Sony Vaio FW quad-core upgrade

Hello, i am the owner of a sony vaio fw51zf(pm45,t9600) and took the plunge and bought a qx9300, only to find out it does not POST at all in my laptop.
The green power led lights up for 5 seconds, then turns off and after 5 second lights up again and the cpu does get warm, but i have no image output on the screen. It now works fine with the old t9600.

According to this old thread: http://forum.notebookreview.com/har...aio-fw-series-cpu-upgrade-will-quad-work.html , some got a quad to work in these series(i have no problem modifying my heatsink if it boots up), and i would like to know how they got it to work.

Does it have something to do with the unused pins on the quads that have to be insulated in dual core motherboards?
I found this picture: http://abload.de/img/quadcore-pinsy1dm5.jpg that described what pins to insulate in order to get the quad to boot on dual core thinkpad t61 motherboard. Could this be of help?

Any feedback would be appreciated.

 

If you already have the Quadcore, then try to insulate the pins, if they do it with succes, then it's electrical safe to do it.
What do you have to lose ?

 

Here's a topic that would help you with your modification. Quick clarification - It's not just the pins, although it would be easier for you since you have the correct chip-set. I have helped a friend to modify his M570TU dual-core motherboard to support quads, soldering and some traces cutting was involved, I think no pin isolation though (but that's what traces cutting did)

 

What i have to lose is either taking chances breaking pins on the underside of a 120 euro chip i just bought, or taking a drill to my motherboard's socket.
Both not so tempting for an upgrade on a whim.

Hey there triturbo, you are not the only one upgrading loved old machines for no actual benefit .
Having read the thread you linked to before posting, i saw that the trace cutting was for the bsel mod, which i dont need since pm45 already runs at 1066mhz opposed to pm965 800mhz.

There are some things about the GTLREF pins enabling the use of all 4 cores, but i figured that for just booting with the quad, only the pin isolation would be necessary.
If i cut the traces of those pins that i would isolate on a quad, dual cores could probably stop working, since those pin function as vcc/vss for them(though they may be doubled).

Will try the pin isolation tomorrow, will have to find a really small drilling tip first. Does anyone know how to safely remove the plastic top of socket p from the underside contacts?

 

How comes no actual benefit?! They run 0.1% faster

As I said, I helped a friend to mod his dual-core PM45 MoBo, they really are wired differently and require some trace cutting here and there and quite the soldering Just looked at your schematics. Your MoBo is designed to be dual-core only, so you'll have to do all of the things that the ThinkPad guys did, minus the FSB mod. Good luck!

I broke one while trying to be careful, so I wont try to remove the top really.

 

Well i couldn't find any small drill,and removing the plastic top seemed like a nightmare, and my hands were so itchy that i cut the 5 pins off the processor with a small flat screwdriver(half an hour wiggling them back and forth).
Still no dice. I get the exact same behaviour. Maybe its bios related, and any help in that department would be appreciated.

Meanwhile a friend was nice enough to let me experiment on his laptop(well he got a free thermal paste change) .
I installed the processor with the cut pins on his toshiba L500(pm45,hd4570,t4200) and it actually booted.
The processor was properly recognised in the bios at the correct speed, but to get it to boot any operating system i had to disable speedstep(no such option on the vaio bios).
I could boot mini windows xp from a recovery flash drive, but the windows 7 installation would reboot the laptop before i could get the cursor on the screen.
Even on windows xp the laptop did shut down unexpectedly some times(it didnt get very hot though). Could it be GTLREF,APIC related or could the 90w power brick not be enough?

On both laptops, the processors' screw mount were covering the back side on the socket, and that made it even harder to reach the vias to perform the GTLREF mod.
Triturbo, if you could describe what mods you did to your friend m570tu in order to support quads, they could be of help. I have to fetch a 120w laptop power supply for now.

 

I have to combine the Thinkpad's mod, with what we did with my friend's Clevo, because you are right about in the middle. You do have the proper chip-set, but you lack the Quad support chain (M570TU has it, just missing components and a few bridges which had to be cut). You've already did like a 1/3 of the job and eventually we'll get you there, if your Sony's BIOS is not as stubborn as it seems to be.

 

I managed to do it. I used mmtool to insert the latest 1067a cpu microcode into bios(the 2010 one, my bios had the 2008 one), extracted the vbios of the ati card and undervolted the card to 0.9 volts at boot time with RBE(i think the power delivery system could have been overloaded at boot time so i tried it).

It didnt seem to work at first(had the old behaviour), but after 3 power cycles i managed to get the bios logo(without the heatsink attached). I then attached the heatsink and had to powercycle the laptop about 50 times(was about to give up ) to get the logo again. After that though it worked fine.

Now the laptop has no problem booting up and windows 7 recognise and utilize all 4 cores(really did not expect that ). The processor idles at 45-50 degrees celsius and i have not tried to test it yet with prime, because the heatsink does not cover the surface of the cores perfectly(covers about 60% of each core). I will modify my heatsink tomorrow, replacing the copper piece with a bigger one, may also attach an extra heatpipe to the northbridge that is passively cooled, and will hopefully have some pictures.

Something odd is that if i try to undervolt it, i get an instant blue screen or reboot. I thought all qx9300 chips were able to run [email protected] volts. It may have something to do with the 2 phase system my mobo uses. I hope the 5 pins that i broke off the processor(that could be unnessesary after all) wont affect the performance or reliability in any way.

Anyways, i am more than happy to have truly maxed out my beloved laptop(have it since 2009) and would be happy to help anyone with the fw series modify their bios and get a quad working on their laptop. With a 3800 passmark score, this chip aint no slouch, it made my laptop useful again for data crunching. Thanks and happy new year to everyone.

P.S. Triturbo, since you had a look in the schematic, how could the laptop see and utilize all the cores? Is the something we dont know about pm45?

 

Since I don't have your board in front of me, I don't really know what's happening. I don't know which schematics revision I have (is it the latest one?), I don't know which board revision you have (is it standard or somewhat special?)... From everything I know, it's safe to say that it's a Christmas miracle We tried with just those pins isolated as well, but not 50 times. I think that at some point your FW was like "If I don't boot he wont leave me alone", and there it goes it worked I'm very interested to see how it performs, is task scheduling working alright and etc. Joking aside, I really don't know what did it for you as BIOS alone is not enough, but maybe BIOS AND pin isolation is what tricks it. Until more tests and results are presented, I don't really know if it's working properly. I mean, I've looked at A LOT of schematics while researching the PM965+Quad project and I can tell you that every Quad supporting motherboard had this extra chain for second core pair recognizing. The only exception are Acer's C2Q MoBos, but they don't support anything other than Q9000, and maybe Q9100.

The two phase power delivery would come into play if you start overclocking it (why did you get Extreme in first place ), slap some heat-sinks on top of the power delivery components and you are golden. As for the undervolting, yes it plays role there as well. Triple phase systems require slightly less voltage for the same CPU speed, Meaker had it tested with the same CPU on 5739G (dual phase) and 7738G (triple phase).

 

The only "bug" i found is that the extra 2 cores start at 1,6 ghz(hey there acpi ) . After tweaking them with throttlestop i have the laptop running stably at 3,06 ghz across all cores at the stock voltage(1,137v).
The system really feels more responsive now, like i really unleashed my evo 840 ssd's speed. Wprime 32m test run in 14,4 seconds, which is the normal for qx9300 at 3.06 ghz. It really is a miracle .

 

I found the sweet spot to be 2,8 ghz for stability(my power brick is about to blow, have to get a 120w one), attached are some screenies . My laptop uses one of the mbx-189 m764 motherboards(the latest and best that fits in a vaio fw), cannot find the exact part with the hd4650 and1 gb gdrr3 memory right now, will post after opening the laptop for modifying the heatsink.

 

You definitely need 120W one Actually even more if you ever decide to modify the cooling and clock it harder.

 

Didn't that unexpected success spark your interest in putting the little quad beast in your 5920g again?
You were the inspiration for me trying this sort of crazy 1% success rate upgrade, and i'm sure you won't put the 5920g endeavors to rest just yet.
(Plus the QHBQ cpu you got can go for a hefty penny on ebay, they are extremely rare nowadays and really sought after)

 

It's in the cards, but my focus is on the 8740w as of now. Sometime down the road, it would be more for fun, than anything else. It is pretty balanced machine the way it is now. I really wanted to make more powerful (and DX11 capable) GPU and be the one and only 7 year old machine to run Watch_Dogs for example (a lot of money wasted in the process) So as I said, the 8740w is my main focus, and the 5920G would get Quad, but it wont be anytime soon. I'm happy that something good came out of a crazy idea, enjoy it

 

I tried to modify my heasink today. What i did was remove the copper piece from the heatpipe with a blowtorch, invert it by 90 degrees so it covers all the cores, resolder it with a vice and the torch and reattach the screw posts(3 of 4) with epoxy resin. I say tried, because the results are not as nice as i expected, the cores are loaded almost equally now, but the total temperature has risen. It goes up to 92 degrees in throttlestop in the short test, but goes down fast, and idles(all cores) at about 53 degrees . Here are the pictures(the first one shows the stock cpu cooling does not cover all cores horizontally). Might get another heatsink and try using just a copper shim next time.

 

Are you sure the pipe makes proper contact with the copper plate? Is the pipe expanded, or still flat?

 

Well, it took a while to put the final words on this upgrade, i guess its time.

The reason the computer had any issues with the quad was because of a northbridge problem (i guess bga solder stuff) which i tried to fix on a local reballing business, that got my motherboard completely fried(and didn't even reimburse me for the damage). Because i love this damn machine , i bought a new motherboard and the quad worked fine,on the first try, without any mods(no new microcodes in bios,no undervolting gpu,fully stock bios,even the pin-mod might have been unnecessary). I got a new copper fin heatsink with fan, used a large and thin copper shim to cover the covers and temps are fine(45c idling,78c max in summer). The only thing that needs to be done on the user end is running throttlestop on startup to adjust cores 3-4 to 2.53 ghz(they are stuck at 1.6 at boot time). As a bonus, i slapped some ram heatsinks on the phase inductors on the mobo, to be completely safe.

From my experience, the vaio fw series are indeed pretty much compatible with quads without any crazy mods. The only thing bothering me is that i can not overclock the processor after i upgraded to windows 10. It solved some minor issues i was having on 7, but throttlestop can now only be used for undervolting and for setting the multiplier to the stock 9,5(running 2,53@1,075 volts right now). If i up the multiplier to 10-10,5 the change do not get applied, and if i up it to 11, the fsb speed gets lowered so that the speed is still 2,53. After 11,5 i get blue screens. I dont have any idea as to what may be causing such behaviour, since i could overclock fine at windows 7. Im still okay with it, 2,53 ghz over 4 cores is still pretty fast for anything i use the computer for, VMs still run blazingly fast. I'm fully satisfied with the upgrade and would recommend it if someone has a similarly specced vaio and can find a cheap qx9300.

 

Dear pchatzop, I love your upgrade for the FW machine. Can you post an image for your new heatsink for QX9300? Thanks!

 

Besides pchatzop, is there any other guy can upgrade your Sony FW notebook to a Quad Core cpu?

 

I have a Vaio FW290 (T9600, PM45), too, since 2009 and still like it a lot. The screen is still a good one nowadays. I would also like to put a QX9300 or at least a Q9100, but it's not clear for me the required changes and I could not find too much info on the internet beside this thread.

pchatzop seems to be the only one who successfully added a quad to a Vaio FW but he mentions some mods at the beginning of the thread and then added that most of them were done due to a northbridge problem and then changed the mainboard.

pchatzop, are you still here to help us with some more information? Is the pins removal still needed? Do we need to change the Bios? I understand that the heatsinks need be enlarged, which is obvious due to the increased TDP.

Thanks a lot!

 

Hi triturbo,
I have a fw56gf since 2009 and somehow recently I decided to upgrade the cpu (which right now is a t9600). I installed the quad one ( which being q9100) but it didn't post. the power button is on, the cpu gets hot (although the cooling system is installed completely), but nothing else. I tried the quad cpu on another laptop (which was a Dell Vostro 1510) and the same result there except for the error code which was showed by flashing lights on the keyboard, which translates to "BAD MOTHERBOARD". :-?
could you please help me with "pin removal" method? or "bios mod" method? or the combination? in fact any suggestion for the laptop to accept the quad cpu.
thanks a lot.

 

Here's the topic. Picture about the pins is in there. Good luck!

 

Thank you all.
I have a Sony Vaio VGN-FW21E with an Intel P8400 and having found an Intel Q9000.
I tried to cut the pins as reported in the previous post.
I replaced the cpu, added a thin copper plate larger than the current one to cover well the new cpu and everything worked on the first attempt, without changes to the bios, the motherboard or the settings of windows 7