AMD’s $599 12-core Ryzen 9 7900X3D is designed to tackle Intel’s Core i9-13900K, which it beats in gaming by 7% on common, proving it has the chops to tackle Intel’s best. Nevertheless, AMD has been uncharacteristically silent about this mannequin — the corporate did not pattern it to press and hasn’t shared any of the usual pre-launch benchmarks we have grow to be accustomed to. Maybe that is as a result of the 7900X3D’s greatest competitor resides in AMD’s personal steady — the extremely spectacular flagship 16-core Ryzen 9 7950X3D that has taken the throne because the quickest gaming CPU available on the market and the finest CPU for gaming for the highest-end rigs.
Like its pricier counterpart, the 7900X3D leverages AMD’s unique second-gen 3D V-Cache to spice up L3 capability to a whopping 128MB, thus boosting efficiency in gaming to previously-unseen heights. The Ryzen 9 7900X3D is geared particularly for players trying to blast by means of CPU-limited video games, and whereas there are some tradeoffs in productiveness work, it nonetheless has 12 cores to assist energy by means of heavy productiveness workloads.
| Value | Cores / Threads (P+E) | P-Core Base / Increase Clock (GHz) | Cache (L2/L3) | TDP / PBP / MTP | |
|---|---|---|---|---|---|
| Ryzen 9 7950X3D | $699 | 16 / 32 | 4.2 / 5.7 | 144MB (16+128) | 120W / 162W |
| Ryzen 9 7900X3D | $599 | 12 / 24 | 4.4 / 5.6 | 140MB (12+128) | 120W / 162W |
| Ryzen 7 7800X3D | $449 | 8 /16 | 4.2 / 5.0 | 104MB (8+96) | 120W / 162W |
| Ryzen 7 5800X3D | $348 | 8 /16 | 3.4 / 4.5 | 104MB (8+96) | 105W |
The primary-gen 3D V-Cache chip, the Ryzen 7 5800X3D, grew to become the go-to chip for players, nevertheless it suffered in some productiveness apps as a result of a restrict of eight cores and its comparatively low increase frequency. AMD’s new lineup is the primary 3D V-Cache household to make use of two compute chiplets, boosting productiveness apps and enabling a lot larger increase frequencies than the previous-gen’s 4.5 GHz. The chips use a brand new thread-targeting method designed to assist apply the positive factors extra evenly throughout a broad vary of video games, however as with the previous-gen chips, there are nonetheless some tradeoffs in productiveness functions.
The $599 Ryzen 9 7900X3D’s shut pricing proximity to the $699 Ryzen 9 7950X3D makes issues troublesome, and the looming $449 eight-core Ryzen 7 7800X3D arrives later in April. That can undoubtedly be the chip to beat: AMD tells us the 7800X3D will provide the lion’s share of 3D V-Cache’s gaming efficiency, nevertheless it comes at a way more inexpensive value. These AMD chips give the 7900X3D loads of inside competitors, however Intel is not sitting idly on the sideline both. Let’s examine how the chips stack up.
AMD Ryzen 9 7900X3D Pricing and Specs
| Road/MSRP | Cores / Threads (P+E) | P-Core Base / Increase Clock (GHz) | E-Core Base / Increase Clock (GHz) | Cache (L2/L3) | TDP / PBP / MTP | Reminiscence | |
|---|---|---|---|---|---|---|---|
| Ryzen 9 7950X3D | $699 | 16 / 32 | 4.2 / 5.7 | 144MB (16+128) | 120W / 162W | DDR5-5200 | |
| Core i9-13900KS | $699 | 24 / 32 (8+16) | 3.0 / 6.0 | 2.2 / 4.3 | 68MB (32+36) | 150W / 253W / 320W | DDR4-3200 / DDR5-5600 |
| Core i9-13900K / KF | $589 (Ok) – $564 (KF) | 24 / 32 (8+16) | 3.0 / 5.8 | 2.2 / 4.3 | 68MB (32+36) | 125W / 253W | DDR4-3200 / DDR5-5600 |
| Ryzen 9 7950X | $579 ($699) | 16 / 32 | 4.5 / 5.7 | – | 80MB (16+64) | 170W / 230W | DDR5-5200 |
| Ryzen 9 7900X3D | $599 | 12 / 24 | 4.4 / 5.6 | 140MB (12+128) | 120W / 162W | DDR5-5200 | |
| Ryzen 9 7900X | $419 ($549) | 12 / 24 | 4.7 / 5.6 | – | 76MB (12+64) | 170W / 230W | DDR5-5200 |
| Core i7-13700K / KF | $409 (Ok) – $384 (KF) | 16 / 24 (8+8) | 3.4 / 5.4 | 2.5 / 4.2 | 54MB (24+30) | 125W / 253W | DDR4-3200 / DDR5-5600 |
| Ryzen 7 7800X3D | $449 | 8 /16 | 4.2 / 5.0 | 104MB (8+96) | 120W / 162W | DDR5-5200 | |
| Ryzen 7 5800X3D | $348 ($449) | 8 /16 | 3.4 / 4.5 | 104MB (8+96) | 105W | DDR4-3200 | |
| Ryzen 7 7700X | $349 ($399) | 8 /16 | 4.5 / 5.4 | – | 40MB (8+32) | 105W / 142W | DDR5-5200 |
The Ryzen 9 7900X3D has 12 cores and 24 threads and drops into the AM5 socket. On the floor, the 7900X3D seems to be a Ryzen 9 7900X with an additional L3 cache chiplet and extra silicon and software program tuning, nevertheless it has loads of distinctive traits in comparison with the opposite 3D V-Cache chips.
As with all different 3D V-Cache chiplets, the 3D-stacked SRAM L3 chip weighs in at 64MB, so the 7900X3D comes with a beastly 144MB of whole cache, with 128MB of that being gaming-boosting L3 cache. The extra L3 cache chiplet is fabbed on the 7nm course of and has a peak bandwidth of two.5 TB/s. You possibly can learn the finer-grained particulars of the second-gen 3D V-Cache know-how and the 6nm I/O Die (IOD) right here.
AMD has considerably elevated the increase speeds with the brand new X3D fashions — the 7900X3D boosts to five.6 GHz, the identical as the usual 7900X however 100 MHz decrease than the 7950X. The 4.4 GHz base clock is 200 MHz larger than the 7950X3D however 300 MHz decrease than the 7900X, a mandatory lodging for the decrease energy envelope.
AMD’s Zen 4 3D V-Cache processors all have a base TPD of 120W and a max 162W PPT, which means the 7900X3D’s rankings are 68W decrease than the 170W/230W score for the usual 7900X. The ability reductions aren’t solely stunning, as the extra cache chiplet ends in barely larger working temperatures that have to be saved inside a secure vary. In truth, the 7900X3D’s most supported temperature is 89C, decrease than the 7900X’s restrict of 95C. The 7900X3D doesn’t include a bundled cooler — AMD recommends a 280mm water cooler, or higher, for the Ryzen 7000X3D processors.
The Ryzen 9 7900X3D’s RDNA 2 built-in GPU comes with two compute models, 4 ACE, and 1 HWS, so it does not have the required horsepower to profit from sooner CPU cores — this unit is solely GPU compute-bound. The iGPU additionally does not profit from the 3D V-Cache tech as a result of it resides on the I/O die and might’t entry the L3 cache, so efficiency is similar because the common Ryzen 7000 processors, as you’ll be able to see in our iGPU testing.
AMD solely allowed overclocking the reminiscence and Infinity Material for the previous-gen 5800X3D however will now additionally enable each the auto-overclocking Precision Increase Overdrive (PBO) and Curve Optimizer. AMD nonetheless doesn’t enable direct frequency overclocking as a result of a voltage limitation for one of many chiplets.
Each the 7950X3D and the 7900X3D have two eight-core Core Compute Die (CCD) chiplets paired with a central I/O Die, marking the primary time AMD has introduced the 3D V-Cache tech to a multi-CCD processor. Nevertheless, the 7900X3D’s two compute chiplets (CCD) have six cores energetic apiece (6+6), whereas the 7950X3D has eight cores energetic per compute chiplet (8+8).
The above picture exhibits that AMD solely mounts a single 7nm SRAM chiplet atop one eight-core CCD, leaving the opposite CCD naked. Provided that each chips have the identical L3 cache capability, the cache-per-core varies as a result of completely different core counts.
| Row 0 – Cell 0 | CCD 0 (3D V-Cache) | CCD 1 (Regular) |
| Ryzen 9 7900X3D L3 | 96 MB | 32 MB |
| Ryzen 9 7900X3D L3-per-core | 16 MB | ~5.3 MB |
| Ryzen 9 7950X3D L3 | 96 MB | 32 MB |
| Ryzen 9 7950X3D L3-per-core | 12 MB | 4 MB |
As you’ll be able to see within the desk above, the Ryzen 9 7900X3D really has extra shared L3 cache per core on the 3D V-Cache geared up chiplet at 16 MB-per-core than the Ryzen 9 7950X3D at 12 MB-per-core. It additionally has extra L3 cache-per-core on the conventional die, too. In distinction, AMD disables L2 cache slices in tandem with the cores, so the 7900X3D has 12 MB of L2 whereas the 7950X3D has 16 MB.
The 3D V-Cache chiplet ends in decrease efficiency as a result of thermal challenges you’ll be able to examine right here, however AMD’s new design permits the chiplet with out the 3D-stacked SRAM to function at full pace, thus delivering the excessive increase clocks we see on the spec sheet for functions that prize frequency. In the meantime, the SRAM-stacked CCD will function at a barely decrease clock charge than the rated increase for the chip however fulfill the wants of functions that reply finest to low-latency entry, like video games. The chip itself has a peak of 1.4V, enabling the excessive increase clocks, however the 3D V-Cache-equipped CCD has a ~1.1V restrict to maintain thermals in examine.
Solely fusing the SRAM onto one CCD additionally reduces manufacturing prices, because the hybrid bonding course of and extra chiplet make this an costly know-how. AMD additionally says that utilizing two V-Cache chiplets doesn’t present sufficient efficiency uplift to justify the additional prices.
In both case, the brand new design requires the mix of a brand new chipset driver and Home windows Xbox Sport Bar to position threads for various kinds of workloads on the right chiplet. AMD’s new thread administration method requires Home windows 10 (1903) or 11 (21H2) and 4 parts — a brand new chipset driver, up to date BIOS, Home windows Sport Mode, and an up to date model of the Xbox Sport Bar (you’ll be able to replace by means of the Microsoft Apps retailer).
Picture 1 of 7
You possibly can learn the deep-dive particulars on AMD’s thread-targeting implementation right here, however the important thing takeaway is that the 4 completely different parts work collectively to mechanically alter the thread assignments into the cores. This method assures the OS thread scheduler can goal the right cores based mostly on present utilization — the 3D V-Cache enabled chiplet for cache latency-sensitive duties like gaming, or the usual chiplet for workloads that reply finest to larger frequencies. You can even override the automated switching between core-ranking modes to customise the settings, however we discovered that the automated mechanisms work properly.
The chipset drivers additionally embody a part that improves efficiency by parking the slowest cores when Sport Mode or Blended Actuality Mode is energetic. In impact, this shuts down the ‘normal’ CCD, thus constraining latency-sensitive workloads (like video games) to the 3D V-Cache chiplet. This improves the cache hit charge and reduces high-latency communication between the 2 CCDs, leading to sooner efficiency for workloads that don’t want entry to (or profit from) all of the cores.
Naturally, each teams of cores are used if the workload is parallelized sufficient. The thread focusing on and core parking solely kick in for dual-CCD Zen 4 X3D processors, so that they received’t influence efficiency with common chips.
Sport Mode is designed to scale back the influence on gaming from background duties, however it’s best on lower-end methods. It does have a historical past of generally unnecessarily throttling background duties on higher-end methods that don’t want as a lot useful resource preservation, which might be problematic with video streaming or different apps that run within the background. We performed our personal cursory exams and didn’t encounter any points, however we’re testing the influence of background apps additional.
Now let’s transfer on to the gaming and productiveness benchmarks on the next pages.