AMD's Renoir APUs rumored to support LPDDR4X-4266 memory

onetheycallEric

Posts: 225   +47
Staff
Rumor mill: AMD doesn't currently offer any CPU that supports LPDDR4X memory. However, if speculation from recent Linux patches is to be believed, that could change with AMD's next-gen APUs. The next-gen APUs will transition to a 7nm process node and feature Zen 2 cores. There's also speculation that Renoir will roll out with Vega 10 graphics, rather than Navi, but we'll have to wait and see.

A pair of Linux patches come with hints that AMD's next round of APUs, code-named Renoir, will boast LPDDR4X memory support. The first patch lists LPDDR4X as a supported memory type, while the second patch specifies the 4,266 MHz frequency.

The patches seemingly suggest that Renoir could arrive with a refined memory controller, which would be a welcome upgrade over the DDR4-2400 specification that Raven Ridge and Picasso support.

It seems German tech outlet ComputerBase and known hardware leaker Komachi_Ensaka were among the first to spot the lines of code referencing the memory upgrade.

Considering that AMD's Renoir APUs will battle Intel's Ice Lake mobile chips, which support both DDR4-3200 and LPDDR4-3733 and come with Intel's beefier Gen11 integrated graphics, AMD likely realized it needed to close the gap in memory support -- especially considering how much APUs benefit from faster memory.

Renoir is slated to be the successor to Picasso, and arrive some time in 2020.

Permalink to story.

 
Is renoir still using vega or navi?
Renoir mostly used the palette consisting of just seven warm and cool versions of primaries, when painting outdoors. Sometimes I just want to know nothing of the modern technology mambo-jumbo, as those names often make no sense at all.
 
Well, it's not 100% certain that it's going to Vega at this moment in time, but as the PlayStation 5 and Xbox 'Scarlett' are both using custom Navi APUs, it would suggest that the designing aspect isn't the problem. Down at the low end of GPU capability, the limitations of the Vega architecture aren't heavily exposed, and using Navi wouldn't make a significant difference (as both will still be memory latency/bandwidth limited).

As to why AMD are still choosing to make new products with Vega is anyone's guess, but I suspect that they're either waiting for RDNA 2 and TSMC's 7+ process to appear before they go full Navi in the APU lineup, or they plan on generating more APU SKUs in the form of a greater differentiated tier system - I.e. 7nm Zen 2 + Vega as a basic model, and 7+ Zen 3 and Navi as a high end APU.
 
Coming from a cynical perspective, it could just be business planning.

1st Gen: 14nm Zen, Vega
2nd Gen: 12nm Zen+, Vega (not much change here)
3rd Gen: 7nm Zen 2, Vega
4th Gen: 7nm Zen 2+, Navi
5th Gen: 5nm? Zen 3, Navi

Roll out the new features in an even cadence to keep the buyers on the train.
 
Coming from a cynical perspective, it could just be business planning.

1st Gen: 14nm Zen, Vega
2nd Gen: 12nm Zen+, Vega (not much change here)
3rd Gen: 7nm Zen 2, Vega
4th Gen: 7nm Zen 2+, Navi
5th Gen: 5nm? Zen 3, Navi

Roll out the new features in an even cadence to keep the buyers on the train.

Completing

1st Gen: 14nm Zen, 14nm Vega
2nd Gen: 12nm Zen+, 12nm Vega (not much change here)
3rd Gen: 7nm Zen 2, 7nm Vega - like Vega VII (Vega VII > Vega 64)
4th Gen: 7nm Zen 2+, 7nm Navi
5th Gen: 5nm? Zen 3, 5nm Navi
 
Would there really be any point to using Navi when the IGPU is bottlenecked by the memory anyhow? Even the 11 cu Vega in the 2400g would most likely be limited by 4266 mhz DDR4. These newer CPUs are rumored to have even more CUs.
So long as the Vega iGPu supports the newest codecs and features, it's just as well to use a lower cost architecture.
 
Back