AMD Zen 5 Technical Deep Dive

[Visible Noise]

It's likely present and functional, but with a much lower achieved range. Or perhaps it was just overlooked in stead of showcasing the flagship processor under a different light - extreme overclocking, curve shaper, etc. which are less boring than just old PBO for that specific SKU.

Nope, it’s in the notes slides.

 

[Dr. Dro]

Nope, it’s in the notes slides.

View attachment 355315

Very odd. I missed that somehow. I wonder if it already comes at the edge of what AMD considers feasible, or if they fear issues with unreliable operation while boosting at high frequencies (some problem arising from operating conditions like we are seeing with the i9's eTVB), and taking a cautious approach.

 

[W1zzard]

Nope, it’s in the notes slides.

View attachment 355315

Nice find, email sent to AMD to confirm

 

[Vincero]

Surprising there's still no plans in this generation to bring the 'c' cores into their main desktop product offering (the 8000g series being both last gen zen4 and also more of a laptop part).

I've often thought AMD could kill two birds with one stone by doing an alternate IO die configuration loosing one of the chiplet infinity fabric channels on the IO die and instead splicing on some zen4c/5c cores - sure they'll end up with a slightly bigger IO die (likely a few mm² bigger) but they could then offer lower cost quad-core AM5 CPUs using the IO die cores only (and AM5 really needs some cheap low-end parts to exist once AM4 really is dead), and have 4c + 8 core desktop products (and whatever derivative SKUs they'd want).
Admittedly the quad-core AM5 using just 'c' cores would be an odd product outside of just being a value proposition - tons of IO capability compared to the 8000g series but not a huge amount of CPU power to exploit it, but I could see home NAS / niche appliance / embedded usage scenarios where the additional 8000g Radeon CUs wouldn't be utilised and a smaller essentially 'monolithic' die with better IO options would be desireable.

I reckon there's a large amount of bare minimum office PCs being sold that utilise new Intel Core-i3 chips instead of AM4 4000/5000g CPUs due to AM4 being seen as older platform, the newer i3's having better IPC and (even though Intel's process tech is behind) comparatively acceptable power usage. Eventually that existing AM4 inventory is going to end or not be used for most new sales.

Obviously they can still make the normal IO die for the 16 core CPU configurations they currently push out - not suggesting they abandon that.

 

[CosmicWanderer]

I'm surprised XDNA 2 isn't making it to desktop Ryzen 9000X processors, especially given all the hype around Copilot+ PCs. We'll probably only see it in Ryzen 9000G desktop processors.

 

[Cifu]

Surprising there's still no plans in this generation to bring the 'c' cores into their main desktop product offering (the 8000g series being both last gen zen4 and also more of a laptop part).

I've often thought AMD could kill two birds with one stone by doing an alternate IO die configuration loosing one of the chiplet infinity fabric channels on the IO die and instead splicing on some zen4c/5c cores - sure they'll end up with a slightly bigger IO die (likely a few mm² bigger) but they could then offer lower cost quad-core AM5 CPUs using the IO die cores only (and AM5 really needs some cheap low-end parts to exist once AM4 really is dead), and have 4c + 8 core desktop products (and whatever derivative SKUs they'd want).
Admittedly the quad-core AM5 using just 'c' cores would be an odd product outside of just being a value proposition - tons of IO capability compared to the 8000g series but not a huge amount of CPU power to exploit it, but I could see home NAS / niche appliance / embedded usage scenarios where the additional 8000g Radeon CUs wouldn't be utilised and a smaller essentially 'monolithic' die with better IO options would be desireable.

I reckon there's a large amount of bare minimum office PCs being sold that utilise new Intel Core-i3 chips instead of AM4 4000/5000g CPUs due to AM4 being seen as older platform, the newer i3's having better IPC and (even though Intel's process tech is behind) comparatively acceptable power usage. Eventually that existing AM4 inventory is going to end or not be used for most new sales.

Obviously they can still make the normal IO die for the 16 core CPU configurations they currently push out - not suggesting they abandon that.

Not to mention the server parts, the Zen4c absent even the Epyc 4004 series, while it's exits in the Siena and Bergamo server CPU's...

 

[stimpy88]

Surprising there's still no plans in this generation to bring the 'c' cores into their main desktop product offering (the 8000g series being both last gen zen4 and also more of a laptop part).

Does it serve any real purpose on the desktop though? We have justification of these slow, cooler, low power cores in the laptop market, and certainly the handheld market, which should all be C-cores if you ask me.

But other than marketing, what would adding 4 or 8 more slow C-cores actually offer the desktop user?

Intel is f'd without them due to not being able to power or cool much more than 8 P-cores, but is AMD in that boat?

 

[wNotyarD]

Does it serve any real purpose on the desktop though? We have justification of these slow, cooler, low power cores in the laptop market, and certainly the handheld market, which should all be C-cores if you ask me.

But other than marketing, what would adding 4 or 8 more slow C-cores actually offer the desktop user?

Intel is f'd without them due to not being able to power or cool much more than 8 P-cores, but is AMD in that boat?

AMD could actually pull all C-core processors for the Athlon and Ryzen 3 lines. In that end of the market, high clocks aren't necessary.

 

[stimpy88]

AMD could actually pull all C-core processors for the Athlon and Ryzen 3 lines. In that end of the market, high clocks aren't necessary.

Does Intel do this?

 

[wNotyarD]

Does Intel do this?

Not quite. The 14th gen i3 are all P-core (4P+0E/8T) though lower clocked than its i5/7/9 siblings and the Intel Processors (what we would earlier call Pentiums or Celerons) are either 2P/4T or 4E/4T, so hypothetical all Zen 5C models (2C/4T and 4C/8T) could match them without much issues.

 

[stimpy88]

Not quite. The 14th gen i3 are all P-core (4P+0E/8T) though lower clocked than its i5/7/9 siblings and the Intel Processors (what we would earlier call Pentiums or Celerons) are either 2P/4T or 4E/4T, so hypothetical all Zen 5C models (2C/4T and 4C/8T) could match them without much issues.

I agree with you that IF it is cheaper for AMD to use Zen5 C-cores on their lowest end products, then it would be a good idea, and no-brainer solution for more profit.

But the C-cores are made on the much more costly 3nm process, (which is baffling to me that it is not done the other way round, as the full fat cores would benefit more from this)... But that might negate the cost savings part...

 

[EatingDirt]

?? No, they do not. Desktop Intel chips are fully monolithic at the moment.

My wording was off. Yes, it's not an I/O die on the Intel CPU, but all the functions of AMD's I/O die are within the Intel CPU. It's still using power all the time to allow the I/O to function.

The eTVB issue is separate and is not why the chips are dying... but no, AMD idle power has been several times that of Intel because they are monolithic chips. This is the same reason the APUs have lower idle power, no IOD to constantly draw power. Zen 5 is great, but this is clearly not something it has addressed thus far. AMD likely cannot address it, at least not with a cost-effective solution yet.

A 5nm I/O die would likely bring Zen's single-threaded power consumption near-parity with Intel.

No one seems to bother with idle power consumption tests anymore. TPU dropped idle power consumption after the 12900k review, and in that review, the idle system draw of Zen 3 was lower than Intel's counterparts, but pretty insignificant when it comes down to it.

The single-threaded power draw of Zen 5 CPU's is a difference of... 2-7w more than their Intel counterparts, sure, that's 10-20%, but like I said, a 5nm I/O die could probably bring that down to under 10%.

Power draw is only an actual issue when it comes to laptops/mobile devices. It's basically meaningless in cost for desktops(it's at most, $20/y, and that's for 24/7 idle operation in high-energy cost countries).

 

[NuttinMuch]

I think the biggest news here is that previous iterations of Zen contained veal

 

[W1zzard]

I think the biggest news here is that previous iterations of Zen contained veal

View attachment 355347

OMG I'm such a noob .. fixed .. thanks for reporting

 

[Vincero]

Does it serve any real purpose on the desktop though? We have justification of these slow, cooler, low power cores in the laptop market, and certainly the handheld market, which should all be C-cores if you ask me.

But other than marketing, what would adding 4 or 8 more slow C-cores actually offer the desktop user?

Intel is f'd without them due to not being able to power or cool much more than 8 P-cores, but is AMD in that boat?

Any sale lost to a competitor is a loss.
My thinking is simply that with proper thread management, the c cores could carry most of the dumb stuff in the same way Intel use the e cores. AMD sacrifice some power draw having both the IO die and core chiplets. Additionally the IO interconnects to the core dies also are a tradeoff - in a performance per watt scenario if Intel were not a process behind then they may be able to consistenly pull better idle/low load power usage.
Unfortunately the reality is that for the majority of the time most people are using their computers have their computer on, a large part of that time is spent with at least one core idling.

Again my main point is to have low end AM5 products - the AM4 train will end at some point - currently AMD doesn't have much flexibility in the proper desktop lineup for low cost solutions (it's either full cores and IO or a laptop/pheonix derived hamstrung part which doesn't reduce cost enough). In reality AMD are probably racking up a few zen4/5 chiplets where less than 6 or 8 cores are functioning - they may be trashing them but I suspect not however there is no SKU that's using them. The problem is a quad-core AM5 CPU with an IO die and chiplet will be more expensive to make and use more power than an i3. Now if it they were combined with a quad c core then AMD could push the power / TDP leverage "because 8 cores" and recoup some money from that crap silicon selling at a percentage lower than the normal Ryzen 7 but offering maybe 75% of the performance.

The other fringe reason I could see such chips (c cores on IO die only with no added chiplet cores) being popular is if they were real cheap, they would be real handy for diagnostic / BIOS flashing purposes...

 

[W1zzard]

Nope, it’s in the notes slides.

View attachment 355315

Confirmed to be supported

AMD Ryzen 9 9950X Supports PBO After All

AMD's upcoming flagship desktop processor, the Ryzen 9 9950X, supports Precision Boost Overdrive (PBO) overclocking. In the comments section of our Zen 5 Technical Deep Dive article, a keen-eyed user noticed a footnote in one of the slides that reads that PBO is supported on Ryzen 9000 series...

www.techpowerup.com

 

[529th]

Wondering if the jump from a 5800X3D to a 7800X3D will be larger than the jump from a 7800X3D to the 9800X3D

 

[rv8000]

Wondering if the jump from a 5800X3D to a 7800X3D will be larger than the jump from a 7800X3D to the 9800X3D

IPC jump from 7000>9000 is slightly higher than 5000>7000, with a negligible improvement of clocks from 7000>9000 the overall difference should be rather similar.

Depending on the rumor to allow overclocking on X3D skus, and what the voltage limits will be, the 9000 generation X3D parts should be more interesting.

 

[529th]

IPC jump from 7000>9000 is slightly higher than 5000>7000, with a negligible improvement of clocks from 7000>9000 the overall difference should be rather similar.

Depending on the rumor to allow overclocking on X3D skus, and what the voltage limits will be, the 9000 generation X3D parts should be more interesting.

So from a 5800X3D to a 9800X3D will roughly be the equivalent of the jump from a 5800X to the 5800X3D, gaming wise? Roughly ~30% ?

 

[theouto]

Please AMD, don't fumble the pricing.
It's looking nice, but getting cocky with bad pricing would be horrible.

 

[rv8000]

So from a 5800X3D to a 9800X3D will roughly be the equivalent of the jump from a 5800X to the 5800X3D, gaming wise? Roughly ~30% ?

I was implying 7800X3D>9800X3D will potentially be similar to the gap between the 5800X3D and 7800X3D; ~15%. So similar to the way you worded that I guess, 5800X3D>9800X3D may be ~30%.

 

[ARF]

Wondering if the jump from a 5800X3D to a 7800X3D will be larger than the jump from a 7800X3D to the 9800X3D

IPC jump from 7000>9000 is slightly higher than 5000>7000, with a negligible improvement of clocks from 7000>9000 the overall difference should be rather similar.

Depending on the rumor to allow overclocking on X3D skus, and what the voltage limits will be, the 9000 generation X3D parts should be more interesting.

So from a 5800X3D to a 9800X3D will roughly be the equivalent of the jump from a 5800X to the 5800X3D, gaming wise? Roughly ~30% ?

I was implying 7800X3D>9800X3D will potentially be similar to the gap between the 5800X3D and 7800X3D; ~15%. So similar to the way you worded that I guess, 5800X3D>9800X3D may be ~30%.

30% is too meh..
Another nothing-to-write-about gen, which will be skipped for various reasons.
Vote with your wallets, save your hard-earned cash.

 

[AnotherReader]

I was implying 7800X3D>9800X3D will potentially be similar to the gap between the 5800X3D and 7800X3D; ~15%. So similar to the way you worded that I guess, 5800X3D>9800X3D may be ~30%.

When gaming with a 4090 at FHD, the 7800X3D is more than 25% faster than the 5800X3D in TPU's review from last year.

 

[ARF]

When gaming with a 4090 at FHD, the 7800X3D is more than 25% faster than the 5800X3D in TPU's review from last year.

Yes, at 480p it would be 100% faster /troll

 

[mkppo]

I agree with you that IF it is cheaper for AMD to use Zen5 C-cores on their lowest end products, then it would be a good idea, and no-brainer solution for more profit.

But the C-cores are made on the much more costly 3nm process, (which is baffling to me that it is not done the other way round, as the full fat cores would benefit more from this)... But that might negate the cost savings part...

I think it's yields, the c clusters are very tiny even with 16 of them in there. Also to fit 192 of them in an Epyc socket probably needs 3mn for size reasons but i'll have to look at Turin again to check if 4nm would've worked.

30% is too meh..
Another nothing-to-write-about gen, which will be skipped for various reasons.
Vote with your wallets, save your hard-earned cash.

I know a whole lot of people who are switching from 13900K/14900K because they don't have to deal with instability anymore and they'll have a nice performance bump all around while saving power. So I guess the timing works pretty well for AMD here. I will certainly get a 9950X simply because it'll speed up my workloads and a potential 20% bump over a 7950x with some nice memory tuning would go a long way without having to go TR.