the EC BIOS, so even if the BIOS is compatible with the hardware, it may not be compatible with the existing EC BIOS and result in a brick due to that. If you want to try it safely there are tools you can buy that clamp onto the BIOS chip and can read and write to it without even desoldering the chip. You just need to make sure you have the correct tool for the correct chip type.

Indeed it's silly. HP RSA protect their bios so users end up having comms redundancy - the system won't support newer wifi cards like say the 802.1ac standard.

I haven't seen a modded Elitebook bios since the Core2Duo days in 2007/2008 which incidentally they backtracked and RSA/bootblock protected as well.

Link to comment
Share on other sites

I've checked the wwan slot's pin, the power supply and GND pin are the same as msata slot, just pin 23,25,31,33 were cut by HP, so I connect them and test with a msata card. Unfortunately startup bios detect it is not wwan and disable it, maybe is whitelisted. And then I cover pin 45,47,49,51, it still stop at bios detect. So i think it also should remove wifi/wwan whitelist from bios.

Other thing to try is moving the HP supplied wifi card into the WWAN slot and seeing if the system boots and detects it since it's in the whitelist. If it does, then those additional lines you soldered are not SATA but rather are pci-e.

If above test is not conclusive,try isolating the mSATA pin20 (W_DISABLE) to see if you can boot with the mSATA card.

If still not conclusive AND you are quite daring, then consider obtaining a dump of a 8470W or 8570W BIOS using fptw64.exe (hit WIN+LEFT_ARROW+RIGHT_ARROW on poweron to unlock the flash descriptor to allow this to work). Either of those systems have mSATA support. Then do the same but flash the bios dump to your 2570P and hope it boots. Risky but could be rewarding not only for mSATA but full 45W i7-quad power limits support.

Further to my message, seems 8470P owners too can't get a test 32GB Samsung mSATA SSD detected here. An official document is quoted showing it does support a 24GB Intel SRT mSATA (HP PN 689954-001). The speculation is then that the mSATA slot is whitelisted.

20130607_891adff472485f738d5eVDLhe6haH2Fy.jpg

Can you get your hands on a 24GB Intel SRT SSD (HP PN 689954-001) to test?

Link to comment
Share on other sites

the system won't support newer wifi cards like say the 802.1ac standard.

I have an Intel 6235 wifi + BT engineering sample that worked in the WLAN slot, it came up as "TEST INTEL" or something like that in the BIOS and booted. Of course the USB lines disconnected, the BT side didn't show. If you can find one (I got mine from eBay), or a different wifi card that you want to use that's an engineering sample, it might work.

When I get some more time I'm going to try and use it and Setup 1.x to unlock the WWAN slot. I don't think it will work because the BIOS halts on startup with a non-whitelist card, but hey, why not give it a shot. I have a USB wifi card with PCI-E interface that works wonders with Linux in vmware.

Link to comment
Share on other sites

I have an Intel 6235 wifi + BT engineering sample that worked in the WLAN slot, it came up as "TEST INTEL" or something like that in the BIOS and booted. Of course the USB lines disconnected, the BT side didn't show. If you can find one (I got mine from eBay), or a different wifi card that you want to use that's an engineering sample, it might work.

When I get some more time I'm going to try and use it and Setup 1.x to unlock the WWAN slot. I don't think it will work because the BIOS halts on startup with a non-whitelist card, but hey, why not give it a shot. I have a USB wifi card with PCI-E interface that works wonders with Linux in vmware.

Good to know that Intel engineering sample wifi cards could potentially bypass our whitelist.

Unfortunately WWAN is USB which I haven't been able to get enabled with Setup 1.x. Only wifi uses the pci-e lines which Setup 1.x has features to unlock. If your into more serious modding then consider running 2 shielded lines from an internal USB port to it instead.

I've also asked BPlus to consider making a ribbon-style cable the interposes between the systemboard and non-whitelisted wifi card so can provide a PCI Reset Delay. Then we could bypass the 104-unsupported wifi BIOS halt message and enable the port using Setup 1.x. That is pretty much the only easy way we could get a newer standard like 802.1ac running on our box.

Well the other would be to run leads that tap off the pci-e lines from our unwhitelisted expresscard slot. Something only an advanced user would attempt.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.