• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

Page 219 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

At least they made some progress which to be honest, is surprising. Guess all those billions in R&D for the GPU were not spent in vain after all.
Not so fast. The AMD laptop was using so-dimm 5600 speed ram. The Intel unit was using LPDDR5x @6400 speed. The AMD laptop might even be using a single ram stick - I can't tell from the review?

So nothing can really be inferred atm from this one review, except that 8 Zen 4 cores beats 16 Intel cores by ~30% if you look at the cpu benchmarks article Phoronix also made.

Edit: In addition the AMD laptop is in a U class config, while the Intel laptop is in a 'HS' config (not sure of the Intel naming for their laptop classes) - see this review for the power config - https://www.notebookcheck.net/Acer-...ra-7-impresses-with-its-AI-core.783349.0.html
 
Last edited:
Not so fast. The AMD laptop was using so-dimm 5600 speed ram. The Intel unit was using LPDDR5x @6400 speed.
I don't understand why laptops use LPDDR5X but clock it at 6400. Why not use plain LPDDR5-6400 ?

Max speed of LPDDR5 is 6400 and max speed of LPDDR5X is 8533. What is the point of using LPDDR5X if you are running it at LPDDR5's speeds?
The AMD laptop might even be using a single ram stick - I can't tell from the review?
LPDDR doesn't come in sticks.
 
Note the gaming benchmarks at notebookcheck;

View attachment 90779

As comparison NBC used the 7840U laptop with the lowest score of all tested 7840U laptops, dunno why they use also a hand held device as second comparison; or is it because it s the second or third worst performing device using a 7840U..?..

They should have used the Framework laptop wich is in a comparable TDP class.

 



At least they made some progress which to be honest, is surprising. Guess all those billions in R&D for the GPU were not spent in vain after all.
1024 shaders are not just for show.
It was deliberate to show that Intel did make some progress, just on the wrong tile 😀

It will be interesting to see if Strix Point will beat MTL's GPU in power consumption.
Why wouldn't It? It will have 33% more CU, so for the same perf you need lower clocks.
 
It's notable because MTL tGPU is made on N5, whereas Radeon 780M is on N4.

AMD has a node advantage, yet Intel beat them?
N5 and N4 are not considered exactly as a full node change, and there are a lot of things to consider, such as area and power budget dedicated to the GPU, number of units, and so on. That and considering that "beating" the 780M outside synthetic benchmarks is debatable if one considers the whole picture.
 
Last edited:
I don't understand why laptops use LPDDR5X but clock it at 6400. Why not use plain LPDDR5-6400 ?

Max speed of LPDDR5 is 6400 and max speed of LPDDR5X is 8533. What is the point of using LPDDR5X if you are running it at LPDDR5's speeds?

It's in the name. LPDDR uses significantly less power at the same speeds, and especially so when idle. Also I think LPDDR5 is currently cheaper than DDR5?

(edit: sorry, misread your post, should have drank coffee first)
 
Last edited:
It's in the name. LPDDR uses significantly less power at the same speeds, and especially so when idle. Also I think LPDDR5 is currently cheaper than DDR5?

(edit: sorry, misread your post, should have drank coffee first)
Not entirely correct.

8 chip memory subsystem for 256 bit bus of GDDR memory uses 40W of power.
4 chip memory subsystem for 256 bit bus of LPDDR5 uses 20W of power, per power draw of Apple M2 Pro chips.

Each memory package uses around 5W of power.

If you use 64 bit memory packages, you will cut the power draw in half, compared to 32 bit memory packages, for the same bandwidth.
 
Not entirely correct.

8 chip memory subsystem for 256 bit bus of GDDR memory uses 40W of power.
4 chip memory subsystem for 256 bit bus of LPDDR5 uses 20W of power, per power draw of Apple M2 Pro chips.

Each memory package uses around 5W of power.

If you use 64 bit memory packages, you will cut the power draw in half, compared to 32 bit memory packages, for the same bandwidth.
M2 Pro actually has only two packages, but each size is x128 bit.
 
Not entirely correct.

8 chip memory subsystem for 256 bit bus of GDDR memory uses 40W of power.
4 chip memory subsystem for 256 bit bus of LPDDR5 uses 20W of power, per power draw of Apple M2 Pro chips.

Each memory package uses around 5W of power.

If you use 64 bit memory packages, you will cut the power draw in half, compared to 32 bit memory packages, for the same bandwidth.
Asus Ally has 4 memory packages for a total of 16GB RAM. With your logic, just this memory would consume 20W of power and we know this is nonsense.
ASUS-ROG-Ally-Handheld-Gaming-Console-_Dave2D-Motherboard-_1-1456x726.png
If what you said was true, then Asus would use only 2 modules to halve memory consumption.
 
M2 Pro actually has only two packages, but each size is x128 bit.
apple_m2_pro_chip.jpg

Does this look like two memory packages?
 
Thats how much power memory uses.

Why is it surpising?
Because It's BS, simple as that.
I didn't put there a picture of Asus Ally just for show.

Here is power consumption of Asus Ally. Link
Screenshot_1.png
Does it look to you that the memory is consuming 20W? If It was consuming that much, Asus would have used only 2 memory modules.

P.S. Was It really necessary to put such a big picture of M2 Pro here? The bigger the better or what?
 
Last edited:
Back
Top