Microsoft's Xbox 360 goes 65nm: Falcon Dissection and Power Consumption
by Anand Lal Shimpi on November 15, 2007 11:55 PM EST- Posted in
- Smartphones
- Mobile
The Red Ring of Death
Earlier this year, Microsoft announced an extension of the Xbox 360's warranty from 1 to 3 years for consoles affected by the infamous Red Ring of Death (RRoD) defect. Microsoft never confirmed what actually caused the RRoD, or how many consoles would ultimately be affected, but the symptoms are very well known. Your console will start to freeze/lock up, eventually followed by three red lights on the front of the system, after which you'll either be able to revive the box for short periods of time or it becomes an expensive piece of modern art.
The present solution to RRoD is pretty simple; you call Microsoft's support hotline, you give the representative some information about your Xbox 360 (he/she will then walk you through some diagnostic steps, nothing too painful), and a few days later you'll find an empty box at your doorstep. Toss in your Xbox 360, affix the pre-paid shipping label (Microsoft even provides tape to seal the box) and about a month later you'll get a refurbed or brand new Xbox 360, as well as a 30-day pass for Xbox Live. While you're without your console for as much as a month, at least there's no cost incurred; overall Microsoft takes care of RRoD victims quite well.
Many have surmised that the reason for the RRoD problems is because of inadequate GPU cooling, resulting in fractures in the lead-free solder between the chip and the motherboard. We haven't been able to confirm this suspicion but we have been able to find evidence that Microsoft ignored many suggestions to improve GPU cooling in the Xbox 360, although we're not sure why.
Simply looking at the Xbox 360's internals you see that there's something wrong with the cooling setup; the heatsink covering the GPU, albeit wide, is barely large enough to cool a low end desktop graphics card, much less the higher powered GPU that's in the Xbox 360. If we assume that the Xbox 360's GPU is at least as powerful as the PS3's, the cooling requirements should be somewhere similar; given that the PS3 basically had a GeForce 7800 GTX under its hood, the cooling requirements should be similar. What would require a two-slot cooling solution in a desktop PC was given a barely adequate heatsink on the Xbox 360 and stuck underneath a DVD drive.
Despite the seemingly inadequate cooling, the Xbox 360 worked just fine - the exception being what seemed to be an inordinate amount of RRoD failures, but since Microsoft extended the warranty it wasn't a huge problem, just more of an annoyance.
It's possible that simple tweaks in the manufacturing process could reduce the likelihood of RRoD, assuming that it is heat related. As yield curves improve over the life of manufacturing a particular chip, it is possible to produce chips that run at lower voltages and are thus cooler. It could very well be that the consoles that fail due to RRoD are simply using higher yield GPUs that run at higher voltages, and thus produce more heat, explaining why the problem seems to affect some consoles but not others. If this correlation were true, as overall chip yields improve, the chances of RRoD go down.
When rumors began creeping up about Microsoft moving to 65nm chips in the Xbox 360, many wondered if this could be the end of the RRoD problems, saving owners the headache with dealing with potential failure. Assuming that the root cause of RRoD is inadequate cooling, it is feasible that moving to cooler chips could alleviate if not altogether fix the problem.
While there's no conclusive way of proving whether or not these new Xbox 360s will reduce the chances of the dreaded RRoD, the geek in us couldn't help but try to go find one of these babies, test it and take it apart.
Facebook
Twitter
RSS
46 Comments
View All Comments
Deusfaux - Friday, November 16, 2007 - link
that the GPU isn't yet, and that a platform with both @ 65nm is still coming down the pipe, expected next year and has the codename of "Jasper"Where do I get this information? Oh, I dunno...
http://www.dailytech.com/article.aspx?newsid=9248">http://www.dailytech.com/article.aspx?newsid=9248
zephyrprime - Sunday, November 18, 2007 - link
It doesn't matter what MS says. Anandtech has physically measured the dies and thus has hard evidence that what MS says is false. What's more, given the die size of the parts, it seems like the Falcon is actually a shrink to 80nm for both processor and GPU, not a shrink to 65nm for the CPU as MS has stated.j00k - Friday, November 16, 2007 - link
i wonder why it takes MSFT so long to make the move in die shrink. CPU's already been in 65nm for quite awhile. By the time they transition to 65nm for gpu next year, everyone would have been on 45nm. im sure the technology's there already but what's the holdup?Deusfaux - Friday, November 16, 2007 - link
ok so MSFT didnt say it but stillj00k - Friday, November 16, 2007 - link
The new xbox 360 mobo also appears to have some solid capacitors, which the old didn't seem to have any at all. I'm surprised that wasn't pointed out as I'm sure it should help quite a bit in keeping the system stable even at high temps. It's certainly a huge deal in the desktop mobo arena.sprockkets - Friday, November 16, 2007 - link
wonder why they did that, seeing as how they do not need to last long anyhow. Well, as long as the other caps were rubycon or japanese, that is ok. Wonder what the cost diff is between them and solid caps?I like the last graph that the consoles use only 2.3 watts and 2.8 watts during the game.
slashbinslashbash - Friday, November 16, 2007 - link
Good to see someone scientifically documenting the differences. I had read before that Falcon had only reduced the GPU size, not CPU. Good to hear that it's both.... I guess I'll probably bite the bullet and finally buy an Xbox 360 this holiday season.yacoub - Thursday, November 22, 2007 - link
well the article doesn't really answer the important questions, like whether or not the GPU and CPU run cooler, so not sure why you find the conclusion that the hardware has changed enough reason to buy when we don't know if the changes actually make a bit of difference toward addressing the pressing issue of RRODs.Orbs - Friday, November 16, 2007 - link
What about heat? Can you put a thermometer through the "flashlight holes" in the case to see if temps improved too? Die shrinks should help but the less intense heatsink may offset that.In terms of the RRoD question, if the cause was heat (which is not confirmed) a test like that would help show if this will in fact increase reliability.
Jopopsy - Friday, November 16, 2007 - link
Yes Anand, please post some information about the heat coming off the Zephyr and Falcon PCs !!! Most of us, from a practical perspective, don't really care so much about the die size as much as we care about the heat (as the heat has been tied to RROD).I for one bought a Elite 4 days ago. It was not a Falcon. It had a LG/Hitachi drive in it and it was definitely a Zephyr board. It was too loud for my tastes, and it was H O T ! ! !. It felt like a blow dryer out the back of the unit and my table was actually hot to the touch right behind my system.
Went back to the store, found a Falcon (Premium 360 HDMI Holiday bundle LOT 739 Team FDOU - components verified w/ flashlight inspection). It has a Toshiba drive in it, and after 2 hours of Halo 3 it was only warm out the back (remarkable differance).
I have some confidance that this Xbox might last (if it RRODs on me anytime soon I'm getting a PS3).