Thursday, December 26th 2024
AMD Ryzen 9 9950X3D Carries 3D V-Cache on a Single CCD, 5.6 GHz Clock Speed, and 170 Watt TDP
Recent engineering samples of AMD's upcoming Ryzen 9 9950X3D reveal what appear to be the finalized specifications of the top-tier AM5 chip. The 16-core, 32-thread processor builds upon the gaming success of the Ryzen 7 9800X3D while addressing its core count limitations. The flagship processor features AMD's refined cache design, combining 96 MB of 3D V-Cache with 32 MB of standard L3 cache. Unlike its predecessor, the 7950X3D, the new Zen 5 architecture incorporates a redesigned CCD stacking method. The CCD now sits above the cache, directly interfacing with the STIM and IHS, eliminating thermal constraints that previously required frequency limitations. The processor features asymmetric cache distribution across its dual CCDs—one die combines 32 MB of base L3 cache with a 64 MB stacked V-Cache layer, while its companion die utilizes a standard 32 MB L3 cache configuration. In total, there is a 128 MB of L3 cache, with 16 MB of L2.
This architectural advancement enables the 9950X3D to achieve a 5.65 GHz boost clock across both CCDs, matching non-X3D variants. The processor maintains a 170 W TDP, suggesting improved thermal efficiency despite the additional cache. AMD's software-based OS scheduler will continue to optimize gaming workloads by directing them to the CCD with 3D V-Cache. Early leaks indicate the 9950X3D matches the base 9950X in Cinebench R23 scores, both in single and multi-threaded tests—a significant improvement over the 7950X3D, which lagged behind its non-X3D counterpart due to frequency limitations. AMD plans to expand the Zen 5 X3D lineup in Q1-2025 with both the 9950X3D and 9900X3D models. Full performance benchmarks and pricing details are expected at CES 2025, where AMD will officially unveil these processors alongside their RDNA 4 GPUs.
Sources:
@94G8LA, via VideoCardz
This architectural advancement enables the 9950X3D to achieve a 5.65 GHz boost clock across both CCDs, matching non-X3D variants. The processor maintains a 170 W TDP, suggesting improved thermal efficiency despite the additional cache. AMD's software-based OS scheduler will continue to optimize gaming workloads by directing them to the CCD with 3D V-Cache. Early leaks indicate the 9950X3D matches the base 9950X in Cinebench R23 scores, both in single and multi-threaded tests—a significant improvement over the 7950X3D, which lagged behind its non-X3D counterpart due to frequency limitations. AMD plans to expand the Zen 5 X3D lineup in Q1-2025 with both the 9950X3D and 9900X3D models. Full performance benchmarks and pricing details are expected at CES 2025, where AMD will officially unveil these processors alongside their RDNA 4 GPUs.
72 Comments on AMD Ryzen 9 9950X3D Carries 3D V-Cache on a Single CCD, 5.6 GHz Clock Speed, and 170 Watt TDP
We already have two identical non cache CCD's that work fine together. They(amd) already have the recipe for this. One CCD with cache already works damn good. They can manufacture a dual cache dual CCD in the same way as they are doing a dual CCD non cache one. They already master this.
But the devil is in de details: manufacture. They literally have to increase the cost by physically adding a second cache die.
I will actually pay more for a dual cache dual CCD. What I will get in return is no more shenanigans regarding the way threads are spread across the entire CPU. This will just work as the simple dual CCD no cache product.
But what does this mean for AMD? They have 0 incentives to do this. The software cost solution for a single cache CPU dual CCD is almost zero compared the hardware one of actually manufacturing and adding a second cache die.
In the end, this is a simple cost reduction business decision from their behalf that has nothing to do with the technicalities of the more desirable dual cache end product.
I think they already have the product, but they simply choose not to release it.
I also wonder, how much money are these companies making out of gamers?
Because us being the most vocal doesnt mean our purchase justify the r&d and other expenses.
And since the AMD name seems to only (or mostly) bring hostility from consumers (even on this cpu related article, the trashing is hot and heavy) then adds to the question, its worth for them to continue?
It's faster and lower latency to run everything off a single CCD with one V-Cache, the instant the game has to hop off that CCD you're incurring latency penalties. Even on non-X3D parts, the scheduler tries to keep games to one CCD. AMD has experience keeping the scheduler running things where they'll be fastest. Do you remember the Ryzen 3 3300X and 3100? On Zen 2, Core Complexes (CCX) were 4-core instead of 8-core (like on Zen 3 onward). So the 3100 was massively slower than 3300X because even though the two CCX's were on a single die, 3300X was 4+0 (all cores in one CCX) and 3100 was 2+2.
See: AnandTech's Zen 4 Review, Core-to-Core Latency
The instant you have to go off-die to the other CCD your latency quadruples, this is a fact of Infinity Fabric. You have effectively lost the latency benefit X3D gives you.
Back when AMD announced the 5800X3D people asked them why they didn't do a 5950X3D, and then when they finally did the 7950X3D and 7900X3D, they were worse for gaming than the 7800X3D and had scheduler issues.
This is why.
I don't want to install and configure any software other than the OS to run my CPU properly. Better?
Hurrrrrrr, TEKNIKALLY, the XBox app and GameBar are a part of a default Windows installation, so checkmate atheists!
No, but really, I think everyone knows that the dual CCD X3D chips are clunky and mostly not worth messing around with. That’s the same energy as people recommending using Process Lasso for Intel CPUs to bypass E-cores. Like, yeah, that do be working, but… I don’t wanna.
Not to mention, if you're a gamer, a 9800X3D will do fine. If gaming is a secondary consideration, the 9950X will give you the oomph you need while still not being half bad in gaming. Being a hardcore gamer who needs the cache for the last FPS drop and also needing 16 cores at the same time is kind of a no man's land, imo.
When 5800X3D came out people wanted a 16 core with 3D cache. AMD gave them two, they didn't liked them because in gaming they where some times somewhat slower than the 8 core part. Now they want each CCD to have it's own 3D cache. AMD will give them this, but they might see that in some cases performance is still lower than the 8 core part, because - don't know - maybe data gets on 3D cache 1 while the CPU core trying to access it is in CCD 2 or something introducing latency. Then they will be asking for a unified 3D Cache under both CCDs and ..... that will never end.Don't know. Progress maybe? Gaming is probably already past 8 threads on P cores and maybe that's why Intel's latest CPUs perform so poorly in some games.