Today sees the AMD Ryzen 9000 series CPUs launching as part of the new 4nm Zen 5 line-up and AMD have decided to launch the range with two dates in mind. First up is the Ryzen 5 and Ryzen 7 reviews which go live today and then, followed by the Ryzen 9 parts which are going live a little later in the month with retail sales coinciding with each release. We’ve managed to get our hands on the Ryzen 7 9700X; a CPU that comes in with the same core and thread count as its predecessor, the 7700X, as well as being 700MHz….slower.
Are we expected to believe that the new Zen 5 architecture and newer 4nm fabrication process will yield some magical results by increasing performance, or have AMD decided to instead, switch their focus into other areas, like improving efficiency by reducing the TDP on the 9700X instead? This leaves us and consumers with a lot of unanswered questions, because if that’s the case, is there really a reason to upgrade if you’re already sporting a Zen 4 CPU? Well that’s what our main focus is going to be on today, because while we’ve tested a whole host of different CPUs from both AMD and Intel, the big question is always going to come down to what uplift, if any can be obtained from this shiny new piece of silicon.
Ok, so first things first. I want to address the elephant in the room. You may or may not have seen that we posted on social media just under a week ago about how we weren’t getting sampled for this launch, and at the time of writing this, it’s less than 24 hours until the embargo launch time, so we’ve had to scramble things together pretty quickly, without looking to reduce our benchmarks down, so you can expect our usual in-depth run of tests as well as a ton of content on this chip in the coming weeks with memory scaling, a full 42 game suite of tests, and some head to heads too, so make sure you stick around for that. That aside, we were able to source a chip and here we are with the 9700X, a CPU that on paper doesn’t seem all that dissimilar from the Ryzen 7 7700X.
When I say dissimilar, it really comes down to the specifications that have been set out for this specific SKU. On paper, it all seems a bit strange with the same core and thread count to its predecessor, while reducing the speed, though AMD’s claims of up to a 16% increase in IPC is certainly an interesting point, but IPC improvements don’t always result in raw performance in every application, and especially not in games, much like how consumers were lead to believe that more cores and threads relayed to huge jumps in framerates, when in actuality, that most certainly isn’t the case.
While gaming is one side of the puzzle, AMD have SKUs directly linked for those use-case scenarios in the form of the X3D line-up, with the 7800X3D proving to be extremely popular, while for the workstation conscious user, the Ryzen 9 has always been touted as the best for the job, if you’re prepared to pay for it. So where does the Ryzen 7 9700X now position itself in a market where AMD are not only competing directly with Intel, but now seemingly competing with themselves? That’s what we’re going to find out today, because from a first glance, this is just a Ryzen 7 7700X with a few tweaks here and there, including a lower power profile, but is that enough to make a splash?
Luckily, there is more to it. When comparing the current Zen 5 microarchitecture to the previous Zen 4 microarchitecture, AMD claims that Zen 5 will see substantial improvements over the last generation, particularly in caching and instruction-fetching. The L2 cache size is doubled to 1 MB per core, the I-Cache is upgraded to 32 KB, and the D-Cache to 48 KB. With dual I-Fetch/decode pipes, Zen 5 can handle more instructions per cycle than Zen 4’s single pipe. Also improved are the Execution Units with 6 Integer ALUs and 6 FP operations per cycle, giving better overall performance and an efficient chip.
AMD has optimised and designed Zen 5 for better data flow and power efficiency, outperforming Zen 4. It has 4 load pipes for 512b AVX-512 workloads and wider L2 cache paths to improve data handling. AMD, for its part, has also changed how it implements AVX-512 in Zen 5: it forgoes one single full path of AVX-512 data. Instead, the processor has two 256-bit paths, making up a more efficient data path layout. Finally, Zen 5 retains SMT, keeping two threads per core for better multi-threaded performance.
The Zen 5 cores themselves are built on TSMC’s N4X process, which, technically speaking, should significantly improve overall capabilities in IPC and ST performance compared to the 5nm that Zen 4 was based on. By that token, N4X provides an 11% increase in performance, a 22% gain in power efficiency, and a 6% boost in transistor density compared to the 5nm nodes. Furthermore, more EUV layers and fewer masks reduce production costs by about 6%.
Edit [1st September 2024]: AMD have now worked directly with motherboard vendors to increase the TDP, in the hope that it unlocks extra performance that may not have necessarily been there at its stock rated 65W TDP. The likes of MSI have rolled out BIOS updates that will increase the TDP back up to 105W, now matching the power targets set by the predecessor Ryzen 7 7700X
Alongside AMD servers, MSI showcased its NVIDIA MGX AI servers and Intel Xeon 6 solutions…
Intel has its Gaudi 2 accelerators available, and Gaudi 3 will be available soon. But…
Intel has just dropped a brand new update for its Arc GPU graphics drivers, but…
The latest keyboard from Epomaker is here, with the Galaxy 100, a $110 fully customisable…
Corsair has just announced the LX-R RGB Series, a new line of reverse-flow cooling fans…
NVIDIA has revealed the new games that support its latest graphics card technologies. We're talking…