• Home
  • Blog
  • Testing the Intel Core i9-14900K processor for the LGA1700 platform

Testing the Intel Core i9-14900K processor for the LGA1700 platform

13.11.2023 16:33

Intel introduced the 14th generation of desktop Core processors codenamed Raptor Lake Refresh in mid-October. At this point, six models appeared on the market, of which three belong to the K series and are intended for enthusiasts. These models have an unlocked multiplier for overclocking and a built-in video core. Three more models belong to the KF series and do not include a graphics core.

Particularly noteworthy is the Core i9-14900K, which is the flagship representative of the new generation and is the company's most powerful desktop processor.

The 14th generation of Intel Core processors uses Raptor Lake crystals, which are improved versions of Alder Lake. These processors use a hybrid architecture with heterogeneous cores, and the main difference is the increased amount of cache memory at different levels. For example, Raptor Lake processors have 2 MB of L2 cache per performance core, as opposed to 1.25 MB for Alder Lake. Also, a quad-core cluster with efficient cores has 4 MB of L2 cache, compared to 2 MB in the previous generation.

Raptor Lake processors operate at higher frequencies, which, combined with larger caches, provide performance gains. However, according to the review, nothing has changed with the 14th generation, and the processors use the same dies produced using Intel's 10nm 7 process.

The tested Intel Core i9-13900K processor, presented for the LGA1700 platform, is the flagship representative of Raptor Lake Refresh. The review notes that Raptor Lake Refresh does not offer anything new and is only an improved version of Alder Lake. Thus, it has not made any changes to the architecture or process technology, but continues to increase clock speeds.

The top model, also designated Core i9-13900K, is an updated version of the processor from the previous generation. This processor has 24 processing cores (8 productive and 16 efficient) with support for 32 threads. The maximum operating frequency is increased to 6 GHz, and the maximum power consumption remains at 253 W, which is comparable to the previous top-end processor.

The flagship 12th and 13th generation Intel Core processor represented a significant upgrade in performance over previous models. However, in the case of the Core i9-14900K, you should not expect the same significant changes. Although the 13th generation of Raptor Lake was similar in many ways to the 12th generation (Alder Lake), improvements were visible in the increased amount of cache memory at various levels, which provided performance gains without changes to the architecture.

Unfortunately, instead of a new generation of Meteor Lake processors, Intel introduced updated Raptor Lake in the 14th generation. However, the 14th generation's benefit is that the 13th generation's old MSRPs will remain the same, meaning that the performance gains will be available for the same money.

It was previously assumed that Raptor Lake would be the last monolithic generation of processors from Intel, and the next ones would be multi-chip Meteor Lake processors. However, despite the plans, Meteor Lake's arrival in the desktop market poses some challenges. Meteor Lake differs in that it uses separate dies with x86 cores, graphics unit and I/O logic on a single substrate, each using a different process technology.

It should also be noted that the 14th generation Core processors will be the company's last series of products under the current naming scheme. Subsequently, Intel processors will switch to a new scheme, without generation numbering, and will simply be called “Core” and “Core Ultra”. Thus, the 14th generation should ensure the competitiveness of the company's desktop processors until the second half of next year.

14th Generation Core Processor Family

Intel continues to use a strategy of gradually launching desktop processors, starting with a small number of models. This tactic allows you to offer more powerful and expensive processors first, and then, if necessary, release more affordable options. There are currently six Raptor Lake Refresh models available, divided into three levels based on power and price. Of this set of processors, half are overclockable (denoted by the letter K), and the other half have integrated graphics disabled (denoted by the letter F).

The 14th Gen Core processors are based on the same high-performance (Raptor Cove) and efficient (Gracemont) core architecture as their 13th Gen predecessors. The new generation does not represent an architectural shift, but an improved version of the previous generation, known as Raptor Lake. The main changes concern frequency characteristics, but not the structure of the cores or the size of the cache memory.

In the first wave of 14th generation models presented, the number of effective cores was increased in only one model, namely the Core i7-14700K. The remaining models retain a similar number of cores compared to similar processors of the previous generation.

So while the 14th generation brought some improvements, it remains an evolutionary continuation of the previous generation, with a focus on optimizing frequencies and performance.

The flagship Core i9-14900K under consideration has 8 productive (P-cores) and 16 efficient (E-cores), this is exactly the same number (8P+16E) as the top processor of the previous generation — Core i9-13900K. Compared to its predecessor, the new i9-14900K has an increased frequency — performance cores have a base frequency of 3.20 GHz and a maximum frequency of up to 6.00 GHz (like the Core i9-13900KS, which was sold for $100 more than the regular 13900K). The efficient cores are also slightly faster, with a base clock of 2.40 GHz and a maximum turbo clock of 4.40 GHz — 200 MHz higher than the i9-13900K's clock speed. The power consumption levels of the flagship processor remained unchanged — 125 W base and a turbo limit of 253 W.

But the Core i7-14700K has an 8P+12E configuration compared to the 8P+8E of the Core i7-13700K, which gives it a clear advantage over its predecessor. Well, the junior model of the Core i5 level from the new generation also did not change its configuration, receiving 6P + 8E — the same as the analogue from the previous 13th generation with a turbo frequency of 5.3 GHz and 3.5 GHz base as for Core i5-13600K. True, the frequency of the effective cores has been slightly increased — it has become 100 MHz more, but this is unlikely to provide a significant increase in speed.

A feature of the now 14th generation of Core processors is the use of the same LGA1700 processor socket, which was known from the 12th generation, this makes it easier to upgrade older systems. Most motherboards based on the Intel Z690 chipset fully support not only the 13th, but now also the 14th generation of Intel processors (you will probably have to update the firmware, perhaps), although motherboards on less expensive B- and H-versions of chipsets do not have this option provide. Some motherboard manufacturers have even decided to release updated top models to support the new series of Intel processors. It is clear that they are based on the same Z790 chipset, but include support for Wi-Fi 7 using separate expansion cards.

In terms of memory support, the 14th generation Intel Core series processors support DDR5 and DDR4, as do the 12th and 13th generations. Memory speeds according to the JEDEC standard are officially supported up to DDR5-5600 and DDR4-3200, respectively. Since DDR5 memory has become more productive over the past year, and prices have dropped significantly, we would recommend using a new type of memory. Possibly, in addition to options for updating existing systems without replacing motherboards and memory modules. Intel also stated that the memory controllers in new CPU models work better with support for higher DDR5 memory speeds, up to DDR5-8000, especially when using 12 GB and 24 GB modules.

When there are no noticeable hardware changes, manufacturers often resort to the good old trick of announcing new software features that are only available for next-generation solutions. And there are such in the case of 14th generation Intel Core processors, and two more at once. The first is intelligent auto-overclocking AI Assist, which does not replace, but complements a previously introduced function with similar capabilities — Intel Speed Optimizer. They are similar, but there are some differences between them — the new technology takes measurements on your system, including data from the built-in sensors, and runs some tests to determine the level of performance and overclocking capabilities.

Data obtained from the new AI Assist auto-overclocking feature in Intel Extreme Tuning Utility (XTU) is fed through artificial intelligence trained by Intel engineers on a variety of systems with different combinations of processors, motherboards, memory and cooling systems. The trained neural network provides optimal overclocking parameters, which can be accepted or manually reconfigured. It is important to note that any overclocking, even using Intel tools, may void the warranty, and if the processor breaks down, the manufacturer may refuse to provide a replacement.

Currently, the AI Assist feature is only available for Core i9-14900K(KS) processors and does not apply to all models of this generation. Intel is working to expand support to other 14th-generation models, but implementation requires extensive testing. The function was trained on hundreds of Core i9-14900K units, based on which the neural network offers recommendations for optimal overclocking. However, it should be noted that AI Assist only affects CPU characteristics, such as clock speeds, multipliers and power limits, without affecting voltage and memory parameters (frequency and timings). XTU also does not save overclocking settings in the BIOS, and to apply the settings you must run the program every time the system starts.

AI Assist initiates a stress test similar to Linpack, which lasts approximately half a minute. Ultimately, a list of recommended overclocking options for a specific system is displayed. It is important to note that the suggested settings may be unstable and cause freezes, as in our case. AI Assist may be useful for less experienced overclocking users, but the recommendations may still require additional adjustments.

For example, AI Assist determined that our Core i9-14900K sample on the test system can run at up to 6.1 GHz with the performance cores active and at 5.8 GHz with all cores in full use. For efficient cores, it is proposed to increase the multiplier by one (+100 MHz), which represents a slight overclock. However, even with these parameters the system was not 100% stable. Some tests and benchmarks were successful, but in some cases, when the temperature limit was reached, there was even a slight decrease in performance, and sometimes the system even froze.

So, while there is general interest in this feature and its potential appeal to those who don't want to delve deeply into overclocking, the technology currently needs some work. For wider use and improved stability, it would be desirable to extend support to other models such as the Core i7-14700K. It is also important to consider the possible consequences of a breach of warranty and the associated risks.

Another software innovation that appeared with the 14th generation processors is the Intel Application Optimizer. This feature is designed to specifically optimize system parameters for specific applications and can likely improve performance in games from a predefined list. The technology may work similarly to optimizing processor parameters for well-known applications in mobile devices to achieve better performance results.

Software optimization technology is only supported for 14-series processors and is not available for 12th and 13th generations, at least initially. It is an extension of Intel Dynamic Tuning technology, which manages the use of hardware resources by applications by distributing threads and managing resources in real time. Most likely, the technology determines the distribution of threads among productive cores, maintaining maximum frequencies for them. Intel claims performance gains of up to 13% in Rainbow Six Siege and 16% in Metro Exodus. Currently, only these two games are supported, but Intel promises to work with developers to expand the list of supported games.

It's important to note that the performance improvements will only be noticeable in specific applications. When applied to desktop processors, it is not entirely clear how effectively this will work, since it requires specific configuration and support from the motherboard. This feature has not been tested at this time and its effectiveness may vary depending on your specific usage conditions.

Flagship model Core i9-14900K

Today we are reviewing the senior representative of the Raptor Lake Refresh family — Core i9-14900K. Compared to its predecessor Core i9-13900K from the Raptor Lake family, the new processor retains the same number of cores for both performance and efficiency. Both of these core types use the same die, with a total of 24 cores and the ability to run 32 threads. This number is comparable to AMD's competitor, the Ryzen 9 7950X, which also has 16 physical cores. Even the best gaming model, the Ryzen 9 7950X3D, offers the same 16 physical cores, albeit at a slightly reduced operating frequency.

The flagship Core i9-14900K model reviewed today is the top-of-the-line Raptor Lake-S die configuration with a total of 36MB of L3 cache, an unlocked multiplier, and top clock speeds. The base frequency of the performance cores is 3.2 GHz, with the ability to overclock to 5.6 GHz using Turbo Boost 2.0 technology. Turbo Boost Max 3.0 technology allocates two cores capable of operating at frequencies up to 5.8 GHz, and when using Thermal Velocity Boost technology in single-threaded load mode, the frequency can reach 6.0 GHz, but provided that the temperature of the processor cores does not exceed 80 degrees .

The efficient cores have a base frequency of 2.4 GHz, which can be boosted up to 4.4 GHz by Turbo Boost 2.0 technology. These values are slightly higher than those of its predecessor. The base and maximum processor consumption indicators remain unchanged compared to the previous generation and are 125 W and 253 W, respectively. There are no time limits for operating at these consumption levels, although on some high-end motherboards these limits may be ignored and the processor may consume more power if the developer's settings allow it.

Overall, on paper, the Core i9-14900K promises to outperform the Core i9-13900K in both single- and multi-threaded tasks thanks to slightly higher clock rates. Theoretical changes include a 3% increase in peak turbo frequency and a 2% increase in effective core frequency compared to the Core i9-13900K. However, in practice, a small performance increase is expected, probably no more than a couple of percent.

Among all the 14th generation models, the Core i7-14700K is of particular interest, which received an additional cluster of four efficient cores compared to its predecessor. We will consider this model in more detail in a separate material. So far, the data for the Core i9-14900K shows little difference from the Core i9-13900K, even though it uses the same chip revision.

Intel's new flagship, the Core i9-14900K, competes with AMD processors, particularly the Ryzen 9 7950X and Ryzen 9 7950X3D. Both processors have the maximum number of processing cores and high frequencies, but there are significant differences in architecture. While Intel's top-end processors have fewer high-performance cores and their efficient cores are less efficient, AMD offers a larger number of homogeneous cores with high levels of performance.

With more efficient cores, the Core i9-14900K can be more efficient in all-core scenarios without hitting the power limit. On the other hand, the Ryzen 9 7950X, with its higher level of per-core performance, can excel in tasks with 9 to 16 active threads. Deciding which processor is best will depend on the specific workload requirements.

It's important to note that the Core i9-14900K supports faster DDR5 memory, which could be an advantage in some scenarios. However, the Ryzen 9 7950X offers a larger amount of cache, which can also have a positive impact on performance in certain scenarios.

Considering the recommended price, the Core i9-14900K looks attractive, especially if retail prices are not much different from the previous model. This could give Intel an edge over the competition, especially given the high price of the Ryzen 9 7950X3D with additional cache. However, more detailed testing is required to fully understand what benefits a 2-3% increase in clock speed can bring in practice.

Performance testing

Test systems and conditions

To test the processors, we used several models of high-performance motherboards for each platform and equipped them with a sufficient amount of RAM operating at the optimal frequency officially supported by all processors, or rather, as close as possible to it, depending on the available memory modules — DDR5-5200 . However, this time we tested games specifically with a different set of memory — more productive, choosing the XMP profile DDR5-6000 with CL36 latencies. All memory settings were taken from XMP profiles, and processor power consumption limits were taken in accordance with their specifications, and not the settings of motherboard manufacturers, which differ from what Intel itself approved.

Since in some tests we were interested in comparing not only the top processors of the latest generations of Intel and AMD, we also added a processor that was the flagship even earlier — Core i9-12900K, and it will be interesting to evaluate the progress of Intel's top solutions over three generations (albeit not in all tests). We took two top processors from AMD’s range, one of which features an additional die with cache memory: Ryzen 9 7950X and 7950X3D.

For most of the tests, we continue to use the previous generation AMD video card since we did not have the new Radeon and GeForce at the time of the start of the series of CPU tests, and the Radeon RX 6800 XT has quite sufficient performance and provides slightly higher rendering speeds precisely in low resolution conditions compared to with competitors produced by Nvidia. But in the new series of gaming tests, which we will briefly consider as part of this material, we have already used the much more powerful GeForce RTX 4090 — it provided the highest possible level of performance so as not to be limited by the capabilities of the graphics core.

Synthetic tests

Memory and caching performance

All processors in our study use DDR5 memory, and it is clearly visible that the efficiency of the DDR5 memory controller for all Intel processors is clearly higher than that of its main competitor AMD. All Intel processors of the last couple of generations in memory tests from the AIDA64 package, which measures the bandwidth and latency of all components of the memory subsystem, are clearly faster than the flagship Ryzen model, which in terms of bandwidth is much closer to DDR4 than to DDR5.

Core i9-14900K
Core i9-13900K
Core i9-12900K
Ryzen 9 7950X

If we compare the Core i9-14900K with the company’s previous models, it is quite logical that they are all almost equal, since the memory controller in Intel solutions of the last three families has not changed. Some differences in memory bandwidth and RAM access latencies are explained by measurement errors and different motherboards that we used in tests at different times. The Ryzen 9 7950X is noticeably inferior to all competing processors in terms of peak throughput when reading, writing and copying data, although it has a slight advantage in latency.

All Core i9s with DDR5-5200 show read speeds of 81-82 GB/s — this is a noticeable increase in memory bandwidth compared to DDR4 memory, and the effective memory bandwidth on these processors is quite high compared to just 65-66 GB/s with reading from a pair of AMD processors with the same memory, differing from each other in that one of them carries an additional chip with a third-level cache.

Because over the past few decades, the growth of computing power has significantly outpaced the increase in memory performance, and therefore processors have used increasingly complex caches to provide increased performance without being constrained by memory capabilities. Currently, Intel and AMD processors use a three-level caching scheme: each core gets a small L1 cache and its own larger L2 cache to get rid of high L3 latency. The last cache level is several megabytes in size and is used by several cores at once. In their case, both latency and throughput are important.

The L1 cache latency of Ryzen is slightly lower than that of the Core — 0.7-0.8 ns compared to 0.9 ns for the Core i9-14900K and 13900K, and in terms of L2 cache latency AMD processors have a significant advantage over Intel solutions. As for the third level cache, in the new Intel processor the latency in this test was measured slightly lower than that of the previous flagship, which can hardly be explained only by the increased frequency of its operation, but it still remained higher than that of L3 -cache of both processors of the Zen 4 family is almost 14 ns versus 9 ns for the Ryzen 9 7950X3D or 12 ns for its gaming counterpart. Memory latency for all processors with DDR5 memory is 76-80 ns, and this is more dependent on the different DDR5 memory parameters installed on different platforms and motherboards.

In addition to cache access latencies, their throughput is also important, especially for vectorized code. The latest generations of Intel and AMD processors have not seen significant changes to the main caches, their bandwidth should remain about the same as in previous generations — but we do see improvements in throughput that can be attributed to optimizations and increased clock speeds. Also, increasing the L2 cache size should reduce the bandwidth requirements of the shared L3 cache. Let's consider a test of the throughput of all levels of cache memory from AIDA64.

In our review of last year's Core i9-13900K, we noted that its cache memory at all levels has become faster than that of the Core i9-12900K — all cache levels took advantage of the increased operating frequency, and in the case of the Core i9-14900K it has grown slightly, mainly as times by 2%-3%, as it should, but there are also a couple of anomalies in the form of writing to the first level cache and copying the L3 cache — the received data is clearly incorrect. The 12900K caches are noticeably slower, and as for the competitor in the form of the Ryzen 9 7950X and 7950X3D, they have a noticeably less powerful L1 cache, but are ahead in L2 throughput and slightly faster in L3 cache — even taking into account its noticeably larger capacity X3D models.

Synthetic tests Sandra

Purely synthetic benchmarks from packages like Sandra and AIDA64 can also be interesting for assessing low-level performance in specialized tasks, although they claim some versatility.

The first group of tests shows relative performance in different tasks and a certain overall score (CPU Overall), calculated from all the results. According to it, the new flagship Core i9-14900K outperforms its predecessor by about 3%, which is what we could expect. However, it still lags behind both the Ryzen 9 7950X and the model with L3 cache, which took first place. Based on individual subtests, it is clear that in some cases the new Core still outperformed Ryzen, but mostly the advantage remained with AMD processors. By the way, there were some oddities here — in the scientific test, the new 14900K suddenly outperformed the 13900K by 19%, which in principle should not have happened.

These tests show computing performance when processing media data, and here the Core i9-14900K showed… almost identical results to the 13900K, the difference between them was not even 1%. And although both of them are noticeably ahead of the 12900K, they cannot compete with the Ryzen 9 7950X and X3D — the advantage of the latter is quite large. However, these are purely synthetic tests with a certain specialization, which are better suited specifically for AMD processors. Let's look at synthetic tests from another universal package.

Synthetic AIDA64 tests

These are also purely synthetic tests that show performance in tasks with a certain specialization. For example, CPU Queen uses integer operations when solving a classic chess problem, and AES uses the encryption speed using the cryptographic algorithm of the same name:

İlk birkaç testte, yeni Intel işlemcinin hızında önceki Core i9-13900K modeliyle karşılaştırıldığında hafif bir artış görüyoruz. Her iki Raptor Lake işlemcisi de tek Alder Lake'ten daha iyi performans gösterdi, ancak AMD'nin DDR5 bellek desteği, artırılmış saat hızları ve güç tüketimi seviyeleri ile en iyi çözümlerinin hala fark edilir derecede daha iyi olduğu ortaya çıktı. Ryzen 9 7950X burada açık ara lider ve 7950X3D de çok geride değil.

In the first couple of tests, we see a slight increase in speed for the new Intel processor when compared with the previous Core i9-13900K model. Both Raptor Lake processors outperformed the only Alder Lake, but AMD's best solutions with support for DDR5 memory, increased clock speeds and power consumption levels still turned out to be noticeably better. The Ryzen 9 7950X is the clear leader here, and the 7950X3D is not too far behind.

The third and largest batch of tests from AIDA64 includes floating point performance tests, including instructions for all SSE and AVX/AVX2 variants. The results of AMD processors in these tests have always been relatively high, and it is clear that the comparison with Ryzen will not be in favor of the new product. The 7950X and 7950X3D show better performance, the advantage over the Core i9-14900K is quite large. But as for the difference between 14900K and 13900K, the new flagship is almost on par with its predecessor, since it is not just one architecture, but physically the same crystal, just with slightly improved parameters, which resulted in a 1%-2% increase, which is very difficult catch taking into account measurement error.

Benchmark CPU-Z

Another synthetic test that we decided to include in this section is closest to rendering tests and is also very convenient for comparing single-threaded and multi-threaded processor performance. In the case of the Ryzen 9 7950X and 7950X3D, a variant of the AVX-512 test was used, which allowed for a slight increase in performance compared to Intel CPUs that do not support the required instruction set.

In terms of peak single-threaded performance, Intel processors have always been strong, this is confirmed by the results of the CPU-Z test — the Core i9-14900K and 13900K are noticeably faster than both Ryzen 9 in such conditions, both without using AVX and with these instructions. In gaming tests, such a difference is usually not in favor of AMD products. Compared to the previous Core i9-13900K, the new top-end Intel processor turned out to be 3% faster in a single thread — the result is not the most impressive, but quite expected — the processors are almost the same, and the difference in frequency is about the same.

Compared to AMD processors, we already see a different result — in a regular test without AVX instructions, the new Core i9-14900K is on par with its main competitor Ryzen 9 7950X, but the more productive version of the test gives the AMD processor a big advantage, and even the slightly slower X3D -the option remained ahead in this case. The predecessor of today's new product in the form of the Core i9-13900K gave it the same couple of percent, which is what we expected — so far everything is more or less consistent with the theory.

General tests
Let's move on to synthetic tests that are close to real problems. They measure the performance of systems in several types of application tasks, at the same time displaying an average value showing the overall performance, as in the PCMark 10 package. This approach has both advantages (ease of evaluation using a single value for an entire area of software) and disadvantages (they try to cover too much many do it imperfectly), but most often processors are still tested in it.

The next flagship Intel processor in this test shows quite expected, although not particularly outstanding results. True, it was sometimes possible to outperform the previous 13900K model by 10% or even more, but this is an oddity of the test, which is too old, unstable and is not able to show the advantage of modern processors, because there is simply no emphasis on multi-threaded computing performance in PCMark.

At the same time, the Core i9-14900K turned out to be slightly faster than its main competitor in the form of the Ryzen 9 7950X (the X3D model is expectedly slower in this test) — probably due to higher single-threaded performance, because the number of computing cores does not play a special role here, except for the gaming test, except that. Another general performance test we'll look at is the 3DMark CPU Profile, which is more related to gaming performance.

In this test, the new Core i9-14900K beats its predecessor by up to 3% in single-threaded mode, which makes sense in theory, and both are far ahead of the 12900K, which is explained by the doubling of the number of effective cores and the increased clock speed of all cores in Raptor Lake processors. due to which they maintain a high frequency, even under high load on all cores. As for the main competitors, the Core i9-14900K is clearly faster than the Ryzen 9 7950X when the load falls on a single core, and just a little faster when the load is multi-threaded.

A few more processor tests from 3DMark — most often these are physical calculations that can use multithreading, but with varying degrees of efficiency. The advantage of the new Core i9-14900K over the Ryzen 9 7950X remains decent, just like the previous Core i9-13900K model. Which, by the way, lagged behind the new flagship by a maximum of 1%-2%. So in the gaming load the advantage of the new product is unlikely to be too great, but it should compete with the Ryzen 9 7950X3D with additional cache.

The last test that we couldn’t include in other sections is the JetStream 2.0 browser benchmark, which measures the performance of JavaScript and WebAssembly code. For tests, we used an updated version of Microsoft Edge on the Chromium engine, and in this test the spread of results is not too large — it mainly affects single-threaded performance, but a large cache also affects the result. The new Core i9-14900K model turned out to be the fastest, and the previous Core i9-13900K lagged behind more than it should in theory. And the Ryzen 9 7950X and its X3D counterpart are not that little behind Intel’s new top model.

Rendering
Rendering tests are some of the most challenging for modern processors due to the multi-threaded nature of the ray tracing workload — while modern processors try to maintain the highest possible frequency, they can consume a lot of power and get very hot. Disadvantages of the cooling or power system (insufficiently high-quality motherboard or power supply) are best revealed in such tests. Very often during the process it is necessary to maintain a stable ambient temperature for the comparison to be fair, since in these tests top-end CPUs quickly reach the maximum possible temperature and can begin to drop frequencies.

Intel and AMD often use the Cinebench benchmark to compare the performance of their processors with competitor solutions — such rendering workloads are better executed with a larger number of cores and threads, which was different from early Ryzen compared to competing Intel CPUs, but now the processors of the second company in terms of the number of cores have caught up and in some places overtaken AMD solutions, and therefore it is already showing the advantage of its flagships in rendering.

In the multi-threaded Cinebench test, we got about what we expected — the Core i9-14900K is almost identical in speed in this test to the 13900K model, since both models rest within the limits of core temperature and power consumption and switch to approximately the same clock speed — despite the increased turbo frequency values are 14900K. Power consumption is very high, and the Core i9-14900K reaches its default maximum temperature limit of 100 degrees almost immediately with an active multi-threaded load, and this leads to a decrease in frequency — even when using a water cooling system. In this case, the frequency of productive cores is reset to 5.4-5.5 GHz.

Compared to the previous generation model, in single-threaded mode the advantage of the new product was an unexpected 6% (perhaps the better selection of crystals allows it to stably maintain a frequency of 6 GHz), but in multi-threaded mode it was only 1%, because the performance of both CPU models was limited by heat dissipation and a core temperature of 100 degrees. If we compare the Core i9-14900K with the Ryzen 9 7950X/X3D, the new Intel processor is faster in both single-threaded and multi-threaded versions of the test — the first is not surprising for a long time, Intel processors are always stronger in the case of such loads, and the second is explained rather by higher power consumption — and we will talk about this towards the end of the material.

Three test scenes in Blender show slightly different results from each other, but we can say for sure that these results have surprised us more than all the others so far — the advantage of the new product over the Core i9-13900K was stable… minus 1.5-2%! We restarted rendering multiple times and nothing changed, the new processor couldn't even match its twin from the previous generation. However, the difference between them is very small and also does not exceed the measurement error.

It is also surprising that the competitors in the form of the Ryzen 9 7950X and 7950X3D were both, although not very much, still faster than the new product. The greatest difference between the best Intel and AMD processors is observed in the case of the most complex scene — classroom, but there was an advantage in the other two as well. It turns out that in similar tasks, either Intel processors (Cinebench) or AMD solutions (Blender) win.

Another rendering test is Corona, and it measures the time it takes to render one frame. If there were no Core i9-12900K in the comparison, then we would have thought that something was wrong with the test, because all other representatives of both Intel and AMD showed an identical result of 36 seconds, confirming our long-standing conclusion that all top The CPUs are very close to each other.

We wanted to consider another benchmark with 3D rendering — V-Ray, but it did not work on the Core i9-14900K, displaying some strange error at startup, allegedly related to the inability to determine the processor frequency (at the same time, it launched at 14700K). We don't think that the comparative results of this test would be noticeably different from what we saw in previous tests in the section, so we didn't lose anything special.

Working with photos and videos
The next test section examines several programs for processing media data — photos and videos. These are very practical tasks, like exporting hundreds of high-resolution RAW images of about 3 GB in Adobe Lightroom Classic — tasks like this that most serious photographers do on a regular basis.

It is clearly seen that in this program Intel processors are clearly more profitable than AMD solutions. Even the Core i9-12900K copes with the task at the level of the Ryzen 9 7950X3D, so it is not surprising that the Core i9-14900K became the best here. However, its advantage over the previous flagship 13900K is only one second, which is barely noticeable in practice. Let's see what happens in the video editor of the same company.

We tested the rendering of a not too complex project in Full HD and 4K formats — many people face this task when preparing an edited video for streaming video services, so the situation is quite realistic. We are already accustomed to the clear performance superiority of Intel processors and this Adobe package. The Core i9-12900K is already inferior to the Ryzen 9 7950X/X3D, but not by that much. Well, today’s new Core i9-14900K outperformed a similar processor of the previous generation by a quite expected and understandable 3%-4%.

Handbrake is a package for converting video data to other formats, and we used the H.264 input video and transcoded it to H.265 format — also a fairly common task faced by users. The new flagship Core i9-14900K processor performed 3% better than its predecessor Core i9-13900K, and this again fits the theory. The competitor in the form of the Ryzen 9 7950X lags behind, although the difference between them is not too great, but it is there.

The second video transcoding test is SVT-AV1, which transcodes video data into the AV1 format, a relatively new open standard. And in this case, the results of the new Intel product turned out to be quite expected — it was ahead of the 13900K again by the same 3%. Competing with the top-end new product, the Ryzen 9 7950X lost a lot to the Core i9-14900K, and the point is in a specific project compiled with insufficient optimization specifically for processors of the Zen 4 family.

The last test in the section is Topaz Video Enhance AI — improving video quality using artificial intelligence capabilities. This compute-intensive task uses Artemis High Quality upscaling from Full HD to 4K. Everything looks completely different in it compared to the previous test — the new Core i9-14900K model showed a 4% increase compared to its predecessor, which is well within the theoretical performance difference between them.

The Ryzen 9 7950X and 7950X3D reveal their capabilities in this test, using the AVX-512 instruction set, which Intel processors do not support, and therefore both AMD processors outperform the Core i9-14900K by almost 50%. That is, we again see how a similar average level of performance can hide completely different strengths and weaknesses of processors from two different companies.

Cryptographic tests
Another important area of processor performance testing is cryptographic tasks. Modern CPUs can encrypt large amounts of information literally on the fly, and some even support special instructions for common algorithms such as AES. The first test is John The Ripper, a free software for recovering passwords using hashes that can take advantage of all the capabilities of modern processors.

Already from the diagram it is immediately clear that the advantage here is clearly with AMD processors, except for the Blowfish algorithm, in which Intel processors are faster. The new Core i9-14900K lags behind the Ryzen 9 7950X and 7950X3D in the first two subtests, losing to them a lot. As for comparing Core i9 of different generations, the advantage of the new product over the 13900K was 2%-5%, which is more or less close to theory. Naturally, the 12900K was noticeably further behind both Raptor Lakes.

VeraCrypt is on-the-fly encryption software that uses different data encryption algorithms and can use hardware-accelerated encryption on the CPU. In the tests, we used a 1 GB buffer and obtained an advantage of the new Core i9-14900K over the previous 13900K model by 2%-5% for different algorithms. As for comparing the new product with the competing Ryzen 9 7950X, the new top-end Intel processor not only turned out to be noticeably faster than the best AMD solution in the first test, but also won slightly in the second, although the 13900K was previously slightly behind. So in this particular case, the release of a new CPU model even made it possible to defeat a competing solution.

And the last cryptographic test is cpuminer-opt. This is a CPU mining program that also uses cryptographic calculations and is very well optimized for execution on modern CPUs. For tests, we chose the x25x algorithm, used in some cryptocurrencies, and for comparison we took the best result from several optimized miner options using instruction sets: SSE2, AVX2, AVX-512, as well as hardware support for AES and SHA.

The Core i9-14900K processor being reviewed today was able to outperform its previous generation model by a maximum of 1%, which may well be explained by measurement error. In addition, the 13900K was sometimes even faster, which confirms their close performance. In this test, the updated Intel flagship is clearly faster than the competing Ryzen 9 7950X processor in SSE2 and AVX modes, but in the most productive test they showed almost equal results.

Compression and decompression
Compressing and decompressing data in archives is known to most users, as are the most prominent representatives of advanced modern archivers, one of which has been WinRAR for many years. We used the built-in benchmark in the archiver, which measures the maximum data compression speed.

WinRAR results again show that the new top-end Core i9-14900K processor is no faster than the previous generation processor when the processors have the same number of cores, the same cache system and RAM parameters. The competing Ryzen 9 7950X processor is slightly faster than today's new product in this test, but the difference between them is very small. But the 7950X3D with cache showed its strength here, becoming the sole leader — a large cache is useful not only in games.

The 7-zip archiver is somewhat less popular, but it is interesting because it supports a more efficient and demanding compression method. If we compare Intel processors of different generations, the results of the new Core i9-14900K are only 1%-2% faster than the 13900K based on exactly the same crystal, but with frequencies that differ by a couple of percent. So everything here is in accordance with theory. But the Ryzen 9 7950X and X3D in 7-zip turned out to be very good, they are helped by a large and fast cache, and their advantage over the Core i9-14900K was clear both during compression and decompression.

Math tests
The section will be small; we included Y-Cruncher, a program for calculating the number Pi, among conventional mathematical problems. Of particular interest to us is this program’s support for the AVX-512 instruction set, as well as the optimization of this software specifically for Zen 4 in the latest version, which we used. Let's check what happened — the calculation time is indicated in seconds:

We tested calculating a billion digits of Pi in single-threaded and multi-threaded modes. The Core i9-14900K completed the first task 3% faster than the Core i9-13900K, and in multi-threading the difference between them was within 1% — due to the fact that both hit the power consumption limit and the temperature limit of 100 degrees.

Not surprisingly, the Ryzen 9 7950X and 7950X3D, which have support for AVX-512 instructions, showed very good results in this test — in multi-threaded mode they are close to 14900K, but they have a slight advantage, and this despite the fact that the thermal package of AMD processors is much higher below. It is also interesting that AMD wins in single thread — perhaps this test uses low-performance cores of Intel processors in this mode, although the difference would then be even greater.

The built-in benchmark in the MATLAB package can hardly be considered a representative test, since it is outdated and runs very quickly on modern CPUs, and its results fluctuate greatly from one run to another. So it’s unlikely that it would show a difference between the Core i9-14900K and 13900K by a couple of percent. Comparing the results of Intel and AMD is also difficult, the spread in different tests is too large, but it seems that the AMD processor is still a little faster in these specific tasks. But it would be better to consider the results of the scientific calculations section from our 2020 test methodology, which we are moving on to.

iXBT Application Benchmark 2020
As additional tests, we also ran a more familiar test set from the 2020 sample testing methodology, which has been known to you for several years. It uses real applications that only partially overlap with the tests whose results you saw earlier in this material.

We have previously noted very similar results from processors of different generations from AMD and Intel: 12900K with 5950X and 13900K with 7950X, which proved the high competition in the desktop processor market. These two companies have been competing almost on equal terms for many years, their solutions are very close to each other, but they have their own strengths and weaknesses. For example, Intel processors perform slightly better in scientific calculations and when creating video content, as well as in data archiving tasks. Well, AMD Ryzen is good at video conversion and digital photo processing, for example. But when it comes to 3D rendering, the best CPUs from the two companies are quite close to each other.

So when choosing a flagship processor, you need to look at specific tasks, since the comparative performance of the best Core i9 and Ryzen 9 models is highly dependent on the task. In some applications, the new fastest Intel processor is ahead, in others the AMD product is in the lead, and the average difference between them according to the iXBT Application Benchmark results does not exceed 5%. But on average, it was the new Core i9-14900K that still turned out to be faster than AMD’s most powerful competitor, but there are also cases like Topaz Video Enhance AI, when support for the AVX-512 instruction set gives Ryzen 9 an overwhelming advantage.

Comparing the new product with its predecessor does not make much sense, since the new flagship processor is essentially the same Core i9-13900K, but slightly overclocked, so we already know all the conclusions. In the entire set of tests, the new product is faster than the previous Core i9-13900K model by an average of only 1.6%, and this difference simply will not be felt in practice — the top model Raptor Lake Refresh differs from the similar version of Raptor Lake only by 200 MHz increased maximum frequency, which is also not always achieved due to the emphasis on temperature and energy limits.

We saw the biggest performance gains in digital photo processing and video conversion, applications that rely more on single-threaded performance. In applications of rendering, video content creation and scientific calculations, the increase was also, although slightly smaller, but in tasks in which there is more emphasis on memory capabilities, for obvious reasons, there is practically no difference — this concerns archiving and text recognition. So in general we can say that the 14900K and 13900K are almost the same product.

Gaming Performance

We will conduct a more detailed study of gaming performance in a separate article — review articles on processors are already too long. But we will show the geometric average results based on the results in several more or less modern games (Cyberpunk 2077, Forza Motorsport, Hitman 3, F1 2022) with support for ray tracing and built-in benchmarks right now, in three resolutions and under three conditions of graphic settings: medium Medium, ultra (maximum) Ultra and ultra settings with the additional inclusion of Ultra RT ray tracing. The latter option can also be useful, since tracing increases the computational load, including on the CPU.

Resolution 1920x1080

We will start with the simplest, but also the most common rendering resolution — Full HD. Most players still use it, despite the fact that there are more and more 4K monitors on sale. It is at this resolution that the dependence on the power of the central processor will be maximum, since on a system with the most powerful GeForce RTX 4090 video card, the rendering speed will not be exactly limited to its capabilities at this resolution. But there is no point in using an even lower resolution, since these conditions will be too artificial — no one plays at a lower resolution than 1920x1080.

On the one hand, with this resolution you can clearly see the difference between processors with different powers. The best samples for gaming (7800X3D and 14900K) show 20%-25% higher performance compared to the weakest options: 7500F and 13400F. And the difference is observed (although it decreases) not only at medium settings, but also at maximum and even with ray tracing enabled, which is quite difficult for a video card.

On the other hand, even if on the fastest CPUs in such conditions you get 350-355 FPS, and on the weakest ones there will be “only” 260-280 FPS, and you can only see this difference on screens with a corresponding refresh rate of hundreds of Hz. It is unlikely that there will be many players in the world with 360 Hz Full HD monitors, and even with an RTX 4090. So, although the advantage of fast and expensive CPUs is obvious, it is more relative. Let's move on to a higher resolution, where everything can be even simpler.

Resolution 2560×1440

There is more work for the GPU, and the difference between the performance of CPUs of different powers has clearly decreased. But with medium settings this is still about 20%-25%, and at maximum with tracing there is already only 6%-7%. In numbers it will be 145 FPS versus 135 FPS in the hardest mode — so what, is it worth overpaying for the top processor model? Let's see what happens in the highest resolution; there may be no difference left at all.

Resolution 3840x2160

Indeed, here there is almost no difference between all processors. If with average graphics settings you can still notice some difference in the frame rate between 220 and 250 FPS, then with maximum settings and using ray tracing you will get 90 FPS on the worst CPUs and… as much as 91 FPS on the best ones — by eye you won’t feel it at all, so you can play in such conditions either on a very weak Core i5-13400F or on the latest top-end Core i9-14900K.

In general, the Core i9-14900K in our new gaming test set was, on average, quite a bit slower (literally 1%-2%) than the competitor's best gaming processor, the Ryzen 7 7800X3D, and was quite a bit ahead of a similar Ryzen processor with twice as many cores — by 2%-4%. But the Ryzen 9 7950X without additional cache memory is already 7%-10% behind Intel’s top new product, and this is noticeable. So the Intel Core i9-14900K processor turned out to be more universal, it is good in both games and applications.

Moreover, the same thing is observed in games as in programs — there are gaming projects in which Intel products have a clear advantage, and there are cases when Ryzen is somewhat stronger. But we are talking about the conditions of the most powerful video card today, GeForce RTX 4090, in resolutions like 1920×1080 and 2560×1440, and at 4K resolution and high graphics settings there is almost no difference left which processor to use — the difference between the best and the worst CPU drops to 5% and below, and you definitely won’t notice this.

Power consumption and temperature

Assessing the power consumption of modern processors is quite difficult — the processor consumption indicators set by manufacturers are arbitrary and are not always maintained. Previously, the TDP or PL1 value indicated the peak power consumption of the CPU, but now, especially in the case of high-end models, numerous overclocking functions are implemented that allow you to go beyond these limits, first temporarily, and then without any temporary restrictions. This all depends on several factors: turbo demand limiter (PL2), variable frequency limits, temperature limits, etc. And CPU consumption can ultimately reach values significantly higher than the nominal ones.

And although Intel itself defines TDP values, such as PL1 and PL2, at a strictly defined level, motherboard manufacturers simply ignore these restrictions with default BIOS settings. Depending on the board and its settings, the maximum power consumption of the tested Core i9-14900K can be 350-370 W or even more! Despite the fact that almost the same Core i9-13900K is content with slightly lower consumption on the same boards. But the difference in performance between these CPUs will be very insignificant, which puts an end to the usefulness of such exceeding consumption limits. It is clear that all motherboard manufacturers want to be the best in performance, even if only by 1%-2%, but they do not pay any attention to consumption. But energy efficiency is a combination of performance and consumption, and in terms of this parameter, the Core i9-14900K is not presented in the best light, often losing even to its predecessor.

If you remove the consumption restrictions from the Core i9-14900K and increase the temperature limit, then it can exceed the energy consumption of 300 W, 350 W, or even 400 W, depending on the settings in the BIOS and the specific motherboard. Compared to the Core i9-13900K, the new processor may consume slightly more power, and it's especially high in applications like y-Cruncher and CineBench. With standard settings for the Core i9-14900K's 253W and 100C specifications, the processor does not exceed these values, but by default all motherboards have their own power limit settings, as they compete with other manufacturers for every extra percentage of performance. Most often, consumption is observed from 250 to 350 W, and this is a fairly wide range of consumption with all cores loaded — it is somewhat wider than that of the Core i9-13900K. You need to take this into account when choosing a cooling system, since although early top-end Intel platforms consume a lot of energy and generate a lot of heat, the new flagship stands out here too, consuming even more energy.

Since it is impossible to constantly chase motherboard manufacturers and their appetites, we decided to put the maximum consumption in the table given the consumption restrictions of Intel itself — that is, it is 253 W for both the 14900K and 13900K, so it is not surprising that these are the results for them meanings. However, the peak consumption of the Core i9-14900K when all limits were removed was around 350 W — specifically on our test board. Under the same conditions, the Core i9-13900K consumed up to 300 W. And 350 W is almost 40% higher than the TDP value for turbo mode of 253 W set for the new flagship. In order to achieve such consumption figures, you need to disable limits in the BIOS settings and run something resource-intensive, like rendering or a computing benchmark. With typical PC use, such high consumption is not achieved on a regular basis; in the same games (even demanding ones) it will definitely be less than 200 W, closer to 150 W and less.

By the way, when rendering in the Cinebench R23 test, the frequency of 5.9-6.0 GHz for P-cores was maintained only for a couple of active threads, 5.7 GHz is maintained up to 16 threads, and then the frequency is only 5.4-5 .5 GHz, falling even lower when overheated. Of course, the PL2 limit can be changed, as well as the default temperature limit of 100 degrees, but even then the processor will quickly reach its temperature limit and will gradually drop the frequency. And operating the processor at a high temperature limit on a constant basis is highly undesirable.

Let's look at the energy consumption of processors separately from the entire system under three scenarios: idle, gaming, and maximum consumption mode, in which Cinebench and Y-Cruncher were used to create the load. In game mode, the game Hitman 3 was launched with the Dartmoor test scene, which loads both the video card and the system's central processor. At the same time, official limits on energy consumption have been set; for top-end Intel processors of the latest generations this is 253 W. It’s interesting that when idle, the Core i9-14900K had lower consumption compared to all other CPUs, but this is not that important.

Under serious multi-threaded workloads, the Core i9-14900K easily consumes power up to the turbo level of 253 W — the same as its predecessor 13900K, which is not at all surprising. But the top-end Ryzen 9 7950X consumes even slightly below the limit set for it in the same conditions — 210 W. So, if we evaluate the energy efficiency of the Core i9-14900K and Ryzen 9 7950X processors, then the AMD product is clearly ahead in this indicator, since the new Intel flagship, although a little faster usually, consumes 20% more. By the way, in terms of energy efficiency, the leader here is rather the Ryzen 9 7950X3D with an additional cache — it has more gentle limits on consumption and temperature, and although it has lost in performance relative to its cache-free counterpart, the difference in consumption between them is much smaller. And if we compare the 14900K and 7950X3D, the processor we are considering today consumes almost twice as much power!

But in game mode, the consumption of all processors is noticeably lower — and even such a CPU-intensive game as Hitman 3 cannot force the new flagship to consume more than 172 W. The new Core i9-14900K model takes a little more energy than the 13900K from the previous generation, with about the same amount of more performance, so they are close in energy efficiency. If we compare it with competitors Ryzen 9 7950X/X3D, then the Core i9-14900K clearly stands out for its worse energy efficiency with similar performance in this case, especially if we take the gaming 7950X3D as a competitor, which consumes significantly less energy.

In rendering tasks like Cinebench, high-end processors like the Core i9-14900K are always hitting their temperature and power consumption limits. But AMD’s flagship has a couple of tens of watts left up to the second limit, and in theory it can work even faster if the temperature limit is raised. In games, the situation is simpler; powerful multi-core CPUs are not loaded to their full multi-threaded capabilities, so even in the most processor-intensive games, consumption remains within 120-180 W, and the core temperature is far from the upper limit — only 65-75 degrees maximum.

Let's talk just about the temperature regime. When idle, the temperatures of all Intel processors are close, but Ryzen heats up a little more than others, and we have already talked more than once about the reasons for the difficulty of removing heat from them. In games, all processors heat up moderately, but the Core i9-14900K is also among the hottest CPUs, even surpassing the Ryzen 9 7950X3D. The difference in games is not too big, but the maximum multi-threaded load mode makes all top processors look like ovens — they always hit the temperature limit, which is the default 100 degrees for Intel processors and 85/95 degrees for Ryzen with and without cache, respectively. And this despite the fact that the results were taken using a powerful three-section liquid cooling system with three powerful fans, although even more productive custom drops are not up to the task of keeping temperatures below the limit values.

And do not forget that all the values ​​obtained above are with the established limitations of Intel. Our experiments have shown that with all restrictions disabled, the Core i9-14900K can consume more and easily and quickly heat up to 115 degrees — the maximum value that can be set in the BIOS settings. So those who want to install a flagship processor in their system, not to mention the ability to overclock it, should definitely take care of purchasing the most powerful cooling system, which will allow them to gain several percent in performance. For such a powerful and hot processor as the Core i9-14900K, in principle, it is necessary to use a liquid cooling system, and preferably as efficient as possible.

Our 360 mm system, even supplemented with very powerful fans, does not have a very thick radiator, which, although it allows the Core i9-14900K to work normally in the default power consumption mode, when removing the consumption limit and increasing the temperature limit from 100 to 115 degrees, this value is reached very quickly in extreme multi-threaded loads, which causes throttling. So when overclocking, you need a custom 360 mm system (preferably 420 mm) with three high-performance fans and a thick radiator. With sufficient cooling, the 14900K will be able to operate for a long time at a frequency of up to 6.0 GHz in a number of low-threaded tasks, but in most tests that load several cores at once, the frequency of the productive cores will in any case be significantly lower. Overclocking the Core i9-14900K is generally quite simple due to the unlocked multiplier, but the main problem is the heating of the computing cores, because even at standard frequencies the temperature quickly rises to the maximum value.

Conclusions

The flagship model Core i9-14900K is positioned as Intel's most powerful solution for high-end desktop computers. Unfortunately, the new product offers only a relatively small increase in frequency compared to the company’s previous flagship, the Core i9-13900K, not to mention the 13900KS option for ultra-enthusiasts. However, its parameters provide some increase in single-threaded performance when the limits of power consumption and core temperature do not limit the capabilities of the new CPU. In conditions of a large number of active threads, the new flagship often runs into the same power consumption limit of 253 W (with default BIOS settings) as 13900K, and the increase in computing speed is even smaller. You can think of the new product as a re-released version of the 13900KS — and we don't think their results will be much different.

In terms of clock speed, the Core i9-14900K is slightly faster than the Core i9-13900K and very close to the Core i9-13900KS. While the 13900K runs at up to 5.8GHz with two cores active and 5.5GHz with more active cores, the 14900K reaches 6.0GHz with two cores and 5.7GHz with multi-threading—that is, 200MHz higher— and this is more than 6.0 and 5.6 GHz of the Core i9-13900KS. But this is true for conditions without a temperature limit, upon reaching which the frequency will decrease even more. There are also limits to power consumption, and the 14900K, like the 13900K, operates at a power cap of 253 W, unlike the 13900KS, which has the bar raised to 350 W, which in theory gives it an advantage under the heaviest workloads, like rendering. 

Let us immediately clarify that in different media the speed increase from 14900K to 13900K may differ; this depends on the motherboards and cooling systems used, which determine the maximum capabilities of the new top-end processor. We tested the new CPU on a non-flagship board, used a factory cooling system that was not the highest performance, and tried not to go beyond Intel's own specifications in terms of power consumption and temperature levels. Therefore, it is quite possible that the speed increase from 14900K to 13900K was not as impressive as it could have been. If you set the motherboard BIOS settings to maximum performance (and power consumption), then the performance of the new flagship will be quite a bit higher — this is literally a few percent, because almost everything has already been squeezed out of a full-fledged Raptor Lake crystal. Our results are close to the reference characteristics set by Intel itself.

In general, in terms of computing performance, the top 14th generation Intel processor is approximately equivalent to that of a similar 13th generation processor. In rare areas, an additional increase in the clock speed of the Core i9-14900K gives some increase in performance, but this is a few percent that you simply cannot feel. A noticeable gain by the Core i9-14900K is achieved in rare multi-threaded tasks, but even then not in all, since they have the same energy consumption limit, and the temperature limit is reached very quickly, which causes a decrease in frequency to approximately the same values. But where the difference is even more noticeable is in tasks with one or two active threads, in which the processor is not limited by the power consumption limit and does not heat up to 100 degrees. After all, it’s not enough to simply increase frequencies to higher frequencies compared to the Core i9-13900K; you also need to keep core temperatures and power consumption within certain limits. And since this is the same crystal, the performance level of the 14th and 13th generation flagships is very close in any case.

Average results across the entire set of tests showed that although the Core i9-14900K provides the highest possible performance among all desktop processors, the increase over the 13900K is small. Compared to the 13900K, it is about 2%-3%, and this is a very small gain for an entire new generation, although we expected this based on theoretical indicators. And if you already have a 13th generation flagship processor, then there is no reason to buy the 14900K, because you will not feel the difference in speed. And the best representatives of AMD are very close — the same Ryzen 9 7950X is slower than the new product by the same 3% -5% percent on average, but the Ryzen 9 7950X3D lags behind even a little more in applications. The Core i9-12900K, released a couple of generations ago, is a quarter behind the 14900K, and this may already be enough of a difference to warrant another upgrade to your system. Moreover, you don’t have to change anything else, neither the motherboard nor the memory.

In terms of gaming performance, the Core i9-14900K is also among the leaders, as is the top-end 13th generation model. Games don't max out the CPU's power and don't push it to its limits, so the 14900K's clock speed advantage can be achieved more often, making the 14900K one of the fastest gaming processors around, so the new 14-series flagship has a slight boost in gaming performance companies. Given the processor's limited rendering speed, Intel solutions often have a speed advantage over the Ryzen 7000, competing only with the “gaming” special series Ryzen 7000X3D. Compared to Ryzen processors without additional cache, the difference can reach 10%-15%, but with increasing resolution and graphics settings it is leveled out. But AMD Ryzen 7000X3D processors have improved performance in games, and the Ryzen 7 7800X3D even performed a little faster in our gaming tests, although the difference between them was 2%-3%, and then only in low resolutions. For gaming tasks, the Ryzen 7 7800X3D processor is generally a much more suitable alternative; it is cheaper and consumes much less energy, although it loses significantly in multi-threaded performance, so it is only suitable for games.

But to be honest, in games in general all the processors mentioned above will provide sufficient performance, and this applies even to relatively slow processors like the Core i5-14600K and analogues from the 13th generation and Ryzen processors — in real conditions there is no significant difference between them differences, especially at high resolutions. Price is much more important when choosing a CPU for gaming, and the Core i9-14900K is unlikely to be the best choice here. Moreover, to support high frame rates, you also need a top-end video card that costs $1000 or more, otherwise you won’t see the difference at all. When using models like the GeForce RTX 4060 or Radeon RX 7700 XT, CPU performance will rarely be the limiting factor, everything will come down to the capabilities of the GPU. And in 4K resolution, the processor becomes practically unimportant at all, and any modern CPU with 6-8 cores will cope with any game, since the main load falls on the GPU, and most CPUs work almost the same. Except for games like strategy, perhaps. So, specifically for games, a six- or eight-core mid-level processor is quite enough, and top models are not particularly needed.

On average, the Core i9-14900K performs slightly better than the AMD Ryzen 9 7950X, but they are still close. This is achieved in different ways — Ryzen has more high-performance cores (it has no others), and Intel uses 8 cores to achieve high single-threaded performance and a larger number of smaller, efficient cores for multi-threaded workloads. And despite all the differences in approaches, architecture and layout (chiplets and single crystal), the difference in performance between the top processors of the two manufacturers is not that great. If we compare the new Core i9-14900K model with its competitors in terms of performance, then it can be called universal, but the Ryzen ones are more specialized. So, although the Ryzen 7 7800X3D is the best processor for games and costs significantly less, it is only the best in games, and will be noticeably slower in applications. The Ryzen 9 7950X3D seems to be universal, but it is slower than Intel's new product in both games and applications. The Ryzen 9 7950X, in turn, is also not universal, since with slightly lower performance in applications it will noticeably lag behind in games due to the lack of 3DV cache. So in this comparison, Intel's new flagship isn't that bad.

Are there any fly in the ointment? One thing is for sure — unfortunately, like the company's previous flagship processors, the Core i9-14900K consumes a lot of power even at the official power limit of 253 W, which it easily reaches under multi-threaded loading. On paper, the new product has a TDP of only 125 W, but this number means nothing, since the processor is allowed to operate at a maximum power of 253 W for a long time, and with the default BIOS settings you can easily get 300-350 W. Actually, the new CPU is also almost equal to the 13900K in this parameter, and if Intel managed to improve efficiency, it was extremely insignificant, especially compared to AMD products. During rendering, the 14900K will consume up to 300-350 W or more, while the competing Ryzen 9 7950X is content with a hundred watts less — with almost the same performance. In games, everything is better, but only when compared with the same 7950X, and if you take the 7950X3D or even 7800X3D, the latter will provide the same gaming performance with half or three times less power consumption! We're not even talking about the difference in the cost of electricity, but about the heating of the entire gaming PC.

The processor temperature quickly reaches maximum values, but this happens more often in heavy multi-threaded tasks, and in real-world applications the 14900K can provide even better performance than the top-end Ryzen 7000, which reaches their maximum temperatures even faster. But at the same time, they do not consume as much energy as the new Intel processor, which is capable of very high power consumption, which makes it not a very energy-efficient product — in this parameter it loses to its competitor in extreme tasks, but may be closer in real ones. It is not surprising that the flagship processor reviewed today can only be cooled using a liquid cooling system, albeit a universal maintenance-free one — we would recommend at least a 280-360 mm cooler from a well-known brand. Even with these coolers, the 14900K reaches 100 °C very quickly, and this can lead to reduced frequency and performance. This is not a problem from a reliability point of view; modern processors simply maintain the target temperature and power consumption, slightly reducing the clock speed of the computing cores — and without a sharp drop in performance, but it will still happen.

Powerful liquid cooling systems will help increase the temperature limit to 115°C, but this will not lead to a significant improvement in performance, overclocking will still be limited, unless we talk about extreme cooling. With a very powerful liquid cooling system, you can try increasing the maximum multiplier for the performance cores — this is usually the most effective method for Intel processors. For the Core i9-14900K, you can reach about 6100-6200 MHz without much difficulty, if you don’t consider extreme methods. In our case, the cooling system was not enough for stable operation at higher values, and we limited ourselves to the reference characteristics of the processor. Speaking of overclocking, the artificial intelligence function for auto overclocking AI Assist seems to work and even gives reasonable parameters, but overclocking in the case of the 14900K will be small, and stability may suffer. Plus, these changes will not be saved in the BIOS, so you will have to apply them to the XTU every time you restart. If we talk about another new Intel technology in the form of Intel Application Optimization, then it is still at a very early stage of development. If Intel releases optimized profiles for new games quickly, then it could become useful, but for now APO only works with two older games that we are not very interested in.

As for prices and competition, the 13th generation Core processors were sold for amounts close to similar-power competitor processors from the Ryzen 7000 line, and the new top-end 14th generation processor at a recommended price does not differ from 13900K — $589. Considering they're the same die, Intel's approach seems pretty reasonable. The 13900K is already cheaper at retail, and you can pay a little extra for a small increase in speed. And if you don’t need a video card (for example, gaming applications), then you can purchase the KF version without an active video core, but at a reduced price of $564 — this is a good saving.

Who can benefit from the Core i9-14900K and who can recommend it? Users of very old Intel platforms 9-11 generations who need maximum performance. Or for those who already have the right motherboard and memory, and the means to upgrade from a 12700K to a new flagship model, for example, they'll instantly have the best-performing system for the next few years. But if it’s important to save money, then you can get by with the 12th and especially the 13th generation Core; there are not many reasons to upgrade to the 14th — the platform is the same and even the architecture is fundamentally unchanged.

Intel's advantages include the ability to use an existing motherboard based on the Z690 or Z790 chipset (you only need to update the firmware), as well as the theoretical possibility of using old DDR4 memory, which avoids the urgent need to replace it, although at current prices for DDR5 memory it is better to immediately buy new — with a new build, for sure. But this is also a minus — the LGA1700 platform is literally in its last years — there will be no more processors for it after the 14th generation, and if when choosing a 12th generation processor this was an advantage, since processors could be installed in the same boards and the 13th and 14th generations, now this is also a disadvantage. Since the competing Socket AM5 platform will also support the next AMD processors, although it is not yet clear how many generations of processors will still be supported. In general, if you need a new motherboard, then the AMD platform has better potential for the future, but if you already have a suitable system based on the Z690 or Z790, and even with memory, then it is quite possible to upgrade to 14900K or 14700K.

Raptor Lake Refresh as a whole is more like a re-release of old chips with different frequency characteristics, which does not quite deserve the title of the new 14th generation — except for the Core i7-14700K model, which received additional efficient cores. It’s clear why Intel needed this — after all, they committed to releasing new lines every year and fell into their own marketing trap. It is unlikely that the “new” generation will greatly improve the company’s position in the market, but a small increase in productivity for the same money is better than nothing. It was possible to reduce prices for the 13th generation to be better competitive against the AMD Ryzen 7000, many will say, but this will bring the company less money than releasing, even nominally, a new generation.

Still, the Core i9-14900K provides very high performance that corresponds to the top level. Compared to the higher-priced rival Ryzen 9 7950X3D, the new product looks quite attractive, delivering about the same performance in games and slightly better performance in multi-threaded applications. The hybrid configuration of heterogeneous cores gives the Intel processor high performance indicators; the processor is also good at multi-threaded workloads, but it is especially effective in scenarios with a small number of active threads. It can be recommended to those who want to squeeze all the juice out of it and overclock as much as possible — in this case, a greater increase is possible compared to the Core i9-13900K. For most users, perhaps, the Core i7-14700K model will be much more interesting, which has more efficient cores compared to last year’s Core i7-13700K processor.

Well, we are waiting for Meteor Lake processors to enter the desktop PC market, although they will provide a maximum of only six powerful cores — this is more than enough for mid-range systems. For top systems you will have to wait for Arrow Lake and at least until the end of next year. Competing Zen 5 plans to release next year, but the list of their improvements is not yet known. In any case, at this point in time, the Core i9-14900K, although it has given almost nothing compared to the Core i9-13900K, is definitely the most powerful and versatile processor for desktop systems today.