AMD Ryzen 9 9950X3D leak reveals 5.65 GHz boost clock and 170W TDP

DragonSlayer101

Posts: 984   +14
Staff
The big picture: AMD is rumored to announce its flagship gaming CPU, the Ryzen 9 9950X3D, at CES 2025 in Las Vegas in January. Ahead of its official launch, multiple leaks have already revealed key features and specifications. The latest leak discloses details about the chip's clock speed, cache, TDP, and other critical specifications.

According to a CPU-Z screenshot posted on X by tipster @94G8LA, the Ryzen 9 9950X3D will feature a boost clock of up to 5.65GHz – slightly lower than the 5.7GHz maximum frequency of the standard Ryzen 9 9950X. Earlier rumors, however, suggested that the 9950X3D might share the same clock speed as its non-X3D counterpart.

The leaked screenshot identifies the chip as revision GNR-B0 with the code name Granite Ridge, confirming its status as an upcoming Zen 5 CPU. It is listed with 16 cores, 32 threads, a bus speed of 99.78MHz, and a maximum CPU multiplier of x56.5. The CPU-Z data also reveals a TDP of 170W, which matches that of the standard Ryzen 9 9950X but exceeds the 120W TDP of the 7950X3D by 50W.

The image also reveals that the CPU will feature 96MB + 32MB of L3 cache, indicating that both CCDs include 32MB of L3 cache, while the 3D V-Cache chiplet adds an additional 64MB layer. This configuration gives the Ryzen 9 9950X3D a total of 144MB of cache, comprising 128MB of L3 and 16MB of L2. This is consistent with the 3D V-Cache setup of its predecessor, the Ryzen 9 7950X3D.

Notably, the chip is marked as an engineering sample, signifying that it is currently undergoing validation testing. This also implies that AMD might adjust some of the specifications, particularly the clock speeds, before the final product hits the market.

In addition to the Ryzen 9 9950X3D, AMD is expected to unveil its RDNA 4 GPUs at CES 2025. According to multiple rumors, the company may forego the Radeon RX 8000 naming convention for its next-generation graphics cards, opting instead to market them as part of the Radeon RX 9000 series. Furthermore, Team Red is rumored to introduce FSR 4 at the event, showcasing its continued advancements in graphics technology.

Permalink to story:

 
Disappointing if this is true - I was hoping they would build the two chiplets with a shared 3d v cache, well - they might still have done that - as the specs doesn't indicate if the cache is shared or not - which would make this cpu a true monster if all cores can utilize it
 
Disappointing if this is true - I was hoping they would build the two chiplets with a shared 3d v cache, well - they might still have done that - as the specs doesn't indicate if the cache is shared or not - which would make this cpu a true monster if all cores can utilize it
You mean both chiplets have V-cache chip, because sharing makes no sense. I think AMD wants to save "all V-cache CCDs" for Threadrippers and Epycs.
I don't know why they don't put these kinds of efforts into GPU and their drivers.
Because market has decided that Nvidia is better even when AMD was clearly better. So it makes no sense to be better than Nvidia, because that makes nothing.

AMD correctly predicted that customers are pissed off with Intel so if AMD has better product, it will also sell better.
 
You mean both chiplets have V-cache chip, because sharing makes no sense. I think AMD wants to save "all V-cache CCDs" for Threadrippers and Epycs.

Because market has decided that Nvidia is better even when AMD was clearly better. So it makes no sense to be better than Nvidia, because that makes nothing.

AMD correctly predicted that customers are pissed off with Intel so if AMD has better product, it will also sell better.
It would if they made an entire layer of V-Cache below the chiplets and basically made both chiplets draw from the same pool - which is then proportinally larger due to the amount of cores that will share it. Both chiplets are "interconnected" anyways.

Can't wait to see what they've come up with :)
 
You mean both chiplets have V-cache chip, because sharing makes no sense. I think AMD wants to save "all V-cache CCDs" for Threadrippers and Epycs.

Because market has decided that Nvidia is better even when AMD was clearly better. So it makes no sense to be better than Nvidia, because that makes nothing.

AMD correctly predicted that customers are pissed off with Intel so if AMD has better product, it will also sell better.
The problem was that AMD GPUs were always presumed to be slower than Nvidias' which was not always true. For that reason AMD never succeeded in flagship segment especially after Titan tier introduction.
I think its good that they gave up top tier.
 
It would if they made an entire layer of V-Cache below the chiplets and basically made both chiplets draw from the same pool - which is then proportinally larger due to the amount of cores that will share it. Both chiplets are "interconnected" anyways.

Can't wait to see what they've come up with :)
That solution would be so slow that it barely makes any sense. Unless cache size is so bit that it costs huge. And then it wouldn't make any sense either. Also power consumption would be an issue. Not to mention AMDs desktop Ryzens are downgraded Epyc/Threadripper CPUs, AMD would not make much different solution for desktop.

It's possible that there are two V-cache chips but I expect only one.
 
You mean both chiplets have V-cache chip, because sharing makes no sense. I think AMD wants to save "all V-cache CCDs" for Threadrippers and Epycs.

Because market has decided that Nvidia is better even when AMD was clearly better. So it makes no sense to be better than Nvidia, because that makes nothing.

AMD correctly predicted that customers are pissed off with Intel so if AMD has better product, it will also sell better.
Hard disagree when it comes to GPUs. I have had dreadful experiences over the years with Radeon products compared to Geforce personally and ive been building and gaming for over 20 years and ive had more Radeons than Geforce cards. Maybe things have changed since I last bought a GPU (currently on a 3070 Ti) but I doubt it. AMD has a long way to go to restore my faith in Radeon at this point. Their small market share makes complete sense to me. What's interesting me is how quickly Intel appear to be taking the value crown from Radeon in the GPU space. Now if only Intel can release a competitive 4K card. There will be no reason at all for anyone to buy Radeon.

Agree on CPUs though. AMD are clearly making a superior product than Intel and the market is responding. Its refreshing because it felt like years and years that AMD would claim they are about to beat Intel and then didnt. Its nice to finally see them win. Im already on a 5800X and im enjoying it, il happily go Ryzen again next time as long as the reviews come in good.





 
Hard disagree when it comes to GPUs. I have had dreadful experiences over the years with Radeon products compared to Geforce personally and ive been building and gaming for over 20 years and ive had more Radeons than Geforce cards. Maybe things have changed since I last bought a GPU (currently on a 3070 Ti) but I doubt it. AMD has a long way to go to restore my faith in Radeon at this point. Their small market share makes complete sense to me. What's interesting me is how quickly Intel appear to be taking the value crown from Radeon in the GPU space. Now if only Intel can release a competitive 4K card. There will be no reason at all for anyone to buy Radeon.
I have never had major problems with Radeon cards. When Nvidia sucked with GTX400-series and market still preferred Nvidia, that's when AMD decided to switch focus on CPUs. So yeah, market and buyers are to blame about situation.

As for Intel, they are "taking" value crown by selling on loss. GPU is bigger and uses more expensive manufacturing tech vs similarly performing AMD/Nvidia GPU, card also has more memory. If Intel looks good, it's just because they are throwing away money. Not sustainable in other words.
 
I have never had major problems with Radeon cards. When Nvidia sucked with GTX400-series and market still preferred Nvidia, that's when AMD decided to switch focus on CPUs. So yeah, market and buyers are to blame about situation.

As for Intel, they are "taking" value crown by selling on loss. GPU is bigger and uses more expensive manufacturing tech vs similarly performing AMD/Nvidia GPU, card also has more memory. If Intel looks good, it's just because they are throwing away money. Not sustainable in other words.
Can you provide any evidence for your ridiculous claim that AMD effectively gave up on Radeon after losing to the Geforce 400 series?
 
Can you provide any evidence for your ridiculous claim that AMD effectively gave up on Radeon after losing to the Geforce 400 series?
Easy. Two points are enough:

1. After 2010 (GTX480 battle), it was clear that market prefers Nvidia, although AMD is better. Then AMD decided to focus on something else. They had one series coming (HD7900 series) and of course they didn't abandon that. But, that also was so far last series AMD even tried to compete with Nvidia on high end. In other words, after 2010 AMD has not seriously developed new high end GPU for consumer market.

2. Zen. AMD put all available (and not available) resources on Zen development that started 2012. Again, it takes some time to prepare even before starting development and timing is perfect to support claim that AMD didn't put major resources for new high end GPU after HD 7000 series. AMD lack of resources when developing Zen is well documented.

There you have it.
 
I don't know why they don't put these kinds of efforts into GPU and their drivers.

They do. They had many innovative products. One that comes to my mind now and I considered really cool:

"The AMD Radeon Pro SSG (Solid State Graphics) is a graphics card that integrates SSD storage directly onto the GPU board."

And many others. Too bad we didn't see amd GPU on mobile devices in smartphones.

Is about market dynamics, not who is better in particular...
 
Back