• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
  • Hey Guest. Check out the NeoGAF 2.2 Update Thread for details on our new Giphy integration and other new features.

AnandTech - Xbox Series X SoC: Power, Thermal, and Yield Tradeoffs

Negotiator101

Member
Jan 24, 2021
654
694
280
In fact, RDNA 2 has increased the IPC per watt, but the IPC is pretty similar with RDNA (just few percentages of increase). RDNA 2 is an upgraded version, with new functionnalities such as RT, MS (etc...) plus optimisation to increase efficiency: reduce the consumption + higher frequency.
What's your background man? You really seem to know your stuff.
 
  • Empathy
Reactions: Lysandros

Superayate

Member
Nov 24, 2020
309
442
240
What's your background man? You really seem to know your stuff.

I'm not sure your post is ironic, and that you just disagree with me. :messenger_grinning_squinting:

I am only sharing my vision for this type of subject, based on datasheets, test results and my experience in the semiconductor industry.
Many people are mixing pure IPC improvements, with IPC per watt or real in game perf / IPC improvements. And mixed also the IPC per units with the overall IPC.
The comparison between RDNA1 and RDNA2, and also GCN vs RDNA1/2 need to be done at ALU level, removing from the equation the nb of Cores (or align the number), the frequency, the possible new process used and some features such as Infinity cache (which in reality, is not a feature of RDNA2, but a feature of NAVI 21 GPU, as shown by PS5 and XsX die using RDNA2).
To have a big IPC steps, you need one (or mixed) of these factors:
  • A bug correction that has affected the previous version (not the case for RDNA)
  • A big bottleneck in the architecture solved (as far as I know, not the case for RDNA)
  • A new architecture (which was the case for RDNA vs GCN, not for RDNA 1 vs RDNA 2)
RDNA2 is clearly not an architecture change, but a very interesting step which has increased a lot the efficiency of this architecture, mainly the IPC per watt (that's why AMD have clearly communicated for such parameters) and the GPU frequency peak. This gave to AMD the possibility to increase the nb of Cores + higher frequency which has increase the overall IPC of Navi 21 against Navi 10, and to help the constraint due to the small bandwidth increase with Navi 21, they have added the Infinity cache.
This optimization was done in parallel of new functionalities that were added (such as RT), but for me, it's clear that RDNA 2 step has not really increased the IPC compared to RDNA 1 (just few percentage).

And to answer you, I'm working in the semi conductors industry since 2007, started working with memory + memory controller design + place and route work. After that, I mainly switched to analog design and associated physical design + digital integration. I am not an expert on digital circuit design, clearly bigger background for analog design, but I work in a mixed team (Digital experts, analog ones, verification, lab, test etc...).
 

prinz_valium

Member
Oct 15, 2013
2,204
2,484
815
So possible variable clocks and throttling?
No. MS doesn't do that. Only the Ps5.
Those slides are the design philosophy behind the Series X and how they come to their results.

The CPU has just a higher heat density and MS addressed that with the overall console design and how high the CPU clocked (fixed still)
 
Last edited:
  • Like
Reactions: longdi

ethomaz

is mad because DF didn't do a video on a video of a video of a video on PS5
Mar 19, 2013
38,979
37,903
1,310
38
Brazil
That quote is a bit weird.

Infinite Cache doesn’t affect IPC.... it affect bandwidth so how fast you have the data to reach the CU processing... in these terms it affect how well you use the CUs.

Now perf/clock are not related to caches... it is how the unit works... how faster it is per clock cycle.

RDNA’s 2 CU perf/clock was improved in “double digit” compared with RDNA CU perf/clock.
 

DaGwaphics

Member
Dec 29, 2019
3,444
4,689
505
RDNA’s 2 CU perf/clock was improved in “double digit” compared with RDNA CU perf/clock.

You've got to watch how AMD phrases that. When you improve "real-world" efficiency as AMD likes to say, you'll get more work out of the CUs.

But the actual number of ops per cycle between RDNA and RDNA2 CUs are identical, this is the real measure of the IPC. Sure adding the Infinity Cache avoids bandwidth issues and keeps the CUs better occupied, where AMD gets their "real-world IPC" improvement from.
 

3liteDragon

Member
Mar 3, 2020
1,624
10,646
610
This goes well with my theory that PS5 seems to be punching above its weight on old games because it can downclock the CPU (since old games are based on Jaguar and Zen 2 can run them in its sleep) while the XSX can't and thermals start cockblocking it. Will be interesting to see what happens once we get actual next-gen games that need all the CPU power.
Nothing's "punching above" it's weight here, the console's doing what it was designed to do.
 

Heisenberg007

Member
Nov 16, 2020
1,139
3,244
350
Can we try not to turn this thread into another childish fight about ones favorite plastic toy?
No one is. Not sure why that's the first thing you thought of. Just wishing Xbox was a bit bigger because I'll always prefer performance over size.
 
Nov 21, 2018
493
1,283
370
Lancaster, England
So if MS did a Sony move (not caring how ugly the console looks) they could have the GPU being the bottleneck? (with a bigger, probably uglier, box)
What would be the advantages over having the CPU as bottleneck?
By adding features like DirectStorage that reduces CPU bottlenecks. Games needs to take advantage of DirectX12 APIs on Windows 10/Xbox Series.

DirectStorage – DirectStorage is an all new I/O system designed specifically for gaming to unleash the full performance of the SSD and hardware decompression. It is one of the components that comprise the Xbox Velocity Architecture. Modern games perform asset streaming in the background to continuously load the next parts of the world while you play, and DirectStorage can reduce the CPU overhead for these I/O operations from multiple cores to taking just a small fraction of a single core; thereby freeing considerable CPU power for the game to spend on areas like better physics or more NPCs in a scene. This newest member of the DirectX family is being introduced with Xbox Series X and Xbox Series S and we plan to bring it to Windows as well.

GPU Work Creation – Xbox Series X and Xbox Series S add hardware, firmware and shader compiler support for GPU work creation that provides powerful capabilities for the GPU to efficiently handle new workloads without any CPU assistance. This provides more flexibility and performance for developers to deliver their graphics visions.

etc...

https://news.xbox.com/en-us/2020/03/16/xbox-series-x-glossary/
 

Elog

Member
Aug 28, 2016
712
2,901
470
Interesting article - basically making sense of the the reduction of the zen2 architecture in the PS5 while understanding that the XSX needed to have those parts due to the dual functionality as a server APU as well.
 
  • Like
Reactions: Heisenberg007

bitbydeath

Member
Nov 25, 2015
14,264
23,730
1,200
This goes well with my theory that PS5 seems to be punching above its weight on old games because it can downclock the CPU (since old games are based on Jaguar and Zen 2 can run them in its sleep) while the XSX can't and thermals start cockblocking it. Will be interesting to see what happens once we get actual next-gen games that need all the CPU power.
That doesn’t make sense since it’s the CPU that’s the bottleneck on the XSX. Your theory would mean the CPU in the XSX could go no higher than Jaguar which I can assure you is not true.