Microsoft Details Its Velocity Architecture Behind the Xbox Series X:

1. Average PC doesn't have 8Gb VRAM
2. Why are you ignoring the SSD's? They are after all, the reason both Microsoft and Sony went with 16Gb RAM.
3. How much extra memory will PC need to make up for the lack of streaming performance when compared to a console?
4. How much of that memory on PC will have to be used to cache data to make up for not being able to stream as fast as consoles?

 

Entire streaming has been addressed before even by me in this very thread. You ignored it because it actually expected you to think. And thinking is not compatible with believing in case there is conflict.

I am out. I admit, I failed as much as others in attempt to talk to you.

 

Remember again the unified scenario. A PC using 6GB VRAM is around 6GB more addressed RAM allocated than a console. Back when I ran CFX this was the reason I had to buy more RAM to avoid hitting the pagefile when gaming, as I was using 2/3GB more RAM with both cards.

 

Yes run away because you are clueless as well as rude and that pathetic attempt at dismissing SSD's on the first page has not addressed anything.

 

Sorry, My 1st post in this thread used too many words which only some can understand as they require actual experience. My 1st PC had 16kB of RAM. I even used mono chromatic screen at time. Data volume, transfer rates, processing power come hand in hand. System is only as strong its weakest part that hits its limits.

But it does not go other way around. Same as having 128GB of ram will not make gameplay of current games any better than having 32GB. Same as having 64 Core CPU will not make current games any smoother than 8 Core CPU. Same as having NVMe that can read 5GB/s sequentially instead of SSD that reads 500MB/s sequentially when both have equal IOPs. (As IOPs is actually main limitation SSD/NVMe-s have.)
That's because actual game limitation is GPU horsepower or CPU single threaded performance (as long as CPU has enough cores) and this depends on resolution.

To those 2 primary subsystems, there is limitation coming from their memory. CPU bottleneck will be worse if you use memory with lower clock or higher timings.
And same applies to GPU when rendered content hits bandwidth limitations.

2080Ti would play games just fine (same performance as one we have available) with half VRAM as long as it would not run out of available VRAM due to high resolution buffers or assets in VRAM. But it would tank badly with 11GB of VRAM having half bandwidth.

Idea that PS5 gets around insufficient RAM by having fast storage would not be that bad if it was not actually limited by memory bandwidth in situation it would need all those data in RAM. If you want to have 60fps+ on PS5, you can only read/write certain amount of data per frame. GDDR6 has 2 channels, but each channel can only read or write, therefore bandwidth is not bi-directional, but sum of read & writes. (More you write, more you eat from read ability.)

That available bandwidth is shared between CPU and GPU in PS5 and that available bandwidth equals to RX 5700 (XT). Do you think that 5700 (XT) will improve its performance or delivers something revolutionary if you give it 16 GB dedicated memory instead of 8GB while filling it instantly with anything you want?

 

Sorry to go off topic, but love the mono chromatic screen reference

 

When i read about people using Star Citizen to test performance of anything i jump out of my chair.
There's tons of optimized released game out there and you pick an unoptimized mess (which i happen to like as an 'evocati' tester)

 

It's the only game I know of that recommended specifications include an SSD. In a way I see why it would be used, but in another, meh.

 

I hope i'm wrong, but I'm still not convinced that an SSD will be a minimum requirement for many if any games next year.

50% faster SSD on paper, in reality 3 second load times instead of 2. Yay!!!

 

All games should be deployed with SSD in mind even if it would not be requirement.
Current issue with game compression is that they use CPU intensive methods to maximize data throughput from HDD. But those methods are mostly unable to decompress using more than 1 thread.

This heavy compression is reducing HDD's read speed bottleneck. But when such data get loaded from SSD, CPU becomes bottleneck. Many lighter compression methods (including free ones) achieve just tiny bit worse compression, but their decompression is light enough to decompress 500MB/s into 800~1200MB/s without saturating single CPU core.
That's big difference in comparison to HDDs.

By this move, HDDs will lose a bit, but SSDs will gain a lot. Modern systems have games on SSDs anyway.

Secondly, when this heavy compression is read from SSD and one CPU core chokes, it can introduce problems on quad core CPUs.

 

Should and will are two different things though.

Developers should, but how many years will it be before they make it mandatory. I think we are still while off.

 

Recommended perhaps.

 

What do you think "scaling" means in this context? I'm using your words, not mine.

I don't know if English is your first language, but you obviously don't know what the word "scaling" means in this context if you think it has nothing to do with LoD.

 

Everything is streamed into memory, including animations. That still doesn't negate the fact that the CPU doesn't read directly from storage when it comes to animation playback, audio or anything else because storage is too slow (even the PS5's)......at least not in modern times that is.

In the old days when memory was much more scarce, that may have been true. I know that the memory deficient Xbox 360 and PS3 would actively stream from optical disc during gameplay, which I actually find quite hilarious as a PC gamer.

 

I think its fine as long as its the same build and location.