Saying that it has been a busy launch season for AMD is an incredible understatement. Not only did we see the release of the high powered, high performing Ryzen 7000X Series, but we also got a full stack of lower powered, but still respectably performant processors in the Ryzen 7000 Non-X Series. And let’s not forget a generational leap for Radeon GPUs!
In all of these releases, we have been waiting for AMD to juice up one of its Zen 4 juggernauts with that sweet, sweet game-changing 3D V-Cache we saw in the Ryzen 7 5800X3D. For the experts, the V-Cache stacks more memory on a chip vertically. For everyone else, number go up. What we’re getting is three options – the Ryzen 7 7800X3D, Ryzen 9 7900X3D, and Ryzen 9 7950X3D.
Launching today (February 28th, 2023), the Ryzen 9 7900X3D and Ryzen 9 7950X3D are the first Ryzen 9 CPUs to get the V-Cache treatment. These CPUs are aimed at the gamer who also happens to be a content creator. There is a twist in that, but we’ll talk about it more later.
For this review, we got to check out the cream of the crop, the Ryzen 9 7950X3D–AMD’s 16-core, 32-thread monstrosity of a CPU.
AMD’s cache-enhanced Ryzen 9 marries Zen 4 power with 3D V-cache technology.
AMD’s cache-enhanced Ryzen 9 marries Zen 4 power with 3D V-cache technology.
Charge multiple devicesThe circular design of the power strip allows for the 15" to coil up inside the body of the product—making it easy to wrap up and throw in a bag to take with you when traveling.
Before we can get into the performance, we have to take a look at pricing, the CPU design, what 3D V-cache is, how extra cache benefits gaming, and how the heck this thing even works! Let’s put the cache on the table first.
AMD’s 3D V-cache technology takes the existing Zen 4 cores, thins them down, and stacks an extra layer of L3 cache on top of the chiplet without increasing the overall Z (or vertical) height of the CPU. This design allows for a more dense connectivity between the CPU die and the cache, providing a faster memory resource than your garden variety system RAM.
In gaming, the extra cache helps the CPU in moments where quick access to random data is essential. For example, if you’re playing a match of Gambit in Destiny 2, your CPU doesn’t know if you are going to bank those motes like The Drifter is screaming about or if you are going to turn around and Thundercrash into an invading player. In other words, player-driven actions are random processes that require rapid computation. In that moment, your CPU is called upon to recall textures, animations, game physics, character placement, and more. When that happens, cache is king.
(You can find AMD’s breakdown of the technology here for a more detailed description.)
Moving on to pricing, AMD did something interesting here. At the top end of the Ryzen 7000 Series product stack, the Ryzen 9 7950X and 7950X3D are priced identically at $699 USD. Both CPUs are 16-core, 32-thread processors, but the 7950X trades 3D V-cache for overall faster base clock frequencies and a higher power draw to match. As things trickle down the stack, we start seeing a $50 USD difference between counterparts and a much slimmer margin of difference. Where the big difference lies is in the design.
The 7950X, 7900X, 7950X3D, and 7900X3D each have a core complex die (CCD) made up of two core complexes (CCX). The X3D processors also have two core complexes, but only one CCX is equipped with the 3D V-cache. AMD’s official word on this is that, in its testing, the company did not see a noticeable enough uplift in performance to justify the extra cost of putting it on both. Fair enough, but is Windows smart enough to know which cores to use for what?
Sort of, but I can only speak for the 7950X3D (while making some educated guesses on the 7900X3D). Because of this, the CPU is dependent on processes like the Game Bar to park the CCX without the V-cache while “whitelisting” or allowing gaming processes to access that CCX with the V-cache. The hit rate for the V-cache CCX can be negatively impacted if you have other background processes running. It will be interesting to see if this is the same with the 7800X3D once it releases.
The AMD Ryzen 9 7950X3D is a 16-core, 32-thread (16C/32T) CPU with a base clock speed of 4.2 GHz and max boost speeds up to 5.7 GHz. While the core and thread counts are the same as the Ryzen 9 7950X, the base clock is slightly (~300 MHz) slower. Base clock speed isn’t the only thing that’s lower, though, as the 7950X3D drops 50W off of the TDP, dropping the Ryzen 9 down from 170W TDP to 120W TDP. That means it draws less power.
Now, in all of this, the comparison between the Ryzen 9s does get a bit trickier. Remember, we have two CCX modules–one with the 3D V-cache and one without it.
Unfortunately, we’re going to be seeing the 7950X3D in isolation from the other X3D CPUs. For our purposes, we are going to show some comparisons with the current flagship CPUs from both sides of the aisle.
In our productivity tests, we used data from the entire Ryzen 7000 Series stack as well as Intel’s 13th Generation Core i5-13600K and Core i9-13900K. We also included data collected by Joanna Neilus from our Ryzen 9 5950X review as a reminder of how far things have advanced across a single generation.
In gaming, however, we narrowed the comparison to the current kings – the Core i9-13900K and the Ryzen 9 7950X. This data was collected utilizing as similar a testing environment as possible, but, as always, your mileage may vary.
As always, we tried to keep our methodology simple in order to give the clearest picture of the performance that the average user can expect to get out of the Ryzen 9 7950X3D. In other words, these numbers represent box-to-build performance with no tweaks. The only advantage we gave them is that we enabled an AMD EXPO profile for low memory latency and speed. We offered the same consideration to our Intel platform by enabling XMP.
Here is the hardware we used to collect our data from the Ryzen 7000 Series CPUs: - Gigabyte X670E AORUS MASTER motherboard- NVIDIA RTX 3090 Founders Edition GPU- AMD Radeon RX 7900 XTX GPU- 32 GB G.Skill Trident Z5 NEO DDR5-6000 RAM- 2TB SK hynix Platinum M.2 NVMe SSD- NZXT C1200 Gold 1200W power supply - NZXT Kraken Z73 360mm all-in-one liquid cooler with three 120mm fans- NZXT H9 Elite mid-tower PC case with two 140mm fans as well as three 120mm fans.
You may notice that we listed two GPUs in our test bench–there is a very good reason for that: we bottlenecked our RTX 3090 in our 1080p gaming test. We are going to include these numbers for comparison, but we went back and ran the same tests on the 7950X and 13900K using the Radeon RX 7900 XTX for a more accurate picture of performance overall. I know that sounds wild, but wait until you see the difference!
In our productivity tests, we ran a suite of software benchmarks to highlight single and multi thread performance in scenarios designed to test how each CPU performed under various computational loads.
Here is what we found:
In Geekbench 5, we calculate the performance of single thread tasks such as image and text compression, web navigation, machine learning, and .PDF rendering. In this benchmark, the Ryzen 9 7950X3D did put on a strong show, but slightly underperformed when compared to the 7950X and the 7900X. Even so, it still managed to edge out the Core i9-13900K by a slim margin in single core performance. These numbers, however, are within a close enough margin that I wouldn’t necessarily declare one of them a winner.
In multi core performance, this is where we see the 7950X3D lag a bit behind its counterpart from the Ryzen lineup while falling further behind the Core i9. It is possible the overall frequency and parked CPU cores could be causing this performance delta, or this could be a scenario where the extra cache just doesn’t hold any power. That being said, the 7950X3D delivers on its ability to be a strong contender in multi core tasks.
We see a similar story in Handbrake:
While rendering a 4K video down to 1080p30, the 7950X3D hangs closely with the 7950X while taking an 11 second lead ahead of the 7900X. It still trails behind the 13900K, but when you look across the landscape, the 7950X3D does well for itself.
That margin got a whole lot slimmer when rendering an image of everyone’s favorite, shiny red BMW in Blender. This image took the 7950X3D just over a minute to render, putting it within 3 seconds of the Core i9’s finishing time and 4 seconds away from the 7950X. Not too shabby at all!
Overall, we did not see an uplift in performance in our productivity benchmarks. However, whether V-cache was leveraged or not, the 7950X3D still performed well overall thanks to the powerful Zen cores.
I’m not going to mince words: this is where things get weird.
When we started running the 7950X3D through our gaming tests, we wanted to keep our testing data similar to our most recent CPU reviews. So we used NVIDIA’s RTX 3090 Founders Edition in our testing bench, like before.
As we got into our benchmarking, we weren’t noticing a difference between it and the 7950X. Not only that, there were actually places where it underperformed. This was frankly disconcerting–especially after seeing what AMD’s 3D V-cache did for the 5800X3D.
It is important to mention that these tests were run at 1080p. While a GPU can impact performance at this resolution, there is much more of a burden on the CPU. Going even lower to 720p would have really pushed the CPU, but this resolution is largely irrelevant to the modern user outside of handhelds.
After doing some digging into Source: Gizmodo