• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

AMD "Zen 3" Microarchitecture Could Post Significant Performance Gains

AMD is going great on desktop, but they still need better mobile offerings than 2 chips based on Zen 1.5.
 
I am waiting. I ended up buying 3700x instead of 3950x as I wanted to wait for this new z3. Hope it’s worth the wait..
 
AMD is going great on desktop, but they still need better mobile offerings than 2 chips based on Zen 1.5.
It's three chips as of last week, but I agree. We'll see how much closer AMD gets when Renoir APUs based on Zen2 arrive in 2020.
 
Intel has now very strong competition and they are still doing nothing. What would happen if they have no competition?
 
Intel has now very strong competition and they are still doing nothing. What would happen if they have no competition?
Its a much higher IPC architecture called Willow Cove. AMD can relentlessly pursue new nodes and incremental enhancements in its eternal quest to equal 4.5/4.9Ghz Skylake parts but ultimately if they drop AMD will be in trouble. People are already praying Zen 4 delivers a significant jump.
 
Do you mean what's the point of AVX-512?
What's the point of changing the architecture so soon. :)
I'm very pro AVX-512 :)
Desktop software is written and compiled towards the x86 ISA, which is microarchitecture agnostic.
"Desktop software"? As opposed to what? :P
The only level of "optimization" in that regard is which optional extensions you choose to use, most of which are supported by both. There is no way to "optimize for Intel" etc., not the way most people think it is.
Nope. Software is compiled for ISA, but it's written using libraries.
Libraries can impact compilation as well.
Intel MKL is usually faster than alternatives (like OpenBLAS) on the same CPUs - utilizing the same instructions. It produces different low-level code.
 
What's the point of changing the architecture so soon. :)
I like progress, even if it's baby steps.
But I do still wonder if too rapid iterations leaves development time for necessary improvements in the long run.

"Desktop software"? As opposed to what? :p
Mobile stuff in general is ARM, embedded software could be ARM, MIPS, etc.

Nope. Software is compiled for ISA, but it's written using libraries.
Your claim was that is written for (Intel's) APIs. While both Intel and AMD does provide some libraries, software in general don't rely on these. And software in general is not more "Intel optimized", if that's what you're implying.

Libraries can impact compilation as well.
Really? Dynamically linked libraries is not compiled into the software you know ;)
 
Its a much higher IPC architecture called Willow Cove. AMD can relentlessly pursue new nodes and incremental enhancements in its eternal quest to equal 4.5/4.9Ghz Skylake parts but ultimately if they drop AMD will be in trouble. People are already praying Zen 4 delivers a significant jump.
AMD has had significant IPC uplift with Zen and again with Zen2. I’m sure they have IPC goals beyond Intel’s current offerings, but today, they have to sell what they are able to produce in volume. How those products compare to Intel simply determines how they are priced.
 
I for one will be irritated if they don't start really pushing their clocks right up into 4.8-5ghz all-core territory, without any of this strange "do they boost to these numbers or don't they" business. I did strongly consider picking up a 3700X but Intel's Comet Lake line-up combined with the lack of m-atx X570 boards has convinced me a slight delay is better even in the face of slight bottle-necking with my 1600.
 
I was about to pull the trigger on an update to Zen 2, but given next year we get Zen 3, Ampere and Navi+ (with ray-tracing), plus the fact the first gen X570 boards are a bit underwhelming and next gen should be better, I'll wait another 12 months. Zen 3 should be a beast, and 4950X would be perfect since I do a lot of CFD simulations, photoshop as well as gaming.
 
CFD simulations, photoshop as well as gaming
Doesn't sound like you need an X570 at all. B450 or the upcoming B550 will do everything you need and leave you with $100 to spend as you please.

Zen3 will be a minor arch tweak but still TSMC 7nm so don't expect huge improvements (5-10% maybe?) and I'm not sure that's worth waiting 6-9 months for. Hell, the 3950X was 5 months later than the rest of the Zen2 chips, so the 4950X could be a mid-2021 product with the 4900X holding the line for a while like the 3900X had to.
 
Last edited:
Its really looking like and even was shown in some presentation by AMD that zen 3 might move to 8 core single die and share L3 cache finally. I think its a sign of things to come and slowly building up. AMD has done great job in their tick tock approach. Zen 3 will likely benefit a lot for gaming.
 
well yeah ultimately we all want that, but Intel is a bit too big for its bridges, a good kick in the nutz is what they need.
Common mistake but, it is: "...too big for his (it's) 'britches'."
capture011.jpg

from Merriam-Webster.com

Sorry, could help it. Live and learn, right?:rockout:

:lovetpu:
 
Its really looking like and even was shown in some presentation by AMD that zen 3 might move to 8 core single die and share L3 cache finally. I think its a sign of things to come and slowly building up. AMD has done great job in their tick tock approach. Zen 3 will likely benefit a lot for gaming.

Well the 3950X matches the 9900K in most games so I would expect that the next chip would beat that too. I am patiently waiting for Intel's real response.
 
Back
Top