sangemaru

I don't envy your situation. Have you found anyone else with your laptop succeeding in the implementation? Sent from my Neken N6 using Tapatalk

I have a recovery partition left over from my HP 2570p installation. I think it's formatted to FAT32, and thus it's selectable in BIOS when setting to UEFI mode, and allows us to define an EFI boot device on it if we copy the EFI folder to the partition. Once it's added, even though the system is booted in Legacy mode, when pushing F12, we'll have the option to boot it. TL;DR: FAT32 partition with EFI folder and the files inside can be booted. EDIT: now that I think about it, I can put all sorts of goodies in that partition. Now why didn't I think about that before...

Sorry, I'd had a bit of a confusion between you and Co11ider. Your PSU should be enough to run 4 750ti's, lol. The fact that your machine does boot off the GPU (showinng the samsung logo on external monitor) gives us a few clues to work with. I'm partial to considering this as either a whitelist or a signaling issue. If it were the TOLUD you'd be able to boot and only then have problems. That's not to say you won't require DSDT patching. Go inside your BIOS and look for any ExpressCard or PCI Express speed selection options (I'm heavily doubting you have any, to be honest). If you don't have the options, and if using EFI variables isn't an option either, than the only way to force Gen1 signal would be to purchase @Tech Inferno Fan's Setup 1.30 to setup your ports to Gen1 speed. Please do the following: Boot your machine as normal (without EXP GDC). Once in Windows, enter Sleep. While the machine is asleep, plug-in the EXP GDC and then resume from sleep. Do this with PTD set to 7s. Do this WITHOUT the external monitor connected. In the best-case scenario, windows would detect the EXP GDC and (hopefully) not try to give it a Gen2 signal. It's very likely that if it does give it a Gen2 signal, your Windows will BSOD (if it is indeed a signaling issue). Assuming you don't get BSOD, use GPU-Z to monitor the GPU sensors and determine PCIe speed. Then you'd have the daunting task of patching your DSDT to extend the PCIe Root Bridge into a 36-bit address (I don't envy the work you have ahead) It could also be that you can only hot-plug your adapter, as I've heard nVidia users being forced to do this before, I think.

Switched thermal paste to Coolaboratory Liquid MetalPad. The heat builds up much slower now. I can make it through 30% of a TSBench 8thread run before I touch 100C. Basically ran into the dissipation bottleneck. During benches like 3dmark the CPU doesn't even go over 87C. 3dmark11 physics score increased by about 400 points, though firestrike remains pretty much the same, lol. Neither are putting 100% load on the CPU. I'll be drilling holes for airflow when I can get a dremel. God, i wish I had two heatpipes. I'm pretty sure I could max frequency on this CPU if I did. 4703 FS P6854 3dmark11 Also finally touching 4.05GHz (not touching 4.1 because of the 99.7 BCLK I think).

Unfortunately I'm the kind of guy who'll go with 'good enough' and keep the bottom cover on (though I'll drill it and use some copper shims when I get the occasion) and use expresscard instead of the x2.2. I like having my WLAN and WWAN both available and my machine easy to pick up and pack up and hook back in. I'm not sure what benches I should be using to compare the card to a desktop-running version. I'll go look some up.

Also possible the cheap PSU can't deliver the rated rail voltage.

Don't mind me, i'm just feeling like tweaking hardware at the moment Besides, you never know who might be interested JuZ0, i think you can use Kepler Bios Tweaker to undervolt your vbios.

Do you already have the DA-2? If yes, then you can flash undervolted vbios to keep it more efficient. But if you don't already have DA-2, consider using an ATX PSU and pushing your card: GTX 680 Volt Mod | Overclockers Forums There's a LOT of upside there (if you already have, or can get a cheap waterblock).

I have a feeling that you can, though it eould be nice to have someonr else confirm. Tbh, at that point i'd just use an atx psu rather than two big weaker ones.

Yes, the vars in the link. I set them as you suggested, var 0x1D4 set to 0x0 and var 0x1D8 set to 0x1. Oh, I noticed I can also boot to the efi setting utility by copying the EFI folder to my recovery partition. No more USB stick required. When setting Internal Graphics to Enabled without setting them as default, the system boots but gives Code43 on the Intel HD Graphics driver. Once set to IGFX default, everything is peaches. Wow, 3.9GHz across 4 cores appears to be stable but gets hot FAST. After 5 seconds of TSBench 8thread 1024 i get immediate throttle down to 3.5GHz... which is still pretty nice, considering the limitations. EDIT: With CPU OC'd to around 3.9-4GHz average and GPU OC to 1200 from 1050MHz core, these are the new scores: P6813 3dmark11 4684 3dmark FS The overclocked r9-270x scores 2512 points in Furmark FHD benchmark, about 70 points below a stock 7970. Considering it's based on a 7870 Pitcairn, that's quite a bit better than expected. This is on a Gen1 link. I'm drooling thinking of a GTX-970 + PE4C v3.0 combo, but considering how broke I am, and that this whole entire setup including the monitor cost me around 600$... I don't think I'm allowed to complain. I think it'll be a LONG time till a TB solution can be had at similar performance / specs for similar cash.

Hi guys. I want to report that I managed to enable both the internal and external display inside my e6430. I used one efi var to force default gpu to iGPU, the other to set it to enabled permanenty. When it boots into windows, the radeon is recognized and begins to drive the external display. Thank you again @timohour for the save. So long as the setting is gen1, there seem to be not many issues (firefox doesn't like being moved to internal screen for example, but I'll experiment to see what works and what doesn't. Chrome doesn't mind, for example). Oh the beauty of a good ol' fashion 1920x1200 24" display with good color reproduction, I'd forgotten thee. Also, the CPU is MUCH cooler when using an eGPU (not to mention cooling pad), so I can look into increasing power limits and multipliers. Once the multipliers are increased through efi vars, is it necessary to do anything extra or just ramp up the multis in throttlestop?

Alright. First things first: the CTD should be in default position (to the left) and PTD in 7s position in order for the eGPU to NOT be detected at BIOS POST, only after. Done like this, your machine should be able to boot, If your machine can't boot with the adapter attached, something's definitely off. What kind of PSU are you using? When you connect the external monitor, where do you get the samsung logo displayed? On the internal or the external one? Do you have control over your PCIe speed? Setting Gen1 to confirm it is or isn't a signaling issue?

GTX-960 consumes much less. I think the DA-2 could handle an undervolted, maybe underclocked GTX-560ti. I really dislike not being able to fix a powercap for the GPU to stay in.

The single heatpipe heatsink is nowhere near enough to cool my 3740QM properly even at stock settings (will regularly flirt with 100C if both CPU and GPU engage, otherwise around 95-98). The dual-heatpipe appears mechanically incompatible with the single-heatpipe machine, so you'd need to get the dGPU version just for decent cooling. Obviously, I won't be doing much CPU overclocking Wow, 3.9GHz across 4 cores appears to be stable but gets hot FAST. After 5 seconds of TSBench 8thread 1024 i get immediate throttle down to 3.5GHz... which is still pretty nice, considering the limitations.

I based the 285 based on real world load benches GeForce GTX 560 Ti Power Consumption and Thermals | bit-tech.net The average power draw is not the same as max load spikes (and you don't want a 280W spike on an under-rated PSU). Sure, your card may spike lower than others, but you still need to be careful. Consider undervolting the card. if you're gonna use a lower-rated PSU.

I'm curious how much power the spare PSU is rated to provide over 12V rails, especially considering co11ider's situation. Connect PSU to EXP GDC, use EXP GDC power cable to feed the GPU. Cheap PSU's should be avoided. That particular PSU is rated to delver 10A over 12V rails, resulting in a Maximum of 220W (like the DA-2 PSU). However, with it being cheap, there's no guarantee it can even fulfill that requirement. Make sure you match an appropriate PSU to the GPU you'll be using. For example, the GTX-560ti can pull up to 285W (let's say 300W) during gaming load (even though the average is lower, you need to be able to compensate for the spikes). This will require a PSU delivering about 25A over 12V rails.

Don't be that discouraged. A Gen2 x1 connection approaches 90% of the GPU potential. You can overcome that difference by overclocking the card, for example. In hindsight buying the 2133 ram that won't even do 2133 was a waste of cash for me, but meh. My EXP GDC was unfortunately unable to maintain a Gen2 link, but even over an x1.1 link, I could still pull off this and this. This being about a 20% performance penalty from scores achieved under desktop use, under Windows 10, DX12. That's a hell of a boost to have, especially if you can get something like the r9-390X. If using a 220 Dell DA-2 PSU, you can use the GTX-970/R9-280X (maybe undervolted a little)/Fury Nano and other powerful cards that you won't really feel the penalty on unless you're comparing them directly to desktop performance. I think the Broadcom can do much higher speeds but I'm not well educated on wi-fi.

Banggood customer service have offered to refund me and allow me to keep the old adapter. Will confirm when the refund is received. In my opinion, this heavily increases the value of the EXP GDC purchased from this seller for budget-minded buyers (particularly europeans), because there are plenty of people having success keeping a Gen2 link, and should the product not be to spec, one can get their money back (and have a backup gen1 adapter just in case).

The power brick can handle t no problem. Yea, you can get 3GHz across 4 cores. In order to max out the iGPU performance, you need 2133 RAM, but make sure it doesn't glitch your iGPU. It's a pity you can't use EFI vars to overclock the iGPU in the 2570p. Yea, smallest battery + 9-cell is also an option. You can also get pretty much any decent ram and just get an eGPU for graphics-related needs.

Unfortunately I've had to be on the move the past 3 days and don't currently have the eGPU adapter with me. I'll be returning home this weekend and try it out then. The seller's customer service has been in communication, but provide the kind of recommendation we've long since taken care of. I've communicated to them the requirement that the adapter deliver clean and stable Gen2 signaling and requested a refund, in which case I'll be getting a PE4C v3.0 (I'm also debating the merits of PE4C v2.1 instead, but to be honest I can't imagine running the machine with wires sticking out of the bottom. I move my laptop around too much for that). We'll see how this goes. Would there be any way to improve the signal? Suppose I soldered the HDMI connector to the port?

Hiya. Sorry if I'll seem abrupt, I wrote a huge answer and then the browser dumped its cache and I lost everything. - Get the cheapest quad-core you can. You won't go over 3.3GHz quad-core no matter what (more likely around 3.3-3.5 single-core and 3-3.1 quad). - 6 cell doubles your battery life, which won't be that great without tweaks anyway. You'll max out at around 4 hours without the 6-cell; - Get the G.Skill RipjawsX PnP 2133 CL11. Multiple users have confirmed they work well in the 2570p and e6430, while others are either not reported (corsair vengeance) or usntable (Kingston Impact). - Consider getting a e6430 instead. It's subtly superior in many ways.

Ah, now I understand. It's a bit confounding to see these design decisions resulting in such wasted potential.

@timohour do you have an opinion on the settings for setting ports PEG0/1/2 do Gen3 Speed with UEFI vars? The PE4C v3.0 is advertised as offering Gen3, right? The expresscard may be gen2 only, but what about the WWAN/WLAN or that mPCIe half/mini port? Wouldn't that raise the transfer rate to 8 Gbps? (probably around 6.5 after 8b/10b encode/decode). Or am I misinterpreting something? What does the Gen1/Gen2/Gen3 setting on the PEG ports refer to?

@Tech Inferno Fan using efi var 0xB4 to force gen1/gen2 instead of the default auto selection. Thanks timohour for the great analysis. That makes a lot of sense. I don't currently have setup 1.30 access (and think it might become redundant if i get a pe4c if the seller refunds me). The best bet at the moment seems to be to use the efi vars to keep igpu enabled with egpu as primary, retaining gen1 speed until i get another adapter. Will try it out later today. Sent from my Neken N6 using Tapatalk

Yep, tried all manner of intel drivers. Sent from my Neken N6 using Tapatalk