timohour

@timohour There's a bug in the latitude BIOS that prevents proper TDP management with an xm cpu (both sandy and ivy bridge). TDP settings only function properly when the BIOS does not detect the power brick ID. When it does not detect the power brick ID, it will throttle the CPU immensely, but you can fix this with throttlestop, and both short and long term TDP values will be whatever was set in the NVRAM. When it does detect a valid adapter ID, you are stuck with the CPU's default long term boost power limit no matter what is set in NVRAM, with short term boost disabled. With an invalid adapter ID detected there is no way to prevent dGPU throttling, but this should not be a problem for you since you use an eGPU.

I can verify that your findings on your XM CPU apply on the QMs and possibly every other CPU.

My Latitude also ignores the Long and Short duration power limit that I set in NVRAM... The only setting that it doesn't ignore is the Long duration maintained, and that's why I got tricked that I fixed the issue. I did some primitive cooling jobs on my E6430 (which I will describe on the E6430 thread) and I can now be sure that I am not temperature limited (95°C top after 15 minutes of Prime 95).

The reason that I got tricked is that XTU may report that the Power Time Window is 128 sec but the actual time is the setting in NVRAM. I set the maximum I could (0xff) which is 255 sec, while the Throttlestop TS 1024M 8 Thread Bench took me just 235.524 (90°C maximum) @ 4C 3,97GHz without throttle because it was inside this window.

I realised that when I tested with Prime95 and after 255 sec of 100% CPU usage I could see that it would instantaneously throttle to x26 for every 5 sec.

Both the power adapters that I have (130W on the docking station and 90W at home) are from Dell (probably valid ID) and I think that I don't have any adapter with invalid ID to check.

Do we have any other options?

...This are my spects

Dell Latitude E6430

i7 3740QM

8 Gb RAM...

@damiancho I can see that you own an i7-3740QM. If you are interested to add +400MHz to your maximum turbo clocks of your i7-3740QM and run it @ 4C 3,9GHz or more, check this thread..

Since you are an eGPU user you are going to see a huge improvement in games/benchmarks with better CPU performance.

My i7-3720QM would score over 1000 points more in 3DMark06 CPU score compared to stock configuration.

I never figured out how to undervolt. The best you can do is use BCLK as high as possible, since BCLK overclocking does not cause an automatic voltage increase, which multiplier increases do.

Thanks!!! That's what I did. x40 would BSOD with an OC over 102 and that's why I finally set @ x39 104,8.

We could overvolt/undervolt our CPU fiddling with the Voltage Regulator but I don't think it would be easy and/or worth the effort

Regarding my non thermal limits I set Long duration power limit 0xB40, Short duration power limit 0xB41 at 100W (0x64) and Long duration maintained Variable 0xB42 to each maximum (0xff) 128 sec.

Is there anything else I need to change?

Awesome! What sort of temps are you seeing running say TS bench at full 4C (unlocked) mode (x38)? Are you being temperature throttled there? Based on temps, is there headroom to consider getting a i7-3740QM (x39 4-core unlocked) or even a i7-3840QM (x40 4-core unlocked).

Oh, and can you link the above post in your E6430 thread? I've headlined a E6430 recommendation over the 2570P as the superior performance platform with not much weight/size penalty at http://forum.techinferno.com/hp-business-class-notebooks/2537-12-5-hp-elitebook-2570p-owners-lounge.html#post33901 linking your excellent thread in the process.

Thermal limits on my i7-3720QM SR0ML

EDIT : Using cooling mods as described here, there is no temperature throttle now even after 6 minutes 4C@3,98GHz Prime95 full load.
The results here remain for reference of the stock dGPU cooler abilities with the bottom lid closed.

From an average of 55C-60C idle temperature starting a 1024M TS bench (8 threads 100% cpu load) until one core reaches 105C and throttles takes:
 

i7-3720QM	_x34_	x35	x36	x37	x38
100 MHz	∞	∞	3' 43"	3' 05"	2' 39"
104,5 MHz	∞	4' 07"	2' 40"	2' 22"	2' 04"

∞ It would complete a 1024M TS bench without throttling.
¹ Used Throttlestop to set the turbo multiplier and HWMonitor to monitor temperature and clocks per core.
² Used freshly applied MX-2 and no other cooling mod. I feel like it would get even better with a small copper shim and/or better Thermal Paste.
³ Test corrected with values from an Air Conditioned Room @ 23°C


i7-3740QM @ 3.8GHz 1024M TS Bench screenshot from XTU
(Throttling shown in the end of the full load graph)

My assumption is that a high binned i7-3740QM (like your first sample) would complete an TS Bench 1024M @ x37 or even x38 without cooling mods... 

An 3740QM would add 104MHz resulting 4C @ 4,07GHz (assuming it will clock as good as this chip) and an 3840QM 4C @ 4,16GHz.
With one of these CPUs I don't know why someone would envy a 4C/8T Desktop CPU...

Even my i7-3720QM won't throttle @ benchmarks or games though since 4C @ 3,8GHz would never be @ 100% load. Most times there would be a dGPU or eGPU bottleneck.

Unlocking i7-37x0QM/38x0QM extra 4 multipliers on a Dell Latitude E6430

For a full guide on how to do this check the first post.

It works!!

Just used the NVRAM tool to set my core limits x40 (1C), x39(2C), x38(3-4C) and my 3720QM and it works!!!



Only problem is that I have to set my long time TDP a little higher cause after around 40 sec anything above x35 (x36,x37,x38) would throttle down to x34,x35.
(Not because of thermal limitatiofns. I could see that my 3630QM would throttle due to thermal reasons when it reached 105C.)

UPDATE: Regarding my long power limit I set [ Long duration power limit 0xB40], [Short duration power limit 0xB41] at 100W (0x64) and my [ Long duration maintained Variable 0xB42] at 128sec (0xff) (not sure if the last is necessary). But I can tell that there is no more TDP throttling now.

TDP throttling kicks in after ~ 6 minutes OF 100% CPU Utilisation! Only solution as mentioned by Khenglish here is to use an invalid adapter and use Throttle stop to overcome the invalid adapter throttle. 


Also seems like max VID for this is CPU is limited to 1,2109V starting @ x37.(check the table on the previous post). This means that there is plenty of room for undervolting here. Any ideas?

Setting single core @ x40 did significally lower my headroom for BCLK OC. It would BSOD if I set anything over 102.4 in XTU. Highest single core OC 4.09GHz!!!



So I decided to limit the CPU up to x39 using XTU for both 1C & 2C, and set my BCLK @ 104.7.
This way I can have a 1C/2C Turbo @ 4.08GHz vs 3.98GHz and a 4C Turbo @ 3.98GHz vs 3.88GHz + I can take advantage of raised BCLK and OC my RAM @2234MHz + I would run similar clocks with lower voltage (BCLK OC does not cause an automatic voltage increase)



3DMark06 CPU score had a great improvement of almost 1100 points compared to the stock chip or an OCed 3630QM and almost 1500 points compared to my 3630QM @ stock.
 

i7-3720QM stock	i7-3720QM OCed

I would definetely love an XM CPU but its pricetag for an OEM product is very high and I am sure that it won't provide much better performance until it reaches its thermal limits (especially in my E6430 which is a small 14" laptop).

A high binned 3740QM or a 38x0QM that i hope it would have a little more headroom in the highest multipliers' voltages meaning that it could do 4C @ 4,07GHz for i7-3740QM/3820QM and 4C @ 4,18GHz for the i7-3840QM + lower temps provided that it could go over 1.21V, would be much appreciated but won't bring huge improvement.

UPDATE: Seems like the highest voltage for all i7-3XX0QMs is limited to 1.2109V. So Practically no actual gain if you go to a 3740QM or an 38x0QM unless it is a superior chip regarding thermal performance. It is possible that every 4C i7 won't be able to do more than 4,1GHz due to insufficient voltage but this needs more testing...

I am very happy with the upgrade. These Latitudes are rare laptops and they provide gaming machine rate capabilities for their pricetag.

I don't think that I have to envy my brothers Haswell Y50-70 and his 4720HQ that he got for 1000€...

I didn't spend more than 400€ (including 3720QM, 8GB 2133MHz, and used laptop) and I got better performance and portability...

There is a guide here on how to overclock your BCLK using XTU.

Hi! my name is Damian and Im from Argentina

This are my spects

Dell Latitude E6430

i7 3740QM

8 Gb RAM

dGPU: Nvidia NVS 5200M

I have a question , I bought a few months ago a pe4h, and recently a Asus Strix gtx 960 4 GB.

And for what I read the pe4h is a 1 gen and my i7 is a 2 gen. It will work anyway?

And there is a guide step by step, to make it work for my laptop? Because I have minimum knowledge in computers

Your PE4H would work with the Dell E6430 but only @ x1.1 (2Gbps). You need newer equipment to use your x1.2 connection your E6430 has.

If you insist using your current hardware and you already own 2 PM3Ns you can connect them both to your laptop WWAN and WLAN port and make an x2.1 (4Gbps) as done here for an x2.2 (8Gbps).

In order to set your EC or mpcie port to gen1 speeds you either have to use setup 1.x or the UEFI variables.

Hello e6430 thread! I'm Miroslav.

Just ordered:

Dell E6430

[email protected]

16GB RAM

no dGPU

GPU: EVGA GTX 960 SC 2GB ITX Version

EXP GDC Beast V7.0

Should I need DSDT Override + Setup 1.30?

@Tech Inferno Fan has a great guide http://forum.techinferno.com/implementation-guides-pc/2747-12-dell-e6230-hd7870-gtx660%40x4gbps-c-ec2-pe4l-2-1b-win7-%5BTech Inferno Fan%5D.html connecting an eGPU to an E6230.

I think this guide also applies to any E6X30 Latitude with iGPU.

If you own a E6430 with NVidia dGPU you have three options

• Use setup 1.x,
  
• Use device manager to disable your dGPU PCI Express Port in Windows.
  
• Use UEFI variables to manually lower your TOLUD to around 2 GB
  

You could also get more info here

I feel that jumping to kepler would be too much. It would definitely need a custom vBIOS.

As for 192-bit I was referencing the GF106 used on the 555m.

I get this.

There is a dead E6430 motherboard on eBay for £36 on eBay. If things change and I can complete a PayPal transaction I may grab it and try. Worst case senario I would try the GF117 mod.

I don't think so since they expect a 128 or 192 bit memory interface.

I may not understand your answer but if this list is accurate it seems almost all Nvidia cards that use the gf108 chip use 128 bit memory interface except for the NVS5200M which uses 64 bit.

The same goes for the GK107 chip which uses 128 bit memory interface on almost all cards except from Quadro 410 and Quadro K500M (64 bit).

example from an gt755M

Also, there is an mobile card that uses one of those two chips (GEFORCE GT640M LE) either gf108 or gk107.

And finally their socket seems identical (at least from the pictures I could find). The only problem that I could find would be the vbios of the card since it is intergrated in my laptop's bios.

Correct me if I am wrong but there is no gk107 chip implementation that uses 192 bit. (unless there is wrong info on the list...)

next is a GF108 core to a GF117 core on an NVS 5200m, which is a full node die shrink of the same core.

Is there any possibility that a upgrade from a GF108 (96 cores 40nm) to GK107 (192 or 384 cores 28nm) would be successful?

They seem to have identical socket.

would it destroy my motherboard if I tried?

Kingston PnP memories running with no tweaks @2133 on a Dell Latitude E6430

UPDATE: I would recommend to purchase 2133MHz CL12 memories. Both sangemaru (3740QM) and me (3720QM) had problems booting with 2133MHz CL11. If you already have CL11 memories and you face instabilities or you are unable to boot, use the UEFI variables as described here to set timings @ 12-12-12-36.

I grabbed a pair of Kingston KHX21S12P1K2/8 @ 2133Mhz. This kit is a 2x4GB @ 2133MHz that I hoped it would have Samsung chips on it but it seems that it doesn't.

The good new are that I didn't have to make any tweak. My laptop would clock @ 2133MHz without changing anything (no XMP settings etc).

Seems like on an E6430 Kingston pnp kits would automatically overclock @ desired clocks (in my situation 2133MHz)

The bad news are that I have to reduce my BCLK OC from 104,9MHz to at least 104MHz.

2216MHz was the highest value I could benchmark this RAM at. Seems like this SKHynix chips that my kit comes with are bad overclockers.

EDIT: This was an issue of my i7-3630QM's memory controller. With my new 3720QM BCLK goes up to 104,9 breaking the 30K barrier.

I am considering trading them with a kit that would have dual rank Samsung chips on it to improve performance over 30K.

dump for anyone interested

Quote

Yes, I cleared the CMOS just in case because I thought that might be what was happening. There are no XMP profiles on this memory.

Ah, I learned something new. I was not aware of this. The Kingston is 8 per side. The Ripjaws worked flawlessly, very stable... just unacceptably slower.

Spoiler

[ATTACH]10907[/ATTACH]

I grabbed a pair of Kingston KHX21S12P1K2/8 @ 2133Mhz cause the shop that I bought them had a picture with dual rank memories, but it seems like these sticks are not like yours...

They are single rank (8 x 512MB) while yours are dual rank (16 x 256MB chips) and they have SKHynix chips while yours have Samsung chips...

The good think is that as soon as I connected them to my E6430 they automatically clocked @ 2133 (no tweak needed).

They are bad overclockers though. I couldn't raise my BCLK more than 104MHz resulting that the best AIDA64 result was @ 2216MHz CL12.

2133	2216

Although I am impressed with the performance gain I am a bit disappointed with the results... I was hoping to go over 30K (dual rank modules) and a bit more BCLK headroom. Maybe I will return them and look for a stick with Samsung chips.

EDIT: This was an issue with my i7-3630QM's memory controller. With my new 3720QM BCLK goes up to 104,9 breaking the 30K barrier @2234MHz.



Time to try some 2400MHz performance.

EDIT2: Still no 2400MHz possible... to be investigated...

Almost everything in the E6530 motherboard was around 3mm off from the E6520 motherboard. I needed to do a lot of dremmel work on the casing to get the board to fit properly. I had to drill holes into the dGPU part of the heatsink to get screws through it, as well as move the fan connector.

This means that such a conversion in terms of fitting is not recommended for the average user.

Did you have any problem connecting other components on the motherboard (e.g. Bt, the Screen, the VGA/ethernet/sound subsystem)?

Quote

Can someone tell me is it possible to use motherboard from E6430 in E6420?

Hello patrykp84!

6430 and 6420 share almost the same chassis. The only problem could be the cpu cooler which seems to be a bit different from a 6420 to 6430. @Khenglish has successfully replaced his dead motherboard on his E6520 with an E6530 one.

Quote

I suspect my 3920xm would overclock fine, but I don't want to switch the CPUs to find out due to how huge of a PITA it was to mount a e6520 heatsink onto a e6530 motherboard.

I have been looking for an old E6420 myself to replace the motherboard with the E6430 one, but then I found a great deal for my E6430 and never gave it a shot.

EDIT : Seems like replacing the motherboard of the Dell Latitude E6420 with the E6430 requires a lot of modifications and is not intended for the average user. Look the following posts for more.

Wow, nice work! I actually have a bigger plan with this cable setup, but that will be for another day or diymodder. Think x16...

I'm intently watching this. Now if an Intel guy could chime in here...

I don't think Intel have something to do with it. I think the E-module II port is a proprietary connector created by Dell.

That's pretty much what I figured. Well, good thing I also ordered a SATA power connector so I should have 3.3V, 5V, and 12V without having to convert, simple enough. I know there are daughter board kits to convert DC jacks if someone wants to go that route.

That's even better. The sata power connectors tend to be a bit fragile though (I have broken a few so far just by plugging and unplugging).

So if you are planning to plug and unplug I would suggest a more robust connector.

It's good to see this thread is live and well tackling that Emodule slot... so onward with my E6430 x4 project:

So I finally settled on my connectors:... So far parts were pretty cheap, the tough part was finding who was selling less than a roll of 7000 connectors. I ended up on Newark's site which luckily gives me blueprints and footprints that I can copy over into Eagle (which I'm also learning how to use, sheesh). Cables are also affordable for this build unlike those proprietary BPlus cables. My 36 pin mini-SAS cable was $10 USD. Once the schematic is done I will run it by you guys before I send off the gerbers to a PCB company. Still lots to do.

Nice to see someone aiming for perfection.

Just a few insights from my experience. You don't have to take them seriously...

• My opinion is that you won't need such a big connector for the 1st port... You can see here that I used only an USB 3.0 cable (8 signals + GND) for my adapter.
  
  
• I guess you already know that but just to be clear for other users, you need to mod the EC to SATA adapter card.The EC to SATA uses a small pcie x1 to sata chip (possibly jmicron or sil) to convert the pcie signals to 1 or 2 sata ports.
  You need to mod the adapter and solder small cables from the TX and RX signals directly on the sata port.Or you could make a custom PCB for the EC to SATA connector.
  You can see more on the EC layout and understand how it works on the EC standard (layout on pg.23).
  
• mSATA to SATA adapters are passive so the only thing you need to be careful is which pin goes where, cause the RX TX signals for the mSATA are different than those of the mini pcie
  
• We haven't found out yet how to enable/use the e-module port 4. Hopefully we do that in time
  
• Also, could you please give us an inside on how are you going to connect the mini SAS on port1?
  

EDIT: Take into consideration that you need power too. Maybe you should consider adding a 24pin PSU connector, so that you can directly connect 3.3V and you don't need to use any 12V/5V --> 3.3V converters

Would you mind explaining that secondary board you have directly wired to the PCIe slot. Knowing you were doing x2 2.0, it looks like you've got some priority wiring for the second lane?

Can you advise then what users who predominantly would like a x1 2.0 (or x1 1.1) link would need to do here?

Post mainly to eliminate any confusion . . .

To avoid any misunderstanding the last image on the Album is for another project regarding x2 2.0 configuration.. I am deleting it so no one gets confused...

Thanks @Tech Inferno Fan for the heads up!

can you post more pictures of the build (adpaterto GPU connection)

More pictures as requested. Sorry for the late response.

Introduction

...

My choice was a Dell E6330. It provides a Gen2 capable express card slot and two free non blacklisted mPCIE-slots (unfortunately not usable for X2).

...

TOLUD change

Suddenly my TOLUD changed from DFA00000 to CFA00000 as experienced by @Tech Inferno Fan for the E6230.

Hello If you are still interested it seems like Dell Latitude E6x30 are capable of x2 2.0 connection as discussed here.

You can also change manually your TOLUD using UEFI variables.

If you are interested for more upgrades on your E6330 check the

http://forum.techinferno.com/dell-latitude-vostro-precision/9690-14-dell-latitude-e6430-performance-upgrades-system-mods.html#post118135

Is the PE4C x2 2.0 capable?

Khenglish said:

I7 3920xm cpu qs 2.9g 3 . 8g 8m e0 step by step-in Processors from Computer & Office on Aliexpress.com | Alibaba Group

He has a cpu-z screenshot there. For some reason it's in xp, but it's legit. I have not tried it in the latitude, but the clevo does load microcode properly.

Well, it seems like I waited too long to purchase. If you find another deal, pls inform us. Thanks.

EDIT: I got a great deal on an i7-3720QM. Not that bad thermally either (compared to the 3630QM)
 

IvyBridge 22nm	x12	x23	x24	x25	x26	x27	x28	x29	x30	x31	x32	x33	x34	x35	x36	x37	x38
i7-3630QM SR0UX	0.8756 12.4	0.9307 20.3	0.9557 22	0.9807 23.8	1.0057 25.7	1.0308 27.9	1.0608 29.9	1.0958 32.1	1.1259 34.8	1.1509 37.7	1.1809 41	-	-	-	-	-	-
i7-3720QM SR0ML	0.8756 11.6	0.8956 18.9	0.9156 20.3	0.9307 21.5	0.9507 23.1	0.9707 24.8	0.9907 26.4	1.0107 28.3	1.0358 30.4	1.0558 32.5	1.0858 35.1	1.1108 37.4	1.1409 40.0	1.1759 43.2	1.2059 47.2	1.2109 48.2	1.2109 49.8

Looking at the steps it takes from x32 to x34 I can say that @ x36 would probably meet its stock Turbo Boost Power Max TDP (~45W) while @ x38 would be under its Turbo Boost Short Power Max TDP limits (under 50W)

had you tried the secondary timings beside the usual first few usually tweaked? I have had good luck on most Samsung chips that have come with my laptops before. Cool J95 uses a real fast CL9 2000mhz

There are not many choises for my machine that I could play with.Even when I am using XMP Profiles it won't comply fully with the XMP Profile settings...

I also don't own TB so I have to flash my RAM on another/older machine... I may don't have the right dumps?

...You have absolutely no way to know beforehand how much farther beyond that your specific chip can go under any given circumstances beyond stock conditions. That's the chip lottery.

Thanks! That makes sense.

I just wanted to know if the problem is the dumps that I use and not the CPU... Is there any way that I can confirm it?

My machine is not exactly an OC beast... Maybe I will try with another CPU...

Ok, the Samsung chip that I have goes up to 2133Mhz (not very stable no matter the timings)! 2000MHz seems to be more stable. It will even go to 2084 with BCLK OC!

Seems though that 2400MHz is not feasible. Even if I set an XMP Profile higher than 2133 (2400, 2666, 2800) it will downclock to 2133. I can't find a reason why this is happening. Maybe my CPU (i7-3630QM)? Why?

Is it possible that i7-36x0 would be locked to up to 2133MHz XMP Profiles?

Also seems like my top limit for Dual Channel is 1866 for now.

The 2GB ADATA chip that I have won't go up to 2133... 2000MHz seems to be its limit and only single channel....

Memory controller???

Any thought would be appreciated.

That would be great,

parts-people.com @ US$ 33 (including shipping) Is this shop legit?

So maybe we need to enable USB3.0 mode with those mod pins (MOD_MD, MOD_SATA_PCIE#_DET), and I tried one of them. And those lanes are direct to CPU, can those be disabled by PCH for example?

Finally got my PE3A to work. So I built an adapter to connect into the Module Bay slot.

used an old mini pcie adapter that I cut so it would fit...

This adapter goes into the modular bay and every TX-RX signal goes on my riser...

Bad News,

It doesn't work for me either.

According to our JSATA2 schematics (below) and as you noticed

there is the MD signal on the SATA power connector. I checked on other schematics and the "official" SATA pinout and I found out that this particular pin (pin 4) is a manufacturing diagnostic pin. Even when I taped this pin my dvd would work.

I took apart an 12,7mm ODD caddy I had laying around just to find out that pin 4 is not connected while pin 1 DEVICE DET# is pulled low? I think.. On the schematics it goes on controller MEC5055-LZY.

MOD_SATA_PCIE#_DET on the other side goes on a ECE5048-LZY_DQFN132 controller (pg 39 UMA) which seems to handle a lot of things. Maybe pulling this pin low could trigger something?

I closed my laptop and I am heading to bed since I am working tomorrow, but I am going to try it in the next days...

Build a $20 Expresscard to PCIe adapter (x1 2.0?)

 

 

Parts needed [URL="http://www.ebay.com/sch/i.html?_from=R40&_sacat=0&_nkw=riser+usb+3.0&_sop=15"] USB 3.0 to PCIe adapter [/URL]starting from $8 or more Expresscard to anything starting from $7 or more. You can even use an old or not working adapter. I used my old PE3A. Flat USB 3.0 cable M-M starting from $5. (optional)(there are cheaper cables but I decided to buy a male-male in case I ruin it. I can also use the other end for a mini pcie adapter :).

My PE3A died a few months back and I have been looking for something to replace it. I didn't want to spent money on more adapters cause my main gaming/video editing machine is an x79 build.

That's why I decided to fix my PE3A-riser setup. But it seems like corrosion took its toll... It wouldn't work and it would also cause my laptop to hang. Since I had a USB 3.0 to PCIe riser laying around from another project I thought that I could solder a USB 3.0 flat cable directly on the ExpressCard pins and then to the adapter. Wiring was one issue.

USB 3.0 to PCIe riser uses 8 signals (& GND). TX+ TX- RX+ RX- REFCLK+ REFCLK- PERST# WAKE#. Detecting which is which was a little bit hard cause the cable that I purchased used a little different colours than the default layout. The following table shows the default colour layout for PCIe TX and RX signals based on the USB 3.0 signal wiring

 

1 RXn

2 RXp

3 GND

4 TXn

5 TXp

 

and the default colour layout for REFCLK PERST# & WAKE based on the USB 2.0 signal wiring.
 

 

1 PERST#

2 REFCLK+

3 REFCLK-

4 WAKE#

 

All you have to do is solder every cable of the above to the correct pin on the EC from the EC standard pg.23.

Don't forget to pull both CPPE# and CLKRUN# to low.

 

 

 

 

 

Can this achieve x1 2.0 signal?

As far as AMD card is concerned, yes. I haven't tested with an Nvidia card but I can be positive that it will work without issues. It seems like with this setup I get better results with my R9-280x than with PE3A + riser setup which seems to had a toll on the signal integrity.

3DMark06 score using an R9-280x and a Dell E6430 along with the custom adapter:



i7-3630QM OC @ 3,56GHz 6GB @ 1675MHz XMP (seems like my 2GB @1333 CL9 ADATA chip doesn't like 1952MHz CL10 :)) R9-280X 3GB STOCK

 

No, I never tried to wire up the modular bay.

That's what we are trying to solve right now. If we could wire up the E-module Bay a fully external x2 2.0 would be possible (using EC port 3 and e-module port4) & maybe an x4 2.0 using ports 1-4.

I finally fixed my PE3A (soldered a flat USB 3.0 cable to the EC connector) and used it with a usb 3.0 pcie riser. Initial tests show that it works with Ethernet Card, Audio Card and an old Nvidia 8400GS I had laying around. Will be testing further to see if it can hold an x1 2.0 connection. If it is stable I will be trying to use E-module for x2.

I cannot justify spending more money on adapters since this is a laptop I use primarily for work. I use my x79 build for gaming/video editing. That's why I work with what I already have.

I heard from Tech Inferno Fan that AMD cards do tend to be more robust with PCI-E 2.0

Seems like a valid point.

@sskillz My dead E6520 mobo still has an intact connector. I could take it off and send it to you.

Have any of you found a schematic of the USB adapter or the modular bay? We could likely narrow down which 4 lines are PCI-E, but to try all combinations on which is which signal would be very tedious.

There is a schematic of the port on pg 28 (UMA) or 29 (Discrete Graphics) of the E6430 schematics. Couldn't find anything closer

Also I have gotten x2 2.0 to work. My 580 allows PCI-E overvolting, so with all socketed conenctions and a PE4H I managed to get to the windows desktop on my latitude. I could move the cursor around and open some folders, but it was not stable enough to run anything 3D. With the cables soldered to the laptop connector it may be stable with the overvolt on the PE4H. I never bothered to get them though since I have the P150EM.

Have you ever tried x2 2.0 with EC and port 4?

I could connect an AMD r9-280x @ x2 2.0 with a custom adapter from ports 1,2 but my Nvidia GTX 460 would hung after a 10-15 sec of benchmarks.