Intel Core i5-10600 vs Ryzen 7 3700X
The same question as always when a new CPU series comes out: Which CPU should I chose?
In this test we compare the new Intel Comet Lake i5 10600(K) against AMDs Ryzen 7 3700X CPU.
Especially gamers want to know where they get the most performance. But the price is also important. That’s why we highlight the CPU with the better performance, for all the hardcore gamers, but also the CPU with the better price-performance ratio.
In the end what do 3% more performance do when this option costs double or triple then the worse performing CPU.
Intel’s 10th generation core CPUs for PCs aka Comet Lake-S have arrived.
These CPUs also still use 14 nm. The Core i9-10900K clearly shows this in the test. It consumes a lot and only becomes really fast when limits are ignored.
In return, the Core i5-10600K proves to be a new edition of the Core i7-8700K.
Intel Core i5-10600 vs Ryzen 7 3700X
Ranking First: Intel Core i5 10600(K)
- Top-notch gaming frame rates for a midrange processor
- Good thermal performance
- 6 cores with HyperThreading
- Potential gains from casual overclocking eclipsed by Ryzen 7 3700X cost
Best performing CPU
With a significant delay, the age of Comet Lake-S officially starts at Intel today, the K-processors should be available immediately.
Led by the ten-core processor Intel Core i9-10900K as the new flagship of the series, the focus is mainly on the smaller models around Core i5 and Core i3, where a lot has changed on paper and one thing in particular has changed: Hyper-Threading is active.
The Core i7 is somewhat in the shade, as it is almost a copy of the previous Core i9. As a test sample, Intel has, not surprisingly, provided the Core i9-10900K and Core i5-10600K.
Core CPUs with Comet Lake-S at a glance
As usual with Intel, not all features are available everywhere.
For example, Thermal Velocity Boost (TVB) is only offered on the ten-core CPUs, Turbo 3.0 only on Core i7 and higher, faster memory also only above Core i7.
Some models are also split into five variants: normal (-), normal without integrated graphics (F), overclockable (K), overclockable without integrated graphics (KF) and power-saving (T).
Ten cores and new turbo modes
With the five Core i9-10900 there are ten-core CPUs for Intel’s mainstream sockets for the first time.
Intel has also adapted how the CPUs use the turbo.
The lowest common Turbo denominator is also for the 10th generation Core in the Turbo 2.0 desktop. This applies to all cores and differs depending on how many are loaded.
In the flagship, the maximum possible all-core turbo is specified at 4.8 GHz, for example. Turbo Boost Max 3.0 is based on the Turbo 2.0 and is only valid for a maximum of two cores.
It raises the clock rate by 100 MHz in both single and multi-core and can be found in the Core i7 and Core i9 series.
According to Intel, the Turbo 3.0 doesn’t need a driver anymore – this was still necessary in the X series.
With four new CPUs, the flagships of the entire family, there is Thermal Velocity Boost (TVB) above the Turbo 3.0, which was introduced in the notebook two years ago and enables an even higher clock rate within strict temperature limits.
In the desktop, it can provide an additional 100 MHz up to a temperature of 70 degrees Celsius – on one or all cores.
But as you can see in the test later, it’s not that easy, all factors must be right.
K models with 125 watts TDP
In addition to new turbo modes and two more cores at the top, Comet Lake-S is being helped to achieve greater performance via the TDP.
Instead of the previous 95 watts, Intel has increased it to 125 watts. In a multi-core load defined by Intel, but not further developed, the new CPUs now draw 125 watts at basic clock rate instead of the previous 95 watts.
The power consumption when using the turbo should also have increased as a result.
Further information about the TDP is provided in the article CPU power consumption: What “TDP” currently means for AMD and Intel.
All K-CPUs use 10-core
The new large 10-core die (stepping Q0) is used for all K-CPUs.
But that’s not all: The same die is also the basis for the Core i5-10400 and Core i5-10400F (Test) versions, which have a soldered heatspreader, but both models are also available with the native 6-core die (G1 stepping) and thus only with thermal paste instead of soldered heatspreader.
The customer can only determine this by the product code (Ordering Code, Spec Code).
The economical Core i5-10400T only uses the 6-core die, below the Core i5-10400 only the 6-core die is found.
The approach is not completely new and economically reasonable, already in the previous generation the 9400/9400F was the limit of the soldering process.
Thinner processor die and new heatspreader
With Comet Lake-S, Intel continues to use the traditional 14 nm production, but has optimized it once again.
The improved production now uses a process that significantly reduces the thickness of the dies.
The now communicated step from 0.8 to only 0.5 mm is surprisingly large.
On the other hand, however, this creates a dilemma that Intel simply solves: in order to remain compatible with all previous coolers, the height of the CPU including heatspreader must be identical.
Intel has therefore made the heatspreader thicker and used the intervention to increase the copper content to an unknown extent.
Copper is known to be a good heat conductor, so it could help the CPUs with the cooling – but Intel does not give exact numbers.
As last, Intel will solder the processor die with the heatspreader on the larger CPUs.
Below the Core i5-10400/10400F, only heat conductive paste is used again, as explained in the previous section.
A new socket and new mainboards
The new platform is built around the new LGA 1200 base.
The base looks very similar to the previous solutions so that all cooling solutions can continue to be used.
However, the possibilities for the future and an even higher power consumption made it necessary to change the socket after many years.
At the beginning only the increased power consumption is exhausted, there are no new features like PCI Express 4.0 – even if motherboard manufacturers advertise this under Intel’s displeasure.
With the Z490 chipset as the first solution, the controller hubs are also updated.
These are followed by the variants H470, B460, H410 and later the Q and W chips for the business sector. Since Techtestreport has already reported about these things many times, we would like to refer to this report at this point:
With the Core i5-10600K with PL2 of 182 watts and PL1 at 125 watts, the behavior cannot be observed and there is a simple reason for this:
Even highly demanding applications require a maximum of 115 watts from the CPU, only in Prime it is up to 150 watts – only there the clock rate drops by 200 to 300 MHz when reaching the threshold EWMA = PL1, because the consumption drops to 125 watts.
In this case, Intel has set limits with the increase in the TDP from 95 to 125 watts, which need not be levered out:
They allow the CPU to always use the full 56 seconds PL2 in everyday use, taking the specifications into account.
This is also shown by the clock and package power curve over three passes in Cinebench R20 with set limits.
What about the Non-K models?
Intel didn’t publish any documentation at the start, so nothing can be said about this topic.
But one thing already seems certain: There are changes here, too.
The Intel Core i5-10400F, which arrived from the shops a few hours ago, shows PL1 correctly at 65 watts and outputs Tau at 28 seconds, but PL2 is at 134 watts.
Consultation with motherboard manufacturers confirm the value as correct. It even corresponds to 2.06 times the TDP. Intel could not or did not want to make a statement again.
Robust benchmarks only without limits
Testing processors with multi-stage turbo mechanisms in a reproducible manner has already been complicated recently, but the fact that Intel’s specifications regarding PL1, PL2 and Tau, which have often been completely ignored for years, are still not ignored everywhere, makes the task even more complex.
This is because a CPU can show completely different faces. Techtestreport has therefore once again decided to test the new CPUs in two configurations:
- Without limits, i.e. always with maximum allowed turbo, which means up to 310 watts package power (Prime95) for the Core i9-10900K.
- With PL2 and PL1 with enough time lag between the benchmarks to exploit Tau
However, the message is clear: even the second test run by the editorial team represents a best case with limits.
If the entire course would be run through with only minimal pauses between the benchmarks, at least the Core i9-10900K would be even slower.
And just the order of the benchmarks would then make a difference.
In the end, in the extreme case two different processors would produce a single Core i9-10900K.
The one that is always allowed to get as much as it wants, the other that falls back to the value directly after 5 seconds above the 125 watt mark.
That something like this has a massive impact on performance is already clear in theory and has been clearly shown in practice in the test.
The editorial staff plans to deliver corresponding benchmarks later.
Yes, 5.3 GHz is available
The Core i9-10900K’s turbo behavior under adherence to the specifications wasn’t an issue for Intel to comment on in advance, but whether the CPU reaches up to 5.3 GHz was already an issue.
After there had been feedback from testers that 5.3 GHz wouldn’t be reached according to common tools, Intel obviously saw a Ryzen 3000 clock debacle coming up in their own house.
Asus even released another BIOS update, which was supposed to ensure that 5.3 GHz was actually reached by optimizing the Thermal Velocity Boost.
For the editors, however, the focus was on understanding PL1, PL2, Tau and EWMA. In this respect, there are only comments to be made on this point:
Yes, 5.3 GHz were achieved with the Core i9-10900K according to HWiNFO and were also used in single-core tests like Cinebench.
The Core i5-10600K with six cores and twelve threads at 4.5 to 4.8 GHz places itself quite far in the golden mean. On paper it corresponds to a Core i7-8700K with 125 watt steroids, games underpin the picture as well as later on the applications.
What six cores were already capable of 2.5 years ago, they are still capable of today.
They are still very well suited for gaming thanks to up to 4.5 GHz continuous load clock, so that the “old” eight-core Core i7-9700 without SMT support has to watch out for the twelve threads here and there.
And even up to the top there’s not too much missing in many titles, only in the frametimes the difference between six and ten cores becomes clearer every now and then.
And so in the end in games it is an expected, albeit narrow success for Comet Lake-S.
As advertised by Intel, Core is now primarily a gaming CPU. On the other front, things look very different, and with much larger gaps.
Benchmarks in applications
10 cores and 20 threads is the new record for Intel’s mid-range solution, but the market has long since seen completely different calibres.
And so the Core i9-10900K is already biting its teeth at the AMD Ryzen 9 3900X, which offers 12 cores and 24 threads with more performance at a much lower price.
What the Core i9-10900K can do thanks to Turbo Boost 2.0, Turbo Boost 3.0 and Thermal Velocity Boost, is Single Core: Here it is undisputedly the king thanks to up to 5.3 GHz.
The Core i5-10600K also has a hard stand against AMD’s Ryzen family, as it only delivers a slightly higher performance than a 2.5 year old Core i7-8700K – and this was already clearly inferior to AMD’s CPUs in applications.
Therefore, the model primarily benefits from the high single-core clock, although the lead of only four percent over the equally priced Ryzen 7 3700X is very small.
This is in contrast to the multi-core rating: AMD’s solution is 32 percent ahead and the Core i5-10600K has to struggle with the Ryzen 5, which costs significantly less than 200 Dollar.
The power consumption
The new Intel processors show their roots in idle:
The entire system, including the GeForce RTX 2080 Ti from Asus, can be operated under the 50 watt mark in idle even on highly bred Z490 mainboards and thus ranks on the predecessor level.
Under load, the differences become clear, and this already starts with single-core load.
Here, clear differences to previous CPUs are visible, the whole 9000 family was much more economical with the older boards and also brought more performance when comparing Core i9-9900KS to Core i5-10600K.
Multi-core load without limits and with PL2 budget
Under full load it depends on whether the CPUs are measured without Intel’s limits or with and in which application they are measured.
The full load measurement is carried out in Cinebench R20 in an unchoked state, this also applies to the AVX test.
With over 280 watts at the power outlet, however, Intel’s ten-core CPUs set a new negative record for Intel processors even within their limit of 250 watts and even clearly surpass Skylake-X, which as a high-end solution provides more cores and delivers more performance in applications.
No temperature explosion
Intel has made the processor die thinner and adapted the heatspreader with a higher copper content to it, which has a positive effect.
Of course, the fact that the ten-core die is larger helps the Core i9, giving it more space to dissipate heat to the heatspreader and later to the cooler.
Thus, the 10900K doesn’t get warmer than the 9900K despite two cores more and increased power consumption.
Because the Core i5-10600K is based on the same 10 core chip, its temperature is significantly lower.
Here, the comparison to the 2.5 year old Core i7-8700K with almost similar performance shows an advantage of almost ten degrees Celsius.
And that even though the 8700K consumes less. The fact that Intel didn’t solder at the time, but only used thermal paste, should play the most important role.
The Core i5-10600K is slower, but reliably delivers this performance even in Intel’s official limits. It pleases much better in the overall package.
It can permanently let off steam at full clock speed in the TDP limit, which has been increased to 125 watts, and goes cool to the point.
His problem is rather the past, because in the end such a CPU has been around for a long time.
The Core i5-10600K corresponds to 95 percent of the Core i7-8700K, which was introduced 2.5 years ago.
And the fact that Intel claims that there are now more threads for the same money is only partly true in this case.
In the spring of 2018, the Core i7-8700K barely cost more than 300 Dollar for months, and that’s exactly how much the trade wants for the Core i5-10600K at the start.
From this point of view, the Core i5 is also at a standstill. If you hit the 8700K two years ago, you can still consider yourself more than happy today.
Conclusion: Best performing CPU – Intel i5 10600(K)
Even before the launch, it was clear that Intel would not stand a chance against AMD’s processors in terms of performance in applications.
And if you add the price, it becomes even clearer.
12 cores from AMD cost only 430 Dollar in the form of the Ryzen 9 3900X and thus 150 Dollar less than the Intel Core i9-10900K. More performance and lower consumption – nothing speaks for Intel.
It’s the same picture in the Core i5 class. For just under 300 Dollar, the Core i5-10600K has to compete with the Ryzen 7 3700X, which is always superior in applications.
Even the Ryzen 5 3600X is that and it doesn’t cost 200 Dollar.
Intel is and remains the measure of all things only in the gaming sector.
But the progress that Comet Lake-S offers is minimal at the top.
How much of it comes from the faster memory alone, the editors could not yet test due to time constraints.
However, Core-i9-9900K(S) owners will not find any weighty reason to change in gaming anyway.
The smaller Core i-10000 could be the most interesting in the end, because there you can potentially get more for the same money.
The editors have already tested the Core i5-10400F (review).
Availability in retail
Starting today, the Comet Lake-S will start in retail, officially with the K-CPU, the rest will follow in a week.
We have been informed in advance that there will be stock on hand.
But traders did not want to comment on the number of units. Surcharges for goods in stock are likely.
The price comparison always provides up-to-date details.
Mainboard manufacturers fear in any case that the delivery situation is extremely tense, which is why they have not yet swung the big advertising club:
Because why advertise mainboards for which there are no CPUs?
The situation is not expected to improve until June. Intel had announced that they would offer all Comet Lake-S from the smallest Celeron to the Core i9 and thus 32 pieces in number from the end of May- but the goal will obviously not be reached.
Ranking Second: AMD Ryzen 7 3700X
- High performance
- Good price
- 65 watts TDP only
- Single-thread performance slightly behind competitive model
Best price-performance ratio – Ryzen 7 3700X
All new, you could say, when you first take a look at the new Ryzen 3000 processors.
Innovative and a pioneer in 7nm manufacturing, AMD was already in the graphics card segment with the Radeon VII and the recently introduced navigation GPUs.
The latter also offer PCI-Express-4.0, which of course the new Ryzen processors also bring along.
In terms of manufacturing and chip design, AMD is taking a different path than before with the Ryzen 3000 series without iGPU. CPU die and I/O die are now housed separately under the heatspreader.
This so-called chip design makes it easier to scale the CPU cores, because as we know, AMD has already announced a 16-core Ryzen 9 3950X.
As a result, it has two CCD clusters, each equipped with two CCX modules.
In our test sample, the Ryzen 7 3700X, there is only one CCD cluster under the heatspreader and combines 8 cores in the two CCX modules.
Since SMT is supported, the user has 16 threads available.
The new CCX modules, which contain the processing units, are manufactured using the 7 nm process, although the I/O die is in the older 12 nm structure width.
AMD itself decided to do this because there would not have been a significant advantage if the I/O die had also been produced in 7 nm.
In addition, they were already familiar with the 12-nm process, which was present at Zen+.
AMD has especially improved the cache of the new Ryzen.
It has not only grown significantly, but is also much faster than the 2000 generation.
AMD took a similar step with Zen compared to Zen+. If you want to own a new Ryzen now, you don’t necessarily have to rely on a new mainboard, because officially all X470 and B450 chipsets support the new Matisse chips after a bios update.
X370 and B350 boards are also partially supported.
However, here it is recommended to have a look at the support list of the respective motherboard manufacturers before the upgrade.
However, if you want to enjoy PCI-Express-4.0 as well, you should consider that it has to be a new motherboard.
Some manufacturers offer PCI-Express-4.0 support for some mainboards, but this only refers to the lanes directly connected to the CPU. Only the X570 chipset can provide additional PCIe 4.0 lanes to allow full bandwidth M.2 SSDs to be used.
AMD has also improved the memory support and recommends using DDR4-3733 CL17 for best operation.
As the graph on the right shows, latencies are lowest when using this memory speed.
This refers to the clock of the Infinity fabric, which runs synchronously with the RAM.
With a higher RAM clock, the Infinity fabric uses an internal divider, which causes it to run slower despite faster RAMs.
AMD Ryzen 7 3700X
The AMD Ryzen 7 3700X is a desktop processor with 8 cores and Hyperthreading (SMT), which is why it can process 16 threads simultaneously.
With its launch on 07.07.2019, it is the second fastest Ryzen 7 processor after the Ryzen 7 3800X, and with 65 watts TDP it is significantly more economical than the fastest Ryzen 7 model.
The Ryzen 7 3700X clocks at 3.6 GHz basic clock speed and can be turbocharged to 4.4 GHz.
When all 8 cores are loaded, the Ryzen 7 3700X reaches a maximum of 4.1 GHz. Thanks to the Precision Boost 2, the clock can be kept at a higher level, as we already know from Zen+.
The performance of the AMD Ryzen 7 3700X is consistently excellent in all applications.
Especially in multi-threaded applications, the native 8-core can show its strengths and despite the low TDP, the Intel Core i9-9900K is within reach.
The two CCX clusters, each consisting of 4 CPU cores, are located in a CCD (4 cores each with their own 16 MByte L3 cache).
This is exactly 74 mm² in size and is manufactured at TSMC using the new 7 nm process.
The I/O die is separated and comes with the older 12 nm structure width, which is still manufactured by Globalfoundries.
The AMD Ryzen 7 3700X can convince in games thanks to the improved IPC.
In games, the Ryzen 7 3700X reaches the performance of an Intel Core i7-8700K.
However, the Ryzen has to admit defeat at the moment, because in most games the Coffee Lake processor offers a bit more performance.
Future support of more cores can quickly turn the tide.
Processor sorbent benchmarks
The AMD Ryzen 7 3700X can show itself from its best side in the CPU benchmarks, especially in the multi-thread tests.
The new Ryzen rung wins the comparison with an Intel Core i9-9900K by a narrow margin.
However, the Intel processor can once again assert itself in the single core test. The archrival still has the nose in front here.
But the gap has clearly shrunk. This is shown by the rates in comparison to the AMD Ryzen 7 2700X.
This processor calculates about 20 percent slower than the Ryzen 7 3700X.
In the multi-core benchmarks, the AMD Threadripper 2920X can stand out in the test due to its 12 cores.
The AMD Ryzen 7 3700X is delivered with the Wraith Prism cooler.
We already know this from the Ryzen 7 2700X.
In the Cinebench R15 multi loop, the cooler manages to cool the 3700X competently.
The values fluctuate a bit in the course of time, but a creeping loss of performance isn’t noticeable.
On average, the Ryzen 7 3700X achieved a score of 2,075 points, although 2,092 points were achieved in the single run.
A good result compared to the 2700X, which scored just under 1,800 points on this stress test.
Gaming performance of the AMD Ryzen 7 3700X
Since the AMD Ryzen 7 3700X is great for gaming, the following results show.
In both the synthetic and gaming benchmarks, the new processor scored better frame rates.
Since the previous tests were done with an AMD Radeon RX Vega 64, we decided to test the Ryzen 7 3700X with this graphics card as well, to rule out any advantages due to a faster GPU.
However, we also used the AMD Radeon RX 5700 XT to demonstrate the full performance of the system with a modern graphics card.
Compared to Intel, the Ryzen 7 3700X is still missing something. In low resolutions, where the graphics card is not the limiting factor, our AMD system has to admit defeat.
On the level of an Intel Core i5-9600K, our test sample can be found in the gaming benchmarks.
In comparison to the AMD Ryzen 7 2700X, the Zen 2 processor can clearly distinguish itself. Here it shows that the changes in the architecture bear fruit.
The Ryzen 7 3700X’s higher IPC provides an overall performance increase of 11 percent in gaming.
AMD Ryzen 7 3700X together with the Radeon RX 5700 XT
In the following we have shown the difference between our current test system based on the AMD Ryzen 7 2700X and the new AMD Ryzen 7 3700X.
An AMD Radeon RX Vega 64 and the newer AMD Radeon RX 5700 XT were used as graphics cards to see if the higher bandwidth had a positive effect on the performance of the graphics card in the benchmarks.
If we take a closer look at the results of both systems based on the AMD Radeon RX Vega 64 at the beginning, we can expect a performance gain of about 9 percent just by changing the CPU.
When comparing both processors and the AMD Radeon RX 5700 XT, we see a performance gain of 11 percent.
We don’t see a direct advantage for the higher bandwidth due to PCI-Express-4.0 in gaming yet.
Verdict: Best price-performance ratio – Ryzen 7 3700X
With the introduction of the new Ryzen 3000 family, AMD is doing a lot of things right and is putting Intel under a lot of pressure.
The buyers are happy about this, because it is well known that competition stimulates business.
At 349 Dollar, the AMD Ryzen 7 3700X costs a bit less than an Intel Core i7-9700K at the time of testing. But the new Zen 2 rung offers considerably more multi-thread performance.
Furthermore, the new processors (without iGPU) offer the more modern platform with PCI Express 4.0 support.
A purchase argument, which cannot be dismissed, because the Navi-GPUs presented at the same time also support this and thus currently give an almost perfect price-performance picture, if it weren’t for the expensive X570 mainboards.
Of course the new Ryzen processors also run on the older boards, but PCI-Express-4.0 is not available everywhere.
However, if you want to rely on a future-proof platform, there is no way around AMD and an X570 board at the moment.
When it comes to multi-thread performance, it definitely offers more than Intel, because with 8 cores, the end is not yet reached this year.
The Ryzen 9 3900X is already available with 12 cores and in September there will even be a 16-core on the AM4 socket. With this, AMD addresses players, streamers and content creators with its current products.
In direct comparison to Intel, AMD must still admit defeat in the single thread performance, although the gap has shrunk considerably.
AMD could also impress in games with the Ryzen 7 3700X.
We were convinced by the Ryzen 7 3700X’s good energy balance.
An 8-core with 16 threads and a power consumption with a maximum of 88 watts is already remarkable.
Verdict: Intel Core i5-10600 vs Ryzen 7 3700X
In the end, it was a really close battle between the Intel i5 10600(K) and the Ryzne 7 3700X from AMD.
Performance wise both CPUs are very strong.
All in all, if you want the best gaming performance without any compromises, you should get the Intel i5 10600(K).
If you use your PC mostly as a workstation and not primarily as a gaming station you will benefit more from the Ryzen 7 3700X, as it is better for applications.
Also: The Ryzen 7 3700X offers the best price-performance ratio between the two tested CPUs.
Considering you only really get 10% more in performance with the i5 10600K compared to the Ryzen 7 3700X, but pay over 30% less for the Ryzen 7 3700X, makes for an easy decision for budget conscious users.
All in all we have to say, that we prefer the Ryzen 7 3700X. 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!