What processor is necessary to avoid suffering a bottleneck with an NVIDIA RTX.
Contents
When making an outlay as important as a graphics card from the RTX range, we must be clear about several points regarding our equipment. It would not be logical to spend more than 1000 euros on a graphics card to not be able to extract its full potential due to components such as the CPU, so what processor do I need to avoid suffering a bottleneck with one of NVIDIA’s RTX GPUs?
We must not confuse some concepts when talking about bottlenecks
Many are the users who confuse the bottleneck with the different performance that the CPUs present.
It may seem obvious, but the truth is that many forums and websites confuse both terms when a CPU does not measure up.
This does not imply that it is being throttled by the performance of the graphics card, simply that it performs poorly compared to its adversaries.
For a bottleneck to occur in a CPU, it must limit the performance of the graphics card in terms of FPS, regardless of the resolution chosen and usually at the same settings used, or on the other hand not allow the GPU to reach a high degree of workload.
If we talk about benchmark, the score should be different, that is, the CPU that produces a bottleneck should give a lower score than the rest of the processors, normally less than 1% difference, considered as a margin of error in the benchmarks. processes.
As if that were not enough, AMD and Intel CPUs behave differently in the face of this fact, mainly due to the differences in architecture and technical characteristics that they contain, so we will treat them independently and based on data that we have collected, some of them they already worked on different articles.
Intel Core vs. Pentium vs. Celeron and AMD Athlon vs. Ryzen
To determine the bottleneck in a CPU, three factors are essential, whether in benchmarks or games: number of cores and threads, turbo frequency, and IPC of the architecture.
The combination between one and the other will mark the limitation in performance and will determine if it is time to acquire a new processor or, on the contrary, if we can buy the GPU that we need or want.
Before starting with the data, we are going to eliminate from the equation and directly several series, both from AMD and Intel, since, as we will see later, they cause a bottleneck in 100% of the cases due to lack of the aforementioned characteristics. .
These series are the Intel Pentium, Intel Celeron and AMD Athlon, most of them are Dual-Core processors with or without HT and SMT, at much lower frequencies than the Core and Ryzen series, achieving a lower IPC, even though they share architecture, where to top it off, their BMIs are normally slower than those of the higher series.
Performance with Intel processors
What we are going to see next are several benchmarks within the 3DMark suite, where each test has two different scores, the highest being the one corresponding to the total system score and the lowest the GPU Score.
The latter is the one that really interests us, since it will determine if with an entire RTX 2080 Ti X-Gaming OC there are notable differences in performance between processors of different ranges and architectures.
If there is a bottleneck, we would have to appreciate a more or less significant drop in performance.
The data shows that the greatest distance, in percentage terms, is 0.6%, which falls within the margin of error that a common measurement can offer.
Therefore, and taking into account that an i3-8350K is a CPU that has already been replaced by processors such as the i3-9100F that provide minimal extra performance, both with four cores, we can say that currently any Intel Core processor will not be a bottleneck. with no GPU on the market at the time of writing this article.
Let’s remember that this does not mean that it does not lose performance compared to, for example, the 8700K or even the i9-9900K, since this is evident and palpable, but let’s remember that we are simply talking about a bottleneck.
What happens if we change the range of GPUs and restrict the cores and frequency of a CPU?
In this case we will see a trend that may be curious but that makes all the sense in the world.
But before explaining it, let’s present the data, but not before saying that the CPU used was an i5-4670K, a CPU that today is low-end but that is perfectly valid to represent the problem that we intend to show.
The way to proceed is simple, the number of cores has been limited to 2 and the frequency will be scaled to see the impact on performance and especially on the CPU load with two different GPUs.
In this case we are facing the worst scenario, two cores and 1.6 GHz with a low-end GPU, where, as can be seen, the GPU limits the CPU load.
If we change the GPU the results change radically, now the GPU is relieved and the processor becomes a clear bottleneck in several cases, where despite having a much more powerful graphics card in the vast majority of cases the FPS fail to take off as expected.
By increasing the processor frequency, the bottleneck is reduced and the FPS begins to rise, even though there is a certain bottleneck depending on which game.
Bringing the processor back to its original frequency the scenario is less bleak, as GPU usage has increased, allowing for better framerate and where processor usage is now lower.
Activating the remaining cores at the same frequency we see how in most cases the FPS skyrockets and the use of the processor drops considerably, leaving the bottleneck aside.
These last examples show what we have seen above with the RTX 2080 Ti, if the processor were to bottleneck the results would be totally counterproductive, but in the case of Intel’s current Core range we will not find a bottleneck in any case.
But what about AMD? let’s see it.
Performance with AMD processors
We discussed this point a short time ago, where we obtained very interesting points of view with what AMD and its processors offer compared to Intel.
And it is that, although the games, resolutions and settings have nothing to do with it, the principle on which everything is based is the same for everyone, since the demand and workload can be extrapolated in many environments.
Under an RTX 2080 Ti, AMD processors pale in some games, where discarding the Athlon 200GE we will focus on Ryzen 3 and Ryzen 5.
Rainbow Six Siege is a very well optimized title and in this case the four cores without SMT are enough to present a battle without a bottleneck.
Project Cars 2 clearly chokes the Ryzen 3 22ooG, while the Ryzen 5 2400G is on the edge.
The former presents a clear bottleneck, so it seems that the sensible choice is the Ryzen 5 2600X.
Hitman is a case very similar to that of Project Cars 2, where the Athlon 200GE, Ryzen 3 2200G and Ryzen 5 2400G are making a clear bottleneck to the GPU, regardless of resolution and where again the minimum required would be the Ryzen 5 2600X .
Shadow of the Tomb Raider once again makes clear what we have been seeing in the previous two games, clear bottlenecks in AMD’s three lowest performing processors.
So the minimum option to choose is clearly the Ryzen 5 2600X.
conclusion
As we have seen, given the same graphics and different scenarios, the minimum recommendations for Intel and AMD are not the same when it comes to talking about bottlenecks.
And it is that, leaving the comparative performance aside, the truth is that with AMD processors we will find more models that can limit the performance of a GPU, especially if it has great power.
The more power in the GPU, the more CPU is required in low and medium resolutions, since the instructions and registers must be processed at a higher speed to keep the framerate as high as possible.
The Ryzen 3 2200G and Ryzen 5 2400G are good examples of this, where at least we will have to opt for a Ryzen 5 2600X within AMD so as not to have a bottleneck in any game or benchmark.
In contrast, at Intel an i3-8350K is enough to not hinder the performance of any graphics card, where even inferior CPUs like an i3-9100F can get the job done without suffering from these issues.
