The bad news is that the fix only worked for ~ 1 month. I now have another E74 error. Obviously the solder points have disconnected again.

Fuck Microsoft and their 99 cent X-Clamps. Piece of shit heat sinks are fucking useless.

The board overheats, flexes, and cracks solder points.

I’m not spending another dime or minute on these sorry fucking junk boxes anymore.

!!!! ADRIAN !!!!

Microsoft uses pathetic heatsink technology on these boxes. The x-clamps are crap.

I’d like to try and reflow my gpu and scalar solders but frankly I’m more likely to burn down my house, than to achieve such a result.

I think it’s time to consider getting a PS3. Although, admittedly I have no idea if PS3s have adequate cooling either.

Now, how does that relate to RROD? The usual RROD case is that the memory sizing fails. This happens quite early in the boot process, when a “virtual machine” is invoked which initializes memory (and a number of other things). The most likely reason for the memory initialization to fail is that one of the address or data lines doesn’t work reliable anymore. There have been many speculations about the cause of this (see http://www.bunniestudios.com/blog/?p=223 for example, and also see how people have been able to make the console working again by applying stress to the pins using XClamps etc.), but let’s just say that they don’t work reliable and that it’s probably also affected by temperature in a very complex way. The memory initialization happens about ~1s after you press the power button, and takes approx. 3 seconds – newer consoles take a bit longer here.

The EDRAM-initialization happens a bit later – usually right before the bootup animation starts. So, why do we see an increasement in those failures now? Possibly not because the EDRAM gets loose, or the connection fails. The more likely reason is that the actual texture memory (GDDR3) is damaged in a way that the test texture is read incorrectly, and thus the checksums don’t match – even if the GPU and EDRAM is working properly. So basically both E-74 and the traditional RROD mean the same thing – the memory doesn’t work.

I can only speculate why the initial memory init doesn’t catch this case – one of the reasons could be that the memory init runs at a completely different temperature level. The CPU is still running at quarter speed, the GPU clocks JUST got initialized into full speed. Temperature (of CPU,GPU,EDRAM) between this point and a few seconds later are likely to change in a range of 10 Kelvin or more – that’s a lot. My guess is that the thermal expansion causes the memory to ultimately fail, but not as early in the boot process as a RROD.

Why is E-74 a new phenomena? I don’t know. To be honest, I haven’t checked the exact implementation in a more recent kernel, so things might have changed here. E-74 itself isn’t a new thing, but the fact that so many users are facing it is.

Anyway… burning hardware… woo!

The xbox has well documented hardware problems, true. But everyone I know who has encountered such a problem has had a repair or replacement outside of warranty without question. I’ve never had to deal with xbox support, but I’ve had similar good service for other MS hardware (most recently a bluetooth mouse). I wish I could have said the same for my PSX, which was only a month old when the video output croaked, not to mention the DRE-plagued PS2. I’m no lover of MS, but when it comes to support I have found them hard to fault. I think that the extension of the scope of their out-of-warranty replacement policy serves to underline this point.