Discussion AMD's Future APU Gone ARM?

[Tigerick]

Highly Confident:

When Reuter reported NV and AMD are going to launch ARM SoC for PC:

Advanced Micro Devices also plans to make chips for PCs with Arm technology, according to two people familiar with the matter.
Nvidia and AMD could sell PC chips as soon as 2025

My first reaction is NV yeah possible, AMD no way. With full lineup of mobile APU coming namely Sonoma Valley, Kraken Point, Strix Point and Sarlak. What is the point? All the above SoCs almost covers all the performance (from 64-bit to 256-bit LPDDR5x) and price points.

However, when I started a thread for upcoming LPDDR6 here, I think AMD might create new lineup for ARM SoC with LPDDR6, more info below:

I contacted my source, turned up there might have some truths there. Apparently, MS Surface which presumably will upgrade Surface Pro X to Snapdragon X Elite this year. Next year 2025 and later, NV and AMD have won the contract for upcoming Surface X series. PS: And I believe here is the source of Reuter's leaks. If Microsoft going to launch new Surface in 2025, the specs must be better than upcoming Surface Pro X w/ X Elite.......or not? See below:



Highly Speculation

My source insisted that AMD will create custom off-shelf ARM core for Microsoft. They won't be used on standard PC lineup, I actually have different opinion. Below is my speculated lineup:

Let's compare Hawk Point APU with my speculated ARM SoC with 64-bit memory bus:

	Hawk Point	ARM 64-BIT MC	X Elite	Apple's M3	Lunar Lake
Node	N4	?	N4P	N3B	N3B + N6
Die Size	178 mm2	?	~171 mm2	~146 mm2	?
Memory Supported	128-bit LPDDR5X-7500	64-bit LPDDR6-12800	128-bit LPDDR5X-8533	128-bit LPDDR5-6400	128-bit LPDDR5X-8533
Memory BW	120 GB/s	102 GB/s	136 GB/s	100 GB/s	136 GB/s
CPU	Zen 4 8P/16T	Cortex 6-8 P?	12P Oryon	4P + 4E	4P + 4E
GPU	RDNA3 12CU	RDNA 8CU ?	Adreno	1280 ALU	ARC 8XE
FP32	4.3 TF	~ 4 TF	4.6 TF	4.1 TF	3.8 TF
AIE (TOPS)	16	?	45	18	45

• I don't know how the core arrangement be. Let's assume 6-8 cores with RDNA 12CU.
• The memory bandwidth of 64-bit LPDDR6 is a bit slower than Hawk Point, that's why I estimate core count to between 6-8 cores. Remember Zen4 comes with HT while ARM likely won't support HT.
• And why don't AMD create x86 version of APU to fit the same memory bandwidth? Cause Hawk Point exist, AMD already has full lineup of solution from top to bottom. BTW, Kraken Point and Sonoma Valley will replace Hawk Point in 2026 for mobile market. Thus, there is no point to create another x86 SoC with similar bandwidth. And that's my reason why AMD would use ARM core to create new lineup with LPDDR6. Also, ARM's die area is smaller than Zen 5 (pending confirmation) and definitely more power efficient. It is possible AMD will support 9W fanless design with ARM SoC.

	Strix Point	ARM 128-BIT MC	Sarlak6 256-BIT
Node	N4P	N3P ?	N3P + N3E
Die Size	225 mm2	?	?
Memory Supported	128-bit LPDDR5X-8533	128-bit LPDDR6-12800	256-bit LPDDR6-12800
Memory BW	136 GB/s	205 GB/s	410 GB/s
CPU	4 x Zen5 + 8 x Zen5c	12-16 P ?	16 x Zen6
GPU	RDNA3.5 16CU	RDNA 24CU ?	RDNA5? 40CU + 32MB IC
FP32	~ 6.1 TF	~ 9.2 TF	?
AIE (TOPS)	48	?	48

Speculated Mobile Roadmap		Memory Bus	2026 DDR5/LPDDR5		2026 LPDDR6	Memory Bus
Ultimate Compute		128-bit DDR5	Zen 6 / N3P / 16C + X3D
Elite Experiences		256-bit LPDDR5X 192-bit LPDDR5X	Sarlak Zen 5? / N4P + N3E / 16C + RDNA3.5 40CU + 50 TOPS		Sarlak6 Zen 6 / N3P + N3E / 16C + RDNA5? 40CU + >50 TOPS	256-bit LPDDR6 192-bit LPDDR6
Premium		128-bit LPDDR5X	Strix Point Zen 5 / N4P / 12C + RDNA3.5 16CU + 50 TOPS		ARM 128-bitMC / 16P + RDNA 24CU + >50 TOPS ?	128-bit LPDDR6
Mainstream		128-bit LPDDR5X	Kraken Point Zen 5 / N4P / 8C + RDNA3.5 8CU + TOPS		ARM 64-bitMC / 10c + RDNA 12CU + >50 TOPS ?	64-bit LPDDR6
Everyday		64-bit LPDDR5X	Sonoma Valley Zen 5c / SF4X / 4C + RDNA3.5 4CU

Highly Confident

• Zen 6 is pretty much confirmed to use N3P and feature 16-core into one die area. And I believe 16-core Zen6 will pair with Sarlak iGPU to utilize LPDDR6 with higher memory bandwidth.
• 32-core Zen 6C (that's the term AMD refer) on N2 process. There's no difference in the architecture, only in the libraries and floorplan used. And Zen 6C would be used to counter upcoming NovaLake-S.
• Since DDR6 is delayed, NVL-S will support 128-bit DDR5. Based on timings, it should support up to DDR5-8000. And that might be the speed AMD would use to support Zen 6C, that's mean AM5 platform will live for another generation.

Model	Zen 4 Cores	Speed	TDP		GB6 1T	Perf Per Clock		GB6 MT	Perf Per Core
R9 7950X	16	5.7 GHz	170 W		2937	515
R7 7700X	8	5.4 GHz	105 W		2906	538
R7 7745HX	8	5.1 GHz	55 W		2686	527
R9 7940HS	8	5.2 GHz	35 - 54 W		2475	475		11670	1459
R7 7840U	8	5.1 GHz	28 W		2102	412		8762	1095
	Zen 5 Cores
R9 8950H ?	4P + 8E		35 W
R7 8950U ?	4P + 8E
Apple's M3	4P + 4E	4.06 GHz	20 W		3084	760		11564	1445
M2	4P + 4E	3.49 GHz	20 W		2634	755		9753	1219
X Elite	12P	4 GHz	23 W		2780	695		14029	1169

	Cortex-X3	Cortex-X4	Cortex-X5	Cortex-X6
SoC Launch Date	Q4 2022	Q4 2023	Q4 2024	Q4 2025
SoC	SD 8 Gen 2	SD 8 Gen 3	D9400	D9500
Node	N4	N4P	N3E	N3P
Clock Speed	2.0 - 3.2 GHz	3.3 GHz
L2 Cache	1 MB	2 MB ?
L3 Cache	8 MB	12 MB
ROB	320	384
Decode	6	10
Int ALU	6	8
Max Memory Speed	64-bit LPDDR5X-8400	64-bit LPDDR5X-9600		64-bit LPDDR6-12800 ?
Memory BW	67 GB/s	77 GB/s		102 GB/s
GB6 1T	~ 2043	~ 2277	~ 2700 ?

 

[gdansk]

I wouldn't be so quick to assume it's for PCs exactly.
AMD semicustom unit would combine ARM CPU IP with RDNA IP for a Sony/Xbox or VR headset maker if that's what was requested.

 

[Glo.]

I wonder who makes semi-custom products for Microsoft.

Nope, nobody from the thread title comes to mind. Absolutely nobody.

 

[FlameTail]

That Reuters article laat year, teporting AMD would make an ARM SoC was one of the bizarre info ever to me. Nvidia I could understand, but AMD?

 

[Asterox]

We have to be realistic, AMD always has various CPU/SoC combinations in preparation or pending etc.

https://x.com/MauriQHD/status/1332601734565416962?s=20

This video is three years old.

 

[soresu]

Based on timings, AMD should use N3P process with Cortex X6 core, I don't know how the core arrangement be. Let's assume 6 cores with RDNA 12CU.

In 2025?

It would have to be late 2025 at least given the fastest cadence we have seen from ARM Ltd IP releases to first SoC's using them.

2025 is likely to be full of X5 SoC's - and now that I think about it, I wonder if the reason that Rockchip and Amlogic have been so schtumm on future higher end SKUs is that they are planning to field a really high end SoC for the first time in a while.

When RK3588 was initially announced it was less than a year after A76's IP release so we could potentially see something similar for X5.

Hopefully though it doesn't take quite so long to land in products as RK3588 which was about 1-2 years late to market.

 

[TheELF]

• 32-core Zen 6C (that's the term AMD refer) on N2 process. There's no difference in the architecture, only in the libraries and floorplan used. And Zen 6C would be used to counter upcoming NovaLake-S.

32 zen c cpu would be great at multithreaded but the c cores can't clock high enough for desktop, it would be a huge regression in many benchmarks due to much lower clocks.
They might do it as a server part if there is demand for it, could be good to counter low power arm servers.

https://zhuanlan.zhihu.com/p/653961282

 

[Shivansps]

That Reuters article laat year, teporting AMD would make an ARM SoC was one of the bizarre info ever to me. Nvidia I could understand, but AMD?

Mendocino would be a good example of something you can replace with an first attempt at a ARM SOC. They kinda need to for low power devices, the only thing stopping something that is already old now, the RK3588, from entering the low end notebook and pc market at the low end is the lack of drivers, otherwise it would be there already.
Next gen ARM SoC may have gpus that may rival at least some of the RMB RDNA2 SoCs (G710/G720).

So it could be possible if they think they cant make a x86 SoC cheap and low power enoght.

 

[Shivansps]

In 2025?

It would have to be late 2025 at least given the fastest cadence we have seen from ARM Ltd IP releases to first SoC's using them.

2025 is likely to be full of X5 SoC's - and now that I think about it, I wonder if the reason that Rockchip and Amlogic have been so schtumm on future higher end SKUs is that they are planning to field a really high end SoC for the first time in a while.

When RK3588 was initially announced it was less than a year after A76's IP release so we could potentially see something similar for X5.

Hopefully though it doesn't take quite so long to land in products as RK3588 which was about 1-2 years late to market.

RK3588 got delayed due to covid.

I think AMD may be aiming for replacing Mendocino in 2025 as a first attempt at a ARM SOC, they dont really need X5 cores, they just need something that has better IPC than Zen2, maybe a combo of X1-X2 cores, and around 6 CUs RDNA3 in order to be on the safe side because the bigger versions of the Malis G710 and G720 will be fast.

 

[FlameTail]

Mali G720 MP12 in Dimensity 9300 rivals the Radeon 780M in raw performance.

 

[mikegg]

Why are people surprised that AMD is rumored to be making ARM SoCs?

Are people not seeing how much better Apple Silicon chips are over AMD APUs? In 2024, AMD still can't make a laptop as good as the fanless M1 Air. They aren't even close.

In order to compete with Apple, and soon to be Qualcomm and Nvidia, they have to go ARM. Period.

 

[itsmydamnation]

Why are people surprised that AMD is rumored to be making ARM SoCs?

Are people not seeing how much better Apple Silicon chips are over AMD APUs? In 2024, AMD still can't make a laptop as good as the fanless M1 Air. They aren't even close.

In order to compete with Apple, and soon to be Qualcomm and Nvidia, they have to go ARM. Period.

Why and be specific...

Apple can't make a chip as powerfull as turin, apple must got x86..... Period.


See how stupid arbitrary points are ....

 

[Nothingness]

Why and be specific...

Apple can't make a chip as powerfull as turin, apple must got x86..... Period.


See how stupid arbitrary points are ....

Yours is particularly arbitrary and pointless: Turin is a server SoC based on Zen5c core, and no one knows for sure its performance. Two good reasons it can't be compared to what Apple does for its laptops.

Which doesn't mean I agree with mikegg point

 

[NostaSeronx]

Should be noted AMD is officially in RISE as a General Member and RISC-V International as a Strategic Member.

Wine-CE => https://gitlab.com/wine-ce/wine-ce
Hangover => https://github.com/AndreRH/hangover/releases/tag/hangover-9.0

Pretty much by the time AMD comes out with a RISC-V core. Linux can run most legacy/new Windows apps on RISC-V.

AMD can avoid the costly fee of having a custom architecture with ARM by using RISC-V. SoftBank is still collaborating a close relationship with Nvidia over other licensees. So, it is best to avoid a toxic relationship. Go ARM then get curbed stomped by Nvidia the Monarch of ARM.

AMD going ARM will be a spectacular failure. Especially, where the biggest market for AMD is punishing x86/ARM manufacturers(CPUs/ODMs/OEMs).

 

[FlameTail]

Simply switching to ARM, will not suffice AMD to catch up with Qualcomm/Apple in power efficiency. Apple/Qualcomm has advantages other than the use of the ARM ISA itself.

1. Custom CPU core designs.

Both Apple and Qualcomm have custom ARM core designs, which as of now, are better than the stock ARM cores AMD/Nvidia will have to use. The next generation of ARM cores (X5/A725) might change that though.

2. A legacy in mobile computing.

Both Apple and Qualcomm have been designing smartphone SoCs for more than a decade. Thus, they have mastered the art of power efficiency, with many tricks and solutions- especially on low-power and idle efficiency. AMD/Nvidia don't have this experience in making smartphone SoCs, and thus lack the knowledge/experience that comes with it.

 

[burninatortech4]

Should be noted AMD is officially in RISE as a General Member and RISC-V International as a Strategic Member.

Wine-CE => https://gitlab.com/wine-ce/wine-ce
Hangover => https://github.com/AndreRH/hangover/releases/tag/hangover-9.0

Pretty much by the time AMD comes out with a RISC-V core. Linux can run most legacy/new Windows apps on RISC-V.

AMD can avoid the costly fee of having a custom architecture with ARM by using RISC-V. SoftBank is still collaborating a close relationship with Nvidia over other licensees. So, it is best to avoid a toxic relationship. Go ARM then get curbed stomped by Nvidia the Monarch of ARM.

AMD going ARM will be a spectacular failure. Especially, where the biggest market for AMD is punishing x86/ARM manufacturers(CPUs/ODMs/OEMs).

I'm thinking we'll see AMD replace the Trustzone A5 cores with RISC-V first. Then stand alone products later.

 

[itsmydamnation]

Yours is particularly arbitrary and pointless: Turin is a server SoC based on Zen5c core, and no one knows for sure its performance. Two good reasons it can't be compared to what Apple does for its laptops.

Which doesn't mean I agree with mikegg point

that was my point to pick something absurd

 

[igor_kavinski]

I wouldn't be so quick to assume it's for PCs exactly.
AMD semicustom unit would combine ARM CPU IP with RDNA IP for a Sony/Xbox or VR headset maker if that's what was requested.

Agree on the 2nd part.

However, if they already fulfilled a design for the 2nd part, why couldn't they just re-use it to fit the PC landscape, considering nVidia would have their ARM SoC too? Why miss out on the chance to compete and grab a share of that pie, however small it may be for the first few years?

 

[Nothingness]

that was my point to pick something absurd

Sometimes I just miss it completely

 

[Shivansps]

Should be noted AMD is officially in RISE as a General Member and RISC-V International as a Strategic Member.

Wine-CE => https://gitlab.com/wine-ce/wine-ce
Hangover => https://github.com/AndreRH/hangover/releases/tag/hangover-9.0

Pretty much by the time AMD comes out with a RISC-V core. Linux can run most legacy/new Windows apps on RISC-V.

AMD can avoid the costly fee of having a custom architecture with ARM by using RISC-V. SoftBank is still collaborating a close relationship with Nvidia over other licensees. So, it is best to avoid a toxic relationship. Go ARM then get curbed stomped by Nvidia the Monarch of ARM.

AMD going ARM will be a spectacular failure. Especially, where the biggest market for AMD is punishing x86/ARM manufacturers(CPUs/ODMs/OEMs).

There is no time for RISC-V, we are going to see a big push for ARM Windows devices next year. RISC-V is at least 10 years behind.

The question here is why AMD is going to pay for ARM license when they can make any x86 cpu they want. And the reson to me is very simple, AMD failed HARD at every attempt of making a low power x86 APU so far. They cant archive low enoght BOM prices to be attractive for diferent OEMs, and they fail HARD at power targets and some times even performance.

If you ever had an older AMD small core you will know what im talking about. When they came out with Ryzen they just given up, Mendocino is there to have something, but ADL-N just dominates this market.

 

[Shivansps]

Mali G720 MP12 in Dimensity 9300 rivals the Radeon 780M in raw performance.

G720 should be MP9 max, G710 is the one that is MP16 max.
But dont expect to be used in that big configurations in many socs, thats big and expensive. And probably need a big memory bus as well.

 

[igor_kavinski]

That Reuters article laat year, teporting AMD would make an ARM SoC was one of the bizarre info ever to me. Nvidia I could understand, but AMD?

Former AMD chip architect says it was wrong to can Arm CPU

Cancelled the K12 CPU? Big mistake. Huge, says Jim Keller

www.theregister.com

Keller left but that doesn't mean AMD fired all those engineers that worked on that project.

It's possible they kept it alive in some form, making tweaks here and there and keeping their relevant engineers abreast of the latest ARM developments to ensure they can jump into the foray when the time is ripe.

 

[yottabit]

Simply switching to ARM, will not suffice AMD to catch up with Qualcomm/Apple in power efficiency. Apple/Qualcomm has advantages other than the use of the ARM ISA itself.

1. Custom CPU core designs.

Both Apple and Qualcomm have custom ARM core designs, which as of now, are better than the stock ARM cores AMD/Nvidia will have to use. The next generation of ARM cores (X5/A725) might change that though.

2. A legacy in mobile computing.

Both Apple and Qualcomm have been designing smartphone SoCs for more than a decade. Thus, they have mastered the art of power efficiency, with many tricks and solutions- especially on low-power and idle efficiency. AMD/Nvidia don't have this experience in making smartphone SoCs, and thus lack the knowledge/experience that comes with it.

You forgot “completely vertically integrated hardware/software/OS” - AMD doesn’t make laptops they just make chips.

A more fair comparison would be to compare the efficiency of another vertically integrated device like a PS5 (powered by AMD CPU and GPU) to an equivalent hobbled together gaming PC built on the console launch day for instance. That discrepancy is huge. At the end of the day it doesn’t matter to the consumer but I’m still bothered this isn’t raised in the many and frequent comparisons to Apple products here lately.

 

[igor_kavinski]

A more fair comparison would be to compare the efficiency of another vertically integrated device like a PS5 (powered by AMD CPU and GPU) to an equivalent hobbled together gaming PC built on the console launch day for instance.

Consoles are sold at a loss. Also, their efficiency degrades in comparison with the evolving PC market. A PC based on 8700G may be more efficient than a console and will get even more efficient with Zen 5 APUs before the current console generation is put on life support.

 

[NostaSeronx]

I'm thinking we'll see AMD replace the Trustzone A5 cores with RISC-V first. Then stand alone products later.

Spec RISC-V TEE stuff isn't finished yet, so Trustzone A5 will still be a thing. To escape Trustzone AMD would need to license SiFive's solution: https://starfivetech.com/uploads/WorldGuard_TEE_REE_App_Note_v1.3.pdf Which means they trade one license for another.

There is no time for RISC-V, we are going to see a big push for ARM Windows devices next year. RISC-V is at least 10 years behind.

RISC-V is expected to be ubiquitous by 2025.
SiFive P670-based chips are expected to be a Q2'24 thing. Microsoft does have a code for RISC-V images/apps.
0x8664 - AMD64
0xAA64 - ARM64
0x5064 - RV64

I won't be surprised if WoRV becomes a thing immediately SiFive chips strike the shores of the USA. Then, everyone (AMD included) can get the ripe rewards of someone else taking the risk.

100% instant ATX boards with RISC-V on em. Microsoft will go Taskforce Five and get Windows on it.