Intel Core i9-11900K PCIe Gen 4 SSD would be up-to 11% faster than Ryzen 9 5950X

Hey, it's on Twitter so it must be true!

 

So in a generation where Intel sacrificed core count in order just to reclaim the gaming king crown by the small margin in special circumstances (w/ high end GPU in FHD) just like Zen3 has over Comet Lake or Comet Lake had over Zen2, etc., Intel is marketing that CPU by being 11% faster using PCI Gen4 NVMe SSDs, which has ZERO relevance in games, as a 550 MB/s SATA3 SSD loads games as fast as an NVMe SSD. This is ingenious!

 

The guy that does the video card reviews had his house burned down in the wildfires in Oregon(I think it's Oregeon) they lost everything and I guess they just never got around to replacing the review stock and what not.

 

Claims stating, "up to ..." are nearly always useless. With the same tests there is usually an equally valid, "as low as" claim that can made. That is also often on some very specialized, niche case where this "big" advantage shows up and nowhere else. This is not just an Intel thing either. Nearly all companies do this stuff.

 

Intel platforms are handling Optane differently, while in AMD or other platforms it's "just" an SSD, correct?

How do you know that?

 

Here is PCIe MP600 benchmarks, AMD with Gen 4 and Intel on just Gen3 and still beats AMD in most tests:
https://www.purepc.pl/pamieci_masow..._test_ssd_na_platformach_intel_i_amd?page=0,4

P.S. BTW I just build my first 5850x AMD on one of the better motherboards x570 MSI Creation and and had issue with Avermedia 4K capture card that it wont pass internal test for 4K/60 that was flying on all my old intels even on chipset slots.
I opened HWmonitor, and all PCIe controllers and switches and USB controllers are made by ASSmedia, a brand I wont be touching no matter what and here its everywhere on AMD and you have no alternative.
AMD is highly limited by using this cheap trash brand in their board designs, they need to make their own like Intel.
BTW it was always like this even on old AMD back when their Athlons were faster then P4, their SATA was slower then Intels in every benchmark.

 

AMD can do direct CPU connect just like Intel, the only real difference is that Intel can also enable Optane cache and VROC (which works great with Optane).

There is a secondary difference when it comes to CPU and RAM speed. Optane 4KQ1T1 LOVES ghz and loves RAM throughput.

If you run Optane on a system with JEDEC timing and low GHZ then Optane 4KQ1T1 performance will suffer and since Intel platforms can reach both higher GHZ and higher RAM speed, Optane will perform slightly better on an Intel system, when the clocks are pushed.

This is what my own personal testing has shown. Overclocking made a huge difference.

Intel with mitigations turned off also sees a huge gain in Optane performance (nearly 20% in some cases), but only in 4KQ1T1 performance, sequential is not impacted.

 

Just a "ssd" on both plattforms

 

Unless you turn on Optane caching. Intel was not honest at all about what was eligible for Optane cache. I tested it and was even able to get Optane acceleration to work on a 900P.



I asked intel about this directly several times and they flat out told me that this does not work, which is crap. Literally anything with 3dXPoint can be used.

 

Nothing Intel, nor Corsair, or ekwb come into my appartement.

 

Cool story bro...

 

I have currently 4 intel CPUs that are in use and 1 which just catches dust. (Weird, but that i7 has no other choice. Unless other CPU dies, it is going to sit and wait indefinitely.)
Nothing wrong with using right tool for right job.

 

All claims are up to even if not stated that way, as there is no universal increase in performance, power efficiency, or realistically anything in any hardware.

 

This is not a healthy attitude.

Returning more to the subject, does anyone have a test of a Zen 3 CPU on X570 vs an Intel platform, comparing I/O performance? They only things I can find are server tests on which Intel gets wrecked by default.

 

Do these both go unused in certain scenarios? IOPS & the overall bandwidth from the PCI-E bus or other different interconnects (in this case PCI-E I suppose) Is that the main contention here? The OS not saturating things or am I missing something else entirely... thanks in advance

 

It was a lot more complicated than that. If it was just price then we would see high end gaming systems using 905P class SSDs. That didn't happen for a bunch of really frustrating reasons:

form factor : "good" Optane drives were M.222110, U.2 or PCIe AIC. People want M.2 2280.
capacity : The only M.2 2280 drives were very small topping out at 118GB.
price to capacity ratio : The only Optane drives that were of decent capacity were so ungodly expensive that absolutely no one with anything approaching a budget would benefit. GPU -> CPU -> RAM and then fast storage.
marketing death grip on sequential : The Optane drive with absolutely dominant 4KQ1T1 performance had less than industry leading sequential performance resulting in a lot of big number obsessed enthusiasts to look to the 970 EVO+ and the like.

All in all it was the perfect storm of failure with a smattering of happy niche enthusiast.

 

Pardon, but does this statement mean that you are loyal to AMD?
Kind off stupid to be loyal to either or any brand for that matter.

 

Have to be honest but if this is the best that Intels stealth marketing can come up with then they may be in trouble especially if AMD drops the 5nm bomb on target or adds a p to the x