16-core AMD Ryzen AI Max+ 395 "Strix Halo" APU Outshines Ryzen 9 7945HX3D in Geekbench

[GGforever]

Ever since AMD introduced Strix Point, enthusiasts like ourselves have been eagerly awaiting details regarding the high-end Strix Halo APUs with integrated graphics that are rumored to be powerful enough for the system to not require discrete graphics at all. Leaks regarding the upcoming performance mobile APU lineup have been trickling out steadily, and a fresh new Geekbench leak reveals the CPU performance of the Ryzen AI+ Max 395 APU, which boasts a 16-core configuration consisting entirely of Zen 5 cores, unlike Strix Point which features a mix of Zen 5 and the smaller Zen 5c cores. And oh dear, are the numbers ever so lucrative.



The APU managed to rake in 2,849 points in the single-core department, and a whopping 20,708 points in multicore. As Videocardz correctly notes, this result is far ahead of AMD's current top-end mobile offering, the Ryzen 9 7945HX3D, which manages around 16,900 points in the multicore test. In single-core, however, the Ryzen 9 7945HX3D does edge ahead, with around 2,900 points. That said, the ROG Flow Z13 laptop that the APU was housed in is most likely still in the testing phase, so it is entirely possible that the final product will sport even better performance. That being said, the Apple M4 Max SoC, however, remains in a league of its own with 3,800 points in single-core and a shocking 25,000 points in multicore. With CES 2025 just around the corner, it's only a matter of weeks before the Ryzen AI Max+ lineup finally sees the light of day and reaches our hands.

View at TechPowerUp Main Site | Source

 

[wolf]

Strix Halo is one of the most exciting upcoming products to me right now. I really want to see how that integrated GPU does with 40CU's and DDR5, theres so much potential, and I hope it's not too badly bottlenecked by memory bandwidth.

 

[trsttte]

I'm not even all that concerned about if the GPU will be great or not, killing all excuses to use a thermal throttled and inefficient discrete gpu is the biggest win of this design.

I just want to see some workstations using these chips, some models from HP already made news on one form or another, can't wait to see something from the other guys. And fingers crossed for some of them to use LPCAMM.

 

[A&P211]

I want to use a laptop with this APU at work, my current one used at work has a 780m. Its good enough for older games.

 

[eidairaman1]

How about running them untethered to the internet and then benchmark them

 

[King Mustard]

The article shows how far ahead Apple is in performance than even CPU companies that have been in the business decades.

 

[A&P211]

The article shows how far ahead Apple is in performance than even CPU companies that have been in the business decades.

It is embarrassing for Intel and AMD.

Apple controls everything from the hardware to the software.

 

[King Mustard]

Apple controls everything from the hardware to the software.

It is not the software making those numbers / scores in those industry-standard benchmarks. It is the chip.

 

[Athena]

It is not the software making those numbers / scores in those industry-standard benchmarks. It is the chip.

software optimization does play a big role in things, and when you control everything, you can basically optimize the heck out of it

There is a long history of Intel for example playing software games with the compiler to favor only them

 

[AcE]

The article shows how far ahead Apple is in performance than even CPU companies that have been in the business decades.

It is embarrassing for Intel and AMD.

Apple gets destroyed by the full AMD processors in data center so no worries. This isn’t AMDs final forme.

 

[LittleBro]

It is not the software making those numbers / scores in those industry-standard benchmarks. It is the chip.

It's chip and optimizations. If you can optimize chip for very specific use, it can became much more efficient and/or powerful.

AMD and Intel chips are universal, that means cross platform. Even though Apple was using Intel chips for long time, they always optimized software for it.
That's the reason when some people were trying to run "hackintosh" (running Mac on non-oficially supported hardware) it required a lot of tweaking.

Anyway, GeekBench is a piece of shit benchmark. Even Lunar Lake had interesting multi core scores in the benchmark and how did it end up ...

 

[kondamin]

It's chip and optimizations. If you can optimize chip for very specific use, it can became much more efficient and/or powerful.

AMD and Intel chips are universal, that means cross platform. Even though Apple was using Intel chips for long time, they always optimized software for it.
That's the reason when some people were trying to run "hackintosh" (running Mac on non-oficially supported hardware) it required a lot of tweaking.

Anyway, GeekBench is a piece of shit benchmark. Even Lunar Lake had interesting multi core scores in the benchmark and how did it end up ...

What’s wrong with lunar lake? it’s a fine laptop chip

 

[LittleBro]

What’s wrong with lunar lake? it’s a fine laptop chip

Well, the benchmark results were somewhat better than reality in multi core.

Same as with Qualcomm's X CPU. In GeekBench it was not bad, but the reality ...

 

[de.das.dude]

The article shows how far ahead Apple is in performance than even CPU companies that have been in the business decades.

It is embarrassing for Intel and AMD.

paying an ARM (pun intended) and leg to buy an ARM processor with an anemic instruction set.. performance isnt something i would really compare.

Also as others stated, when you control everything in the pipeline, it makes it a million times easier. look at spaceX, different industry but same methodology.

x64 consumer stuff has to contend with being able to be run on 100s of thousands of different configurations. Apples new software doesnt even work properly on an apple device which is a few years old. Remember planned obsolescence?

 

[RaphaelOne]

The article shows how far ahead Apple is in performance than even CPU companies that have been in the business decades.

It is embarrassing for Intel and AMD.

Does Apple produce x86 chips? IMO comparing the performance of cores made in different architectures is a bit risky and can give strange results. Usually, if we look at broader tests, we will notice that the performance of different architectures will be different depending on the type of tasks. One synthetic test is not enough to draw far-reaching conclusions.

The previous speaker is right, a lot depends on the compiler and the software optimization capabilities. The x86 sector uses two compilers: (A) from Intel (B) from Microsoft, based on Intel. It was like this for decades, someone please correct me if it has changed. IMO, we should appreciate the long-term success of AMD, which they achieved in very unfavorable conditions. This means that using a compiler fully optimal for ZEN x86 would increase the performance of the physical chip.
For example, in the old days of the AM3 socket, AMD used their own compiler created for the bulldozer architecture and made internal tests claiming that even the mid-range FX6350 (3M/6T) often equaled the i7 SandyBridge (4C/8T) while the FX8350 (4M/8T) outstripped all sand i7s. Yes, changing the compiler can bring noticeable differences in performance and performance/watt of the same physical chip.

I hope that I live to see AMD achieve such high market shares of X86 servers that the AMD ZEN x86 compiler becomes the industry standard. Everything is possible, AMD after introducing EPYC systems started winning tenders for supplying x86 chips to the military, where Intel had a monopoly for a long time. The world is changing.

 

[SRB151]

It's not just the ARM processor with Apple. They have a bunch of customized support "chips" on their SOC. Many of these take the load off of the ARM part. The best parallel I can think of is the Amiga. The 68000 was OK, but the 6 or 7 custom chips that made up the chipset for offloading much of the workload, coupled with complete control of the OS, made a preemptive multitasking, 4096 color PC with a GUI while 286 PC's with monochrome or maybe CGA, running on dos, was the norm. (or a monochrome Mac with the same CPU)

When you have total control of the hardware AND software, like Apple does, you can not only streamline the software, but your free to do as you pleases with the hardware, graphics, sound, IO etc. to make a much more EFFICIENT computer, that doesn't need the latest, greatest, and fastest CPU to get a lot more done.

Not to mention that Microsoft controls the OS, and they seem to have a difficult time killing major bugs. I don't think they spend a whole lot of time optimizing for anything. They spend too much time creating bells and whistles that make thing difficult and slow, hoping users just look at the "shiny object" (Co-pilot, anyone?)

 

[de.das.dude]

It's not just the ARM processor with Apple. They have a bunch of customized support "chips" on their SOC. Many of these take the load off of the ARM part. The best parallel I can think of is the Amiga. The 68000 was OK, but the 6 or 7 custom chips that made up the chipset for offloading much of the workload, coupled with complete control of the OS, made a preemptive multitasking, 4096 color PC with a GUI while 286 PC's with monochrome or maybe CGA, running on dos, was the norm. (or a monochrome Mac with the same CPU)

When you have total control of the hardware AND software, like Apple does, you can not only streamline the software, but your free to do as you pleases with the hardware, graphics, sound, IO etc. to make a much more EFFICIENT computer, that doesn't need the latest, greatest, and fastest CPU to get a lot more done.

Not to mention that Microsoft controls the OS, and they seem to have a difficult time killing major bugs. I don't think they spend a whole lot of time optimizing for anything. They spend too much time creating bells and whistles that make thing difficult and slow, hoping users just look at the "shiny object" (Co-pilot, anyone?)

a good analogy would be to see how efficient the console industry actually is with less powerful chips from AMD.

 

[Punkenjoy]

Well, Geekbench score are just good to compare how good a processors is at running geekbench. It can't be extrapolated too much.

As for Apple M series, just remember that their main design goal is to have a fast and efficient consumer processors. This is a different design philosophy than AMD and Intel where their goals is to design CPU that is going to make high margins in data centers. The fact they sell it as a consumer CPU is more a bonus than their main goals.

Their reals client is the large hyperscaler like Azure, AWS, GCP, etc. it's where they make most of their bottom line in profits.

If Intel and AMD would design a CPU uARCH from scratch specifically aim at the consumer market, the results would probably be really different.

 

[A&P211]

It is not the software making those numbers / scores in those industry-standard benchmarks. It is the chip.

Yes it is

 

[psydroid]

It is not the software making those numbers / scores in those industry-standard benchmarks. It is the chip.

I see diminishing returns in Intel and AMD chips, although they've been increasing the performance again over the past few years. I wonder what the situation will be like in 5 years.