By using this site, you agree to our Privacy Policy and our Terms of Use. Close
CGI-Quality said:
chakkra said:

Do you know exactly what he means by that? I think I have an idea but I´m no expert.

I have some ideas, but that side of tech isn't my thing. I'm better with the details of graphics hardware (such as TMUs, Bus Interface, and CUDA). What can be deduced though is the pipeline is changing and much of the work is going to be offloaded to the SSD (in NVMe form). This is why some people mistakenly expect the consoles to 'beat' the PC next gen, while not understanding that the PC can already do that, and when it is provided with those types of SSDs in the market, will actually do it better.

John Carmack
@ID_AA_Carmack
·
10 apr.
I upgraded to an AMD 3990 system. I was drawn to the 64 cores / 128 threads, but the first performance test I ran is for IO read speed. Previously, I had my data on an older Intel SATA SSD, which I moved over to the new system. The SATA performance is better even with
John Carmack
@ID_AA_Carmack
the old drive, getting 506 MB/s with unbuffered IO. Copying the data to the VP4100 M.2 PCIE4 SSD then reads at 4330 MB/s! Using normal buffered IO, the SATA is at 464 MB/s cold, the M.2 is 2410 MB/s cold, and both are 6500 MB/s from file cache. Unbuffered is now a big deal --
3:44 a.m. · 10 apr. 2020·Twitter Web App
John Carmack
@ID_AA_Carmack
·
10 apr.
Als antwoord op
nearly doubling speed instead of just 10% difference on initial read, and unbuffered striped over two drives should be faster than copying from the file cache, which is pretty remarkable.


 "I think people should define the word crap" - Kirby007

Join the Prediction League http://www.vgchartz.com/predictions

Instead of seeking to convince others, we can be open to changing our own minds, and seek out information that contradicts our own steadfast point of view. Maybe it’ll turn out that those who disagree with you actually have a solid grasp of the facts. There’s a slight possibility that, after all, you’re the one who’s wrong.