Yup, I've had this issue too and it does come back every so often. The first time it lasted nearly a year. The second time, it lasted about 9 months. At some point it will fail to work, or I'll wind up actually damaging the chip and then it won't work anymore.
Come to think of it, my very first intel macbook pro which I had back around 2004 or 2005 when it came out also died of a GPU death - I no longer have this, but I'd bet it would also be fixable the same way. Shame that the last of the 17" Macbooks is so easily killed as they're very collectible as they're the last of the 17" as well as the last of the macbooks to have end user replaceable RAM, drives and even batteries (though you have to open the case to do this, and no, they're not glued in.) Hope this issue doesn't affect the 17" G4 PowerBooks as those are really neat, but so far that one's been working great. I used an inexpensive air reflow workstation (was ~$40 off amazon and it's really just a very precise hair drier) and a lot of time. I started slowly bringing the heat nozzle down from about 6 inches all the way down to about an 1" or 2" away from the chip itself. I think I used 220-250'C on the thing, and then I left it alone for a few hours to slowly dissipate the heat and made sure I didn't touch the table it was on to prevent vibrations from messing up the shape of the solder balls, maybe I was overly protective, but it did work. The sad thing is that even with really good thermal compound and using a fan base on that machine, it still happens. Going above 300'C risks melting plastic parts, delaminating the board, etc. That laptop is poorly designed, and like many of the newer versions today suffer from heat issues. I don't know what the point of having a high end i7 or i9 in a laptop is if you can't use it because it will overheat and it will start slowing itself down. I suspect it's not a "pro" feature as much as a marketing feature. I think, before they turned evil around 2012 or so with soldered in RAM, glued in batteries, and on motherboard soldered in SSD, their stuff used to easily last 10-20 years. Now, they're just littering land fills... You're right about the flex cables, you have to be really patient and very careful. I did break one connector off the motherboard the last time I did it, but it didn't affect operation, not sure what it was for, likely the optical drive that I never use. That's another issue, the plastic gets brittle and fragile and the connectors also break, not just the flex cables. It's also hard to remember which way a plug gets released. Some have a little hinge above them that you have to pop open which releases the cable, others slide in, and others are press-down. The iFixIt manuals help a bit, but they're not always clear on which cable needs what, and it's very easy to use too much force and break something. :( Overall though I think the $40 I spent to get another 1.8 years out the machine was worth the trouble - not that it's performant or anything like that, but because it is the last of its line. Supposedly next year they'll make a 16" one, but I won't be buying it unless they add in DIMM slots, replaceable batteries, standard M2/NVME SSD drives, a headphone jack and multiple ports (not just two measly USB-C ports), and a keyboard with real travel so it doesn't feel like you're typing on a screen. And I know that won't happen, so that's why two years ago, I got a high end laptop (17" 4K display, 1070GPU, i7, 64G RAM) and put Linux on it - and it cost about the same as a 15" MBP with half the specs. I don't think I'm going back. Even running Gallium OS on an $300 Acer R11 Chromebook feels as responsive as macOS on one of their $1000 lighter laptops and lasts for about 10 hours. And even better, it can be used in tablet mode, so no need for an iPad. I guess what it boils down to is that I'm still a fan of "Apple Computer, Inc." but that I never really became a fan of "Apple, Inc." :-) Sad really. In a way it feels a lot like MicroSoft back in the late 90s. Hopefully, like the new Microsoft, they'll wake up and be less evil in the future. And I'm still waiting for them to release the Lisa OS sources. :-D On 7/28/19 4:02 PM, Jerry Weiss via cctalk wrote: > This method is not limited to "vintage" components. > > My MacBook Pro 2011 fails dues to its (famous) problem with the > discrete AMD GPU connections. A reflow restores the laptop, but > inevitably I have repeat the process every few months. Depending on > who you believe, the fault is with the A) poor thermal design, b) BGA > solder used or C) bumps on the AMD GPU itself. The reflow is easy > enough to do, but disconnecting the very fragile cables to remove the > motherboard is not for everyone. Using an inexpensive infrared > thermometer improves the the process. > > Jerry > > On 7/27/19 3:50 PM, Jeffrey S. Worley via cctalk wrote: >> On Wed, 2019-07-24 at 21:24 -0400, Pete Rittwage wrote: >> I did some lookup on the reflow temperatures for various solder >> materials because my gut told me 250 degrees is too low to do any good. >> Turns out this is so. 250 CELCIUS maybe, but Fahrenheit? not. >> >> >> https://www.google.com/search?q=melting+point+of+solder&tbm=isch&source=iu&ictx=1&fir=lNlL1odJeOshrM%253A%252Cdl2_5Te6VgKpAM%252C_&vet=1&usg=AI4_-kRutgIaitoyNNmWoI_dbqyF1P0xmQ&sa=X&ved=2ahUKEwi17--3_NXjAhUK2FkKHZhiCaIQ9QEwAHoECAEQAw#imgrc=lNlL1odJeOshrM >> : >> >> Here's a link to that information. It looks like 220 Celciums is about >> right. So if you were in Fahrenheit then that would explain the total >> failure of the experiment and make it worth retrying. >> >> RSVP >> >> YHOSvt. >> >> ** TNM ** >>> >>> I tried this a year or two back with about 30 x SID, VIC, and PLA >>> chips >>> out of C64's. I heated them in the oven at about 250 for 15 minutes. >>> None of them showed any more signs of life than before I tried it, >>> unfortunately. >>> > >
