AMD reveals Ryzen 7000X3D pricing and availability, starting February 28

[AlphaX]

Something to look forward to: Last year, AMD released the Ryzen 7 5800X3D, featuring new "3D V-Cache" technology. The chiplet cache design allowed AMD to triple the L3 cache found on the processor, resulting in significant performance gains in gaming. After months of waiting, AMD is finally bringing the tech to AM5.

In January, AMD unveiled its first set of Zen 4 X3D processors. While the CES announcement confirmed the specifications featured within the CPUs, AMD did not reveal MSRPs or release dates for the highly anticipated chips. Yesterday, AMD unexpectedly announced those and other details.

First, the 12-core Ryzen 9 7900X3D and 16-core Ryzen 9 7950X3D hit shelves on February 28 for $599 and $699, respectively. Those looking to get their hands on the more-affordable 8-core Ryzen 7 7800X3D will have to exercise patience. It becomes available on April 6 for $449.

While the MSRPs for these processors are relatively close to their non-X3D counterparts, CPUs in the AM5 family have been selling well below anticipated prices. Consumers should expect the new X3D chips to cost anywhere from $100 to $200 more than a processor without the technology.

During the CES keynote, AMD claimed that the new 7800X3D could perform up to 25 percent faster in gaming than its Zen 3 counterpart, the Ryzen 7 5800X3D. Furthermore, the highest-end 7950X3D reportedly performs up to 24 percent faster than Intel's Core i9-13900K processor.

The new Ryzen 9 processors with 3D V-Cache may also bridge the gap in productivity tasks that were heavily prevalent on the previous 5800X3D. While that processor struggled in application benchmarks, the 7900X3D and 7950X3D should compete right at the top, with the latter beating the i9-13900K in some tasks by up to 52 percent, according to AMD.

All three X3D processors feature the same 120W base power draw and a maximum of 162W, which is less than both non-X3D Ryzen 9 models. The chiplet cache design featured on the X3D chips causes more heat to be generated at lower wattages, forcing AMD to limit them. Due to the high temperatures, AMD recommends a 280mm liquid cooler as a minimum for the new processors.

Overall, AMD's new processors are an excellent option for gamers. Introducing the Ryzen 9 models also means buyers will no longer need to sacrifice productivity in favor of gaming performance. Just be sure to have an adequate cooler ready for these chips.

Permalink to story.

https://www.techspot.com/news/97475-amd-reveals-ryzen-7000x3d-pricing-availability-starting-february.html

 

[nnguy2]

Shut up and take my money!

 

[yRaz]

This makes me seethe, they must have some confidence in windows scheduling for the CCDs. Once I saw that only 1 chiplet would have vcache I got angry, but I bought their pitch of "well one chiplet is clocked higher and the other has vcache so you get the best of both worlds" on top of that, you get more cores for a hybrid HEDT setup.

I've been looking forward to these chips for months, I have an x670 motherboard waiting for the 7800X3D sitting next to me. This is a real slap in the face.

The higher end chips are pretty niche.

Microsoft figured out Intel's e core/p core scheduling issue so quickly most people probably don't even remember it so maybe I'm just over reacting. I hope I'm over reacting

 

[kiwigraeme]

Hopefully such innovations becomes standard

 

[Haaky]

Can we really believe what AMD claims?
AMD, NVidia, Intel are all just trying to sell us snake oil.

 

[Irata]

Microsoft figured out Intel's e core/p core scheduling issue so quickly most people probably don't even remember it so maybe I'm just over reacting. I hope I'm over reacting

The same Microsoft that takes forever to support AMD features ? Afair, they never supported AMD‘s heterogenous computing.

But either way, is the dual CCD really an issue for gaming ? There are very few games that show reduced performance.

 

[kiwigraeme]

Can we really believe what AMD claims?
AMD, NVidia, Intel are all just trying to sell us snake oil.

There claims will be correct - but may use very selective conditions -choose games , setup favoring them - we can wait for Steve etc

It's not snake oil as you will get this actual product .
Snake oil is and was a real Chinese product - but hucksters faked it to sell it as a cure-all - original high in =Omega 3 - so probably good in tough conditions back then

 

[takaozo]

Let's see if AMD learned something from latest GPU launch, bring on the reviews!

 

[yRaz]

The same Microsoft that takes forever to support AMD features ? Afair, they never supported AMD‘s heterogenous computing.

But either way, is the dual CCD really an issue for gaming ? There are very few games that show reduced performance.

Issue comes from one CCD have the vcache and the other CCD has a higher clock speed. My concern is that we will see inconsistent performance numbers when windows runs an application in one CCD instead of the other.

And I would like to point out that it MS less than a week to fix the scheduling issue for Intel. I won't pretend to be a hardware expert, but I can't imagine the problems are all that different to solve

 

[Puiu]

About the price I was expecting from these new CPUs.

 

[mbk34]

Microsoft figured out Intel's e core/p core scheduling issue so quickly most people probably don't even remember it so maybe I'm just over reacting.

It might be an interesting article to read how a core is chosen and what difference it makes getting it wrong. There are e cores, p cores, virtual cores on hyperthreaded cpus, cores on separate chiplets, then there's dual processor set ups. How does the system know where to run a process and does it move that process to a different core if it gets it wrong?

I'd find it an interesting read anyway.

 

[yRaz]

It might be an interesting article to read how a core is chosen and what difference it makes getting it wrong. There are e cores, p cores, virtual cores on hyperthreaded cpus, cores on separate chiplets, then there's dual processor set ups. How does the system know where to run a process and does it move that process to a different core if it gets it wrong?

I'd find it an interesting read anyway.

I also think it would make for an amazing article. Unfortunately, I don't know enough about the subject to have a valid opinion on it.

That said, I'm confident that Techspot writers are educated enough to make an interesting article on it!

@Julio Franco

 

[neeyik]

It might be an interesting article to read how a core is chosen and what difference it makes getting it wrong. There are e cores, p cores, virtual cores on hyperthreaded cpus, cores on separate chiplets, then there's dual processor set ups. How does the system know where to run a process and does it move that process to a different core if it gets it wrong?

I'd find it an interesting read anyway.

This is only a short response to this query (it's a good suggestion for an article!) but things like multiple cores on a single die and logical cores in hyperthreaded CPUs are relatively transparent to the process -- operating systems have a thread scheduler that sorts out when threads get processed and takes into account, to a certain degree, when a CPU has hyperthreading/SMT.

The thread scheduler in Windows 11 works with the Thread Director hardware in Intel's hybrid-architecture chips (I.e. the ones with P and E cores) to better allocate threads onto the right core type and/or logical core, depending on the workload, etc. The Win11 scheduler had problems with AMD's dual chiplet CPUs (and still does, to a small degree) so further work is needed to improve in this area.

 

[yRaz]

This is only a short response to this query (it's a good suggestion for an article!) but things like multiple cores on a single die and logical cores in hyperthreaded CPUs are relatively transparent to the process -- operating systems have a thread scheduler that sorts out when threads get processed and takes into account, to a certain degree, when a CPU has hyperthreading/SMT.

The thread scheduler in Windows 11 works with the Thread Director hardware in Intel's hybrid-architecture chips (I.e. the ones with P and E cores) to better allocate threads onto the right core type and/or logical core, depending on the workload, etc. The Win11 scheduler had problems with AMD's dual chiplet CPUs (and still does, to a small degree) so further work is needed to improve in this area.

one, I'd love for you to write such a piece. That said, so much of the infinity fabric has changed from zen1 to zen4 that they're hardly compareable. I'm going to say this as best as I can because I'm not as educated as you, but the interconnects between CCDs has changed significantly. It would be really interesting if someone as educated as yourself could explain to us how and why the infinity fabrics have changed between generations and how that may have implications on performance.

 

[neeyik]

one, I'd love for you to write such a piece. That said, so much of the infinity fabric has changed from zen1 to zen4 that they're hardly compareable. I'm going to say this as best as I can because I'm not as educated as you, but the interconnects between CCDs has changed significantly. It would be really interesting if someone as educated as yourself could explain to us how and why the infinity fabrics have changed between generations and how that may have implications on performance.

Thank you for the kind words. Again, only being brief here, but the main changes over the generations of IF were an increase in the CCX-to-CCX/CCD-to-IOD data path widths between Zen 1 and Zen 2/3, before reducing for Zen 4; FClk (IF clock) was locked to UClk (memory controller clock) in Zen 1 before being uncoupled entirely, but kept to a limited set of ratios for Zen 2 onward; faster overall FClk for Zen 4 than previous generations.

IF is pretty complex as there are different aspects to it, depending on what it's being to connect. For example, a different system is used inside dies compared to the IF that's used to connect multiple packages, and different again to that used in RNDA 2/3.

 

[yRaz]

Thank you for the kind words. Again, only being brief here, but the main changes over the generations of IF were an increase in the CCX-to-CCX/CCD-to-IOD data path widths between Zen 1 and Zen 2/3, before reducing for Zen 4; FClk (IF clock) was locked to UClk (memory controller clock) in Zen 1 before being uncoupled entirely, but kept to a limited set of ratios for Zen 2 onward; faster overall FClk for Zen 4 than previous generations.

IF is pretty complex as there are different aspects to it, depending on what it's being to connect. For example, a different system is used inside dies compared to the IF that's used to connect multiple packages, and different again to that used in RNDA 2/3.

It's a fantastically interesting topic; one that no other tech site is writing about. It could attract a lot of traffic to techspot if someone was so inclined to write about it

 

[McMurdeR]

About the price I was expecting from these new CPUs.

Indeed. A little steep for my liking, but certainly worth keeping an eye out for discounts and price cuts before the end of the year...

 

[Aranarth]

This makes me seethe, they must have some confidence in windows scheduling for the CCDs. Once I saw that only 1 chiplet would have vcache I got angry, but I bought their pitch of "well one chiplet is clocked higher and the other has vcache so you get the best of both worlds" on top of that, you get more cores for a hybrid HEDT setup.

I've been looking forward to these chips for months, I have an x670 motherboard waiting for the 7800X3D sitting next to me. This is a real slap in the face.

The higher end chips are pretty niche.

Microsoft figured out Intel's e core/p core scheduling issue so quickly most people probably don't even remember it so maybe I'm just over reacting. I hope I'm over reacting

So... You want a 7800x3d which has ONE cpu chip in it and that ONE cpu chip will get vcache and you're andgry about it? I
m confused unless you meant the 7900x3d?

 

[yRaz]

So... You want a 7800x3d which has ONE cpu chip in it and that ONE cpu chip will get vcache and you're andgry about it? I
m confused unless you meant the 7900x3d?

I want the 7800X3D so I can avoid scheduling issues. 8 cores and a plenty high clock speed for the girls I go out with. I run VMs and try multiple operating systems, not all have the scheduling support the Microsoft(could) have.

I just see a lot of potential for things to go wrong and we don't live in a perfect world.

 

[Aranarth]

I want the 7800X3D so I can avoid scheduling issues. 8 cores and a plenty high clock speed for the girls I go out with. I run VMs and try multiple operating systems, not all have the scheduling support the Microsoft(could) have.

I just see a lot of potential for things to go wrong and we don't live in a perfect world.

That does not answer my question.

This is a chip with a single CCD that means that this is NOT a dual ccd chip with one CCD with a huge cache and one CCD without it but a faster top speed.

Why would having a chip with ONE CCD and not TWO make you angry in your specific use case?

You have no need to worry about two different chips in one package with different performance depending on which chip is in use.

 

[yRaz]

That does not answer my question.

This is a chip with a single CCD that means that this is NOT a dual ccd chip with one CCD with a huge cache and one CCD without it but a faster top speed.

Why would having a chip with ONE CCD and not TWO make you angry in your specific use case?

You have no need to worry about two different chips in one package with different performance depending on which chip is in use.

The 7800X3D is not a dual CCD chip, the others are. I want the 7800X3D to avoid scheduling issues when going from cache to frequency dependant programs. The dual CCD chips, 7900x3d and 7950x3d could end up performing worse in some workloads because a program is running on one CCD while trying to access the cache on the other. I work with enough unoptimized software, fringe operating systems and virtual machines that this poses a real issue for me. 3d vcache is a great feature but it has to be implemented properly or well see problems similar to first gen threadripper when it comes to scheduling.

 

[Irata]

And I would like to point out that it MS less than a week to fix the scheduling issue for Intel. I won't pretend to be a hardware expert, but I can't imagine the problems are all that different to solve

The keyword here being ‚for Intel‘.

 

[godrilla]

FYI the price of the 7900x3d fell already by $50 at microcenter .