Content
With the announcement of the 13th generation Core chip series, Intel significantly strengthened its position in the desktop segment, confirming the importance of this CPU category. The first processors of the Raptor Lake family were presented back in September last year, while at the beginning of the current Intel expanded the range of new chips.
After the platform update, the top models are usually immediately under the spotlight, so it is not surprising that the main attention was focused on the Core i9-13900K, and later on the flagship of the line Core i9-13900KS with an impressive 6 GHz. At the same time, some interesting chips that deserve special attention remain in the shadows for no reason. Today we are reviewing Core i9-13900F – a representative of the powerful Core i9 line with a declared TDP of 65 W. How do you like this combination? Let’s consider this in more detail.
You can buy a processor from our partners, stores ITBox and Brain.
Intel Core of the 13th generation
Starting with the 12th generation Core processors, Intel uses a hybrid architecture of computing units for its powerful models. In this case, a combination of performance (P-Cores) and efficiency cores (E-Cores) is used. Such a configuration allows, if necessary, to direct all available computing capabilities to the fastest solution of a resource-intensive task, or to distribute available resources according to current priorities, balancing processor power consumption.
The Intel 7 process, which actually corresponds to 10 nm, is used to manufacture the 13th-generation Core chips. However, despite the stated norms, Intel technologists manage to improve the characteristics of crystals every time. For the Raptor Lake plates, the optimized Intel 7 process with third-generation SuperFin transistors with reduced channel resistance is used. This made it possible to reduce the supply voltage by 50 mV at the same frequencies as Alder Lake, or at the identical voltage to increase the operating frequency by 200 MHz. Collectively, these technological improvements allowed for a significant increase in the frequency potential of the CPU. In the case of the top models of Raptor Lake, the difference with predecessors can be up to 600 MHz for productive cores.
In addition, P-Cores with the Raptor Cove architecture received an increased second-level cache. If the 12th generation Cores had L2 with a capacity of 1.25 MB per core, Raptor Lake already received 2 MB for each core. That is, the capacity has increased by 60%. In addition, L2 uses a new L2P prefetching algorithm that attempts to dynamically adjust the prefetching behavior in real time based on the current workload.
However, frequency acceleration is only one of the consequences of technological optimization. Staying within the Intel 7 technical process, engineers managed to increase the transistor budget, using it to increase the number of efficiency cores, as well as the capacity of the cache buffer, including for E-Cores. So, the Alder Lake chips of the Core i9 series had 8 efficiency cores, while the Raptor Lake processors of the corresponding line already received 16 E-cores. And here the developers did not limit themselves to the quantitative increase of the blocks themselves. Cores with the Gracemont architecture received twice the capacity of the L2 cache – each 4-core cluster has 4 MB of L2, instead of 2 MB in Alder Lake. With the exception of an increase in the L2 cache memory, as well as a change in the prefetching algorithm, the architecture of efficiency cores has not fundamentally changed. However, the operating frequencies of E-Cores are also significantly increased here – up to +400 MHz.
After the increase in the number of efficiency cores, the total capacity also increased the cache memory of the third level. In the case of chips of the Core i9 series, this is already 36 MB L3. In addition, a new dynamic mechanism with an inclusive/non-exclusive INI architecture is used for the latter, which, according to certain algorithms, determines which information from L2 should or should not be duplicated in the lower-level cache.
To interconnect internal units, Raptor Lake chips use a Ring Bus, the frequency of which has also increased. In addition, Intel has officially raised the bar for working with high-speed memory modules. Instead of DDR5-5200, characteristic of Alder Lake, older models are guaranteed to work with modules in DDR5-5600 mode. And for chips with an unlocked multiplier, these are far from the limit values. Current world memory overclocking records are set with Raptor Lake chips. Extreme enthusiasts managed to conquer ~ DDR5-11200 modes.
Taking into account the fact that the lithography has not actually changed, while the number of energy-efficient computing cores has increased, as well as the capacity of the cache memory, the size of the monolithic crystal has also slightly increased in the 13th-generation Core chips. The manufacturer usually does not specify the exact dimensions, but experimentally it was established that the area of the Raptor Lake silicon wafer is about 257 mm². While for the older Alder Lake chips, this value was at the level of 215 mm².
Even despite the optimization of the manufacturing process, taking into account the increase in functional units and a significant increase in the operating frequency, the developers had to slightly expand the energy consumption limits. For example, the PL1/PL2 limits for the Core i9-13900K/KF chips are set to 125/253W instead of the Core i9-12900K/KF’s 125/241W.
Despite the fact that, at first glance, Raptor Lake does not offer any deep architectural changes compared to the previous generation of Intel desktop chips, the developers managed to significantly increase the overall performance of the CPU. According to the manufacturer, in single-threaded tasks, the increase can be up to 15%, while under multi-threaded load conditions, the difference can be up to 41%. Given that these changes are happening within the same LGA1700 platform, this speedup is even more impressive. Well, the presence of a strong opponent really becomes a catalyst for development for all players in a certain market segment.
As we can see on the slide, the acceleration in the single-stream mode primarily occurs as a result of a noticeable increase in the operating frequencies of the productive cores. Optimizations at the level of caches and memory subsystems have a certain impact. But the factors of a very significant increase in productivity in multi-threaded mode have several components. Additional computing units have the maximum contribution here. In the case of Core i9, processors get as many as 8 more energy-efficient cores. That is the number of cores/threads within the top line changes from 16/24 (8/16+8) to 24/32 (8/16+16). In addition, the influence of increased frequencies remains significant, because they have increased in both performance and efficiency cores.
The increase in L2/L3 cache memory with a change in interaction algorithms also adds up to more than 10% of the overall score. The impact of improving the memory subsystem is less, especially considering that the 12th-generation chips are also well-optimized for working with high-speed DDR5 kits. By the way, Raptor Lake also retained support for the DDR4 memory standard. In certain cases, this can turn out to be quite a useful option, even though the cost of DDR5 has been decreasing rapidly recently.
The 13th generation Core chips also offer 16 PCI-E 5.0 lanes for connecting a discrete graphics card and four PCI-E 4.0 lanes for direct storage connection.
A hybrid configuration with productive and energy-efficient cores requires more attention to the distribution of the load on all available computing units. It is for such a task that the Thread Director hardware and software mechanism is used, which allows you to monitor the current states of computing cores, balancing active tasks.
In Raptor Lake chips, the “load manager” technology has been improved. The capabilities are most effectively used in conjunction with the scheduler of the Windows 11 22H2 operating system.
The line of Intel Core i9 (Raptor Lake) chips
The top line of Raptor Lake chips – Core i9 – includes several models. The flagship of the series is Core i9-13900KS (Special Edition). The special feature of this “special edition” is the maximum operating frequencies with a peak of 6 GHz, which is generally a record value for the standard operating mode of desktop CPUs.
This model uses pre-selected crystals that can reach the specified frequencies at acceptable levels of supply voltage and power consumption. The developers had to raise these parameters. Maximum performance at any price is what the Core i9-13900KS is all about. The base TDP value is 150 W, while the maximum consumption (Maximum Turbo Power) is declared at 253 W. The flagship of the 13th generation Core desktop chips received a respectable price tag of $699.
Core i9-13900K/13900KF are actually top modifications, with the exception of the “special edition”. Chips with high clock frequencies and an unlocked multiplier, which also allow for additional overclocking experiments. In fact, the manufacturer has already used almost the entire frequency potential of the crystal in the basic specification, but enthusiasts who are ready to fight for every additional MHz will like this opportunity. The additional letter “F” in the model name is a sign that the chip is offered with the integrated video core deactivated. The TDP in both cases is 125 W, and the peak is indicated at 253 W.
Core i9-13900/13900F are models for mass use. Slightly lower operating frequencies compared to those for older versions made it possible to reduce the power consumption of the chips. The basic TDP value of these processors is 65 W, and the maximum declared value is 219 W. The standard versions don’t offer an unlockable multiplier, so overclocking enthusiasts will probably prefer the K-versions. But for everyone who does not overclock, but needs a powerful CPU, basic chips can be very interesting. As in the previous case, the presence of the index “F” in the model name indicates that the integrated video core is disabled.
Core i9-13900T is the only energy-efficient version of the Core i9 series. The model has an additional optimized frequency formula to reduce energy consumption. The parameters of the chip were changed so that the TDP corresponded to 35 W, and the maximum value was set to 106 W. Models of T-modifications are rarely sold at retail, they are usually used by OEMs for systems of compact form factors and monoblocks.
Regardless of the model, all chips of the updated Intel Core i9 line have 8 productive cores (P-Cores) with support for Hyper-Threading technology and 16 efficiency cores (E-Cores). That is, the general functional formula is 8/16+16. It is correct to say that we are dealing with 24-core processors that can simultaneously process up to 32 data streams.
All Core i9 models also use a total of 32 MB of L2 cache memory. Each performance core has 2MB of L2 (8x2MB = 16MB) and each 4-core E-Cores cluster has 4MB of L2 (4x4MB = 16MB). Each CPU also has 36 MB of total third-level (L3) cache memory.
Intel Core i9-13900F processor
For our research, we chose the Core i9-13900F model, which is somewhat overshadowed by the older chips in the series. However, the basic processor of the line has its own characteristics and deserves special attention.
The Core i9-13900F comes in a fairly large box. We are already used to the fact that K-modifications of Intel processors are not equipped with a standard cooling system, so they are offered in compact boxes. But one of the features of the Core i9-13900/13900F is the presence of a basic cooler. Moreover, unlike mass models of younger lines, the corresponding Core i9s are equipped with Intel Laminar RH1 coolers.
As we have already mentioned, the Core i9-13900F is a 24-core, 32-thread processor with a frequency formula of 2.0/5.6 GHz for eight performance cores and 1.5/4.2 GHz for sixteen efficiency computers. Relatively small base values should not be a concern at all, during operation the cores always work at much higher values. We will have time to prove this during practical experiments.
As we can see on the screen, the productive cores have an L1 cache buffer for data and instructions with a capacity of 48/32 KB, and in energy-efficient ones – 32/64 KB. The structure is identical to that of the previous generation models. But L2/L3 capacity in Raptor Lake has increased. For the second level buffer, the configuration looks like 8×2 MB + 4×4 MB instead of 8×1.25 MB + 4×2 MB in the predecessor. The total L3 cache has also been increased from 30 MB to 36 MB. Quantitative metrics have increased, but optimizations in prefetching algorithms for L2 and dynamic inclusiveness for L3 also need to be considered.
The Core i9-13900F chip modification does not support integrated graphics. Silently, the physically corresponding block is present on the crystal, but the video core is disabled in the F-versions. That is, for the system to work with such CPUs, a discrete video card is required. If the capabilities of integrated graphics are enough for your tasks, you should pay attention to the Core i9-13900 model without additional indexes in the name. The MSRP for such chips is ~$25 higher, while retail may vary more significantly.
Just in case, we remind you that the Core i9 processors of the Raptor Lake family without the index “F” in the model name are equipped with integrated Intel UHD Graphics 770 with 256 computing units. The operating frequency of the GPU is 300/1650 MHz with a theoretical computing performance of 800 GFLOPS (FP32).
Unlike older Core i9 models with an unlocked multiplier, which have a base value of the thermal package at the level of 125-150 W, the Core i9-13900F is claimed to have a rather modest TDP of only 65 W. Of course, taking into account the number of computing cores, as well as relatively high operating frequencies, we are obviously talking about certain starting levels of energy consumption. The declared maximum of 219 W for turbo mode already gives reason to pay more attention to this indicator. Moreover, even this value is not the limit for the Core i9-13900F. The manufacturers of motherboards allow you to “play” a little by increasing the energy limit to obtain maximum performance. Details later.
Intel Laminar RH1 cooler
The complete cooler for the Core i9-13900F deserves special attention, which is significantly different from the standard coolers supplied with models of the younger series. Indeed, Intel Laminar RH1 is exclusive to chips of the Core i9 line.
The manufacturer started using this cooler for Core 12th generation chips, however, the range of CPUs included with Laminar RH1 is limited to Core i9-13900/13900F and Core i9-12900/12900F models with a basic TDP of 65 W.
Among the design features of the Laminar RH1, it is worth noting the fairly large radiator unit. The total height of the cooler is 69 mm. The structure uses relatively thick aluminum ribs with decorative molding. This adds monumentality but somewhat reduces the total scattering area.
A copper heat sink is provided at the base for faster heat transfer. A thermoplastic substance is previously applied to the polished surface of the “sole”.
Let’s also pay attention to the fundamental difference in the fastening mechanism. Instead of plastic pins with limiters, in this case, screw fastening is used, which also involves the use of an additional frame bracket with threaded elements.
For a visual comparison, we suggest evaluating the Laminar RH1 alongside the Laminar RM1, which comes bundled with chips from Core i3 to Core i7. The structural differences are obvious. Laminar RM1 is much more compact, besides, the radiator unit of this cooler is fundamentally different from that of the older model. Its dimensions are much smaller, the aluminum structure is located directly under the fan, and the decorative “fins” are made of translucent plastic. At the same time, it should be noted that the RM1 radiator at the base has much thinner ribs to increase the dissipation area. In this case, a copper core is also used in the base.
The design of the fans of both coolers is similar, except that the length of the blades in the RM1 is slightly longer. In both cases, these are 80 mm models. The RM1 version is spec’d to rev between 600-3150rpm, while the older RH1 has a rev range of 1000-3000rpm. In both cases, PWM control is provided.
Despite the significant structural difference, the weight of standard coolers is not so significantly different. The Laminar RM1 weighed in at 345g, while the older Laminar RH1 model weighs 420g (up to 450g on spec).
Another fundamental difference of Laminar RH1 is the presence of additional lighting. And in this case, we are talking about addressable RGB illumination. The translucent plastic sleeve in the central part acts as a light guide, which also enhances the visual effect.
To activate the backlight, you must first connect the cable to the corresponding port on the motherboard (aRGB, 5V). Motherboard models that will be used with Core i9 will probably be equipped with the appropriate connectors. Setting the backlight operation modes depends on the capabilities of the board and is carried out in the company’s specialized software.
In addition to additional RGB illumination, we note the presence of an illuminated white Intel logo in the central part of the fan. During the operation of the system, the inscription remains static. At the same time, it can be independently turned to the required angle.
Additional lighting is not provided for Laminar RM1. Here, the developers limited themselves to only a bright blue ribbon along the contour of the fan.
Design, design features, and lighting are all derived secondary parameters. The main question is how Laminar RH1 will cope with its immediate task – cooling the processor. According to the declared specification, the cooler is designed for chips with TDP up to 65 W, but for the 24-core Core i9-13900F, this is only the initial value. Well, this question will be another stage of our practical experiment.
Test bench configuration
- Motherboard: ASUS PRIME B760M-A WIFI (Intel B760, LGA1700)
- Processor: Intel Core i9-13900F / Intel Core i5-13400F
- Cooling: ASUS ROG RYUO III 360 ARGB
- Memory: Kingston FURY Renegade DDR5-7200 32 GB (KF572C38RWK2-32 2×16 GB)
- Video card: ASUS DUAL GeForce RTX 4070 OC 12 GB (DUAL-RTX4070-O12G)
- Drive: Kingston KC3000 1 TB (SKC3000S/1024G)
- PSU: ASUS ROG STRIX 1000W GOLD
To test the capabilities of the Core i9-13900F, we used the ASUS PRIME B760M-A WIFI motherboard based on the Intel B760 chipset. Taking into account the fact that the processor does not have an unlocked boost multiplier, and therefore is not suitable for overclocking, it is not necessary to use the top-end Intel Z790/Z690 models. Of course, such options are also possible, especially in the presence of specific requests for functionality, however, it is possible to fully reveal the potential of such a CPU on a more affordable board.
Formally, even the simplest Intel H610 board worth $70 is enough to use the Core i9-13900F, but still, taking into account the fact that even according to the declared parameters, the processor power limit exceeds 200 W, it is better to be rational when choosing a specific model. We remind you that the 13th-generation Cores work with both DDR4 and DDR5, so choosing a platform based on the type of memory is not fundamental here. But of course, for a powerful Core i9 system, it is logical to choose a model with support for the new standard.
Please note that the ASUS PRIME B760M-A WIFI in the basic version works with DDR5, while the ASUS PRIME B760M-A WIFI D4 version is also available, which has identical functionality, but is designed for DDR4 standard modules.
The test bench was also equipped with a new dual-channel kit Kingston FURY Renegade DDR5-7200 (KF572C38RWK2-32) with a capacity of 32 GB (2×16 GB). The set deserves special attention, especially considering that it turned out to be excessive for the current platform.
For “insurance”, in addition to the standard cooler, we also used the ASUS ROG RYUO III 360 ARGB liquid cooling system. Such a system will allow to solve the issue of heat removal and check the effect of the efficiency of the cooling system on the performance of the processor.
Platforms based on processors without integrated graphics require the use of a discrete video card, so the ASUS DUAL GeForce RTX 4070 OC 12 GB is quite suitable for these purposes. Well, of course, it will also help evaluate the gaming capabilities of the Core i9-13900F.
During the system testing, a fast drive Kingston KC3000 with a capacity of 1 TB (M.2 PCI-E 4.0 x4; 7000/6000 MB/s) was used, and the “kilowatt” ASUS ROG STRIX 1000W GOLD was responsible for the stability of the power supply of the entire platform.
Intel entered the market of custom desktop video cards not so long ago, and according to the latest statistics from analysts Jon Peddie Research (JPR), it has already managed to occupy 4% of this competitive market. The manufacturer offered several models at once, among which the Intel ARC A750 8 GB looks particularly interesting, the recommended price of which has already been lowered to $249.
Unfortunately, Intel video cards are still not officially available in Ukraine, but we hope that the situation will change soon. Developers are actively optimizing drivers and introducing support for Intel’s proprietary XeSS intelligent scaling system, so this relatively inexpensive model could be a good addition for chips without integrated graphics. We are keeping our fingers crossed, hoping for the start of sales in our retail stores.
Working with DDR5 RAM
As we mentioned, a high-speed dual-channel DDR5-7200 memory kit was used for the experiments. Of course, this is exclusively a set for enthusiasts, which significantly exceeds the specified RAM specifications, and is guaranteed to be supported by the processor.
We will remind you that for the 13th generation Core chips, Intel has officially raised the speed bar to DDR5-5600 mode, instead of DDR5-4800 in the previous generation models. However, processors can usually work stably with much lower-speed modules. This is exactly what we were counting on using the DDR5-7200 kit. Rather, trying to do it.
However, it was found experimentally that the limit mode in which the platform was stable is DDR5-6400. According to the declared capabilities, the ASUS PRIME B760M-A WIFI board should work with DDR5-7200+.
During the refinement of the verified kits, we drew attention to an initially unremarkable parameter. Kits undergo validation in DDR5-7200 mode only when using K-modification processors. That is models for enthusiasts with an unlocked multiplier. Whereas for “non-K” chips, the specified sets are limited to the DDR5-6400 mode. It is unlikely that a potential owner of the Core i9-13900F will initially try to use even faster DDR5 sets, but this feature should be kept in mind when choosing a specific memory kit.
One of the available XMP profiles of the Kingston FURY Renegade DDR5-7200 kit (KF572C38RWK2-32) corresponded exactly to the DDR4-6400 (32-39-39-80) mode, which was eventually used as the main one.
For clarity, we used the AIDA64 utility to estimate the total bandwidth and memory latencies when the platform was running on a Core i9-13900F in three modes. DDR5-4800 is the base level we get with the JEDEC specification without the use of XMP profiles and additional self-debug. DDR4-5600 is a popular mode, as well as the limit for the 13th generation Core according to the manufacturer’s official specification. DDR5-6400 is one of the XMP profiles of the available kit, which turned out to be optimal for the test configuration.
In DDR5-6400 mode, the reading speed is close to 100 GB/s, and writing and copying are supported at a rate of about 90 GB/s. And in general, the operation of the memory controller does not cause any complaints. High transfers will obviously improve performance on memory bandwidth critical tasks.
Overall mode switching delays are also predicted to decrease as the timing formula improves and the operating frequency increases.
As you can see, the use of speed modules makes sense, even on a platform that is not designed for overclocking. The use of DDR5-5600/6000/6400 memory kits will be quite justified on a platform with Core i9-13900F.
Core i9-13900F performance
To evaluate the performance of the Core i9-13900F, we compared its performance with that of the Core i5-13400F. It is this basic model of the new Core i5 line that is an excellent reference point that allows you to visually determine the difference in the performance of chips that initially belong to different categories. Core i5-13400F is a perfect option that we unreservedly recommend as part of optimal gaming configurations in a traditional PC of the Month. This is a good benchmark that reflects the capabilities of modern mid-range chips.
Of course, comparing the performance of Core i9-13900F and Core i5-13400F, it is not difficult to immediately determine the winner and the chip that will come to the finish line second. However, we do not try to arrange a visual “versus”. In this case, we are interested in understanding and evaluating what you can expect if you prefer a top-class processor instead of a mid-range chip. How justified will the additional costs be? Under what conditions will the older model be able to prove itself?
Even before the start of the official tests during the preparatory warm-up stages, we made sure that the base cooler would not be enough to get the maximum performance from the Core i9-13900F.
By activating ASUS Performance Enhancement 3.0 (APE), the motherboard allows you to partially remove restrictions on energy limits for additional CPU acceleration. Of course, in this case, the standard cooler becomes a limiter. To avoid this, we additionally used the ASUS ROG RYUO III 360 ARGB liquid cooling system during the tests.
The motherboard also allows the processor to be used exclusively within the Intel specification. In the case of the Core i9-13900F, this mode provides a short-term possibility of CPU power consumption at the level of 219 W with an increase in operating frequencies under load with a subsequent decrease to the specified 65 W and corresponding values of acceleration of computing cores. Obviously, in this case, the overall performance of the processor will be significantly lower.
We were interested in checking what a processor owner who decides to additionally purchase an efficient cooler can expect. And also understand what limitations should be expected when using a complete cooler, the cost of which is already included in the price of the retail version of the Core i9-13900F. Thus, three modes are provided for this chip in the test:
- CPU cooling using a liquid cooling system with APE 3.0 technology activated in the BIOS. Designation on the diagrams – Core i9-13900F (Liquid cooler)
- CPU cooling with standard Intel Laminar RH1 cooler with APE 3.0 technology activated in BIOS. Designation on the diagrams – Core i9-13900F (Box cooler)
- CPU cooling with a stock Intel Laminar RH1 cooler with APE 3.0 technology disabled. Intel’s base specification. Designation on the diagrams – Core i9-13900F (Box cooler, 65 W)
To add to the informativeness, we also decided to determine whether a radical cooling improvement in the case of the Core i5-13400F makes any practical sense. This processor was also tested with a standard cooler and liquid cooler with APE 3.0 technology activated in the BIOS.
The traditional test with Cinebench R23 rendering in our case becomes very informative and needs a more detailed explanation. In single-threaded mode, the performance of the Core i9-13900F is almost independent of cooling conditions. The 65W limit is also enough to max out one or two cores and get a resulting score of ~2200, which is 26% more than the Core i5-13400F.
The situation changes fundamentally during the multi-threaded stage when all available computing cores are used to build the scene. It is in such conditions that the limited efficiency of the standard cooling system is indicated. When the APE 3.0 technology is enabled, under maximum load, the power consumption of the processor increases to 253 W. The CPU temperature quickly rises to 100C, after which the throttling mechanism is activated, searching for a balance between temperature, power supply voltage, and current frequencies of the computing cores.
As a result, the supply voltage drops to 1.04 V, the frequency of productive cores stabilizes at 4100-4200 MHz, and energy-efficient cores at 3300 MHz. At the same time, the power consumption of the processor remains in the range of 150–160 W. In such conditions, the Laminar RH1 cooler manages to keep the chip temperature at 95–100С, spinning the fan up to a maximum of ~ 3100 rpm. As a result, the Core i9-13900F scores almost 30,000 points in Cinebench R23. In general, a good result, but we are sure that the processor is capable of achieving more, so we repeat the test with a liquid cooling system.
Indeed, the situation is changing significantly. Under the maximum load on all computing units, the frequency of productive cores remains at 5000 MHz, efficiency cores at 3900 MHz. The power supply voltage is 1.24 V. In such conditions, the power consumption of the processor is about 253 W. At the same time, ASUS ROG RYUO III 360 ARGB manages to keep the chip temperature at 75C. Therefore, it is not surprising that the Core i9-13900F in such conditions received 36,563 points, improving the result of the chip with a basic cooler by as much as 22%. Yes, power consumption in this mode has increased significantly, but if you need maximum performance in multi-threaded mode, then the additional power consumption is justified.
The result in the mode with power consumption limitation within the specification is expected, but also quite indicative. We remind that in this case, the Core i9-13900F initially accelerates at frequencies within the power consumption of 219 W, but after a few tens of seconds, the power consumption is limited to the declared 65 W. In this case, all processor parameters are rebuilt taking into account the set limit.
In practice, with the Core i9-13900F, it looked as follows. During a long session with a high load, the frequency of the productive cores was only 2800 MHz, energy efficient – 2300 MHz, with a supply voltage of 0.83 V. Power consumption was indeed fixed within 65 W, and the capabilities of the stock Intel Laminar RH1 cooler were enough to keep the chip temperature at 60–65С even with a fan speed of ~1800 rpm.
Quiet, economical, but not very fast. In the overall score of Cinebench R23, we have 20,615 points. Modest, as for Core i9, especially against the background of previously obtained results. However, here you can say that everything is relative, looking at the performance of the Core i5-13400F.
Evaluating the performance of the Core i5, we immediately note that the standard Laminar RM1 cooler almost manages to cope with the cooling of the chip even with the removal of energy restrictions. The liquid cooling system significantly improves the temperature indicators, but the frequency ones remain almost unchanged. In both cases, the productive cores of the processor work at 4000-4100 MHz, and the energy-efficient ones at 3200-3300 MHz. The supply voltage is close to 1.08 V. Power consumption ranges from 100 to 105 W. In such conditions, the standard cooler stays on the limit – 95-100C with minimal throttling. In such conditions, the Core i5-13400F manages to score about 15,000 points in Cinebench R23 multi-threaded mode.
When replacing the basic cooler with a liquid cooling system, the main parameters of the CPU under load almost do not change, except for the temperature, of course. Therefore, it is not surprising that the overall performance indicators remain almost identical (15,177 points).
The obtained indicators make it possible to initially evaluate the power ratio of Core i9-13900F and Core i5-13400F. With effective cooling, the top-of-the-line processor can be almost 2.5 times ahead of the base model of the Core i5 series. Surprisingly, in such a case, the price/performance ratio is actually preserved, which is usually not characteristic of older processors. Mid-range CPUs often have the best performance here, while additional performance gains come at a significant premium. However, multi-threaded rendering in Cinebecnh R23 creates almost ideal conditions where Core i9 can show its strengths, making the most efficient use of available computing resources. Let’s see how the balance of power changes in the next stages.
The built-in test of the CPU-Z service utility is quite fast, so the difference in Core i9 performance with different cooling conditions is not as significant as last time. We note a 20% advantage of the Core i9-13900F over the Core i5-13400F in single-threaded mode and a 2.5-fold advantage in multi-threaded load.
The processor test from the 3DMark package is also quite fast, so even the basic cooler is enough for the Core i9 to score almost maximum points. Although, when limiting TDP, the overall score drops significantly. Although even in this case, the multi-threaded processor is 60% faster than the Core i5-13400F. When using a liquid cooling system and turning off the limit, the difference is already about 110%.
Another stage with a good level of computational parallelism is the rendering test based on Blender 3.5.0. The overall balance of power and the situation with the cooling effect generally coincide with what we have already seen in the Cinebench R23 stage. We have a perceptible increase in performance when using a liquid cooling system (+13%) and a 2.4 times advantage over Core i5.
We got very similar indicators when checking the capabilities of the processors in the V-Ray test. The only difference is that the regular cooler tried with all its might to “save face” even with increased energy consumption. And it almost succeeded – the difference with liquid cooling is only 6.5%. The advantage over the Core i5-13400F is about 128%.
The general trend is also the indicator of the processor test in the 7-Zip archiver. Using a 32MB dictionary and 32-threaded mode, the built-in benchmark averaged 203,570 GIPS (185,394 GIPS archive, 221,745 GIPS unpack). With the basic cooler, the indicators here drop by 14%. The limitation of energy consumption is even more noticeable. In the confrontation with the Core i5-13400F, there are also no surprises – almost 2.4 advantage of the Core i9.
But the built-in benchmark in WinRAR turned out to be a “quick shooter” that gives the result quite quickly, actually not making it possible to obtain indicators during a prolonged load and entering a stable mode of CPU operation after initial acceleration. As you can see, as a result, the Core i9-13900F has almost the same results in all cases, including when TDP is limited.
We also observe a very similar general picture in the GeekBench 6.0.3 set of synthetic tests. The equality of the Core i9 indicators pleases the eye, but we already know how the situation changes during a long load. By the way, GeekBench significantly increased the performance of Core i5 overall, so even in multi-threaded mode, we have a difference with Core i9 of only 58% and 22% in single-threaded mode.
By shifting our attention from “synthetics” to real practical problems, we get significantly different results. This time we are talking about video processing in HandBrake. We estimated the time it would take to transcode a 5-minute 4K@60 video clip at 25 Mbps with a total size of 872 MB into a 1080p clip with the HQ 1080p30 Surround preset. The Core i9-13900F chip with a liquid cooling system took a little more than two minutes for this task. It took almost 2.5 minutes with the Laminar RH1 cooler, and with the additional TDP limitation, the task required an additional minute.
Even more patience was required when using the Core i5-13400F. Chip managed the task in 4 minutes and 21 seconds. Once again, we draw attention to the fact that even with a high load of the regular Laminar RM1 cooler, it is enough to demonstrate indicators identical to those obtained from SRO.
As for the overall score, as we can see, the Core i9-13900F handled video transcoding about twice as fast as the Core i5. These are exactly the tasks where the advantage of a more powerful processor is transformed into a practical benefit for the user. If you often have to deal with similar tasks in your work, the time saved can be very significant.
Performance in games
Comparing the capabilities of the processors, we could not miss the game stages. Even owners of powerful Core i9 for serious tasks probably do not deny themselves entertainment. So let’s see if there is any practical benefit from a top-end CPU here.
Game test races do not do without a preliminary preparatory stage with 3DMark. For comparison, we used the Time Spy subtest. The results of the comparison of indicators in the processor stage are presented in the diagram. The difference between the cooling modes is noticeable, but the advantage of the Core i9-13900F over the Core i5-13400F is fixed at the level of 60%. But here it is worth noting that we are talking about a multi-threaded load on the CPU.
A hint that in real games you should no longer expect a multiple advantage due to a change in the processor is well demonstrated by the general indicators of the 3DMark Time Spy test, and also take into account the capabilities of the video card. Here, the difference between the Core i9 and Core i5 platforms is already only 8%.
Synthetics give only an approximate idea of the game capabilities of the system, for a more complete disclosure of the issue, we will conduct tests directly in real games. We emphasize once again that for the test platform, we use a GeForce RTX 4070 12 GB video card. Since we are primarily interested in the capabilities of processors, 1080p mode with medium image quality presets was used for all projects. Charts show average and minimum fps.
Starting with Red Dead Redemption II reveals a 14% advantage of the platform with Core i9-13900F over the system with Core i5-13400F. Interestingly, when limiting the TDP to 65 W, the Core i9 performance was almost identical to that of the Core i5.
Forza Horizon 5 reacted most briskly to the appearance of the Core i9-13900F in the system. Average fps increased by as much as 27%. Note that there is no fundamental difference in the performance of the system with liquid cooling and the basic Laminar RH1 cooler. The difference is only 2%, which is almost like a measurement error, especially at absolute values of 270 fps.
Dying Light 2, at first glance, almost did not feel the appearance of a more productive processor, giving about 200 fps in all cooling modes and even with the TDP limitation of the Core i9. However, let’s pay attention to the fact that with the older CPU model, the minimum fps increased by 23%.
But the hungry for processor resources Cyberpunk 2077 became an example of how game performance can depend on CPU power. The difference between Core i9-13900F and Core i5-13400F is already a significant 25%. In addition, this project provides a certain benefit from the use of a liquid cooling system. With a liquid cooling system, the average fps increased by 9%. At the same time, we have a clear dependence on the set TDP limit for Core i9, which was not encountered before. If you limit yourself to the level of 65 W, the performance of the Core i9-13900F can be even lower than that of the Core i5-13400F.
Assassin’s Creed Valhalla when replacing Core i5 with Core i9-13900F added about 12% in average fps and 21% in minimum values. Changing the cooling system and limiting TDP only minimally affected the minimum fps.
Another game where there is an increase when switching to Core i9 was A Total War Saga: TROY. Although here the difference turned out to be relatively small – 6–8%.
Core i9-13900F is a top chip, including a gaming platform. In this discipline, among Intel’s solutions, it will yield minimally, except for the older Core i9-13900K/KS with even higher operating frequencies (but also energy consumption). Of course, in the context of gaming systems, the capabilities of the 24-core Core i9 are even somewhat excessive. From a rational point of view, even the conventional Core i7-13700K (8/16+8) will do well here. However, Core i9 owners who primarily need uncompromising power in work multi-threaded tasks can be sure that their system is also great for entertainment.
Energy consumption
We partially considered the issue of power consumption of processors in different modes during the interpretation of Cinebench R23 results. For generalization and better clarity, we offer a diagram with the obtained indicators.
Despite the TDP of 65 W, on which attention is usually focused, the 24-core Core i9-13900F can consume much more energy, especially in cases where limits are not used, and the efficiency of the cooling system allows you to keep the temperature of the chip within acceptable limits. In such cases, with the maximum simultaneous load of all computing units with high operating frequencies of the cores, the processor can consume about 250 W without time limitations. The values are very important, so if you want to get the most out of your CPU, cooling it should not be taken lightly.
As we have already studied, such working conditions are an overwhelming task for a full-time cooler. It manages to keep the temperature of Core i9-13900F at the level of 90-100C under a multi-threaded load with a chip power consumption of about 150-160 W. Of course, this affects the final performance, although a lot depends on the specifics of the load.
If you limit yourself exclusively to the manufacturer’s specification, rigidly fixing the power consumption mode within the framework of 65/219 W, then only a standard cooler will be happy with such a situation, which will no longer have to increase the speed of its fan to the maximum. Indeed, such a profile can be interesting for situations where high multithreaded loads are only short-lived. However, then a logical question arises, whether in such cases the top and not cheap Core i9 is needed at all. Still, the chips of this line can offer excellent performance, and this is where their maximum value lies.
As for the Core i5-13400F, this chip with the limits removed under load in Cinebench R23 consumed 100-107 W of energy. Moreover, when cooling with a standard cooler, the consumption was several watts more due to the higher temperature.
In general, as we can see, even the energy balance partially coincides with the ratio of the actual performance of both processors precisely in the multithreaded mode. The consumption of the Core i9-13900F increases almost proportionally to the increase in computing power.
The above indicators of power consumption of processors are typical for the maximum load on the CPU. If the nature of the tasks does not involve the simultaneous involvement of all processor resources, then the values will be significantly more modest. For example, let’s provide the indicators during the game in Cyberpunk 2077, noting beforehand that this is one of the most demanding projects, eager for processing power.
When using liquid cooling systems, the processor consumed about 160 W of energy. At the same time, all productive cores worked at 5300 MHz, and efficiency cores were accelerated to 4200 MHz. Hence the maximum fps.
With the box cooler, power consumption was kept at an average of 140 W. The stable frequency formula was already more modest – 4700/3700 MHz. Therefore, it is not surprising that the fps, in this case, decreased by 9%.
Limited to 65 W during the Cyberpunk 2077 game session, the processor generally “rested”. The frequency of the P-Cores cores ranged from 3400-3700 MHz, and the energy-efficient cores worked in the range of 2400-3000 MHz. It is not surprising that in this case, the number of frames/s decreases significantly. And even to a lower level than the forced Core i5-13400F, which, by the way, consumed 68–72 W during daring walks in the Night City.
Price
When launching the 13th-generation Core desktop chips, Intel followed the principle of balanced pricing. Even before the launch of the new CPUs, there were expectations that the manufacturer might raise the price of the Raptor Lake models, but the tangible competition in the segment encourages greater rationality.
Processors of the top Intel Core i9 line are a priori quite expensive chips, even when it comes to the initial model of this series. As you can see in the table of specifications, the recommended price of Core i9-13900F for the American market is $524. Of course, in Ukraine, the cost is somewhat higher, but in this case, the difference is moderate.
You can buy a processor from our partners, stores ITBox and Brain.
According to price aggregators, the current retail price of the Core i9-13900F starts at ~22,000 UAH (~$590). Core i9-13900 versions with identical specifications, as well as activated integrated graphics, are offered from 23,500 (~$625).
Another active participant in the current review is the Core i5-13400F, which has a recommended price tag of $196 and is offered from UAH 8,700 (~$230) on local online sites. As we have already noted, the current cost of chips correlates quite well with the performance of processors in resource-intensive multithreaded tasks.
Loading comments …