Next Generation Hardware Speculation with a Technical Spin [pre E3 2019]
- Thread starter AlNom
- Start date
-
- Tags
- 5 bytes non sequitur
- Status
I've been searching around for footnotes or specifics on this comparison. It's more significant and impressive if compared to Vega 7, and I'd like AMD's position to be fully clarified since it has been known to use misleading start and end points for its comparisons in the past.
Everything we know about it says it's still GCN, it's just being split into "Compute-GCN" and "Gaming-RDNA" with some differences, but the underlying architecture staying same
I 100% agree, however, if the RDNA architecture matches the Nvidia's Turing architecture in performance (in non-RT gaming) or is trading blows (wins) more equally between gaming titles, then its architecture design/heritage/lineage becomes less problematic from PR-marketing standpoint. That being said, if the RDNA architecture is being outperformed handily by Nvidia's Turing architecture (in non-RT gaming), then that's a PR nightmare waiting to happen for a supposed new architecture.
I'm more interested in the watt and performance improvement and if it's comparing it to Radeon 7 that is also manufactured on 7nm.
So if it's a architecture improvement or manufacturing improvement.
Assuming it's an architectural improvement what can we expect for next gen then?
They did say it's going to be powering the next AMD graphics chips for a decade. So I doubt it's just a small change to the architecture.
So if it's a architecture improvement or manufacturing improvement.
Assuming it's an architectural improvement what can we expect for next gen then?
They did say it's going to be powering the next AMD graphics chips for a decade. So I doubt it's just a small change to the architecture.
Last edited:
Bah.
On their slides, it was actually the next chip that was going to introduce a new architecture, and it would be coming out next year or so. If RDNA doesn't perform convincingly against Turing, then NewDNA will be the new shiny next year, and people will bicker about that instead.
nVidia does this all the time with their Kepler-Maxwell-Pascal-Volta-Turing monikers while AMD has used GCN and a version number. AMD would be wise to use these versions for marketing just like nVidia, then they can wave a "generational leap" in front of consumers much more frequently.
D
Deleted member 11852
Guest
I think the posit is that the nextgen consoles could engineer out many of the slow I/O (loading time) architecture bottlenecks that current consoles and PCs suffer. If that was the case some third party developers (and Sony first party developers) may chose to develop games that simply won't work on PC and if that's the case, where does that leave Microsoft's first party lineup as they have committed to support Windows in future.
I don't buy it either but the gamers are fickle and nobody likes loading screens. I buy some multi-platform games on PC because of long-loading times on consoles. Will PS5 have me mothballing the gaming PC hooked up to my TV?
Instant on, sleep/resume and almost no loading? Hell, yes! 
Dr Lisa Su at AMD Computex:
Ike Turner
Veteran
Dr Lisa Su at AMD CES: (Microsoft) is really looking at revolutionizing the console era for the next decade
Dr Lisa Su at GDC: (Google) is really looking at revolutionizing the console era for the next decade
Dr Lisa Su at AMD XXX: (Your favorite company who is partnering with us) is really looking at revolutionizing the console era for the next decade
Jen Sen Huang at whatever: Nvidia invented everything
Amazing how this PR thing works..
It's at around 1:00:28, just a bit after the timestamp of the video
anexanhume
Veteran
Seems about right when you add a 70mm^2 CPU and a little I/O.
AMD claimed a 1.25X perf increase iso power for Radeon 7, so the 1.5X factor would actually be a regression comparatively (1.25X times 1.25X is 1.56X). If that were the case, I would peg it on the memory power dragging it down.
It depends how they compare the architectures because I doubt Navi is running at the same TDP. Same clocks, perhaps? In either event, it should be at a better place on the perf/Watt curve because it’s not pegged at 300W. And we also know it’s not a 300W card because it only beats RTX 2070 by 10% in Strange Brigade, whereas R7 beats the 2080 in that benchmark.
Last edited:
No solid details here, but my gut feel is that's more of a test on CPU capabilities less on GPU capabilities and how the system can handle reading back from GPU memory while sending new data.
Ike Turner
Veteran
RDNA is GCN based....It's just marketing fluff...instead of calling it GCN 6
D
Deleted member 13524
Guest
Did people also say that Pascal is Maxwell based, and the name change was just marketing fluff?
Or Turing is Volta based and it's marketing fluff?
Ike Turner
Veteran
Technically these are all "Fermi based..."
BTW What "people say" is the least important metric ever and you know it Yes, "people" did note at the time that Pascal was basically a die shrunk higher clocked Maxwell. Now, regarding Turing, although it is based on Volta, it does include RT functionality which is a completely new dedicated part of the graphics pipeline, as far as we know.
Regardless of names, did the AMD keynote actually reveal anything that would give us some info on what will be in the next gen consoles? From where I was sitting, it looks like the PS5 will get a 3700 with some kind of Navi GPU, which look quite expensive right now, so what am I missing?
I guess if we get July 2019 tech in 2020, then that’s a bit of a cost save...
I guess if we get July 2019 tech in 2020, then that’s a bit of a cost save...
- Status
Similar threads
- Locked
