SCE confirm 80GB PS3 in Japan (Oct. 30) has 65nm RSX

A better comparison (cropped/resized to match RAM chip dimensions)



That's a pretty big difference.

edit: not the most accurate, but if I crop further, rotate and put the 65nm on top of the 90nm...




It'd be nice if the people who actually took the pains to uncover the die measured the bloody dies with rulers :devilish:
 
A better comparison (cropped/resized to match RAM chip dimensions)



That's a pretty big difference.

edit: not the most accurate, but if I crop further, rotate and put the 65nm on top of the 90nm...




It'd be nice if the people who actually took the pains to uncover the die measured the bloody dies with rulers :devilish:

Is it normal to actually change the shape of a die on a shrink? The bigger one is more equilateral, it seems.
 
Wanted to chip in. I have a UK 80GB, bought a month back, CECHM03, and I am pretty sure it's a 65nm RSX. Why? I stuck a Killiwatt meter on it, and idle it was 90w, under load 120w. Much lower than my original launch 60GB (90nm/90nm) and lower than my brothers 65nm Cell/90nm RSX). I don't see to many other ways of reducing that amount of power consumption aside from a die shrink.

Also worth mentioning, the unit is damn quiet, whisper quiet, Sony have done a fantastic job of engineering these things.

Could you provide some more hard numbers for your brothers PS3? It would be interesting if you put them on a scale as well.

@Alstrong: You may be right in that there is a more notable difference than what I first thought. Where did you look up the RAM sizes?
 
Could you provide some more hard numbers for your brothers PS3? It would be interesting if you put them on a scale as well.

My original 60GB PAL PS3 (90/90) varied between 175 -195w
My brothers 40GB PAL (65/90) varied between 145 = 155w
My new 80GB PAL (suspected 65/65) varies between 95 = 115w

The numbers do jump around quite alot, I have listed the peaks here. If you want something more RMS, I would say 180w, 150w, 110w.

Looking at these numbers, it's very clear to me, that the "M" models in the UK are also 65nm/65nm I do see how else Sony can keep the same functionality/performance and make such a large improvements in power consumption and thus heat/noise.
 
Wanted to chip in. I have a UK 80GB, bought a month back, CECHM03, and I am pretty sure it's a 65nm RSX. Why? I stuck a Killiwatt meter on it, and idle it was 90w, under load 120w. Much lower than my original launch 60GB (90nm/90nm) and lower than my brothers 65nm Cell/90nm RSX). I don't see to many other ways of reducing that amount of power consumption aside from a die shrink.

Well, there are other ways though. Improvements in the optical drive, PSU efficiency, and improved yields (in terms of process maturity, voltage requirements) on existing 90nm RSX, 65nm Cell production would all yield lower power requirements.

Remember that die shrinks don't lower wattage because they just do, they lower power requirements because they tend to lower the voltage the chip requires to run at the same given frequency.

Anyway if you really wanted to know I would just find/locate any open images of that SKU and see if the heat-spreader/mobo layout matches the one provided above by One. But honestly if you see how the tested wattage numbers above are nearly identical from the
CECHH00 to the CECHL00, I would take that as indication that perhaps your own tested numbers aren't truly able to differentiate between whether it is a 65nm RSX or not. It could simply be a newer CECHL00-style revision, which has a similar wattage profile (within single digits) to the 65nm RSX units.

So is this worldwide, or just Japan? My sister just bought an 80GB PS3 here in Canada.

One thing I'm sort of curious about is why people are actively looking for 65nm RSX's in terms of retail availability. I mean, I put myself at the top of the list in terms of folk interested in discussing process shrinks and knowing what path things are taking, but when it comes to what I buy at retail, recognize that for the consumer there is very little difference for this particular change. I mean if we were seeing a significant double-digit wattage drop with this revision then I do think it would warrant consideration - as the original shrink to the 65nm Cell did - but it looks like a pretty modest return to the end user thus far. I would say that later 65nm RSX units may be better/improved in their heat profile as the fabbing process improves, but overall it looks like the gain might be fairly nominal period. 45nm on Cell will be the next step to watch for.
 
Last edited by a moderator:
The physical size of the ram didn't change... Just scale the images to match the size of them (roughly) in both dimensions.

The length / width ratio of the RAM modules seemed to have changed between the two revisions judging from your comparison picture.
 
The length / width ratio of the RAM modules seemed to have changed between the two revisions judging from your comparison picture.

As I already mentioned, it was a rough scaling; they're close enough that you can see there's a difference in the GPU size and that it isn't negligible. The other source of scale reference would be the resistors on the chip itself.

The second image with the overlay wasn't meant to be used for precise reduction % as it was meant to show you yourself the approximate size difference wasn't negligible.

edit:

Ok, I see what you mean now. But now using those blue resistors as the scale reference makes the photo of the 90nm die slightly bigger. ;)
 
I've pre-ordered a PAL 80GB PS3 bundled with LBP. This is apparently scheduled for release here in the UK on 28th November. I'm hoping to receive a 65nm/65nm combo and I'll take some power draw readings once I receive it - should be early in December.
 
You haven't been following the hardware! 65nm GPUs are only just appearing for all consoles.

Right i wouldn't be surprised to see 45nm cells early next year in the ps3s. They will most likely use the 65nm rsx and 45nm cell chips to get their next price drop which i'd say is in febuary.
 
Right i wouldn't be surprised to see 45nm cells early next year in the ps3s. They will most likely use the 65nm rsx and 45nm cell chips to get their next price drop which i'd say is in febuary.

We have no indication that the 45 nm Cell has gone into mass production yet. It usually takes > 6 months from that until units hit retail. But we don´t always get that information so they may very well be in production at this very moment.

@22psi: There are probably working prototypes of the 45 nm RSX as well. When these chips go into mass-production is very much dependent of the price of the process and the yield achieved. The production of more mature processes may be considerably cheaper which may compensate for the larger die. There is no point moving to a smaller process unless it saves you money in some way.
 
I bought a LBP 80GB EU bundle on Saturday and measured it's wattage with Kill-A-Watt.
98W (Idle) - 118W (MGS4 Demo).
 
I decided to cancel my order for the LBP bundle. Amazon put the delivery date back about a week and, more pertinently, they've just changed their deals available and VAT has dropped here in the UK.

After cancelling my order for the original LBP bundle, I have now re-ordered a PS3/"300" movie bundle for a cheaper price and I also get LBP, the Blu-ray handset and a HDMI cable free of charge! I've also ordered the new Tombraider game for just a tenner as part of the deal. Only problem is that I'll have to wait a couple of weeks to receive LBP but I can live with that! ;)

Hopefully I find the opportunity to measure power draw on the 'movie bundle' PS3 before the end of the week. If not, I'll report back early next week.
 
Back
Top