Intel’s Alder Lake-S processors should launch later this year with support for both DDR4 as well as DDR5 memory, although we wouldn’t suggest pairing them with the former. A new benchmark of a 16-core/24-thread Alder Lake-S CPU has surfaced, and while normally it wouldn’t be worth a second look, this particular SKU was tested with DDR4-3200 memory instead of DDR5. This lets us analyze the impact of the newer memory standard on CPU performance. Before we begin, please keep in mind that both the chips are engineering models, and as such, the final performance will greatly vary:
The single and 2-core performance is largely unaffected (as expected), but the quad/octa-core CPU performance takes a hit of 8-12% with DDR4-3200 memory. Scaling it up to 64 cores, the benchmark shows a steep decline of nearly 50%, but I reckon that’s a bit of an over-exaggeration, something very common with Userbenchmark.
The memory latency of the two configurations is rather interesting. Although DDR5 kits (at least for now) come with much higher timings (CL30+), the DDR4-3200 kit still has a 50% higher latency penalty than the former.
The DDR5-4800 kit reports a max latency of 85-90ns, while the DDR4-3200 modules top out at just under 140ns. It’s worth noting that both setups (likely) feature dual-channel memory, with two DIMMs populated. This shows that gaming performance with DDR5 should be quite a bit higher than existing DDR4 modules. This should result in some fairly large leaps in gaming performance with Intel’s Alder Lake processors. Will be worth seeing how AMD responds to this, and whether the 3D V-Cache alone will be enough to make up for it.
Source: Twitter (Apisak)