After a recent review of the Ryzen 5 5600G, it occurred to me that it might be interesting to test how much memory affects the performance of the integrated Vega 7. The original plan also included overclocking the iGPU part, as the core ticks at a maximum of 1900MHz and at 2400MHz. The Vega 7 in the 5600G has only 448 shaders and the effect of RAM throughput will be lower here than with the Ryzen 7 5700G, which has an iGPU with more active CUs. Unfortunately, I was slightly disappointed with the motherboard, for some bizarre reason it was not possible to overclock the iGPU and for this reason I did not try even more memory and try at speeds exceeding 4000 MT / s. It was not possible to overclock the iGPU on the tested Gigabyte X570 I AORUS PRO WIFI motherboard for some strange reason. Overclocking did not work directly in the BIOS in the standard OC menu, the buried menus in the AMD CBS / PBS menu also did not work. I tried it directly in Windows through MSI Afterburner, but even that was not successful. Changing the PTT limit also didn’t help. So take the test more from the point of view of the factory non-overclocked R5 5600G with different memories.
I’ve already returned the processor, so I can’t try another motherboard, there was no time for that.
As for the test set, the basis is the same for all memories, ie Gigabyte X570 I AORUS PRO WiFi miniITX motherboard with Ryzen 5 5600G. The processor cools the Noctua NH-U12S SE-AM4 heatsink with the Noctua NF-F12 iPPC-2000 fan. The set is powered by the Corsair HX1200 power supply and the primary operating system is Windows 10 Pro v20H2. Windows is installed on a Gigabyte AORUS 7000s Gen4 2TB SSD, which is located in the primary M.2 slot next to the X570 chipset.
Now for the different memory configurations, I finally chose six different configurations, so in all the graphs I only distinguish the memory configuration, because the processor is the same everywhere. In the BIOS, I chose the “Game Optimized UMA” option, which sets 2GB of memory for the iGPU if the system has at least 16GB of RAM. If one 8GB module is used, it will be reduced to 512MB, however iGPU can take more dynamically.
G.Skill TridentZ RGB 2x8GB DDR4-3200
I also used this memory kit in the original review and it is a relatively standard memory kit, there are of course worse 3200 MT / s memories with worse timing, but I would say that this set is enough for an idea. To make the test a bit more interesting, I also tried to leave only one module in a single-channel connection, as this is something you can come across in the real world.
HyperX Fury RGB 2x8GB DDR4-3600
Another test set is among the rather cheaper 3600 MT / s memories with worse timing, memories of this type are relatively cheap and it may be interesting to compare faster paper modules with slower 3200 MT / s, which have better timing.
Gigabyte AORUS RGB 2x8GB DDR4-3933
I also received this AORUS memory kit from Gigabyte, which also contains two humorous demo modules, which are basically such memories without memory, only a PCB with a heatsink and an RGB LED. The memories support the speed of 3600 MT / s with the XMP profile, so I decided on some manual overclocking, in the BIOS I then used the “AORUS Memory Boost” option, which performs a kind of automatic overclocking. I eventually drove the memory to 3933 MT / s stably, the problem was rather that I could not overclock the iGPU, the increase in performance compared to 3800 MT / s was very small.
Kingston KVR24E17S8 ECC 2x8GB DDR4-2400
These memories are standard green bits with ECC support, I used them as an example of slow cheap memories without an XMP profile to spice up the whole test, I left only one module on the board, which I deliberately slowed down to 2133 MT / s.
Source: Diit.cz by diit.cz.
*The article has been translated based on the content of Diit.cz by diit.cz. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!
*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.
*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!