Install a 1066MHz FSB CPU on T61 (including Core 2 Quad)

[RMSMajestic]

@el-sahef
if you go to throttlestop panel and find 'UNLK' button, click it and everything's ready to go nuclear bomb is assembled
looking forward to your test results

P.S. on my 14.1" 4:3 T61p, the heatsink is exactly the same as that on a 15.4" T61p, I don't know why. But which one is better?

@Qing Dao,
I had thermal pads fitted on both CPU chokes and voltage regulator, I'll see if there's any overheating occuring

[FryPpy]

Today I stumbled on this topic: http://forum.thinkpads.com/viewtopic.ph ... 37#p670993 outdated but very interesting research - hacking for performance on T60.
Nowdays we are modding our T61 / frankenpads for performance but better test is super PI.
So lets try our ones with test from that old topic. I is clearly noticed that only raw CPU power is bottleneck for that test. May be flash player of nowdays can use nvidias, but i doubt it.

[triturbo]

@ triturbo: I know that thread and the tests that Ricki Berlin posted there with his HDX9000. Since he got quite far, it encouraged me to get involved with the quad core mod subject on socket p thinkpads, first on mainboards from T500 and R500 (link) which was unsuccessful because of the BIOS, later on the T61. If he is still active on NBR, mabe you can post him a link to this thread: http://thinkpad-forum.de/threads/172581 ... ndervolten
It is more detailed and since he is German too, he will understand everything. I think it should be also doable on the HDX9000, but it is very likely that one needs to use the APIC table loading method via GRUB2 bootloader on these machines (unless someone creates a mod bios with the quad APIC table for these machines too).

He encouraged me as well, we also had quite of a chat. In the end I'm glad that you made it first since it's much better than what I thought and I would never guess that the APIC table needs modification. I'll post it, I guess he would be VERY happy I've already posted a link to here.

I had thermal pads fitted on both CPU chokes and voltage regulator, I'll see if there's any overheating occuring

Thermal pads? Copper heat-sinks would be better. I've improved the stability of my machine with them. From bluescreening every hour or a bit more machine, to rock solid all day long torture I guess this is where your problem lies. A friend of mine with modified dual-core (with my help ) M570TU board got QX9200 to 3.6GHz with only 2 (TWO) phases. So yes, you should be able to run it at least 3GHz stable.

On a side note, we have pretty similar cooling setups Here's mine if you want to check it: https://www.youtube.com/watch?v=mnVi5h1oZkw I'm waiting for new fittings since one of old ones is leaking, so I haven't tested it with water yet.

[RMSMajestic]

Today I stumbled on this topic: http://forum.thinkpads.com/viewtopic.ph ... 37#p670993 outdated but very interesting research - hacking for performance on T60.
Nowdays we are modding our T61 / frankenpads for performance but better test is super PI.
So lets try our ones with test from that old topic. I is clearly noticed that only raw CPU power is bottleneck for that test. May be flash player of nowdays can use nvidias, but i doubt it.

This is what I got with QX9300 only running at 82% of it's original speed
http://forum.51nb.com/viewthread.php?tid=1478474

with external GTX 645

3DMark11 P2781
3DMark13 Ice storm 50831
3DMark13 Cloudy Gate 6367
3DMark13 Fire Strike 1867

Not too bad

[el-sahef]

if you go to throttlestop panel and find 'UNLK' button, click it and everything's ready to go

That does not work for me. I can set multipliers higher than 9.5 in throttlestop and click on unlk but at 100% load with prime95 the multiplier stays at 9.5 according to CPU-Z. This is on 64 bit Windows 7.
Do you have 64 bit or 32 bit Windows and which Windows Version and Throttlestop version did you use? In what order do I have to set mutiplier to lets say 10.5, click "unlk" and "turn on" ?

Also, do you have lenovo power manager installed and what power plan do you use?

BTW: Those @ 51nb having strange stability problems with a quad can test if these are gone when they use a dual core CPU made for FSB 1066 e. g. T9800, E8335 or P8600 if they have one. If the problems persist with such a CPU, the problem is on the RAM side and not on the CPU side. It might also be the case that systems with nvidia GPU instead of intel integrated graphics have a higher probability to be unstable because of the additional power draw and heat of the GPU.

[RMSMajestic]

I use windows 7 64-bit ultimate edition sp1, pwrmgr installed with CPU performance "adaptable", pwrmgr version 3.67. I used throttlestop 4.10, clicked "turn on" then "UNLK"
I didn't try overclocking while undervolting, too dangerous. although this might be what you really want on a T60p casing


P.S. 1. the instability issues @51nb was solved by underclocking the graphics card by 50%, it is a power supply problem.
amwrdfe has also noticed that in device manager there are only two CPU cores, no matter under windows XP or 7, and that occurs on mine as well

2. On my laptops, the sudden restart problems have disappeared for almost one week now, maybe this is a breaking-in period problem? 0_o So confused

[RMSMajestic]

Thermal pads? Copper heat-sinks would be better. I've improved the stability of my machine with them. From bluescreening every hour or a bit more machine, to rock solid all day long torture I guess this is where your problem lies. A friend of mine with modified dual-core (with my help ) M570TU board got QX9200 to 3.6GHz with only 2 (TWO) phases. So yes, you should be able to run it at least 3GHz stable.

On a side note, we have pretty similar cooling setups Here's mine if you want to check it: https://www.youtube.com/watch?v=mnVi5h1oZkw I'm waiting for new fittings since one of old ones is leaking, so I haven't tested it with water yet.

Thanks for your information. But it is really hard to find space for the copper heatsinks in a T61 or T60p. I used thermal pads to conduct the heat to the main heatsink, might come up with a better idea in the future. And now the sudden restart problems "suddenly" disappeared. 0_o Will settle with the settings now, it's summer time, bad time for overclocking

P.S. 1. nice modding there!!!! Sadly there isn't enough space in T61/T60p
2. how did you solder the copper water tube onto the heatsink? what solder did you use?
3. where did it leaked?

[triturbo]

Nice to hear. BTW what kind of adapter you are using? If it's 90W you definitely must think about 120W. My X9100 E0 alone hits 118W under stability testing (Intel Burn Test), but averages around 85W in real world scenarios, and this is without anything graphics intensive.

Thanks
I haven't used solder since it would have been way too difficult to align everything properly. I used ArcticSilver Thermal Adhesive. I was hoping to get my hands on NanoFoil, but it's rather expensive ($193.50 for tin plated 10 pieces of 1.0" x 1.0" (without delivery)), that's why I turned to the Thermal Adhesive.
Outside, but I didn't want to risk. It's a drop every five seconds or so, but better safe than sorry. The new fittings should arrive this, or the following week.

[QWERTY Andreas]

Hi
I just did the quadcore mod on my Frankenpad, however i am experiencing a major problem

When i boot the thinkpad, it just hangs on the BIOS screen.

If i press "F1" it gets stuck on the "Entering BIOS setup utility" screen
If i press "F12" it gets stuck on the "Loading boot devices" screen

Otherwise, the fan spins - the LEDs Work - the thinkligth Works... Can't really figure out what the problem is

Sometimes the screen just goes Black, and nothing happens. I can enter BIOS with the X9000, however.

[el-sahef]

I think the RAM is the most probable reason for this fault.

You can try to solder in the purple connection on this image:
http://abload.de/img/fsb200k8qj4.jpg (225,73 KB image link)

This will force the FSB down to 200 MHz with the quad core. If it works stable with this FSB, then the RAM is your problem (e. g. timings too low).

[QWERTY Andreas]

I think the RAM is the most probable reason for this fault.

You can try to solder in the purple connection on this image:
http://abload.de/img/fsb200k8qj4.jpg (225,73 KB image link)

This will force the FSB down to 200 MHz with the quad core. If it works stable with this FSB, then the RAM is your problem (e. g. timings too low).

Thank you

accidentally i could get into the BIOS with the quad installed. Now it just keeps giving me a Black screen... But sometimes the screen turns on.

I think i will try working a Little more to figure out what it is.

[RMSMajestic]

@Qwerty Andreas
I have encountered exactly the same problem twice, once occasionally and the other time when my QX9300 overclocking failed and I changed RAM just before the next reboot.

I reckon it's the problem with BIOS configurations which you can't change, or the APIC tables.

How I fixed it was replacing the QC with a dual-core 1066mhz CPU, enter the OS, shutdown, the put the QC back, then everything works as fine. I hope this works for you.

Maybe my insulation to the pins on that Q9100 was not firm (it's hard to tell cos the nail top is transparent), sorry ><


@el-sahef and @triturbo and maybe @Qwerty Andreas you wanna look at this as well
I've find out the reason for sudden restart/power loss after my power meter arrived.
Here's a few statistics about power comsumption

T61(Q9100(4*2.26Ghz@1.1625V)+6GB+128GB msata SSD+SXGA screen@highest brightness)
doing nothing: about 30W, occasionally goes to 40W
everest: 75W
everest+furmark: 85W-90W
prime95(large FFT): 85-90w
Intel burn test: This is where the problem starts, with standard level, total power consumption goes up to 100W, then after the first test, laptop restarts itself.
On the second time, before the expected restarting, I've connected 4 HDDs (3 USBs and one ultrabay) to my laptop and run furmark, I should expect an at least 10W increase in the total power consumption, but the total power never went over 105W, and this did not change when a 135W power supply was used


T61(QX9300(4*2.53Ghz@1.1625V)+6GB+64GB SSD+UXGA screen@highest brightness)
doing nothing: about 30W, similar to Q9100
everest: 75-80W
prime95(large FFT): 87-92w, however the temperature is much higher and cannot be controlled without water cooling
Intel burn test: Similar, pass the 1/10 test and restarts, and the total power never went over 105W
On the second try, it just restarts itself

Both of my T63 passed all of the tests above when CPU is undervolted to 1.05V, and highest power comsumption reached 90W on IBT

So, I guess we can conclude that T61 does have some limit in total power consumption. And this is confirmed by amwrdfe, who found the following part on T61 circuit diagram
https://fbcdn-sphotos-d-a.akamaihd.net/ ... 9074_n.jpg
according to him, the total power is limited at 150W, however my tests states a different value (100W). maybe this is the difference between intel graphics and nVIDIA graphics

Another thing which I've noticed, is that throttlestop does not overclock my QX9300, unless the voltage is higher than the original voltage

[el-sahef]

Nice findings, RMSMajestic!

With your hint regarding the voltage, I was finally able to unlock my QX9300 as well. This is how it always works for me:

1) Start CPU-Z.
2) Start Throttlestop 6.00.
3) Choose an unused profile (for me e. g. number 2 "game").
4) Choose multiplier 9.5 and voltage 1.2625 and set the "Set multiplier"-hook.
5) Click "UNLK".
6) Click "Turn On". CPU-Z will show the default voltage (1.175V) despite this.
7) Click the "arrow-up"-button next to the voltage value 1.2625. The value in Throttlestop will not change, but CPU-Z will now show 1.188V.
8) Click the "arrow-down"-button next to the voltage value until it is 1.175. CPU-Z will now show 1.175V
9) Click the "arrow-up"-button next to the multiplier value to set it to 10.0. This had no effect before but if you run a stress test like prime95 now, you will see that the setting has been applied this time and the QX9300 will run at 2.666 MHz. Now you can choose all multipliers and voltages and they will be applied and shown in CPU-Z accordingly.

I had it running at 3.066 MHz (11.5 and 1.175V) but even at fan level 64 the temperature got over 95°C with prime95 in-place large FFTs and the system became unstable. If the cooling would be better, I think these settings would work well. With higher voltages (e. g. 1.250 V), the system seems to crash much earlier (although the temperature has not reached 90°C yet), so I think that keeping the voltage and power consumption down will give better results than trying to crank up the voltage to achieve higher speeds.

But to be honest, the default clock is sufficient for me. I had the CPU running at 1.050 V on all multipliers from 6 to 9.5 for the last three weeks and it was stable. I will most likely leave it like that.

With your hybrid-watercooling and fan level 64 you might be able to run 3.066 GHz stable. For me, mobility is a key feature, so it has to be stable on air cooling. This week, I got a T61-4:3-nvidia-cooler and was able to swap its fan against the one from the T500-ATI-cooler (as announced here and already done here). I removed the two black plastic parts next to the fan since I had the impression that they rather block airflow than guide it.
http://abload.de/img/cimg2189sek03.jpg (47,06 KB image link)

Also, I drilled six holes into the base cover so that the fan gets more fresh air from the outside.
http://abload.de/img/cimg2180zvjjq.jpg (93,94 KB image link)
http://abload.de/img/cimg2184xek79.jpg (88,41 KB image link)

To mount the fan assembly, I sawed out the lower right mounting point for the T61 cooler mounting bracket of a T61 14,1" 4:3 structure frame and glued it to the T60 15" 4:3 structure frame, so that I did not have to modify the fan at the metal bloc in the middle but could use the T61 cooler mounting brackets instead .
http://abload.de/img/cimg2166dukxk.jpg (121,37 KB image link)

At 2.533 GHz (multiplier 9.5 and 1.050 V) and fan level 64, the maximum temperature with prime95 in-place large FFTs is around 78°C.
http://abload.de/img/temps7kyd.jpg (screenshot, 794,21 KB image link, those with slow connections, beware !)

So, I guess we can conclude that T61 does have some limit in total power consumption. And this is confirmed by amwrdfe, who found the following part on T61 circuit diagram

I had a look at this. I do not think that the overall current limit, which is set at MAX8765, is the only problem. The CPU current limit, which is set at ADP3207, might also play a role.

Refer to the second equation on the right on PDF page 18 in the MAX8765 datasheet. If I did not make a mistake, the overall input current limit should be around 133 W (which would make sense, since the most powerful AC adapter has 135 W) because RS1 is 0.01 Ohm (R380 in the T61 schematic) , V_ref is 4.096 V, V_cls should be 3.624 V regarding the resistive divider with R157 and R904 in T61 schematic on PDF page 76.

The CPU limit should be (again, if I did not make a mistake) around 49 A, but I am not 100% sure here because I do not know if the 13 mV/µA from the design example in the ADP3207 datasheet is always the right value for A_lim. Refer to equation 25 on PDF page 24 of the ADP3207 datasheet. R_o is 2.1 mOhm in their design example and according to figure 4 in the intel mobile C2Q datasheet, the loadline slope is -2,1mV/A, so that 2.1 mOhm from the design example should be ok. R_lim is 215 kOhm according to the T61 schematic (R246 on PDF page 82) and if we assume that we can take the 13mV/µA from their design example (I do not know if this assumption is good), then the current limit for the CPU would be 49A.

According to the intel C2Q datasheet, the ICC design target is 64 A for these CPUs. For the Socket P merom C2D CPUs, the ICC design target was 44 A. I am sure that the CPU current limit is the reason for your shutdowns at default voltage with intel CPU burn test (assuming that the cooling was sufficient while running the test) because the "CURRENT-LIMIT, SHORT-CIRCUIT, AND LATCH-OFF PROTECTION" seems to be able to cause such restarts if the CPU current limit is exceeded over some period of time.

After a current limit is hit, or following a PWRGD failure, the
SS pin is discharged by an internal sink current of 2 μA. A
comparator monitors the SS pin voltage and shuts off the
controller when the voltage drops below about 1.65 V. Because
voltage ramp (2.9 V − 1.65 V = 1.25 V) and discharge current
(2 μA) are internally fixed, current-limit latch-off delay time
can be set by selecting the external SS pin capacitor.

On W700, R_lim is 121k which would give a current limit of 87 A if the other values would be the same. This seems a bit high to me and it is very likely that the value for A_lim is a bit lower for W700, since W700 has a three phase design and according to the datasheet, the number of phases play a role here, but I did not fully understand what the impact is on the equation.
Lets assume that A_lim is a bit lower, e. g. 10 mV/µA, then we would get a current limit of 67 A, which seems more reasonable to me.

Conclusion: R_lim (R246 in T61 schematic) should have a lower value to have a higher current limit for the CPU if one wants to overclock it. Another possibilty would be to entirely disable the current-limit latch-off function:

To
disable the current-limit latch-off function, an external pull-up
resistor can be tied from the SS pin to the VCC rail. The pull-up
current has to override the 2 μA sink current of the SS pin to
prevent the SS capacitor from discharging down to the 1.65 V
latch-off threshold.

Otherwise, the best thing to do is undervolt the CPU to 1.050V (or the lowest possible value where the CPU is stable) for all multipliers with throttlestop to lower power consuption and heat.

[RMSMajestic]

@el-sahef
I've had QX9300 running at 80 degree celcius with prime 95 @3.06Ghz @1.175V for about 6 mins on hybrid cooling.
How about drilling holes on keyboard bezel? cos from my revious tests, with and without keyboard makes a more than 10C difference in temperature
after my pump arrives, I might try to make a more efficient and smaller watercooling system with the TEC semiconductors.


Regarding current limit:
I've once pushed my X9100 to 4.2Ghz@1.5V
http://forum.51nb.com/thread-1382851-1-1.html
And did not have any restart problems

According to Intel's datasheet, X9100/X9000 has a peak current of 57A
QX9300 has 64A and the number is the same for Q9100 and Q9000
based on my experience, a X9100 @4.2Ghz is far much hotter than QX9300 @3.06Ghz (I even used liquid air poured directly from an air duster for 4.2Ghz X9100)

But still this question: If I change the R246 to 121k on T61, will it definitely help?
And still, the CPU power limit cannot really explain why 4 harddrives connected to my laptop results in no increase in the total power consumption.

P.S. even a T6670 is more than enough for me But if there's a chance to put a i7 on T61, I'll go for it
P.P.S. I like your third mini PCI-e slot. can you put a HD decoder in there?

@QWERTY Andreas
Have you fixed the problem yet?

[600X]

@el-sahef
How about drilling holes on keyboard bezel? cos from my revious tests, with and without keyboard makes a more than 10C difference in temperature
after my pump arrives, I might try to make a more efficient and smaller watercooling system with the TEC semiconductors.

I don't know if it was just my imagination, but my T60 ran hotter when I installed a solid steel backplate keyboard as opposed to the more modern keyboard with holes. Might be something worth trying out, you even get a lighter ThinkPad that way but you sacrifice keyboard quality.

[el-sahef]

How about drilling holes on keyboard bezel? cos from my previous tests, with and without keyboard makes a more than 10C difference in temperature

Might help a bit, but I am not going to try it, because it will not look that nice . On the bottom side I do not care, but at the upper side I do not want to make any changes to the appearance of the thinkpad.

after my pump arrives, I might try to make a more efficient and smaller watercooling system with the TEC semiconductors.

Looking forward to the pictures of this setup .

But still this question: If I change the R246 to 121k on T61, will it definitely help?

It is an assumption, but there is no guarantee. Taking your experience with the X9100 into account, that idea seems more unlikely to me now. I can not be of much help here, since I do not have this problem. My T61 restarted only once when I raised the voltage and overclocked the processor, but this was obviously related to the overclocking.
Do you cool down the FETs and VRM providing the power for the CPU?

even a T6670 is more than enough for me But if there's a chance to put a i7 on T61, I'll go for it

With the T62 board, this might become possible, at least for Chinese citizen. Before I built my T61 frankenpad, I hoped to get one of the T50 boards, but with the C2Q mod for the T61, it is not that attractive for me any more. There are way too few T50 boards made to serve the demand and there is no possibility for foreigners to get one. I hope that it will be different with the T62 boards if they ever come.

I like your third mini PCIe slot. can you put a HD decoder in there?

Should be no problem to put a BCM70015 card in there (which is a PCIe based card). I tested the slot with a PCIe based WLAN card and a USB based TV card.
http://thinkpad-forum.de/threads/176720 ... Mainboards
USB based cards work without any further changes. For PCIe based cards, you have to connect REFCLK+ and REFCLK- from the PCIe slot for the WWAN card to the third PCIe slot because lenovo seems to disable the REFCLK outputs for the third PCIe slot via I²C at the clock generator. The CLKREQ# signal is shared anyway between the WWAN slot and the third PCIe slot. It also worked when there was a PCIe based WLAN card installed in the WWAN slot and another PCIe based WLAN card in the third PCIe slot at the same time.

On a 14,1" 4:3 T61, only the modem plug and the screw threads for the modem card need to be removed to be able to solder in the third PCIe slot. On a 15" 4:3 frankenpad, the structure frame has to be modified with some parts cut out of a 14,1" 4:3 T61 structure frame and the slot itself has to be trimmed at the left.

[QWERTY Andreas]

@QWERTY Andreas
Have you fixed the problem yet?

Not yet. But the truth is that i dont have a lot of time to really look at it

But the truth is that im a Little afraid that the motherboard is toast;

Most of the time, the screen remains Black when i try to power it up.
Sometimes i get the splashscreen
Very few times i can get into the BIOS, but that doesnt bring me any further

When i insert the X9000 there is a higher chance i get into the BIOS

The truth is i think the mobo is toast, and i have to get another, since i have NO idea of what could cause this behavior (since it does not do the same Things every time)

So unless someone have a magical solution, ill get another board

[el-sahef]

Did you cut off the five pins on the quadcore or did you enlarge the five pin holes in the socket? When you enlarge the holes in the socket with a drill, you MUST hang out the top with the holes in it and put something between the two parts (e. g. a thin sheet of wood) to protect the socket contacts from being damaged by the drill. If you tried to carefully enlarge the holes with the drill without hanging out the top, then you did most likely damage the socket contacts with the drill and the parts of these may now cause shorts inside the socket.

[QWERTY Andreas]

Did you cut off the five pins on the quadcore or did you enlarge the five pin holes in the socket? When you enlarge the holes in the socket with a drill, you MUST hang out the top with the holes in it and put something between the two parts (e. g. a thin sheet of wood) to protect the socket contacts from being damaged by the drill. If you tried to carefully enlarge the holes with the drill without hanging out the top, then you did most likely damage the socket contacts with the drill and the parts of these may now cause shorts inside the socket.

Well, i guess i have my answer. I tried to drill it, but it didn't Work out very well (The insulation was just a Little too large for the holes)

i have now just cutted the pins. Thank you for the clear answer - i think ill get a new board and try over (without socket drilling)

[el-sahef]

If you do not need the GPU Power of the FX570M, you should get a board with intel graphics (or NVS 140M and undervolt it to 0.95V or even 0.9V via mod BIOS), because it would be extremely difficult or rather impossible to cool down the quadcore and the FX570M. Also, you will not have the risk of failure because of the nvidia bug since there is no nvidia GPU that could fail. There was someone at 51nb.com that had problems because of too high power consumption with the FX570M and a quad. He solved these by downclocking the FX570M.

[QWERTY Andreas]

If you do not need the GPU Power of the FX570M, you should get a board with intel graphics (or NVS 140M and undervolt it to 0.95V or even 0.9V via mod BIOS), because it would be extremely difficult or rather impossible to cool down the quadcore and the FX570M. Also, you will not have the risk of failure because of the nvidia bug since there is no nvidia GPU that could fail. There was someone at 51nb.com that had problems because of too high power consumption with the FX570M and a quad. He solved these by downclocking the FX570M.

I like to play games once i a while, so the FX570M is the only choice. I will, however, undervolt it (instead of the insane high clock i had on the old board). I also have a source for post-08/08 boards so ill just jump on that once Again.

[RMSMajestic]

Not yet. But the truth is that i dont have a lot of time to really look at it

But the truth is that im a Little afraid that the motherboard is toast;

Most of the time, the screen remains Black when i try to power it up.
Sometimes i get the splashscreen
Very few times i can get into the BIOS, but that doesnt bring me any further

When i insert the X9000 there is a higher chance i get into the BIOS

The truth is i think the mobo is toast, and i have to get another, since i have NO idea of what could cause this behavior (since it does not do the same Things every time)

So unless someone have a magical solution, ill get another board

The thing with X9000 is, that after Quad Core modifications, if you put it in, it will be overclocked at 3.72GHz, and if IDA is enabled, 4.00Ghz in IDA mode, which is almost impossible. (When I had my X9100, it needs 1.425V to run stably at 4Ghz)

Also have you tried my method? I have a spare P7350 which I can ship you for just the postage.

Similar things happened to me and I fixed it by putting a T9400 in and then put the QX9300 back. Worth a try

[McDonnell]

I have a few CPUs here and was thinking about switching my T61 4:3 to maybe P8700 or P9500, do you guys think there would be a temperature difference from a T7300 and T7500? I have a few unit just waiting for frankenpad mods etc, but thinking a cooler CPU might spark new life into it, if they will run as cool as on T400.

I have a T61 4:3 with T9300 but wont go ahead with that machine as it is my main machine at the moment, and that I am quite satisfied with its temperature and performance.

[RMSMajestic]

@McDonnel, P8800 is cooler than T8100 and T7100, and should be cooler than T73 and T75 as well. But I haven't used any 6MB P9xxx.
a T60p heatsink can keep the temperature of a P8800 below 70C

@el-sahef
I've received the pump but I gave up the liquid cooling circulation. cos the pump is even noiser than 20 GTX 480s or R9 290x running together.
The pump I got was this:
http://www.ebay.co.uk/itm/301179794848

Any better and cheap suggestions?

[el-sahef]

Any better and cheap suggestions?

I think that quiet and cheap exclude each other . There are pumps for water cooling of desktop computers, e. g. from eheim or laing which are somewhat quiet but not cheap. Other sources for alternative pumps might be fish tank accessories (e. g. filters) or camping accessories.
The other possibility would be to run such a pump like you bought it at the lowest possible voltage and see if this is quiet enough. But this reduces water throughput.

A submersible pump should also be quieter than a pump that runs outside the water.

I got the R61 heatsink today which I ordered some weeks ago. It has a bigger surface and the air throughput is also good but as expected the heatpipes are in the way of the keyboard and I do not have an idea how to solve this. The sheet for the chipset is also not in the right position since this cooler is intended for a 14,1" widescreen unit but this could have been solved by cutting that off and use another passive heatsink for the chipset.

[RMSMajestic]

Any better and cheap suggestions?

I think that quiet and cheap exclude each other . There are pumps for water cooling of desktop computers, e. g. from eheim or laing which are somewhat quiet but not cheap. Other sources for alternative pumps might be fish tank accessories (e. g. filters) or camping accessories.
The other possibility would be to run such a pump like you bought it at the lowest possible voltage and see if this is quiet enough. But this reduces water throughput.

A submersible pump should also be quieter than a pump that runs outside the water.

I got the R61 heatsink today which I ordered some weeks ago. It has a bigger surface and the air throughput is also good but as expected the heatpipes are in the way of the keyboard and I do not have an idea how to solve this. The sheet for the chipset is also not in the right position since this cooler is intended for a 14,1" widescreen unit but this could have been solved by cutting that off and use another passive heatsink for the chipset.

Decided to suspend the TEC-watercooling idea until someone has cracked the power limit.
QX9300 @2.53 Ghz@1.05V doesn't realy need water. Air is enough to keep it under 80.

What I've found in the past month:
1. adjusting "clock modulation" in throttlestop gives a BSOD,
and the following might provide a hint for the solution to many problems
2. When disabling "CPU multi-processing" in the BIOS, only one core will be disabled and the rest three will still run (one at 2.53Ghz and the other two 1.6Ghz)
Considering also that only two CPUs being shown in the device manager, I guess other parts/modules of the BIOS does not detect the existence of other two CPU cores (such as the part controlling multipliers and the part reporting/detecting installed hardwares). could also be an explaination for the 2*2.5+2*1.6 problem.

Question: How does throttlestop controls multipliers? How is it different from the default mode?

Or, maybe it's due to the lack of some control circuit.

Any ideas?

[Qing Dao]

Have you tried pin-modding the multiplier and seeing where that leads?

So on the default multiplier everything works fun, but when you try to change it you get problems?

[RMSMajestic]

@Qing Dao
No. I haven't pin-mod the multiplier. The key is to understand how throttlestop adjust multiplier and that might gives a solution/answer to most problems. (as I doubt this is a BIOS problem)

I get problems only when I tried to change the clock modulation in throttlestop, and this problem did not occur on dual cores

Also , I doubt that the VT-d is turned on on the other two cores

[Qing Dao]

@Qing Dao
No. I haven't pin-mod the multiplier. The key is to understand how throttlestop adjust multiplier and that might gives a solution/answer to most problems. (as I doubt this is a BIOS problem)

I get problems only when I tried to change the clock modulation in throttlestop, and this problem did not occur on dual cores

Also , I doubt that the VT-d is turned on on the other two cores

Well, throttlestop works with the QX9300 on different platforms, so I really am curious to know what the difference is. I agree that it is not a BIOS issue. The BIOS code supporting quad cores has been implanted. The problem has to be something with the chipset, FSB, and processor and how they interact with each other. If a Core 2 processor has something going wrong with the FSB or multiplier, it locks at 6x, which is what it seems to be doing for you. For some reason the software is unable to tell both parts of the core what they should do. There is probably something missing from the board that would allow that. I really don't know though.

I think you should try something. Run the computer at the original 800Mhz FSB and try to adjust the multiplier with throttlestop to see if it exhibits the same odd behavior or if it runs fine. You can also try a different C2Q processor, one that isn't an extreme edition, and see how it behaves when you change the multiplier.

There is one other option. There is the Intel Extreme Tuning Utility (XTU). However, it can only be installed if there is BIOS support, so that would require finding what is required in another manufacturer's BIOS, and making another modification to the T61's BIOS.

[QWERTY Andreas]

So im back Again....

I found out that my RAM was 533 MHz and not 800 MHz as the seller said once i bougth them

Anyway; Currently it is starting well. BIOS is working, i can boot from different medias and everything. However i cannot load Windows. When i try to boot from a installtion USB it gets stuck at the screen with the Windows logo.
When i try to boot from the SSD with Windows 8, the same. Neither can i boot a Windows 7 install disk, Ubuntu or anything beyound the low level installation part.

Splash screen: http://i.imgur.com/yfcOV2K.jpg
This is DOS, so something Works: http://i.imgur.com/BKUkD7C.jpg
This is where it is stuck on Windows 8 installation AND boot: http://i.imgur.com/vggu9Yd.jpg (same screen)
This is the BIOS image: http://i.imgur.com/W3LZq1R.jpg

If i try to run a Memtest86+, i can somehow only run it with the "one CPU" option...?

Does anybody have an idea about what migth be wrong?

Best regards