Intel's latest 12th-Gen Alder Lake chips add another factor into your PC build equation: RAM. Naturally, you'll want the best RAM for your shiny new Alder Lake builds, but with Alder Lake's hybrid support for DDR4 and DDR5 memory, you have to contemplate whether to stick with the declining DDR4 standard or jump on the new DDR5 bandwagon. Unfortunately, as with any cutting-edge technology, early DDR5 memory kits are pricey. In addition, the ongoing semiconductor shortage has hobbled DDR5 production. With a lack of DDR5 supply and excessive pricing, picking between DDR4 and DDR5 becomes more troublesome.
Alder Lake officially supports DDR4-3200 and DDR5-4800 memory modules. There are no considerations with the former since DDR4-3200 is the standard across all configurations regardless of whether you're using single-or dual-ranked memory modules or the number of memory slots you plan to fill. With DDR5, however, there are certain caveats to take into consideration.
The hybrid desktop processors support DDR5-4800 on motherboards with two slots, irrespective of the nature of the memory modules. With motherboards with four slots, the supported data rate drops down to DDR5-4400 when you fill two memory slots. In a scenario when all four memory slots are populated, Intel only guarantees DDR5-4000 with single-rank memory modules and DDR5-3600 with dual-rank memory modules.
DDR5 Frequency (MT/s) | Slots Per Channel (SPC) | DIMMs Populated Per Channel (DPC) | Rank (s) |
---|---|---|---|
4,000 | 2 SPC | 2 DPC | 1 R |
3,600 | 2 SPC | 2 DPC | 2 R |
4,400 | 2 SPC | 1 DPC | 1 R |
4,400 | 2 SPC | 1 DPC | 2 R |
4,800 | 1 SPC | 1 DPC | 1 R |
4,800 | 1 SPC | 1 DPC | 2 R |
Intel introduced its Gear Mode with its previous generation of Rocket Lake processors, and Alder Lake retains the exact mechanism. In a nutshell, the Gear mode is the ratio between the memory and processor's memory controller. Gear 1 is when they both are on the same level in a 1:1 ratio, whereas Gear 2 finds the memory at twice the speed of the memory controller (2:1 ratio). The latter opens the door for the processor to support data rates way above Intel's specifications, but it comes at the cost of higher latency. With Alder Lake, there's even a Gear 4, further dropping the ratio to 4:1 for future high-speed DDR5 memory.
By default, Alder Lake supports DDR4-3200 on Gear 1 across the product's entire lineup. However, golden chip samples can hit DDR4-4000 in Gear 1 without problems. At the same time, Alder Lake only supports DDR5 memory on Gear 2 and receives the penalty of higher latency.
As a reminder, Gear 1 is best for low-latency applications like gaming and lightly-threaded apps. Gear 2 results in higher data transfer rates, which can benefit some threaded workloads, but also results in higher latency that can lead to reduced performance in some applications — particularly gaming.
Alder Lake Test System and Setup
We tested six different frequencies for DDR4, ranging from DDR4-2133 to DDR4-3600. Unfortunately, we had to stop at DDR4-3600 because our retail Core i9-12900K sample that we used for this testing refused to run DDR4-4000 with the Gear 1 setting. Our sample size is small, but two out of our three Core i9-12900K samples spread out among our different labs could successfully do DDR4-4000 on Gear 1.
For DDR5, we evaluated five frequencies, starting from DDR5-4800 up to DDR5-6200. There wasn't any choice here as we had to cling to Gear 2 to adhere to Intel's specifications.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
We used a bunch of memory configurations to assess the impact of memory ranks. Our DDR5 tests included 8GB (1x8GB), 16GB (2x8GB) and 32GB (4x8GB) setups, while we performed our DDR4 tests with 16GB (2x8GB), 32GB (2x16GB, 4x8GB) and 64GB (4x16GB) configurations.
Processor | Intel Core i9-12900K |
Motherboard | MSI MAG Z690 Tomahawk WIFI |
Row 2 - Cell 0 | MSI MAG Z690 Tomahawk WiFi DDR4 |
Memory | G.Skill Trident Z5 RGB DDR5-6000 C36 2x16GB |
Row 4 - Cell 0 | Crucial DDR5-4800 C40 2x8GB |
Row 5 - Cell 0 | Trident Z Royal DDR4-4000 C14 2x8GB |
Row 6 - Cell 0 | Trident Z Neo DDR4-3600 C14 2x8GB, 2x16GB |
Storage | Crucial MX500 500GB, 2TB |
CPU Cooler | MSI MAG CoreLiquid K360 |
Graphics Card | MSI GeForce RTX 2080 Ti Gaming X Trio |
Power Supply | Corsair RM650x |
Case | Streacom BC1 |
Operating System | Windows 11 Professional |
Display Driver | Nvidia GeForce Game Ready 511.23 WHQL |
We ran our RAM benchmarks three times to minimize the margin of error and used the median value as the final result.
We updated the firmware for our MSI MAG Z690 Tomahawk WiFi (DDR4) to the latest publicly-available revision. In addition, we used a fresh 64-bit installation of Windows 11 Professional on the software front with all available updates. Furthermore, we updated our test system's drivers, benchmarking programs, and game clients to the latest versions at testing.
Intel Alder Lake Memory Ranks
Going from two memory ranks (2x8GB) to four (4x8GB) awarded us a 3.4% performance increase in our cumulative measure. Of course, you'll want to avoid running single-channel memory as there was a 15.1% difference between running a single-channel setup (1x8GB) instead of a dual-channel (2x8GB) one.
Specific workloads benefitted more from four memory ranks in comparison to two. In Adobe Premiere, 4x8GB offered 11.5% more performance than 2x8GB. In 7-Zip compression workloads, the former was up to 12% faster than the latter.
Due to the short supply of DDR5, we had to pair four 8GB DDR5 memory modules together to make up four memory ranks. In an ideal scenario, you would opt for two 32GB DDR5 memory modules instead. Remember that with DDR5, the 16GB memory modules have a single-rank design. With four filled memory slots, Intel only guarantees to support up to DDR5-4000, so higher data rates could require manual tuning.
When it came to gaming performance, the difference between two memory ranks and four was negligible. For example, the delta between a single memory module and two memory modules was only 3.8% in a gaming environment.
With DDR4, we saw a 4.6% cumulative uplift between two memory ranks (2x8GB) and four memory ranks (2x16GB). Since DDR4 memory kits are not as scarce as DDR5, there was more flexibility to get to four memory ranks. Of course, you can go with a 2x16GB or 4x8GB configuration, but the final result is comparable.
Again, Adobe Premiere demonstrated that it was worth maximizing memory capacity and ranks. For example, 4x16GB delivered 12.7% more performance than 2x16GB and 25.5% than 2x8GB.
Memory ranks with DDR4 were irrelevant in gaming. As a result, we didn't record any significant performance differences between the four configurations that we tested.
Intel Alder Lake Memory Scaling
DDR5-6200 was 3.1% faster than baseline DDR5-4800 under the exact primary timings. Therefore, the margin will be more prominent when comparing memory kits with different timings.
Adobe Photoshop responded well where DDR5-6200 showed 4.5% better performance over DDR5-4800. Lightroom followed suit with DDR5-6200 scoring 3.2% over DDR5-4800. However, the most considerable margin came from 7-Zip compression, which saw DDR5-6200 outperform DDR5-4800 by 10.8%.
Faster DDR5 memory speeds didn't contribute to improving gaming.