Intel Core i5-10600 vs Ryzen 9 3900X
As always when a new CPU generation comes out you will have to ask yourself at some point the question “Which CPU is best for me.”
In this review we tested the new Comet Lake Intel i5 10600(K) vs the AMD Ryzen 9 3900X CPU.
One question remains when it comes to chosing a high quality CPU: Which CPU offers the best performance for Gaming or for Applications, like in a workstation?
Also, which CPU gives more performance for the money, i.e. how good is the price-performacne ratio?
All of these questions will be answered in this review.
You can find our final verdict at the end of the article down below.
Test Result: Intel Core i5-10600 vs Ryzen 9 3900X
Ranking First: AMD Ryzen 9 3900X
- Support for PCIe 4.0
- Incredible performance
- Compatible with X470 motherboards
- Beats Intel at same price
- Requires expensive X570 motherboards for PCIe 4.0 support
Best performing CPU
Comparing the Ryzen 9 3900X powerhouse CPU to the Intel i5 10600K might not be very fair.
That’s why we also compare the Ryzen 9 3900X to the i9 9900K to get a better sense for the performance.
How fast are AMD’s Ryzen 3000 and ninth-generation Intel Core in games when overclocked in combination with memory-OC in CPU limit released on games?
This question is discussed again and again in the community.
But already at the initial question about the right test approach the heated debate starts: May or must the CPUs be overclocked? And which RAM settings are the right ones?
Both processors can be further accelerated in different ways – provided the right motherboard is available.
The AMD proband allows overclocking the cores (to a small extent, usually not useful for games), the infinity fabric and the RAM.
The adjusting screws for more processor performance with the Intel processor are the core clock, the cache clock and the memory clock and memory timings.
And when it comes to RAM, AMD has an advantage on paper with DDR4-3200 to DDR4-2666, but Intel’s memory controller usually allows much higher RAM-OC clock rates.
In this article the CPUs therefore compete against each other in completely different states:
Once strictly according to the manufacturer’s specifications, once with overclocked memory suitable for everyday use and once with maximum achievable CPU and RAM clock rates.
Why fast processors?
For a gaming PC, only the best possible graphics card that fits the budget is important! Or is it? Not quite.
Although the graphics card naturally plays a large part in the gaming performance of a computer, the processor used also plays a decisive role.
The processor must supply the graphics card with the necessary data.
Depending on which games are to be played, it can even happen that the CPU mainly determines how many frames per second appear on the screen.
This is often the case, for example, in real-time strategy games or in graphically less complex e-sports titles.
Why faster RAM?
If the processor is slowing down, it could often perform better if it were supplied with more data faster.
Then you can give it a leg up with faster RAM.
The speed of the RAM is made up of the memory latency (How long does it take to bring data from the RAM to the processor?) and the memory bandwidth (How many data can be transported at once?).
The memory bandwidth is mainly influenced by the memory clock, the memory latency primarily by the so-called memory timings.
A detailed explanation of what memory has to do with the processor, how it can influence the performance of the CPU and what memory timings are can be found in the article about memory OC on Intel Core (X) processors.
But nobody plays in 720p!
The goal of this review is to get an overview of the performance of the Core i9-9900K and Ryzen 9 3900X in a selection of eight games.
So it is mainly a CPU test.
Since the render resolution has nothing to do with the processor, but only puts a load on the graphics card, the test is carried out in the lowest possible resolution (usually 720p).
This way it is determined for how many frames per second the processor can supply the graphics card.
This means: Even in higher resolutions, the maximum number of frames per second can be calculated.
The only disadvantage of testing in low resolutions is that the frametime values potentially suffer a bit.
If you want to know which processor you need for how many frames per second in a given game, you have to consult CPU tests.
By the way, many e-sports titles have options set by default for a reduced internal render resolution, so some players may unconsciously enjoy titles that are rendered internally at a much lower resolution than the output image.
An example from the course would be Rainbow Six Siege.
How should the values be understood?
To determine how many frames per second are calculated in real life with weaker graphics cards, a GPU test of the same game can be used.
The minimum average frames per second that processor A achieves with graphics card B from manufacturer C in 720p and the average frames per second that graphics card D from manufacturer C achieves in GPU testing give a good indication of what can be expected from a system with these components.
Be sure to check the graphics card series and the graphics card manufacturer.
How the driver works can have a large influence on where the CPU limit starts. You can see this in Assassin’s Creed Origins, for example.
RAM-OC can affect data integrity
Even more than for overclocking the processor or the graphics card, where an unstable configuration usually leads immediately to application or system crashes, caution is required with RAM-OC.
Even without an application or system crash, an unstable configuration can lead to corrupt files, rendering an operating system installation unusable and causing creeping data loss.
To avoid this as much as possible, time-consuming stability tests such as HCI Memtest, Karhu or Gsat are necessary.
For memory overclocking, at a certain point the memory temperatures are also important, so a top blow cooler on the processor, a dedicated fan above the RAM or even a RAM cooler can help.
As always, overclocking success cannot be guaranteed, the gains achieved depend heavily on the memory bars used and the memory chips installed on them, and faster memory is usually more expensive than less fast.
In the border area, the processor, RAM and the mainboard used are potential limitations.
A useful indicator of memory quality within a chip type (Samsung B-Die etc.) is the CAS process duration. The lower this is, the better the bin (i.e. presorting by the manufacturer) of the memory.
A well maintained overview of recommended memory kits can be found on Techtestreport.
The AMD processor was overclocked exclusively via “Precision Boost Overdrive” (PBO).
In games no significant clock gain (about 50-100 MHz per core) could be achieved with this.
Also the power limit was not reached to any extent, mostly the power consumption (read out by the sensors, according to which the mainboard controls) was below 100 watts.
Consumption measurements were deliberately omitted, because the expensive equipment required for this and the associated effort are far out of reach for this test.
The first tested memory configuration corresponds to the respective manufacturer’s specifications, the middle two correspond to popular memory profiles and the fourth memory configuration represents a proper overclocking within the scope of household cooling methods and suitability for everyday use.
The tested Ryzen 9 3900X did not allow a 1,900 MHz clock on the Infinity fabric.
For the second and third configuration, all primary and secondary timings and the memory clock were fixed.
There are timings like tREFI that are only available on one platform (in this case Intel) – these were not fixed and were automatically trained by the motherboard. In configuration 4, all memory timings were set and optimized manually and both processors were overclocked.
All given results are average values from three runs of the respective test.
The original measured values can be offered to interested parties for viewing.
The result labeling consists of processor, graphics card and memory timings.
For example, “3900X/R/3200CL14” stands for a result of the Ryzen 9 3900X together with a Radeon RX 5700XT in memory configuration 2 or “9900K(OC)/G/4133CL17” for a result of the overclocked 9900K together with a GeForce RTX 2080 Ti and memory configuration 4.
Orange and red and blue and green bars belong together, so they are comparable to each other.
Performance ratings per graphics card are used as a rough classification for the test results.
As already indicated, different ratios can be observed with the two graphics cards in the performance rating.
This is mainly due to the fact that the graphics drivers work differently and thus put different loads on the processors, because all tests were carried out within the CPU limit.
If you look at the performance ratings per graphics card, it is noticeable that the Ryzen is slightly more difficult with the Radeon graphics card than with the Nvidia model.
The AMD processor still manages to stay close to the 9900K with memory configurations 1, 2 and 3. The 3900X proves to be especially competitive in the frametime rates.
In memory configuration 4 with overclocked processors, the 3900X doesn’t stand a chance and the Intel processor clearly takes the lead.
As always, it is worth taking a look at the individual results.
It can be seen that, for example, Anno 1800, Far Cry 5 and Metro: Exodus are generally not particularly suited to Ryzen.
Here, the Intel test person is much more decisive than in the ratings. In World War Z, the Ryzen is the winner. A respectable success!
In Assassin’s Creed Origins, Rainbow Six Siege and Tom Clancy’s The Division 2, the results are balanced, or rather the Ryzen is slightly stronger than the ratings suggest.
In Shadow of the Tomb Raider, it depends on the graphics card used how well the Ryzen performs.
Together with a navigation graphics card, it is very competitive as long as the Intel processor is not overclocked.
If the Intel CPU is heavily overclocked, the Ryzen is clearly at a disadvantage, especially when it comes to frame times.
Together with a Turing-based Nvidia graphics card, the AMD processor only performs according to average with JEDEC memory.
As soon as faster RAM is used, the AMD proband is clearly at a disadvantage in terms of both average frames per second and frame times.
Performance ratings per processor
Since each processor was tested with two graphics cards, the measured values can be used to generate performance ratings per processor as additional information.
These can show trends in the influence the graphics card had on the performance of the processor, as the speed of the graphics card played a minor role in the game tests.
In each individual test, the processor limited the calculated frames per second.
Among the eight games tested, there was only one that the Radeon graphics card simply did not like:
Assassin’s Creed Origins.
If you leave it out of the ratings, there are on average only differences in measurement inaccuracy between the two cards.
Verdict Ryzen 9 3900X: Best performing CPU
The comparison of AMD Ryzen 3900X and Core i9-9900K up to the limit in terms of CPU and RAM-OC has produced a number of interesting results.
The first one already concerns the starting point completely without overclocking, because even in the absolute CPU limit in games provoked in this test, there are on average only significant differences between the 3900X and the 9900K when the 9900K is overclocked to the extreme.
Two OEM systems, which are configured the same except for the processor, should hardly differ from each other in normal gaming operation.
The same applies to ordinary homemade systems.
Another interesting aspect is the highlighted influence of the graphics card used: With the GeForce, both processors respond to the faster memory in about the same way, with the Radeon, the Intel processor takes up the faster RAM much better.
In every game the performance was improved by faster RAM.
Noticeable are the noticeable improvements of the frametime values through RAM-OC.
These improvements can only be achieved by faster processors, but not by faster graphics cards.
For those who want to have smooth frametimes, it is worth taking a closer look at the available memory kits.
For enthusiasts who are willing to delve deep into the subject matter and are usually within the CPU limit – be it through the choice of games or the demands on the calculated frames per second – an Intel processor with fast RAM can still make sense.
For people who don’t want to deal with OC or who are mostly in the GPU limit anyway due to high resolution, an AMD system with suitable memory is very useful.
Especially the often much better price-performance ratio is and remains a strong argument for Ryzen.
However, if only a few games are played or if there are only clear requirements for a small number of games in terms of frames per second and frame times, a closer look at individual results and dedicated CPU tests, at best with one graphics card each from AMD and Nvidia, is worthwhile in order to make the right purchase decision.
Games like Anno 1800 or Far Cry 5 on the CPU side and Assassin’s Creed Origins show this particularly clearly when it comes to the influence of the used graphic card on the CPU limit.
All in all, the Ryzen 9 3900X is our performance winner in this test against the i5 10600K, which is not surprising, as comparing these two CPUs is not really a fair fight.
Ranking Second: Intel i5 10600K
- Impressive single-threaded performance
- Consistently fast in games
- Now includes Hyper-Threading
- Chipset has upgraded networking
- AMD remains faster in multi-threaded workloads
Best price-performance ratio CPU
With this Intel opens its Comet Lake processors for the market, because from today on the first models are available on the market.
The core competencies of the new processors, especially on the top model Core i9-10900K, lie in the up to ten cores, which in some cases are supposed to reach a clock speed of 5.3 GHz. The Core i5-10600K rather covers the bread-and-butter business for Intel’s new series.
Intel, or rather Comet Lake-S, certainly didn’t have an easy start.
The ultra mobile Comet Lake processors (U and Y models) already started last year.
However, there were no real innovations here, as the six cores can hardly develop their power at 15W or 25W.
This is already somewhat different with Comet Lake-H. Intel presented its new gaming and workstation processors for mobile use at the beginning of April and with a thermal velocity boost of 5.3 GHz, it also gave a preview of what to expect on the desktop.
But for Comet Lake-S, the desktop version, it had to be a bit more than that. Up to ten cores based on the Skylake architecture, but still manufactured in 14 nm, a boost of up to 5.3 GHz, Hyper Threading for all models (also for the Core-i5 and Core-i3 processors) and a few more points should again encourage desktop users to consider an Intel processor.
Apparently Intel is slowly but surely feeling the pressure that AMD has built up over the past few years, not least with the Zen 2 processors.
Whether Intel will be able to do something against AMD’s Ryzen 9 3900X and Ryzen 9 3950X in the multi-threaded segment is certainly the most exciting question.
The core processors range from the i3, over the i5 and i7 to the i9 models.
Intel makes the distinction based on the existing cores.
All Core i9 processors have ten cores, the Core i7 processors have eight cores, with Core i5 there is the leap to six cores and the Core i3 processors have four cores.
All Comet Lake S processors can process twice as many threads as they have cores, so hyperthreading is always active.
The K and KF models have a thermal design power of 125W.
Compared to the previous Core i9 processors, the TDP has been increased from 95 to 125 W.
The Core i9-9900KS already came to 127 W, but is rather not comparable as a special model.
The F models and those without letter addition manage with 65 W.
Intel makes a further distinction in memory support. DDR4-2933 is only available on paper for the Core i7 and Core i9 models.
All others have to make do with DDR4-2666. Intel justifies the choice for DDR4-2933 and not DDR4-3200 with the necessary validation for the faster standard.
This also applies to the smaller models when sticking with DDR4-2666.
At least partly new for Comet Lake-S is the use of Turbo Boost Max Technology 3.0 (TBMT) in a further expansion stage.
Introduced was the TBMT 3.0 2016 with the Broadwell-E processors.
With the Skylake architecture, Intel enabled the technology to clock two cores much higher instead of just one.
With the Cascade-Lake-X processors Intel brought the Turbo Boost Max Technology to a new level.
Four instead of two cores are selected as “superior cores” and work with the highest boost clock rates.
For Comet Lake-S, the TBMT 3.0 addresses more than two cores.
The Thermal Velocity Boost (TVB) makes an additional boost possible, which is applied for two cores.
However, for Comet Lake-S there is only one step of +100 MHz if the CPU temperature is below 70 °C.
Above this, no additional boost is applied.
In Comet Lake-H, Intel made another step of +200 MHz at temperatures of 65 °C and less.
Between 65 and 85 °C the additional 100 MHz was applied and above 85 °C no thermal velocity boost is applied. Intel simply sets the Max Turbo 3.0 a bit higher for Comet Lake-S so that the +100 MHz of the Thermal Velocity Boost is sufficient to get to the same level.
So whether you can call up the TVB depends largely on the temperature. We’ll also deal with this in the course of the test.
We have two models with an open multiplier with the Core i5-10600K and Core i9-10900K.
Therefore, it’s naturally obvious that we’ll take a closer look at the overclocking.
It’s mainly about increasing the all-core clock rate significantly.
The Core i5-10600K offers a boost clock rate of 4.8 GHz in games using fewer cores. We speak of 5.3 GHz for the Core i9-10900K.
It already looks a bit different in the all-core clock with 4.5 and 4.9 GHz respectively – here we see the greatest potential for optimization.
For this test we used the H150iPro AiO water cooling.
The benchmark was run through five times in direct succession, the points were averaged and the temperature was recorded over all cores during the last run.
To get from an all-core clock from 4.5 to 4.8 GHz, we had to increase the voltage from 1.128 to 1.275 V.
The Core i5-10600K’s temperatures increase relatively strongly. In the next steps to 4.9, 5.0 and 5.1 GHz, a further increase of the voltage is necessary.
We didn’t want to go higher than 1.4 V, but couldn’t guarantee stable operation at 5.2 GHz even with this.
The temperatures of the individual cores already reach values of 100 °C at 5 GHz.
The Core i9-10900K works without an OC with an all-core clock of 4.9 GHz.
Accordingly, our first step was to bring the ten cores to 5.0 GHz.
For this we had to increase the voltage to a bit over 1.2 V and the temperatures already rose significantly to over 90 °C. 5.1 GHz are still just as portable and stable at 1.243 V, but there is already a throttling due to the temperatures.
For 5.2 GHz, the choke takes effect so strongly that the performance values already drop again.
The Core i9-10900K is a real challenge for any cooling without its limits with its 225 W.
As there are many among our readers with a custom water cooling, we naturally also took a look at how Intel’s new top model behaves with such a cooling.
We used a custom water cooling system from alphacool to ensure sufficient cooling.
Verdict Intel i5 10600K – Best price-performance ratio CPU
Intel basically shows us an infusion of what was brought to the market 2.5 years ago.
In the case of the ten-core processors, two more cores were added. On the models with eight cores and less, clock speeds have been increased due to improved manufacturing and raising the TDP.
At least all models now offer hyperthreading, which frees up otherwise idle resources, especially in multi-threaded applications.
The Comet Lake S processors show on the one hand the strengths of the current series, but also the weaknesses.
AMD and Intel are on a very similar level when it comes to the IPC.
But Intel can call up a significantly higher clock rate from its architecture in combination with a very mature production in 14 nm.
However, the arguments for better gaming performance have shrunk more and more with AMD’s Zen 2 processors.
But let’s start with the new platform:
The mainboards with Z490 chipset and LGA1200 offer hardly any improvements compared to the predecessor.
There are still only 16 PCI Express lanes available for connecting a graphics card. All other lanes must be connected via the chipset.
The DMI 3.0 connection between the processor and the chipset can become a limiting factor here.
PCI-Express 4.0 does not play a role for Comet Lake-S and will probably only be introduced to the desktop by Intel with Rocket Lake-S.
It remains to be seen whether this will then be the case on the current Z490 mainboards.
But an establishment of 2.5GbE could be credited to the new platform.
So the new platform doesn’t offer any revolutionary news and should therefore not be a reason for an update.
A DDR4-2933 is certainly not the decisive argument compared to a previous DDR4-2666.
Intel is blocking the way of the group of buyers for whom an upgrade of the platform would be the change argument.
Without an independent self-commitment of the new LGA1200 platform, Intel also doesn’t draw a picture of future security.
The situation is already different with the processors.
With the Core i9-10900K, Intel offers a new ten-core model, which is supposed to accelerate in both single- and multi-threaded applications.
Without question, the Core i9-10900K with its up to 5.3 GHz is always fast when not all cores are used.
However, whenever it is necessary to use as many cores as possible, the Core i9-10900K is of course superior to its eight-core predecessors, but still has to subordinate itself to the Ryzen 9 3900X with 12 cores and the Ryzen 9 3950X with 16 cores.
Even 16 “old” cores of the Core i9-7960X are superior to the new ten-core model in this respect.
Operated within the Intel specifications, the Core i9-10900K offers excellent gaming performance, but has to pull the brake after 56 seconds in multi-threaded applications.
This is good for temperatures and consumption, but not for a longer period of performance. Outside the specifications, the Core i9-10900K becomes a real hothead.
More than 200 W isn’t easy to dissipate in the long run.
The processor also runs constantly at the pain threshold under continuous load – at least as long as no custom water cooling is used.
The Core i9-10900K is priced at 589 Dollar, which makes it 160 Dollar more expensive than a Ryzen 9 3900X.
Only the Ryzen 9 3950X is a bit more expensive with 769 Dollar, but then also offers good gaming performance and leaves the 10900K clearly behind in multi-threaded applications.
The Core i9-10900K retains the crown in games, but can’t hold a candle to the Ryzen processors with many cores if these cores are also to be used.
One point can still speak for the Core i9-10900K: It still has potential for overclocking.
You won’t get the single-core clock much higher than 5.3 GHz, but there is still some room for improvement in the all-core turbo – assuming a powerful cooling.
In any case, the inclined buyer can still let off some steam here.
Intel Core i5-10600K:
The Core i5-10600K is already worth a closer look for many because of its lower purchase price of about 310 Dollar.
Unlike its predecessor, the Core i5-9600K, it now offers hyperthreading – which applies to all Comet Lake S processors in the Core series.
Six cores including Hyperthreading are more than sufficient for the everyday use of many users.
In games, it can’t quite keep up with the top models, but in the rarest cases, CPU performance is the limiting factor.
The higher the resolution and the more detailed the display, the more important the graphics card becomes.
Probably the most important counterpart of the Core i5-10600K is the Ryzen 7 3700X with eight cores.
It is even a few euros cheaper at 290 Dollar.
In games it maintains the usual Intel advantage, but the Ryzen processors pass by again in multi-threaded applications.
Two years ago the Core i5-10600K would have been a real recommendation.
But today, AMD’s Ryzen processors are far too good an alternative for a Core i5-10600K to be recommended without reservation.
If you feel more comfortable with an Intel processor than with an AMD model, you can certainly access this, but you shouldn’t expect performance miracles.
Verdict: Intel Core i5-10600 vs Ryzen 9 3900X
All in all, it was a really close battle between the Intel i5 10600(K) and the Ryzen 9 3900X from AMD.
Performance wise both CPUs are very strong.
All in all, if you want the best gaming performance and also application performance, i.e. for a workstation, without any compromises, you should get the Ryzen 9 3900X.
If you use your PC mostly as a workstation and not primarily as a gaming station you will benefit more from the Intel i5 10600(K), as it is better for gaming and less so for use in a workstation.
Also: The Intel i5 10600K offers the best price-performance ratio between the two tested CPUs.
Considering you only really get 20% more in performance with the Ryzen 9 3900X compared to the Intel i5 10600(K), but pay over 30% more for the Ryzen 9 3900X, makes for an easy decision for budget conscious users.
All in all we have to say, that we prefer the Ryzen 9 3900X. It just offers amazing performance for a great price.
There are also rumors that at the end of 2020 AMD will announce the new Ryzen 4000 series.
If this series convinces, it could be a deathblow for Intel, especially when AMD keeps its pricing strategy.
We will keep you updated!