How is DDR5 different from DDR4?

Architectural differences between DDR4 and DDR5.

In terms of architecture, the DDR5 standard is noticeably different from its predecessor.

  • The first and most obvious innovation is a separate controller that is responsible for managing memory power, regulating voltage and synchronizing the operation of the sides of the module halves. In the case of DDR4, the processor and motherboard are responsible for this.
  • The configuration has changed: the connector for motherboard connection has not changed (288 pins, the same as DDR4), but the key has moved to the left, which makes DDR4 and DDR5 modules physically incompatible.
  • DDR5 has doubled the number and density of memory banks, making it possible to seamlessly increase memory capacity within a single module. This has already made it possible to increase the maximum memory module capacity from 32GB to 48GB and 64GB.
  • In terms of frequencies and bandwidth, the DDR5 standard is noticeably ahead of its predecessor. DDR4 memory, which started at a modest frequency, later received the official JEDEC profile of 3200 MHz, and modules from 3600 to 4400 MHz are considered overclocking. In turn, the only official JEDEC DDR5 profile so far is 4800 MHz. And overclocking modules start from 5200 to 6200 MHz.
  • DDR5 does not feature rank division; rather, it comprises two virtual memory modules on a single physical module. This configuration presents itself to the operating system as dual-channel memory for one module and quad-channel memory for two modules.

Also, DDR5 has simplified the overclocking process due to an increased number of profiles in XPM/EXPO, a built-in ODECC error corrector has appeared, and the operating voltage has decreased to 1.1 – 1.2 V. You can learn more about the difference between the two formats from our last year comparison.

As for compatibility, DDR5 was only supported by 12th generation Intel Alder Lake processors at the start, later joined by 13th and 14th generation chips. At the same time, some Intel motherboards and processors can work with both DDR4 and DDR5. AMD, for its part, a year ago began the active introduction of Ryzen 7000 processors and AM5 socket, which is exclusively designed for the new standard.

Comparison of DDR5 vs DDR4 performance in synthetic tests

Test system:

Since the JEDEC DDR4-3200 standard is the most popular for DDR4 sticks, we chose it as a reference for subsequent tests. The most popular Kingston Fury Beast DDR4 2x8Gb set with CAS 16 timings helped us in our tests. Of course, you can also find much more productive overclocker-level sticks on sale with operating frequencies up to 4400 MHz, but in most cases they are either expensive or inadequately expensive, so and are considered enthusiast products. We will compare with the same mass-market set from Kingston, which included 2 DDR5 modules with a base operating frequency of 4800 MHz and CAS 38 timings.

3DMark Benchmark (overall score)
DDR5 10215 score
DDR4 10138 score

In the first synthetic tests 3DMark Time Spy, which evaluates the overall system performance, both sets showed almost identical results: the DDR5 set scored 10,215 points, while the DDR4 set lagged behind literally by a third of the case, scoring a final 10,138 points. In the tests for graphics and processing load the results were also almost identical, such a level of error can be seen even when comparing components of the same class by comparing identical models of different brands.

UL Procyon Benchmark for Adobe Premiere Pro (average score)
DDR5 7856 score
DDR4 7653 score

The UL Procyon benchmark, which simulates a workload in Adobe Premiere Pro video editing software, showed that the increased bandwidth may give a small performance boost, but it's so insignificant: 7856 points for DDR5 vs 7653 points for DDR4.

7-Zip compression and decompression benchmark (average score)
DDR5 135636 score
DDR4 115455 score

But in tests of data compression and decompression in 7-Zip, we finally saw the real advantage of the new DDR5 architecture. When compressing a single 10 GB file, the fifth-generation memory set was almost 25% faster than DDR4. Final scores for all tests: 135,636 points for DDR5 vs 115,455 points for DDR4.

Performance comparison in real applications

Since the launch of DDR5, many manufacturers have announced new product lines for artists, designers, and musicians.

Transitioning from theory to practical testing, let's compare the performance of DDR4 and DDR5 memory in professional software applications such as DaVinci Resolve, Adobe Premiere Pro, Blender, and others. In a specific test scenario using DaVinci Resolve to render a 10-minute 4K video, the DDR4 memory completed the task in 4 minutes and 17 seconds, while the DDR5 took slightly longer at 4 minutes and 19 seconds. Interestingly, further experimentation with different timings and frequencies yielded similar results, with variances of no more than 5 seconds. These findings suggest that professionals working with DaVinci Resolve may find greater benefit in investing in a more powerful graphics card rather than upgrading to DDR5 memory, as the GPU ultimately has a more significant impact on performance.

DaVinci Resolve: rendering a 10-minute MP4 video in 4K resolution
DDR5-4800 4:19 min
DDR4-3200 4:17 min

In Adobe Premiere Pro tests, the situation is quite similar to what we have seen before: DDR4, despite the bandwidth limitations, demonstrates almost identical results compared to DDR5, and the determining factor is again the graphics card. To be honest, we were expecting DDR5 to take the lead here and the slight advantage of DDR4 was a pleasant surprise for us.

Adobe Premiere Pro: rendering a 13-minute MP4 video in 4K resolution
DDR5-4800 3:28 min
DDR4-3200 3:20 min

The results of the third test conducted in Blender reinforced our concern regarding the high latency of DDR5 modules, which frequently counteracts the advantages of their higher clock speeds. In this test, the DDR4 set successfully handled a complex and intricately detailed Junkshop scene in 2 minutes and 33 seconds, while the DDR5 completed the task only 4 seconds faster.

Blender: Junkshop Demo in 4K
DDR5-4800 2:29 min
DDR4-3200 2:33 min

Both in Photoshop itself and in its subsidiary benchmark PugetBench the situation is the same, the difference in rendering is literally 3 – 5 seconds, so we suggest to move on and see if we will see any significant difference in games.

Does DDR5 give a performance boost in games?

When using DDR5 in Cyberpunk 2077, the average performance increase is between 1 and 3%. The average FPS in 1080p resolution with the graphics settings turned to maximum is 88 frames per second. Once again, the graphics card and processor have a much stronger influence on the final result.

Cyberpunk 2077: 1080p, Ultra settings, average FPS
DDR5-4800 110 FPS
DDR4-3200 108 FPS

In Red Dead Redemption 2 with ultra graphics settings and 1080p resolution, the difference in results also does not exceed 3%. To be honest, by this point we had already despaired of finding any noticeable difference, but it was worth a try.

Red Dead Redemption 2: 1080p, Ultra settings, average FPS
DDR5-4800 144 FPS
DDR4-3200 142 FPS

Having reached Far Cry 6 with its large and open game world, it was decided to add several identical memory modules with different clock frequencies to the test. And this time it turned out a little more interesting, but nothing radically new: the difference is still tiny, but now a slight advantage is on the side of DDR4. Overall, the results are as follows: DDR4-3200 shows similar performance compared to DDR5 4800 MHz, DDR4-3600 is on par with DDR5-5200, and DDR4-4000 is almost as good as DDR5-6000.

Far Cry 6: 1080p, Ultra settings, average FPS
DDR5-4800 133 FPS
DDR4-3200 136 FPS

Honestly, at this point it feels like quoting the iconic villain Mando from Far Cry 3. "Insanity is the exact repetition of the same action over and over again,

in hopes of change." So in order not to clog the atmosphere with dozens of graphs, let's just note that in Baldur's Gate 3, Horizon Zero Dawn, Doom Eternal, COD Modern Warfare and Callisto Portocol the situation is identical. In fact, DDR5 really beat DDR4 only in Shadow of the Tomb Raider, Resident Evil 4 and Star Wars Jedi: Survivor. Moreover, in the latter, the advantage of DDR5 reached 30%! However, this is rather an exception to the rule, since the game is known for its excellent optimization and slowdowns even on graphics cards of the GeForce RTX 4080 level. More graphs from TechSpot — https://www.techspot.com/review/2777-ddr5-vs-ddr4-gaming/.

Finally, we would like to note that only high-level sets with clock speeds over 7000 MHz provide a stable advantage in games. For example, in Baldur's Gate 3, when comparing two DDR4-4000 and DDR5-7200 sticks, the difference in performance was 12% in favor of the latter. However, DDR5 modules with the frequency of 7200 MHz and higher are much more expensive than DDR5-4800 RAM, so here again we have to state the fact that it is better to invest this money in a more powerful processor and graphics card.

Afterword: is it worth updating or not?

DDR5 modules with a frequency of less than 7200 MHz do not yet provide a significant advantage when working with professional software.

Before conducting the tests, we were confident that DDR5 would emerge as the clear winner in performance, with the only uncertainty being the margin of victory. However, our findings have prompted us to reconsider our assumptions. As observed, DDR4-3200 and DDR5-4800 memory sets perform similarly across all scenarios, including benchmarks, professional software applications, and gaming. Only DDR5 modules with frequencies of 7200, 7600, and 8000 MHz offer a significant advantage, but their high prices render them unsuitable for budget and mid-budget PCs. Nevertheless, for sub-flagship and flagship gaming or work PCs, such high-frequency DDR5 memory could represent a compelling option, providing a solid foundation for future performance enhancements.