AMD Ryzen AI 9 HX 475 & 470 "Gorgon Point" APUs Surface in Shipping Manifests

[AleksandarK]

Last month, a confidential presentation in China gave AMD's laptop partners their first glimpse of what's coming next. The slides hinted at modest clock-speed improvements and the addition of more entry-level models, all under the codename "Gorgon Point." Not long after, NBD's shipping manifests today show a series of new FP8, FP10, and FP12 product codes. These identifiers don't match any existing "Strix Point" or "Kracken Point" chips, so it's clear AMD is gearing up for a Ryzen AI 400-series refresh set to roll out around mid-2026. Despite the new name, Gorgon Point sticks with the same winning formula. It still uses Zen 5 and Zen 5c CPU cores alongside RDNA 3.5 graphics and XDNA 2 neural accelerators. Core counts haven't changed, so you'll see configurations ranging from 4 to 12 cores just like before, with up to 4x Zen 5 and 8x Zen 5c cores.

What's different, according to the leaked partner slides, is a slight bump in boost clocks. The top-end Ryzen AI 9 HX 475 and 470 chips are now rated for up to 5.2 GHz, a slight but welcome increase from the previous 5.1 GHz, while maintaining a 28 W default TDP. AMD is also broadening its reach into budget laptops. In addition to the Ryzen 9 HX upgrades, the company will introduce new Ryzen 5 and Ryzen 3 models. Early leaks mention parts numbered 440 and 430, as well as a mystery Ryzen 3 SKU. These entry-level chips will use the same Gorgon Point silicon but will be tuned for cost-sensitive devices. Branding will likely follow AMD's recent pattern. Given how AMD has renumbered previous families, slotting Gorgon Point into familiar retail channels should be straightforward. Until AMD makes an official announcement, these shipping manifests and partner leaks are the best clues we have about the performance and efficiency gains in the next wave of AI-accelerated laptops.



View at TechPowerUp Main Site | Source

 

[kondamin]

I hate it when they do that, it's the same product .
Just keep the same name

 

[HOkay]

I hate it when they do that, it's the same product .
Just keep the same name

I came here to say the same thing, this is standard marketing rubbish to confuse consumers into thinking they're getting "next gen" when in fact it's a very slight power efficiency gain at best.

 

[Nostras]

I hate it when they do that, it's the same product .
Just keep the same name

They tried that with the XT suffix. That didn't work out and since this just works and there's no (notable) backlash...

 

[Daven]

Rebrands are the worst!

 

[Rightness_1]

Why does AMD cling on to their awful RDNA 3.5, when RDMA 4 has proven to be a worthy architecture?

 

[Daven]

Why does AMD cling on to their awful RDNA 3.5, when RDMA 4 has proven to be a worthy architecture?

Some of it may have to do with the presence of an NPU doing some of the things RDNA4 does well. There also may not be enough space or power to fit an RDNA4 chip into an iGPU. Also the main selling point of RDNA4 is RT improvement which doesn't make much sense to implement in an iGPU. Finally, one of the big things about RDNA4 are the very high clocks. Can’t do that in a mobile chip.

Actually the iGPU clocks are quite high.

 

[TumbleGeorge]

Still highest version of PCIe for this series is 4.0. Pass!

 

[Tyler-98-W68]

But but but, platform longevity right guys??????? AMD is the best for that (Rebrand after rebrand)

 

[Daven]

Everyone, this is a mobile chip meaning you can’t connect a discrete graphics card without an eGPU case connected through USB4 or TB. It is not upgradeable as it will be soldered to the motherboard. The iGPU+NPU combo targets different usage cases than straight PC gaming.

Therefore, RDNA4, PCIe 5 and platform longevity don’t make a lot of sense here.

What’s wrong here is the rebrand. Shame on any company that rebrands or refreshes with the idea of fooling a customer out of their money. If you don’t have anything new then you don’t get to release a new product until you do.

 

[Panther_Seraphin]

Some of it may have to do with the presence of an NPU doing some of the things RDNA4 does well. There also may not be enough space or power to fit an RDNA4 chip into an iGPU. Also the main selling point of RDNA4 is RT improvement which doesn't make much sense to implement in an iGPU. Finally, one of the big things about RDNA4 are the very high clocks. Can’t do that in a mobile chip.

Actually the iGPU clocks are quite high.

Normally you finish designs of the Desktop/server class hardware, then filter them down into dGPUs for laptops by both redesigning the card layout to suit with a laptop and then bin parts to suit power draw/temperature constraints. Finally you tend to design down the core architecture to suit iGPUs etc.

I honestly suspect there was a cancellation of all RDNA 4 development when the high end chip failed to pass tape out/design phase and AMD went all in on UDNA development. I havent even seen anything about dedicated laptop class parts let alone iGPU variants and from the little I have seen about Zen 6 the IO die also seems to be RDNA 3 based.

 

[Daven]

Normally you finish designs of the Desktop/server class hardware, then filter them down into dGPUs for laptops by both redesigning the card layout to suit with a laptop and then bin parts to suit power draw/temperature constraints. Finally you tend to design down the core architecture to suit iGPUs etc.

I honestly suspect there was a cancellation of all RDNA 4 development when the high end chip failed to pass tape out/design phase and AMD went all in on UDNA development. I havent even seen anything about dedicated laptop class parts let alone iGPU variants and from the little I have seen about Zen 6 the IO die also seems to be RDNA 3 based.

This sounds like a good assessment. I haven't heard much about RDNA4 laptop discrete GPUs either. When it comes to high end laptop form factor GPU performance, I think Strix Halo type SoC is the path AMD will take going forward. Probably the next version will use UDNA.

 

[Panther_Seraphin]

UDNA on paper looks amazing, lets see how the it is executed as I dont suspec it will be cheap but it could be massively powerful/scalable looking at the MI350x design language.

 

[Daven]

Still highest version of PCIe for this series is 4.0. Pass!

I should have clarified that Fire Range is the laptop version you are looking for. It has PCIe 5.0 for connecting internal discrete laptop GPUs and SSDs. This 28W laptop SoC is not really designed for discrete GPUs and there may be enough of a power penalty to justify avoiding PCIe 5.0 SSDs for the target slim laptop form factor.

 

[A&P211]

Some of it may have to do with the presence of an NPU doing some of the things RDNA4 does well. There also may not be enough space or power to fit an RDNA4 chip into an iGPU. Also the main selling point of RDNA4 is RT improvement which doesn't make much sense to implement in an iGPU. Finally, one of the big things about RDNA4 are the very high clocks. Can’t do that in a mobile chip.

Actually the iGPU clocks are quite high.

I use one of those AI HX 370 chips for a plex server, the clock speeds for the igpu is 2900mhz.

But but but, platform longevity right guys??????? AMD is the best for that (Rebrand after rebrand)

Everyone does it.

 

[Tyler-98-W68]

I use one of those AI HX 370 chips for a plex server, the clock speeds for the igpu is 2900mhz.


Everyone does it.

How many rebrands has Intel done for their mobile cpus? They haven't. 12th 13th 14th Core ultra series 1 core ultra series 2, you could argue 12 to 13 and 13-14 was a rebrand but you got additional clock speed, cores and performance, not identical chips reused.

 

[Panther_Seraphin]

How many rebrands has Intel done for their mobile cpus? They haven't. 12th 13th 14th Core ultra series 1 core ultra series 2, you could argue 12 to 13 and 13-14 was a rebrand but you got additional clock speed, cores and performance, not identical chips reused.

2nd to 6th Gen intel comes a knocking.

13th and 14th were also identical parts with clock speed/production benefits being the only difference.

 

[Tyler-98-W68]

2nd to 6th Gen intel comes a knocking.

13th and 14th were also identical parts with clock speed/production benefits being the only difference.

Interesting so there was nothing different between the 2nd to 6th gen Intel mobile parts? No process shrink nothing?

 

[Panther_Seraphin]

Interesting so there was nothing different between the 2nd to 6th gen Intel mobile parts? No process shrink nothing?

Sure 32nm to 22 in 2nd to 3rd gen then 22 to Intels re-occuring 14nm at the 5th gen.

But when you look at performance gains and especially core growth its horrendous and even with the node reduction the power requirements were basically the same part for part.

2860QM
4 Core 8 Thread part
45 Watt TDP

Base	Turbo (1C/2C/4C)

2.3 GHz​

3.4/3.3/3.3 GHz

Release date - January 2011

6920HQ
4 Core 8 Thread part
45 Watt TDP

Base	Turbo (1C/2C/4C)

2.9 GHz

3.8/3.6/3.4 GHz

Release date - September 2015


So nearly 5 years of development and there has been 2 node decreases and near 0 benefit in power but between 100-400mhz benefit and this is BEST CASE. There was a mid life addition to the 2nd gen that narrowed the growth to 100-200mhz at most.

2860QM

2.5 GHz

3.6/3.5/3.3 GHz

Sounds like a perfect way to spend 300+ dollars on new CPU while requiring new RAM and Motherboard at the same time.
This was prime time Intel profit gouging as it wasnt too long after when suddenly 4 cores were not the be all and end all and 6/8/10 core parts arrived in very short order. The MOMENT Zen 1 releases Intel "magically" has 6 and 8 core parts on offer to the public within 6 months and less than a year after the 7th gen release.

 

[Lionheart]

Stop rehashing the same shit AMD, couldn't even give us 8 Zen 5 cores & 4 Zen5c cores?? Plus the ZenC core naming scheme sucks! Why couldn't you just copy Intel! There's my whiny rant for today.

Interesting so there was nothing different between the 2nd to 6th gen Intel mobile parts? No process shrink nothing?

Omg bro how low do you have to go to suck Intel corporate D lmao, who the hell cares, these companies don't give a rats ass about us.

 

[A&P211]

How many rebrands has Intel done for their mobile cpus? They haven't. 12th 13th 14th Core ultra series 1 core ultra series 2, you could argue 12 to 13 and 13-14 was a rebrand but you got additional clock speed, cores and performance, not identical chips reused.

Lots of the i5 intel processor are rebrands with the same amount of cores.

 

[Daven]

How many rebrands has Intel done for their mobile cpus? They haven't. 12th 13th 14th Core ultra series 1 core ultra series 2, you could argue 12 to 13 and 13-14 was a rebrand but you got additional clock speed, cores and performance, not identical chips reused.

Intel literally called one generation, Coffee Lake refresh. Oh wait they also called another generation Raptor Lake refresh. You really don’t know your chip history.

Edit: oh and they rebranded 14th gen models into Core 1 and 2 series mobile chips recently. That’s a real slimeball move.

Edit2: BTW as some have already said, all chip companies refresh and rebrand. ALL of them. As someone else said, companies are not your friends. They care only about money.

Edit3: oh and I forgot Haswell had an infamous refresh as well.

Intel Core 'Haswell' Refresh processors - ASBIS

BUY ONLINE 4th generation Intel® Core™ processor family Built in the LGA1150 package and compatible with motherboards running 8-series chipset with BIOS updates, the chips are accompanied by an armada of motherboards based on the latest 9-series chipset, which support the chips out of the box...

news.asbis.com

 

[Tyler-98-W68]

Sure 32nm to 22 in 2nd to 3rd gen then 22 to Intels re-occuring 14nm at the 5th gen.

But when you look at performance gains and especially core growth its horrendous and even with the node reduction the power requirements were basically the same part for part.

2860QM
4 Core 8 Thread part
45 Watt TDP

Base	Turbo (1C/2C/4C)

2.3 GHz​

3.4/3.3/3.3 GHz

Release date - January 2011

6920HQ
4 Core 8 Thread part
45 Watt TDP

Base	Turbo (1C/2C/4C)

2.9 GHz

3.8/3.6/3.4 GHz

Release date - September 2015


So nearly 5 years of development and there has been 2 node decreases and near 0 benefit in power but between 100-400mhz benefit and this is BEST CASE. There was a mid life addition to the 2nd gen that narrowed the growth to 100-200mhz at most.

2860QM

2.5 GHz

3.6/3.5/3.3 GHz

Sounds like a perfect way to spend 300+ dollars on new CPU while requiring new RAM and Motherboard at the same time.
This was prime time Intel profit gouging as it wasnt too long after when suddenly 4 cores were not the be all and end all and 6/8/10 core parts arrived in very short order. The MOMENT Zen 1 releases Intel "magically" has 6 and 8 core parts on offer to the public within 6 months and less than a year after the 7th gen release.

So I said rebrand, you said intel's 2nd to 6th gen was a rebrand, but then go on to say there was a process improvement (actually going from 32nm to 14nm) So that's your idea of a rebrand?

Interesting choice of words. So you say the performance gains are horrendous?


So a 23% increase in single thread and 36% increase in multithread performance is bad and not good enough?

What about on the desktop? 30% and 38% increases, but you said it was a rebrand, seems to me they were working quite well.

Also remember this was during the "tick" "tock" era of intel introducing new CPU's.

What was AMD doing at that time? They had absolutely nothing energy or performance competitive

Intel literally called one generation, Coffee Lake refresh. Oh wait they also called another generation Raptor Lake refresh. You really don’t know your chip history.

Edit: oh and they rebranded 14th gen models into Core 1 and 2 series mobile chips recently. That’s a real slimeball move.

Edit2: BTW as some have already said, all chip companies refresh and rebrand. ALL of them. As someone else said, companies are not your friends. They care only about money.

Edit3: oh and I forgot Haswell had an infamous refresh as well.

Intel Core 'Haswell' Refresh processors - ASBIS

BUY ONLINE 4th generation Intel® Core™ processor family Built in the LGA1150 package and compatible with motherboards running 8-series chipset with BIOS updates, the chips are accompanied by an armada of motherboards based on the latest 9-series chipset, which support the chips out of the box...

news.asbis.com

I'm quite well versed in chip history, since I"ve owned both intel and amd systems (desktop and laptops) over the last 25 years.

Haswell refresh..........yup one hell of a refresh, 4770k to 4790k, it was a huge refresh because the 4790K could overclock huge compared to the existing haswell CPU's

I could do 4.7ghz on air without anything exotic. But that wasn't much of an improvement right, not a tangible increase in performance? The OP's article referenced a 100mhz clock speed increase.....

Stop rehashing the same shit AMD, couldn't even give us 8 Zen 5 cores & 4 Zen5c cores?? Plus the ZenC core naming scheme sucks! Why couldn't you just copy Intel! There's my whiny rant for today.


Omg bro how low do you have to go to suck Intel corporate D lmao, who the hell cares, these companies don't give a rats ass about us.

I see you can't form proper sentences and have to speak in "bro speak" I guess that means you haven't been weened off AMD's tits yet either?

 

[Lionheart]

So I said rebrand, you said intel's 2nd to 6th gen was a rebrand, but then go on to say there was a process improvement (actually going from 32nm to 14nm) So that's your idea of a rebrand?

Interesting choice of words. So you say the performance gains are horrendous?


So a 23% increase in single thread and 36% increase in multithread performance is bad and not good enough?

What about on the desktop? 30% and 38% increases, but you said it was a rebrand, seems to me they were working quite well.

Also remember this was during the "tick" "tock" era of intel introducing new CPU's.

What was AMD doing at that time? They had absolutely nothing energy or performance competitive


I'm quite well versed in chip history, since I"ve owned both intel and amd systems (desktop and laptops) over the last 25 years.

Haswell refresh..........yup one hell of a refresh, 4770k to 4790k, it was a huge refresh because the 4790K could overclock huge compared to the existing haswell CPU's

I could do 4.7ghz on air without anything exotic. But that wasn't much of an improvement right, not a tangible increase in performance? The OP's article referenced a 100mhz clock speed increase.....




I see you can't form proper sentences and have to speak in "bro speak" I guess that means you haven't been weened off AMD's tits yet either?

And what sentence's that I can't form you referring to? All because I said "bro" you have to stick me under some made up category of your's "bro speak" Lmao what the hell even is that? Great comeback "BRO!" and yeah I love sucking on AMD teet, if you're smart enough & depending on what country you live in they got the best bang for buck, Mmm feels good.

 

[Panther_Seraphin]

So a 23% increase in single thread and 36% increase in multithread performance is bad and not good enough?

What about on the desktop? 30% and 38% increases, but you said it was a rebrand, seems to me they were working quite well.

Also remember this was during the "tick" "tock" era of intel introducing new CPU's.

For 5 years of development you were happy with that?

Christ we ridicule AMD/nVidia for their small gains per generation but at least they are hitting this after 1 year not 5.
Imagine if we went from 1080ti to 2080ti performance after 5 years of development. That is the same scale of performance gain that you are deeming "quite well"

Lets even look at CPUs

We went from Zen 1 to Zen 4 in the same amount of time
Intel went from 7700k to 12900k

So yeah Intel being happy as larry to charge you $300+ per generation for single digit increases as well as new RAM + Motherboard required every two. If it were up to Intel I can imagine something like the 10900k still being king of castle for consumers today if we were to follow their growth plans and did not have to answer to AMDs challenge.