Intel DDR5 wakeup check

[SaltyPretzel]

As Linus has stated in his videos about RAM speed increasing performance for AMD CPUs, will the introduction of DDR5 seal the win for AMD? I've always wanted to ask this question since RAM speed for Intel doesn't help with fps. Intels future tiles or not I assume they will begin to build their CPU's dependent on RAM speed to keep up with AMD.  Anyone ever had thoughts on this? DDR5 is right around the corner.

[Skiiwee29]

<-- Moved to CPUs, Motherboards, and Memory -->

[patiencehasleftthebuilding]

We can only speculate. We will know once the technology comes to consumers. I am certain Linus will do a video on this

[Mark Kaine]

10 minutes ago, SaltyPretzel said:

As Linus has stated in his videos about RAM speed increasing performance for AMD CPUs, will the introduction of DDR5 seal the win for AMD? I've always wanted to ask this question since RAM speed for Intel doesn't help with fps. Intels future tiles or not I assume they will begin to build their CPU's dependent on RAM speed to keep up with AMD.  Anyone ever had thoughts on this? DDR5 is right around the corner.

no, I think what will seal the deal is if intel finally manages to make modern products or not,  at this point you basically kinda have to expect they'll just pull out nilly willy... haha, can you imagine! 

 

^in other words,  DDR5 is largely irrelevant and so is AMD, they're too small. 

[SaltyPretzel]

I talked with a computer engineer and he told me intel was the best performer for the longest time because they knew how to cram more transistors into a die compared to AMD.  He said they have spent to much R&D and time trying to keep the same density of transistors moving to 10nm and 7nm. I'm glad we finally have competition in the market to help keep intel prices at reasonable levels. My speculation being AMD in the spotlight for a whole new RAM generation which if Intel does nothing will leave them in the dust. Using 3400mhz RAM compared to 2133mhz RAM can double your fps at high resolutions with AMD.  Not sure when Intel will have an answer.  We also can't forget those M1 ARM chips that are blowing out AMD and Intel. The wind will be changing directions in 2-4 years mark my words! I'm not an Apple/Intel fanboy I like to see competition!

[dizmo]

Uhhhh. I don't buy your claim of doubled FPS when going from 2133 to 3400 at all. Maybe it'll see an improvement here and there, but it's far from double across all titles. Intel also gets benefits from higher speed RAM, the days of it not mattering were in the 4000/6000 series era. Times change.

 

27 minutes ago, Mark Kaine said:

no, I think what will seal the deal is if intel finally manages to make modern products or not,  at this point you basically kinda have to expect they'll just pull out nilly willy... haha, can you imagine! 

 

^in other words,  DDR5 is largely irrelevant and so is AMD, they're too small. 

I mean, even with their older processes, they're still holding their own. Unless AMD figures out supply, Intel will still have a leg up there as well.

[Guest]

Intel's IPC should greatly increase when they hit 7nm. 

Skipping 10nm, maybe for the practice of building chips, Alder Lake and 16 cores looks promising. 

 

Since both AMD and Intel will be using TSMC fab, we should see very close numbers in IPC.

It'll be a matter of time to see the differences in the actual architecture and layout differences because that's where a lot of the performance will differ from. 

 

So 7nm and 5nm TSMC process nodes with AMD vs Intel Architecture. We may see only 5% or could see 50%, but we want to increase IPC either way. 

The next 5 years will be interesting. I'm interested in a longer time period of 10-20 years from now. Looking back that far.... quite a lot of things have changed in computing. We are much more complex today that 1999. And back then, much more complex than 1979.

 

When your 600$ FX-55 is now worth 40 bucks..... lol. It was Sooo fast man!

[SaltyPretzel]

There will be a point though in 20 years when physics limits our ability to shrink transistors. Sure a university achieved 2nm but the average size of an atom is roughly 150 picometers or .150nm.  Until someone can figure out how to manipulate an electron we won't see transistors shrink much more in the coming years. We'll be moving to quantum computers at this rate and have all the processing done in the cloud.