Today, the gaming benchmarks are a bit more reduced, firstly because Ubisoft had broke my Far Cry 6 installation with Uplay and secondly because in the resolution “4K” or 3840 x 2160 effectively no differences between the configurations are measurable. The reason for this is, of course, the bottleneck that moves more and more towards the GPU at such high resolutions. Thus, even an Nvidia RTX 3090 reaches its limits and the CPU-RAM combo in front of it becomes almost unimportant.
The graphics card was set to maximum performance and temperature limits with MSI Afterburner to shift the focus in the render pipeline to the CPU and RAM as much as possible. The benchmark in Cyberpunk 2077 version 1.31 is our 1-minute drive and run parkour around and through Kabuki Market, while Shadow of the Tomb Raider uses the built-in benchmark. The performance data is recorded with Nvidia Frameview 1.2, based on the open source software Presentmon.
Quad HD – 1440p
In both Shadow of the Tomb Raider and Cyberpunk 2077, the ratio looks very similar at 1440p resolution. DDR5 is just ahead of DDR4 and the respective dual-rank config is even closer ahead to single-rank. In Cyberpunk, the two DDR5 configurations even come to the exact same Average and 1% Low FPS, which is not an error in the data. And even though DDR4 single-ranks performs better than dual-ranks here, these are just the Average FPS, not the 1% Low which is often even more important. All in all, the differences amount to about 2 FPS in extreme cases – as often measurable, but probably not perceptible.
The ranks are more important than the difference between DDR4 and DDR5 when it comes to the frame time variance, which can be used to track possible “stutters” while gaming. Even though we’re talking fractions of percentages here, dual-rank offers a smoother gaming experience than the single-rank configurations. To be fair, we are already very close to the actual measurement tolerance. Here you can see again why 4K tests would not bring any added value in this case.
Full HD – 1080p
In Full HD or 1920 x 1080 pixels, the differences now become increasingly larger, but so do the FPS numbers in total. The order is the same as 1440p and while the differences are limited to a maximum of 3 FPS in both titles, the impact is greater in percentage terms in Cyberpunk 2077. Interestingly, DDR4 Dual-Rank is behind Single-Rank in Average FPS, but ahead of it in 1% Low Fps, just like in 1440p. And even though the top bar is noticeably longest with DDR5 in Dual-Rank, that’s only a few percent, which will probably never be experienced in reality.
In terms of frame time variances, the image is again similar to 1440p. Dual-rank is again just ahead of single-rank here in terms of the smoothness of the image. I’ll spare you the other graphs because the differences are so small that you can see even less difference between the lines there. For those wondering about efficiency, I can tell you the CPU has the same power consumption with any RAM config. As a result, with the faster RAM configs, efficiency naturally increases, at least in theory.
I couldn’t reliably measure how much power the RAM allows itself to consume. Because even if DDR5 integrates a software monitoring with the PMIC on the RAM modules that can also be read out with HWinfo, this does not work for DDR4. And also a measurement over the cables from power supply to mainboard is rather difficult, because the voltage regulation for the RAM is supplied over the 5 V rail of the 24-pin ATX connector. But the bottom line is that the power consumption per module should be in the single-digit watt range and effectively compared to CPU, GPU and other background noise in the system this is only a very small part.
19 Antworten
Kommentar
Lade neue Kommentare
Mitglied
Veteran
Urgestein
Veteran
1
Veteran
Veteran
Neuling
Urgestein
Urgestein
Neuling
Veteran
Neuling
Mitglied
Urgestein
Veteran
Mitglied
Veteran
Neuling
Alle Kommentare lesen unter igor´sLAB Community →