Tech Inferno Fan

I would appreciate if you could upload the version 8 for Series-7 chipset systems to say mediafire.com and link for me to look at. Thank you.

Have a a m11xR2 + [email protected] implementation here using a PE4H+PM3N. m11xR1 should be just as easy.

Yes, x1 1.0 on an older Intel chipset with an AMD card gives a choppy/stuttering gaming experience. REF: Kizwan's HD5870@x2 and x1 game benchmarks. You really need at least x1E mode to see things improve. Not sure if switching the LAN card from D0->D3 state would then allow it to not act as a second lane if you switched port5 into x2 mode. Certainly finding the power pin on lan and cutting it would take it off the pci-e bus. Or else x2E mode using the 2x mPCIe slots + DIY eGPU PE4H 2.4a hardware is the other performance fix. The aforementioned Lenovo T420s/T430s have four+ performance benefits for eGPU implementation over your T61. This makes the gaming experience considerably better than either the onboard iGPU or the NVS4200M/NVS5400M dGPU options Lenovo provide. 1. a much faster CPU 2. a pci-e 2.0 expresscard/mPCIe slot giving double the bandwidth over your pci-e 1.1x one 3. negotiate a full duplex link with a AMD card, again giving ~ double the bandwidth 4. Have an Intel HD3000/4000 iGPU. NVidia's Optimus driver then engaging pci-e compression on a x1 link AND allows a transparent internal LCD mode. If using an AMD card, the iGPU allows use of LucidLogix Virtu to render the image using the eGPU but display it on the internal LCD. 5. i7 T430s has a Thunderbolt port. It's 10Gbps downlink giving slightly more than double the bandwidth of expresscard 2.0. Only problem there is the $180 BPlus TH05 has been recalled so cheapest eGPU enclosures then start at $320 (Sonnet Echo Express SE, OWC Helios). Those cannot host dual-width cards so would need to rip the internal board out and patch ATX power through. Also add $50 to purchase the Thunderbolt cable.

Great mod there. You note the dips in performance here and there. There are two solutions I can present to deal with that, netting noticeably better performance out of your HD7750: <strike>1. Since the dock is using port5, you can use the Setup 1.x software to switch that port into x1E mode and see 15-30% better performance. The theory is that older Intel IO chipsets such as your ICH8M only negotiate a half-duplex link with a AMD card. Running x1E appears to fix the problem and get full duplex performance.</strike> <-- miro_gt correctly confirms below this won't work since LAN is wired to port6. 2. Use the DIY eGPU hardware to do a 2 mPCIe x2E configuration using port1 (WWAN) and port2 (wifi). REF: http://forum.techinferno.com/diy-e-gpu-projects/2504-diy-egpu-x2-lenovo-t61-possible-help-3.html#post36334

INFO: Battery life tweaks Substantial battery life gains can be had by doing the following. I use Battstat or BatteryBar to measure instantaneous power consumption. * Using a ramdisk for browser cache (phillofoc) : to extend battery life * if have Intel 6205AGN installed then load the latest Intel driver. I'm using http://downloadmirror.intel.com/23186/eng/wireless_16.1.5_Ds64.exe and now get a nice 2.5W drop in power consumption when switch from AC to DC. Prior to that I was using HP's 2570P wifi driver which would only go into low power after a AC to DC switch if I disabled it in Device Manager. However, re-enabling it kicked power consumption back up. The HP driver is best avoided. * better yet, use a US$34.95 Broadcomm BCM4352 802.11AC wifi card, suspected of being more efficient than Intel cards as documented at http://forum.techinferno.com/hp-business-class-notebooks/2537-12-5-hp-elitebook-2570p-owners-lounge-43.html#post79977 . * enable ASPMs as discussed below. Decreases idle power consumption from 8W->6W (2W less). That's a lot. * if enabling ASPMs then consider also setting card reader port3 and wifi port4 to run at x1 1.0 instead of x1 2.0 as discussed to see a minor power saving. * enable AHCI Link Power management for SSD/HDD with HIPM+DIPM set. * disable the USB port that hosts the WWAN device while on DC and not in use. Can be automated as a batch file by using MS' devcon utility. * disable the Intel(R) Display Audio. It sucks about 0.3W and isn't generally used. * set LCD brightness to be min+3 increments for DC mode in the power profile * set CPU max to be 99% in power profile for DC mode in the power profile. That disables the power hungry turbo boost. Consider extending this further with reg hacks to enable "Processor Performance Core Parking Min Cores" (0cc5b647-c1df-4637-891a-dec35c318583), "Processor Performance Core Parking Core Override" (a55612aa-f624-42c6-a443-7397d064c04f) and "Processor Performance Core Parking Parked Performance State" (447235c7-6a8d-4cc0-8e24-9eaf70b96e2b) as described * set iGPU for max battery in Intel control panel and power profile * set wifi for max battery in power profile * set 'Time to Power Saving' to 5s in the SRS (Sound) system tray icon. * install Flashblock browser add on so blocks out power sucking flash content * Use a ramdisk for the browser cache as discussed to reduce power-sucking SSD/HDD reads * If have a SSD use SSDtweaker to disable the pagefile, hiberation amongst other tweaks * A further 0.3W saving can be had by using Throttlestop to decrease lowest max multiplier from x12 (1.2Ghz) to x8 (800Mhz) for the CPU and disable USB 3.0 as discussed at http://forum.techinferno.com/hp-business-class-notebooks/2537-12-5-hp-elitebook-2570p-owners-lounge-43.html#post79848 . Can insert the code below into the 2570P-PM.bat further down in this post to do this using Setup 1.x pre-boot software instead: echo Disabling 2570P's USB 3.0 . . . pt MEM write 1 0xFED1F41B 0x1F * replace a HDD with a power efficient SSD like a Samsung 840 Pro. HP rate their SSD models as providing 15% better battery life than HDD ones. Implemented by Tech Inferno Fan here using a Dell Samsung SM841, a rebadged Samsung 840 Pro. * apply cooling system mods to decrease fan engagement With the all above implemented except the last three I track to precisely 6hrs of battery life as measured with batterybar during wifi websurfing from a 62Whr 6-cell. Lowest idle power consumption seen at 90% battery life point being 5.4W. That's with a 500GB 7.2k HDD installed. I'd get 0.8W less with one of the efficient SSDs shown here. INFO: 33% longer 2570P battery life by enabling ASPMs and L0s L1 I was disappointed with the 2570P out-of-the-box battery life compared to my previous 2560P. An optimized 2560P can idle at < 5W as shown with a SSD. After all optimizations the best my 2570P can do is 8W idle giving a pathetic 4.5hrs of web surfing battery life. This falls far short of HP's claim in their 2570P Quickspecs that they get average 8.02W power consumption with Windows 7/64 and up to 9 hours and 15 minutes of total battery life with a 6-cell. The relevant snippet is shown in the spoiler. With my 62Wh 6-cell, my 2570P would need to idle at 6.7W to meet that spec. However, the lowest I've been able to get it to idle with the stock bios is at 8W which would give 7 hours and 45 mins of total battery life if sitting idle, doing nothing. So what was the problem? Upon investigation I found the ASPMs and L0s/L1 power saving features weren't enabled. The first can be identified by running powercfg -energy in Win7/8 then reviewing the output HTML file. There will be an entry showing this problem: Platform Power Management Capabilities:PCI Express Active-State Power Management (ASPM) Disabled PCI Express Active-State Power Management (ASPM) has been disabled due to a known incompatibility with the hardware in this computer Doing the fix below corrects this issue. The result being idle power consumption was less than 6W. It even dipped as low as 5.4W. That's the level I observed with a similarly configured Ivy Bridge Dell E6230 I had previously battery optimized. The fix saw my historic battery life increase from 4.5hrs to 6hrs using a 62Whr 6-cell. An impressive 33% increase in battery life. Not bad An initial con was the SD/MMC controller disappears off the pci-e bus (and in Device Manager) about 5mins after switching from AC->DC. However, inserting a SD card into the reader sees it reappear. The same behavior occurring if this tweak is applied or not. Thanks phillofoc for picking that one up. The fix: enable ASPMs in the ACPI FACP and set L0s L1s in the pci-e configuration Obtain DIY eGPU Setup 1.x software and install it as a disk image. This will require a MBR type Windows install. Check this by reviewing Device Manager->Disk Drivers->[select]->Volumes->populate. Check the Partition Style setting says Master Boot Record (MBR) rather than GPT (EFI). The Setup 1.x disk image will be accessible as the drive V: in Windows after installation. Create a v:\config\2570P-PM.bat file containing @echo off echo Enabling 2570P ASPMs L0s L1 power savings ... echo PROs: Reduced power consumption from ~8W to 6W at idle echo CONs: none ::********************************************************************** :: FACP: pci-e ASPMs enabled in IARCH_BOOT :: :: PER http://web.archive.org/web/forum.notebookreview.com/hardware-components-aftermarket-upgrades/606100-enable-aspm-t61-running-windows-saving-2w-power-consumption.html#post7902424 :: FACP address identified using 'r-w everything' software :: Note: FACP address is 0xB9CF5000 for bios F.34 :: FACP address is 0xB9FFC000 for bios F.40, F.41, F.42 as used below ::********************************************************************** :: Change FACP IARCH_BOOT ASPMs on bit 4: 0x13 -> 0x3 pt MEM write 1 0xB9FFC06D 0x3 :: Now add 0x10 to checksum correct 0x10 deducted above to a unused byte pt MEM write 1 0xB9FFC019 0x10 ::********************************************************************** :: Enable ASPM L0s L1 power management :: Per http://wireless.kernel.org/en/users/Documentation/ASPM ::********************************************************************** :: The expresscard slot (0x0->0x3) setpci -s 0:1c.1 50.b=0x3 :: The SD/MMC controller: 50.b (0x42->0x43), 90.b (0x42->0x43) :: Note: these devices disappear of PCIe bus when IARCH_BOOT ASPMs are :: enabled. It makes no difference if change below to 0x40 (ASPMs off) setpci -s 0:1c.2 50.b=0x43 setpci -s 23:0.0 90.b=0x43 setpci -s 23:0.2 90.b=0x43 :: The wifi card: 50.b (0x40->0x43), f0.b (0x40->0x43) setpci -s 0:1c.3 50.b=0x43 setpci -s 24:0.0 f0.b=0x43 Set the contents of v:\config\startup.bat to be call 2570P-PM call chainload mbr Boot the system via DIY eGPU Setup 1.x. At the Windows bootmenu select "DIY eGPU Setup 1.x". At the DIY eGPU Setup 1.x menu, select "automated startup via startup.bat". That will take you back to the Windows bootmenu but in this second iteration it has the altered pci-e/FACP configuration. So then select Windows there to boot your OS. It's possible to eliminate this looping and boot into Setup 1.x directly where it does the ASPMs fixups and then chainload to Win7/8. That gives a fully Win7/8 boot requiring no user intervention. Do that by: i. Boot into Setup 1.x and change Chainloader->mbr to be either Chainloader->bootmgr or Chainloader->mbr2 ii. Remove the Setup 1.x bootitem from bootmgr bootmenu with c:eGPUuninstall-disk-image iii. Download BootICE v1.30 or newer iv. Configure the options as shown here Reboot and you'll see that Setup 1.x will now always be the default boot option. As you disk image is still configured to use 'chainload bootmgr' (or chainload mbr2 in most recent builds) wit no additional Win7/8 bootmenu items, when you boot into Setup 1.x and selected "automated startup via startup.bat" it will boot straight into Win7/8. Emergency Recovery or restoring original MBR: if you want to restore the original Win7/8 MBR then can either do it in BootICE (set HDD Windows NT 6 MBR) or boot a Win7/8 CD/USB stick and run 'bootsect /nt60 SYS /mbr' . Disable ASPMs per device if there is a problem enabling it system wide As described at Active State Power Management in Windows Vista Disabling ASPM The driver developer must place the following entry in the device driver’s INF file: [DDInstall.HW] Include=machine.inf Needs=PciASPMOptOut Enabling ASPM on Pre-1.1 Devices The driver developer must place the following entry in the device driver’s INF file: [DDInstall.HW] Include=machine.inf Needs=PciASPMOptIn Log a case with HP to fix this in their bios. That would eliminate all the steps above. Likely ALL their Ivy Bridge Elitebooks and Probooks are affected. You could otherwise seek a refund based on misleading advertising in their 2570P Quickspecs. With their unoptimized stock bios there is no way this system can achieve the battery life and power consumption levels they've given in their Quickspecs. REF - Enable ASPM in T61 running Windows for saving ~2W of power consumption - ASPM - Linux Wireless

Your T410 is the model without the iGPU active so you cannot deactivate the primary video device, the NVS3100M dGPU. Means x1.Opt (Optimus compression) doesn't activate on your eGPU so you only get 3dmark06 < 6k. I believe there were some later T410 models that did have Intel HD + NVS3100M in an Optimus configuration. Those models would allow the dGPU to be used. So really there are three options for better performance: 1. See if x2 is possible on your system. That means combining expresscard+mPCIe or mPCIe+mPCIe. You need to have those devices as being [port1+2], [port3+4], [port5+6] or [port7+8]. x2 mode requires additional eGPU hardware and would require a less convenience 2 cable configuration via underside panels. 2. Swap your systemboard out for one with the iGPU or iGPU+NVS3100M model. 3. Use an ATI HD7xxx video card instead. It will outperform a NVidia card in lower bandwidth scenarios. Note: older chipsets show only half-duplex performance with AMD cards. Not sure if yours is or isn't affected. 4. [recommended] offload the T410 for a Sandy/Ivy Bridge system with an iGPU + expresscard. Then you can get Optimus pci-e compression AND doubling of bandwidth due to the expresscard being pci-e 2.0. You will need the EC060A cable to get reliable pci-e data transmission. See PE4H (PCIe passive adapter ver2.4) . Budget systems being a Lenovo E420/E520, Dell Vostro 3450/3550, HP Probook 4530s/6560b. A s/h 12.5" HP 2560P offering a great mix of portability and performance potential (socketted/upgradeable CPU) at a low price.

They are certainly enthusiastic to give people refunds along with coupons. Maybe they are doing all they can to please whoever is driving the recall (Intel/Apple?). I'm might entertain the idea if it was a US$386 paypal refund so could purchase a OWC Helios or Sonnet Echo Express SE. Though then their basic 60W/75W enclosure to get Intel certfication would just a nuisance. I'd rip it apart, take the board out and figure a way to run a ATX PSU through it in the exact same way a TH05 works. Though then it would lack the PERST# delay setting needed for some of these Macbook implementations. So a TH05 was a much better product for my needs to begin with, even if it delivers 89% of the performance of the Helios/Sonnet. Why does whoever is driving this recall (Intel/Apple??) need to be such a fun-wrecking control freak? Now if BPlus moved their August 2013 timeline for the TH04 to be Feb/March 2013 then I might consider it. The refund + BPlus voucher would presumably get me a TH04 + enclosure solution that works for me since it has external ATX PSU input. Just hope the Intel certification doesn't mean they need to remove the PERST# delay. Doing so would then force the pci-e initialization problem back on Apple to fix their UEFI firmware so it would work in the plug-and-play manner that Intel Certification requires.

A CPU socket adds extra thickness, weight and cost to the board. It's more often the case that ultraportables, ultrabooks and thin-and-light machines have the CPU BGA soldered. The exceptions with socketted CPUs being a 11.6" Asus U24E, 12.5" Asus B23E, 12.5" HP 2570P/2560P, 13" Fujitsu T901 tablet. If you still really wanted a CPU upgraded then would need to have a BGA soldering outlet remove the existing CPU and solder on the replacement you desire. With notebook prices being as low as they are it just doesn't make economic sense to do that unless you are in that line of business and can do it yourself at no or low cost. If they have a dGPU option then it's a low spec one to keep heat and power consumption down. That too will be integrated into the systemboard as a soldered chip. You do however have the RAM slot upgradability. Some systems like the Lenovo X1 carbon even solder that on the board giving no RAM upgrades either. That system then has a certain functional lifespan and due to no upgradability, becomes a consumer throwaway item.

The CPU has two cores, but with hyperthreading enabled appears as 4 cores in Device Manager. Per the TM2T-2100 Maintenance and Service Guide both the HD5450 and CPU are soldered onto the systemboard. There are RAM slots so 2x8GB could fit. You'd need to ask HP if the bios can make use of it, or ask a computer shop if you can test it. Only way of getting faster GPU performance is as a DIY eGPU (mPCIe), swapping out the wifi card. It's accessible via the underside memory cover as shown below: The system is only pci-e 1.x capable HM55 (Series-5) chipset and the lack of an iGPU means no Optimus/Virtu is possible. The best performing eGPU for this system would be a HD7xxx series attached to an external LCD.

Forgot to mention these: * 15" Sony SVS, i7-3612QM, GT640LE, 1920x1080 IPS, 2kg. Review Sony Vaio SV-S1511X9E/B Notebook - Notebookcheck.net Reviews * 15" Asus Zenbook U500UVZ, i7-quad, GT650M, 1920x1080 IPS, touch display, 2.2kg: Review Sony Vaio SV-S1511X9E/B Notebook - Notebookcheck.net Reviews

From http://forum.techinferno.com/lenovo-ibm/2785-overclocking-lenovo-x230t-hd4000.html#post38060 . A reply that may be important for our other readers: I'm only aware of the T901 tablet as having a socketted CPU. Supposedly, the newer Ivy Bridge T902 doesn't have it anymore. Seek a refund from NBR for lack of disclosure? If you were guided by NBR for your purchase then maybe you could ask them for a refund since you now have evidence from my NBR ban of them deliberately hiding that information from users? Disclosure of their primary sponsors would be helpful too. Based on their recommendations in the What Notebook Should I Buy area? (WNSIB) it looks like Lenovo is first and Dell are the second highest paying sponsors. It explains too why NBR don't have comparison charts of systems but rather place a moderator and other gophers in the WNSIB area whose intention is to steer users to their sponsors machines. I wonder what kickbacks they get? A Sandy Bridge comparison table was made for eGPU users. Smallest i7 quad-capable eGPU systems For eGPU purposes, the smallest-most powerful quad-capable systems are 13" Sony SVZ (mPCIe), 12.5" HP 2570P/2560P (EX), 13" Fujitsu T901 (EX), 14" Lenovo S430 (TBT) and 15" MBP (TBT). The SVZ and 2570P are RAID-0 capable too. Then the soldered dual-core CPU systems: 12" Lenovo X230 or Dell E6230 (EX), 13" MBP (TBT), 14" Lenovo i7 T430s (TBT). Given the huge recent uptake in tablets recently (Android/IOS), you'd think there would be interest from the notebook ppl to show what all-in-one unit can do. Sure, these systems are larger than a Android/IOS tablet but still small enough to kick around, yet powerful enough (when attached to an eGPU) to function as a desktop replacement. The 2560P particularly a bargain for the penny pinchers. Unfortunately we see that there are vested interest groups that want to maintain the three different markets: Android/IOS tablets, notebooks and desktops. Intel have also been locking down features to control markets. Eg: it's harder to overclock notebooks to get free performance, we can't do x2/x4 DIY eGPU implementations anymore, they gave us 10Gbps Thunderbolt down a 20Gbps-capable link. There's talk too of their future desktop systems having BGA soldered CPUs. I assume too that Intel screwed us by being responsible in the background for the TH05 recall.

Answered at http://forum.techinferno.com/diy-e-gpu-projects/2109-diy-egpu-experiences-%5Bversion-2-0%5D-66.html#post38064 as this may be of interest to other users.

XTU is useful if you have an Extreme CPU (i7-3920XM) and an unlocked bios. Then you can ramp up the multipliers as much as you like to overclock the CPU to the point of instability or you hit thermal/power limits. Y our CPU has no overclocking ability other than the standard Turbo Boost Intel provide. The HD4000 might see some performance boost if you overclock your RAM. Eg: instead of running it 1600Mhz, run it at 2133Mhz by flashing new timings using Thaiphoon Burner. If you want a highly modifiable tablet then consider a 13" Fujitsu T901. It's Sandy Bridge but users have installed a 45W i7-quad as have a socketted CPU. 12.5" HP 2560P/2570P also with a socketted CPU so can accomodate a 45W i7-quad CPU, but they aren't a tablet.

startup.bat looks good. Your pci.bat was created using 36-bit PCI compaction and did not include the eGPU. That is incorrect. Reason being it can only work if you have a 36-bit root bridge something most bioses don't enable. You'd need a DSDT override to do that. Nor is it necessary for you since you can disable the dGPU to free up 32-bit PCI space. So boot in Setup (menu-based), ensure the eGPU is detected (F5), disabled the dGPU with Video cards->dGPU[off] , perform a 32-bit PCI compaction on the iGPU+eGPU and proceed to then manually add the I/O address per T10. Even after successful 32-bit compaction I still get a error 12 in Windows if you still get an error 12.

+1 for the Dell M4700. Just note, the IPS LCD option for the M4700 disallows Optimus from engaging since the iGPU can't run a 10-bit LCD. So if battery ife is important option it with the TN panel instead. It has a 97Wh 9-cell battery option if want to again extend battery life plus an mSATA slot so can keep the primary + optical drives. If he wants to cheap out then can get a Dell Latitude E6530 with FHD LCD. They can be had for great prices from the Dell Outlet. A big problem with the 15" MBPr is it can't have the iGPU engaged in Windows. Meaning, battery life will be pretty poor there. The iGPU however can be engaged in MacOS. The battery is a sealed 60Wh unit on a MBPr, So overall, not a good unit to meet the OP's battery life requirements. A 13" MBPr does have the iGPU enabled, though it misses out on the quad-core GPU and GT650M dGPU. A better unit for battery life. If wanting the smallest, most powerful unit in a ultrabook-thin chassis, then look at a 13" Sony SVZ. FHD LCD, quad-core, raid-0 SSD storage, PMD with external BD and HD7670M.

More than happy to give you a hand. Would mean we'd get a Trinity system on our leaderboard. The GTX460 and HD6750 both good ones to get the ball rolling with.

For your purposes, the PE4L-EC060A 2.1 that HIT will provide will gain you pci-e 2.0 performance, doubling what your previous PE4H 2.4 provided. According to BPlus' website PE4L-L060A-PM3N is a Gen2 capable product with the mHDMI socket on the PM3N end. The cable is soldered on the PE4L end. That's the configuration that gives the most convenient implementation on a Sony SVZ as you can leave the PM3N in the system and just plug in the mHDMI cable as you need it. This also eliminates wear on the SVZ's mPCIe socket.

Hi HTWingNut, Welcome aboard! Would love to see your first eGPU implementation of a AMD Trinity APU equipped a HP 6475b. You are coming aboard at a pivotal moment as we are discovering the PE4H 2.4a is Gen2 compliant. That's what I'd recommend for the eGPU hardware. I can see the GTX460 would be a left over from the upgrade. Certainly, would give some decent performance but I've discovered that AMD cards do better in a limited bandwidth configuration. See http://forum.techinferno.com/diy-e-gpu-projects/2747-12-5-dell-e6230-gtx660%40x1-2opt-hd7870%40x1-2-pe4l-ec060a-2-1b.html comparing the NVidia and AMD x1.2 results. The x1.2Opt results do not apply to you as you have no Intel iGPU to activate that mode. Only issue with the AMD card is you'll need a unified driver set for your AMD iGPU+eGPU whereas NVidia would present no such issue. That can either be hacked as INF file entries or can look for a modded driver set that does that already, eg: AMD-DNA.

This is what I got back from Masaharu @ BPlus RE: Gen2 compliance of PE4H. Still waiting for a reply about buying just the Gen2 cable ends for our users with existing PE4H 2.4a units. A Gen2-compatible PE4H 2.4a is definitely a better way to go. Well worth the extra $20 over a PE4L 2.1b for it's free-standing nature and interchangable ends. Consider too getting just a EC060A so you could plug it into a expresscard-slot equipped notebook too. First stumbling block I see is you'll need to contact BPlus to order a PE4H 2.4a + PM100A/PM200A. They only list a PE4H 2.4 + EC060A as a buy-it-now component. Going a mPCIe implementation on the Sony SVZ will be a less painful and far more cost effective approach than your previously planned MBPr + $950 netstor TBT enclosure. The netstor enclosure lacks the user settable PERST# delay (as was found on the not discontinued $140 TH05) needed to run a BIOS mode MBPr TBT eGPU installation.

With the 36-bit DSDT override and for systems with low TOLUD, pci-e address space won't be an issue. However, we did have a user way back a bit try to get a GTX590 to work and failed. No error 12, but other error "device cannot start". May have something to do with the internal XFing of the device. In addition, a XF/SLI device on a expresscard link is going to be severly crippled by the bandwidth. It's not worth getting such a powerful card. XT3 and 2760P are both Sandy Bridge. You do realize you can buy 2x startisback licenses for US$3 and get an authentic Win7-type start button on Win8? There's also a multi-touch screen option for a Lenovo T430s, the i7 version with both an expresscard and Thunderbolt port. Most upgradable convertible tablet I've seen is a Sandy-Brudge Fujitsu T901. Has a socketted CPU so users are putting in 45W i7-quads into it. REF: http://forum.notebookreview.com/fujitsu/663890-upgrading-t901-results.html#post8529374

Sandy/Ivy Bridge notebooks have the port link width setting in the flash descriptor and that is locked down by the manufacturer. A utility was leaked for Sandy Bridge systems ME Firmware 7.x available here that allows port settings. Same connectivity requirements apply, you need to have [port1+2], [port3+4], [port5+6] or [port7+8] accessible as mPCIe+mPCIe or EXP+mPCIe for x2 mode or [port1+2+3+4] or [port5+6+7+8] for x4 mode. I believe the Dell Latitude E6x20 series would be able to do x2 and x4. Problem is nobody has publically explained how to unlock the flash descriptor on a Dell system. Khenglish has investigated this for his E6520 finding no solution as yet in his thread http://forum.techinferno.com/general-notebook-discussions/2091-lets-enable-overclocking-all-6-7-series-laptops.html#post26529 . The complexity to getting it work is such that we can say x2/x4 mode is no longer a viable option on Sandy/Ivy Bridge systems. Otherwise, yes, a AMD HD7xxx@x2E card would do better on your G71V than the GTX560Ti and would even be better on your E330 if you decided to have a crack at overcoming the whitelisting by hotplugging after booting with wifi.

Thunderbolt products get certified if they are plug'n'play. You can see that on the MBPr, the EFI framework doesn't allow plug-n-play where the iGPU is enabled alongside the eGPU. kloper resorted to going back to bios mode, using PERST# delay and Setup 1.1x to get it all functional. There the PERST# delay was necessary to allow the eGPU to get detected after the firmware has booted. May I suggest you ask them to at least provide a toggle switch that lets you manually toggle PERST#. Something that is easy enough for them to manually add if necessary. You shouldn't have to solder in for a $950 device? Without it you might have trouble getting internal LCD mode working. Good thing is the netstor will give you full 10Gbps performance, approx x2 2.0 + 12.5%, so is a little faster than a TH05. Other additional benefit being it allows daisy chaining of TB devices.

This thread is getting some serious readership so glad to help out to make it happen. @maven1975, could you ask netstor if their product has a user settable PERST# delay? That delay is necessary in bios mode to boot say into the Apple option screen (ALT). If it's set too low or there isn't one then upon poweron, the firmware running in bios mode will power the system down upon detecting the eGPU on my 13" MBP. Likely does the same on other Macbooks. Would netstor be willing to sell us just the Thunderbolt-to-pcie board with say a floppy PSU connector? That would be cost-effective solution for users that want to use their own PSU and run the GPU without an enclosure. Did you want to disclose how much you are paying for the 250W netstor ready-to-go TB chassis? Last ebay link you sent had it at ~US$950.

The fan control in setPLL works with HP 2530P and probably other Montevina (last C2D) and Calpella (1st gen i-core) HP Elitebook units. No Dell unfortunately. However, try the the FN+15324 trick explained at diefer.de - The 15324 trick and one happy E4300 owner . I can see too that i8kfangui doesn't work based on your comments at E6520 Owner's Thread - Page 64 so speedfan looks to be it.

Latest Virtu MVP (2.1.221) trial allows adding programs via the management panel. The version I hacked (2.1.220) to work has that greyed out but can edit the XML/ini file with an editor to add. Yes. You have a NVidia chipset so get the equivalent of x1E with Intel chipsets. Posting your RE5-var-dx9-1280x800 benchmark would let us see how if fares in the leaderboard. Something is wrong when the eGPU is placed under load. Check your PSU. Yes, could even be a bad PE4L or PCIe cable. Your setup will work with a newer AMD card so long as those devices (PSU, PE4L, cable, EC2C) are all working OK.