I dunno, man. I never knew anyone who had a yellow light PS3 and the only ones I read about were from people who had kept them in enclosed cabinets. I also watched a very in depth 2 hour documentary on the 360 rrod and it wasn’t due to ati.
Comment on Xbox 360/PS3/(to a lesser extent) Wii owners represent
heythatsprettygood@feddit.uk 3 weeks agoWii was mostly okay, but boards with a 90nm Hollywood GPU are somewhat more likely to fail than later 65nm Hollywood-A boards (so RVL-CPU-40 boards and later), especially if you leave WiiConnect24 on as it keeps the Starlet ARM chip inside active even in fan off standby - most 90nm consoles will be okay due to low operating temperatures, but some (especially as thermal paste ages and dust builds) are more likely to die due to bumpgate related problems.
PS3s did crap out with yellow lights of death, although not as spectacularly as 360 red rings (lower proportion due to beefier cooling and different design making the flaws less immediately obvious, but still a problem). NVIDIA on the RSX made the same mistakes as ATI on the Xenos - poor underfill and bump choice that could not withstand the thermal cycles, which should have been caught (NVIDIA and bumpgate is a whole wild story in and of itself though, considering it plagued their desktop and mobile chips). The Cell CPU on there is very reliable though, even though it drew more power and consequently output more heat - it was just the GPU that could not take the heat.
360s mostly red ringed due to faulty GPUs, see previous comments about the PS3 RSX. ATI had a responsibility to choose the right materials, design, and packaging partner to ship to Microsoft for final assembly, and so they must take some responsibility (they also, like NVIDIA, had troubles with their other products at this time, leading to high failure rates of devices like the early MacBook Pros). However, it is unknown if they are fully to blame, as it is unknown who made the call for the final package design.
ColeSloth@discuss.tchncs.de 3 weeks ago
DacoTaco@lemmy.world 3 weeks ago
Ok so, let me set this all straight.
The wii had nothing to do with the gpu but with the die of the gpu that nintendo had designed and kept secret.
Inside the gpu die is both the gpu (hollywood) but also an arm core called starlet. It runs the security software and thats where (rarely but happened) things went wrong, as it was always running code, even in standby. This had nothing to do with ati.
And the ps3 was not what you said. The ps3’s problem was that the ihs wasnt making a good enough contact to the core so the heat of the cpu didnt transfer well into the cooler. You can fix this, but is very tricky and is easy to permanently damage the ps3 in doing so ( you have to cut the silicon under the ihs without touching the die or the pcb, remove the silicon and do reattach it with less glue ). This could be contributed to the manufacturer i suppose
heythatsprettygood@feddit.uk 3 weeks ago
Your description of the Starlet is more accurate, yes. However, its heat output consequently caused some of the issues with the ATI designed parts of the Hollywood, as it exacerbated the thermal issues the 90nm variants had, and that a better designed chip would have been able to handle.
The PS3’s IHS was not the problem. There was decent contact and heat transfer, maybe not as perfect as it could have been (there’s thermal paste instead of it being soldered into place, which is why a delid and relid is fairly essential if you have a working 90nm PS3 due to aging thermal paste), but definitely not big enough of a problem for a properly designed chip to cook itself at the operating temperatures of the PS3 (75-80 temperature target on the RSX on an early variant at full load). The Cell B.E. next to the RSX that uses more power (consequently outputs more heat) has a similar setup for its IHS, but IBM did not make the same design mistakes as NVIDIA, and so we see very few reports of the CPU cooking itself even in those early PS3s.
DacoTaco@lemmy.world 3 weeks ago
Ye no, ive have a ps3 that ylod which i reflowed back to life. After it was working again i started digging and the temps the core was reporting wasnt even close to the ihs that i measured with thermal couple. Also the thermal paste is on top of the ihs, not under it and it wasnt soldered in place. Early ps3’s did cook themselves. Less than 360 by a long shot, but they still did!
Also, side note, its funny how some 360’s rrod was not due to the heat issue but can also be caused by power supply failure or the plug being faulty. Thats how i got and fixed my 360 🤣
heythatsprettygood@feddit.uk 3 weeks ago
If it’s a reflow, your PS3 is running on borrowed time. A reflow heats up the chip enough that parts of it expand enough to make the GPU work again temporarily (the solder bumps on the bottom of the silicon attaching it to the interposer line up their cracks again), but eventually you’ll be back to square one. The real fix is to replace the 90nm GPU with a later 65 or 45nm variant that has the fixed design (search up “PS3 frankenstein mod” for more). There is thermal paste both under and above the IHS - the one under for taking the heat from the silicon up to the IHS, then the top layer for taking it to the heatsink. Here’s an image of a delidded RSX and Cell to show (apologies for the quality, was the best one I could easily find). Image
PS3s did cook themselves, but not to the extent of the 360.
It is funny to see how there are probably so many misdiagnosed 360s out there with bad power supplies that have been subjected to being bolt modded (shudder) or something. It doesn’t help that the three red lights just mean “general hardware fault” without doing the button combination to get further information. I guess at least more helpful than the PS3, whose diagnostics were only made available recently due to a key being cracked.