After yesterday’s launch of the RX 5500 XT and today we learned about the price reduction of some NVIDIA models, the question remains: is it worth paying more and choosing the NVIDIA model? 8 GB vs. 4 GB? We are talking about a chip that is pure mid-range and which in itself comes with a slightly higher price, so what is the real difference between the two models?
Due to the fact that we’ve been able to access two different models of the same chip, the comparison can’t wait, especially considering the step that developers are taking regarding the so-called RAM cache that The latest engines do.
Therefore, already ending 2019 and fully entering 2020, it is more interesting to see and understand the effects of VRAM and its size compared to performance per euro investment. In addition, the best-selling range of graphics cards, with more adjustable features and a focus on the player with a 1080p screen, makes the comparison even more interesting if possible.
With that, let’s meet the two main characters and start with performance and VRAM data.
Sapphire Plus RX5500XT 4GB vs Power Color RX5500XT Red Dragon 8GB
The two cards have many similarities, such as the ITX PCB, the same watch rate, a double fan design, the backplate, and a single 8-pin connector. In addition, they are at the forefront of this series of GPUs, so the hostility from a performance standpoint is maximum.
We’ve already analyzed the Sapphire card, so we know it well, and its 4GB has behaved more politely in most scenarios. But as we have pointed out in its respective reviews, its performance is somewhat inconsistent, which has led us to think on the basis of different performance metrics that VRAM size, terms of configuration, performance This may be the reason for the change.
To confirm this, we are going to compete against these two natural rivals in 5 different games, with different engines and under our joint testing system, which consists of:
- Intel Core i7-8700K (No Deal, Stock, AS5)
- ASUS Maximus X Formula
- Corsair Vengeance Pro RGB DDR4-3200MHz Cas16 (Samsung B-Die, XMP ON)
- ASUS STRIX RTX 2060 O6G (stock)
- Corsair AX1200i Platinum
- Corsair MP510 960GB
- ASUS STRIX Helios
- EK Vardar EVO RGB (x 4)
- Tacklif HM02 (thermometer and hygrometer)
- Ckeyin DNM-51 (sound level meter)
- Windows 10 1909 64 bit
The games that have been tried are Ashes of the Singularity: Classic, Assassin’s Creed: Odyssey, Battlefield V, Ghost Recon Wildlands, Shadow of the Tomb Raider. We thought of introducing other topics like Metro Exodus, Call of Duty: Modern Warfare or Breakpoint, but the drivers in all three did not finish the job, where we have everything from blue screens to PC restarts. Was
Used drivers are the latest available (Adrenalin 2020 19.12.2 December 12 and WHQL) so all conditions are very favorable for this comparison.
The data presented for the same game belongs to the same pattern, so they are recoverable, for example, it can be seen between a framerate and a frame time, where it is more clearly defined.
Ready? Let’s go there
Ashes of uniformity: Classic
A title made purely in DX12 is a must-have test object for any tester and will test the performance of each graphics card. As we can see, the difference between the two models is minimal, where the minimum, maximum and approximate is found.
We have a lot of information here, because, although the minimum, maximum and average data reflect a certain equation, the agreement with the things seen in the graph is practically the opposite, which we will see later. I will see this in the following graph.
In any case, the 4GB version seems to have better smoothness in this game.
Equality is such that even a small percentage is attached to the nails, who will take the cat to the water?
Here we see something very interesting: the 8GB version gets less FPS deviation, but the 4GB version gets more smoothness, exactly what the frame time shows. As we can see, the differences are still minimal, so in the Ashes of the Singularity: the 4GB version of the classic prevails, although the relevant ways to consider choosing one model or the other. Not from
Assassin’s Creed: The Odyssey
The Odyssey’s analog graphics engine starts to make a difference even under the DX11. Even with the same chip and the same frequency, the difference in performance is just spectacular: 11.7 FPS. Also, as the power color version (8GB) gets better, less and more, the graph is clear about the difference in this frame rate.
The frame time is interestingly similar, with a lot of smoothness on both cards. Aside from the performance peaks that damage FPS, it can be seen that 11.7 FPS is not ridiculed in this section, as they are separated by an average of 9 ms and where at least the same There are.
Battlefield V demonstrates another similar scenario, where texture loading fines the performance of the 4 GB version. The average performance difference is at 17.2 FPS, where the shortcomings in this case skyrocket in favor of the larger version of VRAM.
The DICE engine is probably the most suitable for AMD architecture. This is exactly how the smoothness is maximized while playing and where at the same time, the sapphire has more problems maintaining the type, with the addition of some quite annoying MS which we have already framed. I have seen
It can already be considered as the famous “jerk” in the game, fortunately there are only two and they certainly correspond to moments of high structural load.
Division 2 is one of the better titles than many AMD’s, so we expect little difference in performance. Something that is demonstrated in the frame rate, where we can see that the difference in FPS is really very small between the two models.
Frame times are just as great, it’s hard to nail the data as much as possible, where smoothness is undoubtedly the central character when playing. There’s hardly a leap here and when they do they are minimal, just 3-4 ms up, the parity is still there.
Nothing to highlight in the low percentage, again the maximum equality between the two versions.
Here we have the first difference which is really noticeable because the FPS variable is better in the 8 GB version, where it is very surprising that both the cards have eliminated the variability in terms of stuttering, which we have so far. Did not see.
Does VRAM matter? Is it decisive for 2019 and 2020? Is it worth investing in a VRAM model? At least on these RX 5500 XTs the answer to all three questions is a resounding yes. We live in a time where everything in system RAM is loaded to a minimum, something that in general terms indicates that it is going to be an easy task.
The developers seem to have this clear: Don’t have enough VRAM on your GPU? No matter, the game engine will export textures into the system RAM and you will be able to play even if you lose performance. Improve at another time.
This pushes the market into a complex trend. We’ve seen how low VRAM forces us to have as much RAM in our PC as possible so that everything is ready with the least possible problems, it has to do with performance.