Jump to content

Lwerewolf

Registered User
  • Posts

    8
  • Joined

  • Last visited

About Lwerewolf

  • Birthday 06/21/1992

Lwerewolf's Achievements

Curious Beginner

Curious Beginner (1/7)

10

Reputation

  1. Now I'm just hoping for somebody to do a 3:2/4:3/etc version of this. Also wondering if the core M will limit performance (heavy throttle?) and if there will be an option for 16 gigs later on... As a tablet I think it's slightly behind the surface (to a lesser extend - viewing angle, to a way bigger extent - screen ratio, no clue about the pen yet) but it's a far more versatile otherwise... simply because of the dual thunderbolts (and probably the better keyboard - we'll see). I hope somebody else follows suit as well
  2. So, seeing that both the SP4 and SB don't have any system bus out (well PCI(-E) out) ports (except for the proprietary SB keyboard connector, probably), on to the question... From what I've read thus far, the problems are the following: 1) http://forum.techinferno.com/diy-e-gpu-projects/8875-those-hoping-thunderbolt-microsoft-surface-pro-4-a.html ...nevermind that this can be mitigated with an IOMMU, but: -AFAIK driver developers still bypass the DMA APIs with https://msdn.microsoft.com/en-us/library/windows/hardware/ff554464%28v=vs.85%29.aspx and https://msdn.microsoft.com/en-us/library/windows/hardware/ff554489%28v=vs.85%29.aspx, -Windows doesn't support IOMMUs, at least it was the case in December 2013: https://www.osronline.com/showthread.cfm?link=251064 (message 7) 2) the Iconia w700 case - supposedly both Intel and Microsoft forced Acer to remove the thunderbolt port. So, does anybody know of any (upcoming) eGPU-suitable tablets, and/or any other problems contributing to the lack thereof? note: to whoever moved my thread to a post here - thanks, wasn't aware of this discussion
  3. General setup: HP 8740w, 820QM, 4x4GB, some intel wifi mpci-e card, 1TB M550 GFX: HP M7820 (5870m, stock), Dell M6000 (7870m supposedly) What happens: Boot Black screen, I can get it to boot to linux (mint x64) reliably, windows seems to BSOD with a "general hardware error" (pci-e, probably a driver doing a no-op?) [*]Linux Open-source driver (radeon): Working as normal, even fired up warsow Dmesg shows errors trying to raise clocks pretty bad performance [*]FGLRX driver (AMD Proprietary binary blob): Black screen SSH-ing to the laptop + dmesg | grep fglrx reveals the following: [ 5.103269] <6>[fglrx] vendor: 1002 device: 6825 count: 1 [ 5.104278] <6>[fglrx] ioport: bar 4, base 0x5000, size: 0x100 [ 5.104603] <6>[fglrx] Kernel PAT support is enabled [ 5.104630] <6>[fglrx] module loaded - fglrx 13.35.5 [Mar 12 2014] with 1 minors [ 6.625545] <3>[fglrx:hal_read_VBIOS] *ERROR* Invalid AMD VBIOS at bus 1! [ 6.625549] <3>[fglrx:hal_init_gpu] *ERROR* Failed to read VBIOS! [ 6.625551] <6>[fglrx] device open failed with code -1 [ 6.626712] <3>[fglrx:firegl_get_vbios_image] *ERROR* Invalid image size requested [ 6.627593] <3>[fglrx:firegl_cail_query_parameter] *ERROR* CAIL_QueryParameter failed [*]Windows (7, ultimate, x64, catalyst): Was able to successfully boot and remote desktop to it only twice: First time, code 12, hd7800 series Second time, video function wasn't even enumerated in device manager ?! Lots of unsuccessful tries between the first and second time, plus afterwards MEMORY.DMP viewed with windbg reveals that it's a WHEA_UNCORRECTABLE_ERROR (124), pci-e error (no-op again?) [*]AtiFlash (ms-dos removable usb stick, recorded logs with msdos piping, missed the fact that atiflash can save logs to a file): -i adapter bn dn dID asic flash romsize test bios p/n ======= == == ==== ==== ======= ======= === ======= 0 01 00 6825 Verde R600 SPI 10000 fail - [*] -ai Adapter 0 (BN=01, DN=00, PCIID=68251002, SSID=1520103C) Asic Family : Verde Flash Type : R600 SPI (64 KB) No VBIOS [*]-p (attempting to program an m6000 vBIOS taken from techpowerup's site) Failed to read ROM ERROR : press '1' to continue [*]-s Failed to read ROM [*]-unlockrom ROM Unlocked Notes: The m6000 does have a vBIOS chip on it The 8740w's BIOS appears to be overriding the m6000's pci subsystem id to an HP value maybe the proprietary drivers get confused by this and access wrong BARs / wrong offsets / something like that? Questions: What is the open-source driver doing differently to make it work? What do you think is causing the issue in the first place? What can I attempt to make it work? DSDT overrides / forced PCI realloc, I guess? Maybe the windows driver is causing error 12 because it can't find the vBIOS? [*]If I want to get into ACPI/DSDTs and etc, are there any intermediary steps or should I go straight to reading the ACPI spec? I have some knowledge on how PCI-E and device drivers work in general (well, I program device drivers, but not video drivers ). Thanks in advance for ANY insight in what might be going wrong!
  4. Think I just hit a dead end with my current setup here. Port 4 is the expresscard, port 3 is most likely the 3G/GPS device (i.e. modem), port 2 is wifi, port 1 and 5 are free (those are the dock's PCI-E plus... something else, dunno what). Is there a way to do the x1e tweak on an even port, if the adjacent odd ports are free?
  5. Update: Got the PE4L 2.1 (expresscard) + SWEX. Immediately tried hot-plugging, worked flawlessly. 3DMark 06: ATI Radeon HD 4870 video card benchmark result - Intel Core 2 Duo Processor P8600,LENOVO 2767W2E Now I guess I need the setup 1.x to get the x1e and... oh s#@$ can't PM yet. All in due time I guess Let me know what other details I should post so that the info may be added to the performance/compatibility/etc table.
  6. Attempted to do that, but it looks like PCI-e extender cables are not to be twisted. In other words, severed pretty much all the connections at the end joints. They were beginning to deteriorate anyways, but still. I guess it's time to move on from the dock I guess the next step is to read on the PE4L stuff, how it works, what should I pick, etc. A new laptop probably won't happen anytime soon, as I'm a student at the moment... Plus there's a mountain bike at home that needs some love as well Regardless, thanks for the help, and I'll keep you updated on my findings EDIT: OK now I'm confused. Does PE4H support Gen2? From what I've read it's either "supports it with PCIEMM-060B custom cable" or "does not support it". Asking because I'm still thinking about doing the x2 1.0 setup. I'll just need to figure out how to get that cable out of the chassis... I have that strange feeling that I don't need that external VGA port anymore
  7. Only one problem with the dock. Forget about installing anything other than a radeon on it, and then it'd have to be HD5k or older. Haven't tested with hd6x, but my 7850 wouldn't post whatever I did (I have a reasonably beefy setup back home, this is my student thing... that's why it's kinda budget). Other people have tried with a variety of nvidias, won't work. I guess that's some kind of white-listing, even thought the hd5k were released a fair amount of time after the T400. Also, the power supply moment. Right now running at 650/650/1.1v and there's INSANE coil whine coming from the dock when I'm running any game. CS1.6 locked to 76fps? QuakeLive? Dark Souls (that's actually the heaviest game in terms of how much the GPU draws)? Whines like hell. I might have to retrofit the dock's power supply at some point as well, or just to get a powered PCI-E riser cable (most likely the latter, unless I want to make it an uber-dock). Right now my HD4870 is crippled even before considering the half-duplex issue If you want to have some more WTF moments... In order for the dGPU to boot at all, I have to enable both GPUs in the BIOS, and to enable OS detection for switchable graphics*. I can also make it load windows by turning it on with those settings, then going into the bios and changing the gpus to internal only + disable OS detection for switchable. That way the gma4500 is still running (supposedly), but it's not driving the internal LCD, and it shows one VGA port being available (none work, tried). The moment you shut it down and try to start it again... System won't even POST. I have to remove the laptop from the dock, switch to switchable gfx, then it posts. Also, shutting down really means holding down the power button, because usually it freezes on restart/shutdown (right when the sound card switches off), so I guess there's a race condition there. Fun stuff * NOTE: OS switchable graphics setting is supposed to check whether you have win7/vista and then either enable them both or only the dGPU, but I guess it just checks if you're on XP, because I've had multiple linux distros and both cards stay on. EDIT: Converted my pci-e riser cable to a powered one, the 4870 is now at 800/900/1.265v: http://www.3dmark.com/3dm06/17382884 Can still pull more out of this setup I guess...
  8. Time to post my system as well I guess. Also, hello to all T400 w/ [email protected], 4G RAM, GMA4500+HD3470, etc etc missing only fingerprint reader for some reason (got it 2nd hand). Sapphire HD4870 512mb undervolted to 1.083 and underclocked to 650/650, all in bios (due to the dock's power supply being unable to handle more). Thinkpad Advanced Dock - that's where I get the PCI-E from. x1 1.0 I think. Some FSP power supply unit. 3Dmark06 score is 7305: ATI Radeon HD 4870 video card benchmark result - Intel Core 2 Duo Processor P8600,LENOVO 2767W2E Now, on to some questions: It seems like the t400 does support an x2 1.0 setup, and by the GPU scaling charts I assume that performance loss with that on an HD4870 will be minimal. However, I'll need to purchase the adapter for that, but this way I'm ensuring that I'll be able to get the GPU back to its original clocks, which would mean even more performance. Any suggestions on how should I proceed? A new laptop is kind-of out of the question for now, and this one is serving its purpose perfectly otherwise. Cheers
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.