Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Oct 28, 2020.
I'll bet AMDs guidelines to reviewers will be to avoid any Intel CPU.
How the hell did we went from fans not buying into corporate PR to communism is beyond me ...
Its a 4k to be honest i9 would match those benchmarks.
Any 5000 series on 500 motherboard will suffice. And performance will improve further as games get optimised patches.
And they will do so, given same optimization which going to be used on Xbox under DX12U.
Well if the 5900x is better than a 10900 then AMD should not have to say anything. GPUs are traditionally reviewed used the best CPU on the market.
It is to no AMD's fault. DX-R is around for very long time. And if CDPR bent knee to nVidia instead of going for standard, they have one fewer sale.
They promised raytracing for Wither 3. We'll see soonish where that lands too.
This may be case of older games too. nVidia loves doing their stuff on side of standard and locking anyone out. Their mentality of: "If you are not with us, you are against us." is exactly why I do not buy their HW together with Huang who damaged gaming industry long time ago. And anyone ever complaining about ecosystem where there are only 2 companies making dGPUs can thank him and his dishonest practices.
As for raytracing on RDNA2. TMU does hit checks as it is almost same thing as texel work. SIMDs are doing logic (decision making). Ray performance = TMU performance as SIMD work is not holding TMUs back. (This is what was in AMD's patent.)
Performance will be sufficient for time being, but not good enough to actually deliver great benefits. And same applies to Ampere.
Those dynamic shadows/illumination shown in video are nothing special. Anyone can have that in F2P game like Path of Exile at 60fps+ even on RX 580. (And it looks more impressive there too.)
Are you saying those Nvidia numbers in the slides were not with a 5900x?
At 4k you basically always gpu limited. It would not make any difference.
Yeah I think it could work but there's potentially this last step from DXR through RTX for the current RTX ray-tracing titles and effects and how these work.
Vulkan utilizing extensions but I think Khronos has uplifted many of these as standard 1.2.140+ defaults so AMD can support these unless NVIDIA has some exclusive NV ones too for some titles like Quake2 RTX.
I'm a bit concerned but especially concerned about AMD and NVIDIA splitting Direct Storage in a similar way with NVIDA RTX IO and AMD going with MS DS whatever this is going to be called.
I don't mind a bunch of extra effects good as ray tracing will eventually be and a step over rasterization though it won't take over just yet.
I do mind if we're going to be hampering a core performance and load time and streaming system that might be a key thing especially for Windows and how it's bottlenecked with IO.
EDIT: But we'll see it's too early to tell.
Windows 10 21H1 isn't happening until April or May at the earliest and then support for Win10 Direct Storage and implementation from start to familiarity and using it good.
Plus unlike ray-tracing (Which is now swapped around and called RTX due to NVIDIA's lead start here.) AMD's in it from the start now so that could be a good counter for having something more standardized going forward once this functionality lands in the next Windows 10 version.
Farewell my 1080Ti, you have been a great friend but a 6900XT will be coming soon, and my 3800x might find itself being replaced as well.
Looks like my son will be getting a 1080Ti and 3800x into his loop ha ha ha ha
That seems like a pretty meek revolution. How about AMD's best card costs less than $800 and whips the 3090's ass? Force nvidia to completely dump their pricing. Total price war. That would be the kind of revolution I could get "excited" about. Not expecting it, just saying...
Holy Moly AMD.
Think they showed what the best the competition could provide, meaning Intel and Nvidia, versus complete AMD system. That way it can throw some level of performance benefit provided by the extra interaction provided by the complete AMD setup
Nvidia numbers were likely picked from Nvidia marketing material or a recent product review. There is no "feature" performance benefit to Nvidia by using 5900X.
If Zen 3 is faster on gaming than Intel, then ofc reviewers SHOULD use AMD CPUs on all reviews from now on.
When we asked for more AMD CPUs on GPU reviews we were shutdown with "Intel is faster on gaming, AMD is irrelevant on games". Same rules SHOULD apply now.
Till then nVidia has chance to call their implementation of Direct Storage whatever they want, but have Direct Storage in background.
I/O really is more important than raytracing. I am kind of angry that loading times of almost all games are same on SSD and NVMe due to heavy compression used without supporting multi core decompression. And that many games can't even saturate SSD's read speeds.
SSDs are around for damn long time by now, and software lags behind even today.
What I didn't liked:
And this in games that advantage them - nothing unusual here, Nvidia does this also.
But what happens if you don't have a Ryzen 5000 CPU and new motherboard for RX 6800 and 6900XT?
Great to see Nvidia rivaled, but we really need a new 250-350$ card . I don't feel confortable spending 500+ on something that is obsolete no matter what in 2 years. CPUs are a different story, a good high tier CPU can last you 5+ years
Absolutely. But its the Smart Access Mode that requires a 5xxx series Ryzen to provide the results we are seeing. Which is great for those building new rigs, but what what about 3900x or 10700 owners who dont want to upgrade their whole system? 4k performance was virtually immune to CPU variables but now SAM may be a factor that affects that. If so, a 3080 in the eyes of 3900x or 10700 owners may be a preferable choice. Of course we wont know the extent of it until the reviews, so we'll just have to wait and see.