Latest AMD AGESA Hints at Ryzen 7000G "Phoenix" Desktop APUs

[btarunr]

AMD is preparing to launch its first APUs on the Socket AM5 desktop platform, with the Ryzen 7000G series. While the company has standardized integrated graphics with the Ryzen 7000 series, it does not consider the regular Ryzen 7000 series "Raphael" processors as APUs. AMD considers APUs to be processors with overpowered iGPUs that are fit for entry-mainstream PC gaming. As was expected for a while now, for the Ryzen 7000G series, AMD is tapping into its 4 nm "Phoenix" monolithic silicon, the same chip that powers the Ryzen 7040 series mobile processors. Proof of "Phoenix" making its way to desktop surfaced with CPU support lists for the latest AGESA SMUs (system management units) compiled by Reous, with the AGESA ComboAM5PI 1.0.8.0 listing support for "Raphael," as well as "Phoenix." Another piece of evidence was an ASUS B650 motherboard support page that listed a UEFI firmware update encapsulating 1.0.8.0, which references an "upcoming CPU."

Unlike "Raphael" and "Dragon Range," "Phoenix" is a monolithic processor die built on the TSMC 4 nm foundry node. Its CPU is based on the latest "Zen 4" microarchitecture, and features an 8-core/16-thread configuration, with 1 MB of L2 cache per core, and 16 MB of shared L3 cache. The star attraction here is the iGPU, which is based on the RDNA3 graphics architecture, meets the DirectX 12 Ultimate feature requirements, and is powered by 12 compute units worth 768 stream processors. Unlike "Raphael," the "Phoenix" silicon is known to feature an older PCI-Express Gen 4 root complex, with 24 lanes, so you get a PCI-Express 4.0 x16 PEG slot, one CPU-attached M.2 NVMe slot limited to Gen 4 x4, and a 4-lane chipset bus. "Phoenix" features a dual-channel (4 sub-channel) DDR5 memory controller, with native support for DDR5-5600. A big unknown with the Ryzen 7000G desktop APUs is whether they retain the Ryzen AI feature-set from the Ryzen 7040 series mobile processors.



View at TechPowerUp Main Site | Source

 

[Space Lynx]

took long enough, but honestly when a Intel N100 mini-PC can run two 4k videos side by side without issue and only cost $179 for the entire system (bought my dad this mini PC recently) AMD just can't compete on price to performance for the casual user for someone like my Dad and his older age group who just need something that works.

and if you want gaming, I mean as much as these will most likely cost you would be better off building a cheap mini itx rig with a dedicated card at around the same price.

so imo this is dead on arrival

 

[TumbleGeorge]

I was still hoping for further rework for the 7000G series(desktop APUs) and support for faster DDR5 as standard. Still, it will be almost a year, if not a whole year, between the release of the desktop version, after the mobile version, and today for DDR5, the prices of modules faster than 5600 are quite bearable for the budget sector of the market.

 

[Shou Miko]

took long enough, but honestly when a Intel N100 mini-PC can run two 4k videos side by side without issue and only cost $179 for the entire system (bought my dad this mini PC recently) AMD just can't compete on price to performance for the casual user for someone like my Dad and his older age group who just need something that works.

and if you want gaming, I mean as much as these will most likely cost you would be better off building a cheap mini itx rig with a dedicated card at around the same price.

so imo this is dead on arrival

I am still hoping for gaming to see a APU in my life time before I get too old and grumpy that can do games not at 720/1080p.

 

[ymdhis]

I am still hoping for gaming to see a APU in my life time before I get too old and grumpy that can do games not at 720/1080p.

Strix Halo will be that (up to 2560 shaders), but it's supposed to be mobile only and chiplet based. Strix Point is said to be monolithic but max at 1024 shaders.

If they brought Strix Halo to desktop it would kill all discrete graphics below the x700 category.

 

[Shou Miko]

Strix Halo will be that (up to 2560 shaders), but it's supposed to be mobile only and chiplet based. Strix Point is said to be monolithic but max at 1024 shaders.

If they brought Strix Halo to desktop it would kill all discrete graphics below the x700 category.

That I know, otherwise we have to get Minisforum to help which will be like a expensive option.

 

[Denver]

Strix Halo will be that (up to 2560 shaders), but it's supposed to be mobile only and chiplet based. Strix Point is said to be monolithic but max at 1024 shaders.

If they brought Strix Halo to desktop it would kill all discrete graphics below the x700 category.

The problem with APUs is that DDR5 does not offer enough bandwidth, note that the difference in performance in iGPUs with 6CU vs 12CU is not even close to being equivalent to the increase in shader count.

I thought that with DDR5 we would have 128bit (dual Channel) per module, and then quad channel(256b) with two modules. But the industry decided to stagnate with dual-channel.

 

[john_]

In the ancient times, on AMD motherboards there was something called "Sideport memory". Just a RAM chip on the motherboard, connected with the North bridge where the iGPU was, with a 32bit data bus (I think), to give some extra memory bandwidth to the iGPU. You could use the memory chip alone, living all the system RAM to the CPU, ignore it, using only the system RAM or combine that chip with whatever part of the system RAM was given to the iGPU. I wonder when they will bring that back(probably never). Now that the iGPUs are strong enough for the memory bandwidth to become a bottleneck, even with fast memory, this would have been a nice solution.

AMD HD 4200 - 256MB System Memory

AMD HD 4200 - 128MB Sideport Memory + 128MB System Memory - 5% performance increase, going up to 12% in Game 4, that was the most demanding (Nature).


P.S. Any Intel engineer reading this? How about YOU do it? Because I don't expect AMD to do it.

 

[zigzag]

Are they jumping from v1.0.0.7c to v1.0.8.0?

 

[Shou Miko]

Are they jumping from v1.0.0.7c to v1.0.8.0?

It looks like it from twits I seen on Twitter.

 

[Kapone33]

Finally

 

[AnotherReader]

took long enough, but honestly when a Intel N100 mini-PC can run two 4k videos side by side without issue and only cost $179 for the entire system (bought my dad this mini PC recently) AMD just can't compete on price to performance for the casual user for someone like my Dad and his older age group who just need something that works.

and if you want gaming, I mean as much as these will most likely cost you would be better off building a cheap mini itx rig with a dedicated card at around the same price.

so imo this is dead on arrival

The Phoenix APUs probably would be able to use faster memory than the regular Zen 4 cores, but for most use cases, your hypothetical mini ITX rig would be better. Only the most power constrained cases where a dGPU can't be used would qualify.

 

[TheinsanegamerN]

took long enough, but honestly when a Intel N100 mini-PC can run two 4k videos side by side without issue and only cost $179 for the entire system (bought my dad this mini PC recently) AMD just can't compete on price to performance for the casual user for someone like my Dad and his older age group who just need something that works.

and if you want gaming, I mean as much as these will most likely cost you would be better off building a cheap mini itx rig with a dedicated card at around the same price.

so imo this is dead on arrival

If price is your ONLY object, then duh. But if you value tony computing, then your comment makes no sense. Why would I build an itx rig with a dGPU that will be 4x as big as an API only solution?

I am still hoping for gaming to see a APU in my life time before I get too old and grumpy that can do games not at 720/1080p.

Any tech that benefits APUs will also benefit dGPUs, and game swill become more demanding to take advantage of that.

Llano was supposed to be capable of 1080p gaming. Then it was the vega 7, and now it's strix point. There will always be new demand around the corner.

If you want a current API to reliably play modern games above 1080p, that is a pipe dream that will never happen.

 

[Shou Miko]

I saw this twit and the user is also good to follow as I have said before here on TPU

https://twitter.com/x/status/1709704045827662023

Any tech that benefits APUs will also benefit dGPUs, and game swill become more demanding to take advantage of that.

Llano was supposed to be capable of 1080p gaming. Then it was the vega 7, and now it's strix point. There will always be new demand around the corner.

If you want a current API to reliably play modern games above 1080p, that is a pipe dream that will never happen.

It will, with the power demand to lower it for private people it will be the biggest leap in history if a 150W apu could do 1440p I might not be around to see it because I might not live to be 100+ but it will come.

Because if you can get a good solid pc gaming experience at 1440p at 150W instead of about 500-1000W it will be a good for environment too.

 

[HD64G]

DDR5 with latest BIOS will allow frequencies of at least 8GT/s in near future, so those APUs will not have any problem getting worked out. As for cost, the best in its class always is expensive relative to inferior solutions. Basic APUs as 5600G with AM4 board and 32GB DDR4 3600 are ultra cheap now for anyone of basic needs.

 

[ymdhis]

If price is your ONLY object, then duh. But if you value tony computing, then your comment makes no sense. Why would I build an itx rig with a dGPU that will be 4x as big as an API only solution?


Any tech that benefits APUs will also benefit dGPUs, and game swill become more demanding to take advantage of that.

Llano was supposed to be capable of 1080p gaming. Then it was the vega 7, and now it's strix point. There will always be new demand around the corner.

If you want a current API to reliably play modern games above 1080p, that is a pipe dream that will never happen.

The Vega iGPUs were outdated from the start, but AMD kept using them for way too long. Vega was always kneecapped by lack of memory bandwidth, RDNA2 can do a lot more with lot less, which is why Phoenix is so much faster at equal shader count (and on top of that it's on DDR5).

 

[mechtech]

MORE PCIe LANES!!!

4TB m.2 (top performance) are still too much, so I need to get three 2TB for the same money, so I need at least 6 m.2 slots, preferably 8 slots.

thanks

 

[Regenweald]

I am still hoping for gaming to see a APU in my life time before I get too old and grumpy that can do games not at 720/1080p.

Same, I've been waiting since Llano for that killer desktop product and I feel like AMD has just been half assing the space. Now it's been over a decade and Intel is actually a desktop APU contender, AMD dragged so much ass that Intel had the time to build an entire GPU arch and build proper APUs. Corporate decisionmaking is a unique animal.

 

[Shou Miko]

Same, I've been waiting since Llano for that killer desktop product and I feel like AMD has just been half assing the space. Now it's been over a decade and Intel is actually a desktop APU contender, AMD dragged so much ass that Intel had the time to build an entire GPU arch and build proper APUs. Corporate decisionmaking is a unique animal.

It's also demand, if the demand is not there things don't evolve but look at us in europe with flexing prices on electricity I feel the there is a need for it, instead of being locked into a gaming laptop.

 

[Minus Infinity]

The problem with APUs is that DDR5 does not offer enough bandwidth, note that the difference in performance in iGPUs with 6CU vs 12CU is not even close to being equivalent to the increase in shader count.

I thought that with DDR5 we would have 128bit (dual Channel) per module, and then quad channel(256b) with two modules. But the industry decided to stagnate with dual-channel.

Zen 5 will support much faster DDR5 out of the box than Zen 4, rumours say it is being tested with DDR5 7200 support as standard. Every bit will help.

 

[ADB1979]

Any tech that benefits APUs will also benefit dGPUs, and game swill become more demanding to take advantage of that.

Llano was supposed to be capable of 1080p gaming. Then it was the vega 7, and now it's strix point. There will always be new demand around the corner.

If you want a current API to reliably play modern games above 1080p, that is a pipe dream that will never happen.

The problem there is IMHO in two parts. (1). The "claims" made by the company that produces the product (iGPU / APU). (2). The games that "reviewers" "choose" to test and review with.

If the reviewers test the claims (1), and do so fairly and the product does not live up to those claims, then they SHOULD give reviews that honestly state this, but also what
the part CAN do, and what is can NOT do.

If the reviewers (2) decide to test the latest and greatest games, even at 1080p their performance would be awful and unusable.

All we need is honesty from the manufacturer about the product, and realistic, and honest reviews.