AMD Renoir Lands In Mesa's RadeonSI - Further Pointing To Vega, Not Navi

[tuxd3v]

Don't know about the TDP part, but the closest laptop that I could find:

https://www.notebookcheck.net/Finally-Asus-ZenBook-14-SKUs-with-AMD-Ryzen-3500U-and-3700U-options-are-coming.430320.0.html

But probably next year we will see better APU implementation from OEM

I will look into..
There are some negative reviews online about blanks screens, green, and such, that on windows..now imagine in Linux..
I don't know..

I wanted too, but Ryzen2, probably will be better, but it only comes to next year

In the mean time I searched for slimbook, and I felt in love by this machine:
Slimbook PROX
14"( in the size of 13" ), Intel i7 boot 4.6Ghz, NVIDIA MX250
It seems very nice..but it has dual graphics..

In its category, the best Linux Laptop in the world

But...they don't have my keyboaard... and I also wanted keyboard in black..well the shell( case ) also black..
I will email them with my personal request( fingers crossed.. )..

Seems a very nice laptop, what you guys think?

Regards,

[Snaipersky]

Since the transfer rate of the (DDR4-)memory of APUs limits the 3D performance, a stronger graphics unit might be useless. In addition, Vega would be smaller, so more chip space could be saved. Maybe there are more cores?

Consider faster cores allowing for fewer cores to meet or exceed compute performance of existing options. Fewer cores would lead to higher per-core bandwidth.

[DavidC1]

Consider faster cores allowing for fewer cores to meet or exceed compute performance of existing options. Fewer cores would lead to higher per-core bandwidth.

That's not how graphics works. The total Flops is what determines bandwidth needs. Whether it does it by using more cores at a nominal frequency or fewer cores at a higher frequency it doesn't matter, as long as it achieves the same Flops rating.

Streamlining the process node for all GPGPUs, APUs, CPUs reduces cost and for RDNA puts GCN in the rear-view mirror finally. AMD has never capitalized on their APU advantage, but Navi gives them a much lower power consumption and thus attract more manufacturers than with Vega. Vega for Macbook Pros are all under powered due to the design of Vega never being for portables. Navi on the other hand has the embedded to workstation design by default.

I don't think we have everything about AMD's decision making process to conclude things so quickly. Things like TTM(Time to Market) is important too. And cost.

[smitty3268]

That's not how graphics works. The total Flops is what determines bandwidth needs. Whether it does it by using more cores at a nominal frequency or fewer cores at a higher frequency it doesn't matter, as long as it achieves the same Flops rating.



I don't think we have everything about AMD's decision making process to conclude things so quickly. Things like TTM(Time to Market) is important too. And cost.

And opportunity cost. If they're using 7nm fabs to manufacture APUs, they're not using them to crank out 7nm server CPUs. Take a guess which one is going to make them more money.

[scineram]

Renoir is 7 nm.

[sykobee]

I'll get the ball started, this product is DOA.

You want to capture the laptop market you must jettison Vega from the APU.

It's an APU. Vega might have desirable features for an APU - the CUs might be physically smaller (on the same process, and Renoir is 7nm), you don't need to reach super-high clocks, and you've got bottlenecks elsewhere - memory bandwidth being the main one. It's performed great in the 14nm APUs.

The main rumour, vague as it is, suggests a 15CU GPU in Renoir. That's 960 shaders, which at 1.5GHz would be 2.9 TFLOPS, and a lower-power mobile clock speed could still provide around 1.5 to 2 TFLOPS, which compares well with the 1 TFLOPS that Intel has been crowing about.

But unless AMD provides a mobile variant with LPDDR4X support, even this GPU will be strangled. So you can see why there's no need to push absolute performance with Navi.

[juno]

I'll get the ball started, this product is DOA.

You want to capture the laptop market you must jettison Vega from the APU.

The best is not good enough?

[wizard69]

...
But unless AMD provides a mobile variant with LPDDR4X support, even this GPU will be strangled. So you can see why there's no need to push absolute performance with Navi.

Even with LPDDR4X the APU will be strangled. This is the fundamental issue people seem to be missing in this discussion.

I can actually see AMD addressing the bandwidth problem in a number of ways. An APU with HBM would be one approach but there may be other paths they could take. One is to leverage the shrink to 7nm to implement a huge cache maybe even dedicated video memory for the GPU.

The other reality is this, the chip may not be NAVI per say but that doesn't mean that AMD can't improve the older generation. Again the shrink to 7nm offers a lot of possibilities. So maybe we will see a VEGA+ type. That is an en

[Nille_kungen]

Navi would have been great but the Vega gpu in my 2500u based laptop is actually pretty good.
The zen2 progress in cpu should also be great and noticeable.
The only thing that i think really needs a real improvement is memory support, the 2400mhz limit on ryzen laptops will be a big limit in 2020.
I don't think it will bring enough for an upgrade this time around, if it had navi i would have upgraded my laptop.

[tuxd3v]

Renoir Renouveau, sounds like it will be just refresh on a GPU side of things
Since i don't remember there was GCN 1.0 APU, it started with GCN 1.1.
Sounds like deja vu to me All acording to expectations, no surprises.

If it is same price, that would be even more expected... surprises tend to cost more anyway

Yeah, I waited for a APU to have heterogeneus Computing, and such.. and kavery/Carrizo was forgoten..

AMD see APUs has something with low value, which is a shame, for something they created..
But in the Mobile Market, its their best chance..!
I don't understand why not a 7nm 8 core part with at least 8MB $3 cache..

If they want the Mobile Market, then they should show some aim, and make it hapen...that will not happen with current strategy!!

All ryzen+ cpus all are 4MB $3 cache..its a big problem since for graphics, APUs depends on RAM speeds..
I think they should double #3 cache, and also increase $2 cache to improve SMP a bit/Graphics Performance..

Intel parts have 8 MB $3..
AMD already is doing 7nm in Desktop,they should go with Navi, for High end market, at least.. why waiting for Q1/Q2 2020??
Process not mature enough for optimisation?