Question [Anandtech] AMD Details Renoir: The Ryzen Mobile 4000 Series 7nm APU Uncovered

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
How is Renoir Zen2 IPC 25% greater than Zen based APUs? Could this mean that a Desktop Renoir(full Zen2 monolithic chip) be 25% higher IPC than desktop Ryzen 1800X?
 
  • Like
Reactions: ao_ika_red

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
How is Renoir Zen2 IPC 25% greater than Zen based APUs? Could this mean that a Desktop Renoir(full Zen2 monolithic chip) be 25% higher IPC than desktop Ryzen 1800X?
Zen2 and 7nm HD allow for the possibility that large decreases in power will produce only minimal decreases in performance. This also means that large increases in power don't produce much increase in performance. So I'd say a full Zen2 monolithic chip isn't likely to produce as many gains over 1800X, I'd still expect 15% IPC gain over original Zen.
 

borandi

Member
Feb 27, 2011
138
117
116

thetrashcan

Junior Member
Jan 13, 2020
4
16
41
That seems more like an artist's impression - the same style that we see on Nvidia's "die shots". For a real die shot we'll probably need to wait a few months for Fritchens Fritz to get their hands on one.

The part that you were looking at appear to be the dual CCX's. The proportions of that block are consistent with what we know about from Rome's CCX die, with each CCX being around 3.2 mm wide (same as a CCX for Rome) and 6.2mm tall (which is about what we'd expect if we removed 75% of the L3 form a Rome CCX).
 

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
I really wish it wasn't basically the apocalypse out there, this should be bigger news than it is. It was agonizing watching AMD release bogus '3000' series mobile SKUs that were just Zen+, quite a letdown after watching Zen2 hulk out and wreck the desktop world. Better late than never at least. Although I hate that the naming convention is sort of busted now. Zen+ mobile should have had a 2000 series family name. Perhaps with a 50 added in (2650U etc or whatever) as a middle ground so they didn't have to make it this offset naming convention.
 

Mopetar

Diamond Member
Jan 31, 2011
7,797
5,899
136
I really wish it wasn't basically the apocalypse out there, this should be bigger news than it is. It was agonizing watching AMD release bogus '3000' series mobile SKUs that were just Zen+, quite a letdown after watching Zen2 hulk out and wreck the desktop world. Better late than never at least. Although I hate that the naming convention is sort of busted now. Zen+ mobile should have had a 2000 series family name. Perhaps with a 50 added in (2650U etc or whatever) as a middle ground so they didn't have to make it this offset naming convention.

The companies building the laptops and desktops that these APUs wind up in don’t like doing a xx50 release and want a whole new name scheme because the customers really get sold on having a 3000-series instead of a 2000-series CPU.

This is really just a lamer version of the megahertz wars. Bigger number equals better!

For the high-end in desktop it doesn’t matter as much. People buying those chips will do more research.

Just be glad that we’re at least getting new parts with these generations. With mobile GPUs it hasn’t been too uncommon for the low end parts to be rebranded stuff from two generations ago.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
No, the GPU is the 8 long rectangular structures to the right of that.
However, that is in general a bad spot to place the GPU.

If Renoir was a Bobcat/Jaguar product, that would fly, maybe. But, this is Llano -> Trinity -> Kaveri -> Carrizo -> Raven lineage. The GPU is usually always ran to the edge. Where the WGPs are should be the 8 CUs. Where the NGCUs are at is where the CPU is at. As there isn't a command processor next to it.

¯\_(ツ)_/¯ This "Renoir" die is all sorts of messed up. It doesn't even have ECC, just 16x8 instead of 18x8 like Raven.
 
Last edited:
  • Like
Reactions: Zepp

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
I’m interested in how well the Vega cores compare to older iterations. Specifically, if IPC is higher independent of frequency. Would also love to see sustained multicore boost frequencies on the 4800U.