Announcement

Collapse
No announcement yet.

The Performance & Power Improvement Of Steam Deck OLED's 6nm APU

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    Originally posted by catpig View Post
    See the links - "6nm" N6 and "7nm" N7+ have virtually identical transistor density, whilst "7nm" N7 and N7P have quite a bit less.
    What makes you think Deck LCD uses N7+ Process? I did a quick search and found N7+ was supposed to be used from Zen3, so Zen2 should use normal N7 process. So going to N6 should result in 12% more density.

    Comment


    • #22
      Originally posted by Mathias View Post
      So going to N6 should result in 12% more density.
      But we have actually 25% more density, might be more advanced high density librarys (see Zen 4c).

      Comment


      • #23
        Originally posted by catpig View Post



        ​See the links - "6nm" N6 and "7nm" N7+ have virtually identical transistor density, whilst "7nm" N7 and N7P have quite a bit less. Don't ask me why TSMC's marketing chose to label them in this fashion, I'm sure they had their reasons. In the end the "7nm" and "6nm" processes are usually grouped together anyway, I suppose (and same with "5nm" and "4nm").

        As for the reduction in area - maybe I should've been clearer, I didn't say there's no improvements between them. There's many ways to improve production process - and chip design - other than making things smaller.
        However, the end result in terms of perf and power is, as the benchmarks have shown, only marginally different. Whether 131mm² at "6nm" is cheaper than 163mm² at "7nm" we'll probably never know, and since the OLED deck has many other changes the retail price of the overall device can't help either. I'm assuming you quoted the chip sizes correctly.

        Edit: I forgot, our host DID write "die shrink" Although it's noteworthy that the shrink on this chip has been quite a lot greater than what would be expected solely by the change in process, going by the numbers cited on wiki. But then, of course, there's a lot more to average density than production process alone.
        N6 provides around 20% better density than N7 and original SoC die size was reduces around that much. So I really don't understand WTF are you trying to prove here.

        Comment


        • #24
          Unless you play in bright conditions (and therefore need to increase screen brightness), I think an OLED is perfectly fine for mobile devices. There are two key takeaways about the Deck:
          1. It's hardware is bound to get rapidly obsoleted. That means by the time the OLED gets any appreciable degradation, the device might be good to be replaced anyway.
          2. It's user-maintainable. If for some reason your OLED is degraded and you don't feel the need to upgrade the whole system, it shouldn't be that great of an expense to swap out the display. I'm not sure how much it costs to replace the panel but I imagine in a few years (when people are likely to replace it), it shouldn't cost more than $50. You can get an external touchscreen of the same size and resolution for about $80.

          Comment


          • #25
            Originally posted by Anux View Post
            But don't we agree that they only made it smaller here?


            I would say it's the same with some error tolerance. But if they are running the same frequencies with the same voltages it is to be expected.
            I meant individual transistors/transistor parts. I'd say the benchmark difference is a bit large to call it error tolerance, but you're quite right that that alone does not plausibly justify the cost of doing all the necessary changes.
            And yeah, it being cheaper probably is the reason, or at least a reason. Though I can think of two more plausible reasons: switching to the new process reduces load on the chip factories, allowing larger volume of chips coming out. Assuming AMD makes a profit on each chip (and why else would they agree to keep making it) that would still be a benefit, as it allows them to either make more chips for the deck, or more chips for other markets. The reduced area usage might also have been useful in allowing the larger battery and/or shorter traces to memory etc. But I'll pre-emptively concede that I'm starting to stretch "plausible reason" with that last sentence

            Comment


            • #26
              Originally posted by Mathias View Post
              What makes you think Deck LCD uses N7+ Process? I did a quick search and found N7+ was supposed to be used from Zen3, so Zen2 should use normal N7 process. So going to N6 should result in 12% more density.
              I didn't say that, sorry if it looked like I was implying it. I was just saying that the "nm" marketing labels don't refer to any physical size nor density, as the table on wiki illustrates by showing "7nm" N7+ having virtually identical density to "6nm" N6.
              Also, AMD uses different processes for the same core type all the time - e.g. according to https://www.anandtech.com/show/17584...dna-2-graphics the "normal" Zen2 APUs used "6nm".
              Whilst the original deck is a semi-custom Zen2 APU, yet it used some form of "7nm". The new deck still uses Zen2, but is "6nm". There's many ways to optimise area usage other than making individual transistor parts smaller, and that's what N6 and/or additional work put in by AMD did here.

              Comment


              • #27
                I recall N7 and N6 have the same design rules ... just replaces DUV with EUV on a few layers. Not really considered a shrink. TSM just likes to lower that number and keep you guessing. This was from old articles on fuse.

                Comment


                • #28
                  am I allowed to post this link?
                  An update on TSMC's upcoming 5-nanometer process technology.

                  Comment


                  • #29
                    Physical dimensions of a N6 transistors are irrelevant for this discussion. The fact is N6 offers 17-25% better density than N7/P and that is the only reason why SoC of OLED version is smaller. HW IP blocks have not changed, GPU/CPU core config has not changed, I/O has not changed. "AMD did something unknown in SoC design to make it smaller" is a total bullshit.

                    Lithography is a king. It was always the case it will be always the case.

                    Comment


                    • #30
                      Originally posted by jayN View Post
                      am I allowed to post this link?
                      Yes looks like you can. Thanks, I haven't read this before, some nice details in there.

                      Comment

                      Working...
                      X