Jump to content
NotebookTalk

AstroMan

Member
  • Posts

    10
  • Joined

  • Last visited

Everything posted by AstroMan

  1. One thing to also consider is power draw. MXM-B is much more power hungry. If I remember correctly my Clevo GTX 980m drew about 125W maximum from the slot, which is quite a lot to ask, even from alienware laptops (I think it was its actual limit in 17 R1 Ranger stock). It's definitely a lot of stress for the motherboard and PSU (if it isn't upgraded). I want to also point out that with those chokes on the side of newer gen gpus (or top of the card if connector is pointed downwards), there are heatsinks made for AMD which accommodate for them as AMD FirePro M4000 also has the big chokes in similar spot. I guess would be a good spot to start, just be careful as to not short things out as you will kill motherboards power rail and gpu itself Edit: upon closer inspection, the placement of the chokes on the new cards seems different so it's a moot point
  2. When it comes to unpopulated components, its usually RAM chips, related power circuits (coils/inductors/chokes, mosfets and caps) and related caps and resistors. Really its just reusing (cost saving) the same pcb for different products. A good example in Maxwell is GTX 970m 6GB vs GTX 980m 8GB. In 970m you have clearly missing one power supply (among other things) and 4 GDDR5 512MB RAM chips, making it... surprise surprise 6GB. If one is to try and add such components to juice up the gpu, you would also require modded vbios addressing the change. I have attempted this about a decade ago on 860m 4GB but sadly without success. At some point I was left to my own devices and learning everything was very time consuming for a young lad like me so I de-soldered my mods and left it back as it was. Just one issue less to deal with haha I don't mean to discourage of course, but there is an art to vbios modding and memory timing tables and all that jazz. I've done my fair share of modding but I'm by no means an expert, so if someone could chime in with juicy details, that would be great.
  3. I made a mistake in my post, what I meant to say is that you don't have to choose "big" (B type) cards if you want to go with Tesla architecture gpu. I guess there is some benefit with encode/decode, maybe efficiency and some open source drivers but that comes at a steep cost of DC screen at the moment (and no guarantee it will work btw), oh and heatsink mods too. But yeah, with less TMUs, cores, rops, memory and TDP, I don't think T1000 will be quite the... showstopper, even compared to 965m. as always, the answer is: it depends
  4. If you check docs, M17X R4 (Voyager) also has DP_D on eDP. This might sound crazy buuuuut unplug the eDP screen on M17X R4 and see if it outputs to external via HDMI and/or at least stops beeping. Maybe the screen is the issue all along. 8 Beeps in AW means more like "dont wanna" instead of "the gpu is dying of death". Baby steps.
  5. Those diagrams and schematics are probably made for engineers and hide behind NDA Unless someone knows and shares, it's usually people like us who tinker and probe the traces in the darkness haha Outside of learning how to do all the vbios VBT magic (which takes time and effort), step below is to get your hands dirty and plunge money down the drain by buying gpus, flashing them and reporting what works in what combination. No guarantee if it will ever work, it wasn't designed to from factory. Not that this was ever an issue
  6. You don't have to go with a large MXM-B with GTX 980m or mod interface to 3.1 to have RTX 3000/5000 go fast. NVidia made Quadro T1000 (Turning) which are MXM-A and go for about 250$ a pop. From what sellers tell me, they require eDP, so that's a thing.
  7. Whether a mxm card works or not is a mixture of compatible firmware (vbios, efi etc) and hardware (traces). DP_C stands for Display Port "channel C" sooo, if your hardware (laptop) has motherboard traces for channel C and is configured in firmware to do so, it will work with a graphics card that's also channel C DP enabled. To get it working in M17X R4, you probably need 120Hz screen (eDP cable screen), PEG mode (bypass intel - default for 120Hz), pure UEFI (no CSM - at Maxwell and later cards) and some luck with a card (compatibility). You can always get a ch341a @1.8v and backup the vbios prior to opening that can of worms (flashing vbios). Mind you, it's risky and just because you flash "compatible" for your laptop vbios, doesn't mean that the vendor implemented all the traces into the GPU pcb. Also ch341a extracted vbios will be different to gpuz extracted vbios.
  8. Thank you!! That's seriously awesome! Would love to replicate it for 8560w, seems logical and 100% compatible. Funny those old laptops are more modular than modern modular brands. Great laptops really. Anyone tried these adapters for 2230/2242 nvme to expresscard 34? Kind of makes sense for hotswap storage or test system drive, hopefully not at 2.5Gbit speeds of Expresscard
  9. Honestly, lovely discussion. It all depends on the vbios (or config within it) with these mxm modules. Depending on your config (firmware + hardware), you will be able to use some cards over others (stock). With eDP (the moded cable), you get access to new gpus as long as they fit the heatsink. Some new display options are quite nice too. @GuitarG For those who can't access ibmnb forums, could you please repost the MXM 3.1 - 0 ohm resistor mod here?
  10. Really impressive work GuitarG. So impressive in fact that I had to make an account just to say thanks for all your work. I have been following your work across forums, really really neat stuff, especially the MXM 3.1. I do wonder whether all these upgrades also apply to 8560w, its older brother with QM67. As far as I can tell, they are almost identical outside of the motherboard.
×
×
  • Create New...

Important Information

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. Terms of Use