PC of the Month (September 2025)

The first half of autumn is usually a good time for those who want to build a powerful gaming system. Component suppliers are preparing for the start of the active business season, trying to deliver the latest innovations, while getting rid of possible remnants of outdated components. In addition, practical needs for new PCs are growing. The new season means new work tasks, training plans and a whole series of game announcements. Let's analyze the current situation in our market and, within the framework of the traditional "PC of the month", determine how to more effectively spend money on the purchase of powerful desktop platforms in three price categories.

Entry-level gaming PC of the month ($700–850)

Accessories for the “Basic Gaming PC” from our partner HATOR

Processors. This time, within the basic gaming configurations, we will try to reduce the total cost of the systems, but to the point where they will still remain professional and can cope with heavy projects. To save money when choosing a processor, we return to the Core i5-12400F (6/12; 2.5/4.4 GHz; 18 MB L3). Despite everything, this is the most affordable 6-core chip that can meet the needs of entry-level gaming platforms. Complete with a standard cooler, the Core i5-12400F can already be purchased for the equivalent of $110. This version of the supply is justified in this case, because almost the same money is asked for OEM modification.

Maybe there is something more affordable? Yes, the Core i3-12100F (4/8; 3.3/4.3 GHz; 12 MB L3) is offered for $70 in a boxed configuration and even for $60 in a tray version, but 4-core chips have already left the "gaming chat". Even despite the support for Hyper-Threading, which allows the processor to simultaneously process up to 8 data streams, this is not enough for resource-intensive projects. So the Core i5-12400F with a functional formula of 6/12 can be considered a starting position for entry-level gaming PCs.

If you want and have the ability to make some extra computing power, the next step is the Core i5-14400F (6/12+4; 2.5/4.7 GHz + 1.8/3.5 GHz; 20 MB L3). Among the fundamental differences of this model is the presence of 4 additional cores. Despite the fact that these are energy-efficient computers operating at relatively low operating frequencies, the so-called E-Cores significantly increase CPU performance in multi-threaded loads.

The appearance of the Core i5-110 processor (6/12; 2.9/4.3 GHz) in Intel's range did not go unnoticed by the community, which is actually a copy of the Core i5-10400 for the LGA1200 platform. Such a 14-nanometer "novelty" of the Comet Lake family for $200 obviously seems strange in September 2025. However, this is more of a solution for terminals and specific systems with an extended service life, so we probably won't see them among the offers in retail. And that's for the better.

But what chips would you still like to wait for, it is the Core Ultra 3 205 (4+4; 3.9/4.8 GHz + 3.3/4.2 GHz). At least to satisfy academic curiosity. The model was nominally announced by Intel at the beginning of this year, but very quickly the mention of it disappeared from the manufacturer's official website. Since then, the fate of the "three" has remained unknown. Finally, there is hope for the official launch of the CPU soon. Korean enthusiasts have even managed to conduct the first tests of the Core Ultra 3 205. As a result, the integrated graphics (2Xe) allowed the chip to score 1125 points in 3DMark TimeSpy, which is 75% more than the Core i3-14100 (643 points). In the purely processor-based multi-threaded rendering of Cinebench R23, the new chip was ahead of its predecessor by 45% - 13,394 vs. 9044 points. For such tasks, eight physical cores (4+4) are probably better than a 4/8 combination, but we will talk about the potential success of the Core Ultra 3 205 and the possibility of getting into recommended configurations only after the official launch and actual retail prices. Will LGA1851 finally appear in basic gaming systems? We leave this possibility open, but we make no promises.

Choosing an affordable solution on the AMD platform, we stay on Socket AM5. The cheapest chip for this platform is still the Ryzen 5 8400F (6/12; 4.2/4.7 GHz; 16 MB L3), the capabilities of which are generally sufficient for this position. For ~$105 we get a decent 6-core processor with a single-chip architecture and 16 MB of L3 cache. But such a price tag is relevant only for the tray version of the CPU supply. For a retail Box modification with a cooler in the kit, they ask for $145–150. If we are talking about maximum savings, you can consider the tray option with a simple cooler for $15–20.

A slightly more interesting chip for basic configurations on AMD will be the Ryzen 5 7500F (6/12; 3.7/5.0 GHz; 32 MB L3) - higher operating frequencies, larger capacity of the third-level cache, support for PCI-E 5.0. But this chip in a tray configuration costs from $145–150, and if you add the cost of the cooler, it turns out to be a not-so-affordable solution. In addition, when using entry-level video cards, the actual difference in performance will be insignificant. There were certain hopes for the Ryzen 5 7400F (6/12; 3.7/4.7 GHz; 32 MB L3), but now they are actually offered at the price of the 7500F, which excludes the younger model from the list of candidates.

When considering options with a reduction in the cost of the system, it is worth mentioning the Socket AM4 platform. Especially considering that AMD quite regularly reminds of this long-liver. The latest example is the recent announcement of the Ryzen 5 5600F (6/12; 3.0/4.0 GHz). The new model belongs to the Vermeer line and differs from the iconic Ryzen 5 5600 (6/12; 3.5/4.4 GHz) only in slightly lower operating frequencies. The designation of this CPU is somewhat surprising. Usually, the presence of the "F" index in the model name indicates the absence (deactivation) of integrated graphics, but Vermeer chips do not have it initially. Presumably, the developers would have used the name "Ryzen 5 5500" if possible, but it is already reserved for the existing model of the Cezanne family. Because of this, you have to experiment, which does not add logic to the labeling system.

In general, what can really interest AM4 now? The aforementioned Ryzen 5 5600 processor is offered for ~$90. Thanks to 32 MB of L3 in games, it will not be much inferior to the Ryzen 5 8400F, and in exceptional cases it may even have a certain advantage. Although here too the "equalizer of potentials" will be the video card. The difference in the price of chips is not so fundamental that this factor will be decisive when choosing a platform. Perhaps the Ryzen 5 5600F will be even more affordable, but it is unlikely that the price will be more attractive immediately after the start of sales. In addition, its recommended cost has not even been announced yet. A certain drawback is the reduced operating frequencies. With an unlocked multiplier, this does not seem to be a problem, but you will have to fiddle with the settings, and this is not to the liking of all users.

From economical options, you can look at the Ryzen 5 3600 (6/12; 3.6/4.2 GHz; 32 MB L3). The price tag of $55 is really very attractive. But in this case we are dealing with the Zen 2 architecture, so even despite the rather good characteristics and 32 MB of third-level cache, a certain lag from the newer generation models can be observed even on a platform with entry-level video cards.

Socket AM4 can still offer the Ryzen 5 5500 (6/12; 3.6/4.2 GHz; 16 MB L3) on the Zen 3 architecture. A fairly capable single-chip 6-core for $60 competes with the Ryzen 5 3600 and is even somewhat more economical. But its main drawback is support for only PCI-E 3.0, which can be a problem when using basic video cards with a PCI-E x8 interface. And recently, graphics adapters in this category have been using exactly this bus configuration.

So, more modest alternatives still encourage you to go for the Ryzen 5 5600, meaning you need to increase your budget to $90 + the cost of a cooler. But if you carefully study the range of chips for AM4, it's hard not to notice the 8-core 16-thread Ryzen 7 5700X (8/16; 3.4/4.6 GHz; 32 MB L3) on sale for $115, which breaks all the molds and will make even the most frugal user hesitate when choosing. But again, this is not a story about spending less.

Maybe Socket AM4 will allow you to get by with a more affordable motherboard? Indeed, in the range of models based on the AMD A520 and AMD B450 chipsets, the cheapest models are offered at a price of $55–60. Of course, these are the most primitive power subsystems, the simplest component base and trivial general equipment. Even if you close your eyes to all this, there remains one, but weighty argument "against" - they support only the PCI Express 3.0 bus. Even if the processor has a PCI-E 4.0 controller, the video card and drives will be limited to PCI-E 3.0. If with SSD it is unpleasant, but tolerable (transfers up to 3500 MB/s), then the outdated standard for connecting a video card with PCI-E x8 can be a certain problem. Let's not exaggerate its degree - it is not critical for all games, but in certain projects the fps losses can be quite significant.

To get rid of this "flaw", you will have to look towards AMD B550 models, and this starts at $90 for typical inconspicuous models with something like additional VRM cooling. For this price, you can already buy AM5 boards, so there is no special saving here either.

For a long time, the AM4 platform was generally cheaper due to the use of DDR4 RAM. But recently, prices for modules of this standard have risen sharply, so now a dual-channel DDR4-3200/3600 kit with 32 GB (2 × 16 GB) costs only slightly less than a DDR5-5600/6000 kit - $75–80 vs. $90. Memory chip manufacturers have significantly reduced their DDR4 offerings, refocusing on the production of more modern chips. The situation is generally typical, it happens every time there is a global change in RAM generations. However, too rapid a jump in prices raises suspicions of some kind of agreement between chip/module manufacturers.

As a result, Socket AM4 remains a fully functional platform, and the availability of relatively inexpensive Ryzen 5000X3D chips provides prospects for a possible effective upgrade if necessary. But even here there are some peculiarities. Recently, Ryzen 7 5700X3D have almost disappeared from sale in Ukraine. Back in August, there was information that AMD was cutting back on supplies of this model, so the lack of a corresponding position in the price list may not be a temporary phenomenon. It will be a shame.

In June, AMD introduced the Ryzen 5 5500X3D (6/12; 3.0/4.0 GHz; 96 MB L3). At an adequate price, this could be a rather interesting option for relatively inexpensive gaming systems. Even despite the 6-core layout and reduced operating frequencies that are not adjusted (overclocking is not supported). However, at the end of September they are still not available for retail sale in Ukraine. Considering the distribution region of this model indicated on the manufacturer's website - LATAM, it is not a fact that it will even reach our shores in commercial quantities.

Considering the above nuances, Socket AM4 does not offer options for significant savings at the start. This option is rather worth considering if certain components were purchased for a nominal fee. For example, you already have a processor or a set of DDR4 memory. If the PC is being assembled from a "clean slate", even in the case of basic configurations on AMD, it is more logical to stick to the more modern Socket AM5 platform, which no longer requires significant overpayments. Instead, it offers better prospects - generally more productive processors, PCI-E 5.0, DDR5, USB4 (optional) and, at a minimum, guaranteed support for future CPUs on the Zen 6 architecture.

Speaking of motherboards, let's decide on rational models for LGA1700 and AM5. For the Intel platform, we suggest choosing a solution with the Intel B760 chipset with support for DDR5 standard modules. We definitely need the ability to accelerate RAM beyond the modes specified by the specification. Intel H610 boards do not offer such an option, so in the case of the Core i5-12400F we would have to be content with DDR5-4800, which is an obvious loss of fps compared to DDR5-5600/6000. Models with DDR5 are slightly more expensive than versions with DDR4 with the same equipment, so we will have to focus on costs of ~$120. Despite this, we do not consider options with DDR4, especially after the increase in the cost of memory of this standard. The final price difference is minimal, and DDR5 with much higher bandwidth will provide a certain increase in frames/c.

For AM5 configuration, a simple AMD B650 board is suitable. They are more interesting compared to the completely trivial models on the AMD A620 chipset, and also potentially have the opportunity to experiment with CPU overclocking at least at the initial level. However, in case of significant financial difficulties, devices on the AMD B620 cover the basic needs, offering a PCI-E 4.0 x16 slot and at least one M.2 PCI-E 4.0 x4 port. With the general peripheral wiring there, everything is very simplified due to the presence of chipset lines only of the PCI-E 3.0 standard.

The cost of noteworthy models on AMD B650 starts at the equivalent of $110–120, boards on AMD A620 can already be purchased for $85–90. Subjectively, solutions on AMD B650 are worth the extra money. There, the VRM is usually better (maybe even with MOSFET cooling), and the presence of two M.2 PCI-E 4.0 x4 ports is the norm, and the set of interface ports is more interesting.

Video cards. For basic gaming configurations, this time we suggest considering a new product – GeForce RTX 5050 8 GB. The model very quickly burst into the market of entry-level video cards, immediately "unhooking" several alternatives in this class. According to the results of practical tests, the solution based on the compact GB207 graphics processor barely (~5%) falls short of the GeForce RTX 4060 8 GB. And this is generally a frankly good level, as for a model with a recommended price tag of $249 (in the US market, of course).

We have 2560 compute units, a 128-bit bus, and 8 GB of memory. A typical set for a budget solution that could be significantly improved using the new GDDR7 memory chips. However, unfortunately, for the desktop model RTX 5050, the developers decided to limit themselves to GDDR6. As a result, the memory bandwidth is 320 GB/s, instead of 448 GB/s, which could potentially be obtained on new chips with an effective 28,000 MHz. But for contrast, let's recall that the GeForce RTX 4060 is content with 272 GB/s. This is not self-censorship, but a reminder of how quickly we get used to something good.

So, with the results of +50–60% to the GeForce RTX 3050 8 GB and -5% to the GeForce RTX 4060 8 GB, the new NVIDIA model literally occupies the niche of entry-level video cards. The older model of the previous generation (RTX 4060) is already offered only as remnants and is leaving the market quite quickly. Let's recall that at the peak of its popularity it was offered in Ukraine from $340–350. The new model falls under the pressure of the Radeon RX 7600 8 GB. At a similar price, the RTX 5050 offers slightly higher performance and better economy. We don't even focus on the 25–30% advantage when activating RT - modes with rays are not for this class of video cards.

So AMD's alternative solutions in the price category up to $300 are actually not presented at the moment, or, to be more correct, not relevant. The company offered the Radeon RX 9060 8 GB, but the announcement in early August was quite ordinary and without an official indication of the recommended price. Initially, it was even said that this model would be available only to OEMs for some time, however, the appearance, albeit a very moderate number of models in the assortment of key partners, still leaves hope for wider distribution.

According to the first test results, the Radeon RX 9060 will compete more with the GeForce RTX 5060 8 GB. But if it were to compete with the RTX 5050 in price, it would be a potential hit in this category. In general, the model uses the same Navi 44 graphics processor as the older Radeon RX 9060 XT, but in a modification with 1792 compute units (2048 in the 9060 XT) and a proportional reduction in the number of associated functional units, except for ROPs (64 in both cases). The memory bus remained 128-bit, the bandwidth at the same level - 322 GB/s. The Radeon RX 9060 is offered only in a modification with 8 GB of memory, so it will obviously be in the camp of entry-level solutions for Full HD games. We are waiting for it to appear on sale.

For the base system, the "ageless classic" - GeForce RTX 3060 12 GB - also remains relevant. Despite the fact that at the end of August it lost the first place in the popularity rating among Steam users to GeForce RTX 4060, the latter is already actively disappearing from sale, and the demand for RTX 3060 12 GB remains. The increased memory capacity is decisive. In the confrontation with the GeForce RTX 5050 8 GB, it is not so easy to determine the winner. The novelty is on average 10–15% more productive, if 8 GB of memory is enough, of course. We can mention the full support for DLSS 4 Multi Frame Generation. In addition, it is 40–50 W more economical and even cheaper. The cost of GeForce RTX 3060 12 GB does not fall below the equivalent of $300, while the GeForce RTX 5050 8 GB is already offered from $280. The GeForce RTX 3060 beats with increased memory capacity, and in certain cases this is a killer argument.

For $300, the system can also be decorated with the Intel ARC B580 12 GB, which in the usual rendering will be ~5% ahead of the RTX 5050, and if you enter the 1440p territory, the difference will increase to 15% in favor. Yes, a certain exotic, but also an opportunity to try something new.

A few days ago, AMD quite unexpectedly expanded the Radeon RX 7000 line of video cards, introducing the Radeon RX 7700 16 GB model. The adapter uses the Navi 32 graphics processor in a configuration with 2560 compute units, 160 texture units and 96 ROPs. At the same time, the Radeon RX 7700 got a 256-bit bus and 16 GB of memory with a bandwidth of 622 GB/s. As a result, we have a kind of hybrid that has as many as 35% fewer stream processors than in the case of the RX 7700 XT 12 GB (3456), while the memory subsystem configuration is identical to that of the Radeon RX 7800 XT 16 GB.

On the description page of the new model, AMD has indicated approximate performance indicators in games, using values for 1440p mode. According to this information, the Radeon RX 7700 is expected to be 10-20% slower than the Radeon RX 7700 XT 12 GB, so it is obvious that it will also have a lower level of performance than the newer Radeon RX 9060 XT 16 GB. In this case, questions arise about the general positioning of the new product. Let's assume that the Radeon RX 7700 16 GB will become a competitor to the GeForce RTX 5060 8 GB. Despite the fact that this role has already been assigned to the Radeon RX 9060 8 GB, theoretically the Radeon RX 7700 could become an alternative with 16 GB of memory for about the same money. It is with the latter that the intrigue remains - AMD has not yet announced the recommended price tags for the "new" model.

The Radeon RX 7700 16 GB will probably cost less than the Radeon RX 9060 XT 16 GB ($349), but will the difference really be so significant as to make a choice in favor of the slower, less technically equipped model with a higher level of power consumption? As for the latter, the specifications on the developer's website indicate a Typical Board Power (TBP) of 263 W. We hope this is a mistake, since this value is relevant for the Radeon RX 7800 XT 16 GB, and the GPU for the RX 7700 has one and a half times fewer activated compute units. Although, the presence of two 8-pin PCI-E connectors in the photo of the first models is frankly confusing.

In light of the Radeon RX 7700 16 GB, AMD will also have to decide the fate of the Radeon RX 7600 XT 16 GB ($329), which, despite its solid memory capacity, is much less suitable for 1440p gaming due to its 128-bit bus and bandwidth of only 288 GB/s. We are waiting for bold steps from AMD, which now really needs effective measures to maintain user interest. The price factor has almost always worked flawlessly.

Optimal Gaming PC of the Month ($1200–1450)

Accessories for the “Optimal Gaming PC” from our partner HATOR

HATOR Pulsar 3 Pro Wireless Gaming Mouse

HATOR Icefall Pro Wireless Gaming Keyboard

HATOR Hyperpunk 3 Wireless Gaming Headset

Processors. After the cost of the Core i5-14600K (6/12+8; 3.5/5.3 GHz + 2.6/4.0 GHz; 24 MB L3) decreased so much that it was possible to use it as part of an optimal configuration, it is doubtful that this position will change even in the medium term. There is nothing better than ~$200 on the market in principle. Last time we argued in quite detail for this particular choice. It is difficult to say how long such a "promotional offer" will last, but it is worth taking advantage of the moment.

Despite the fact that the Core i5-14600K does not belong to the top series in its generation, the chip has a considerable level of power consumption and under maximum load requires up to 180 W. So in this case, it is worth paying special attention to CPU cooling. The complete set of processors of the enthusiast lines, which have the "K/KF" indices in the name, does not include a standard cooling system, even if we are talking about retail box versions of the chips. So the choice here is entirely at the discretion of the user. We advise you not to try to save as much as possible, a rational approach would be to purchase a mid-level cooler for ~$50–70.

As for the motherboard, we rely on models with the Intel B760 chipset and DDR5 support. Yes, they do not allow overclocking the processor even with an unlocked multiplier, but in this class of systems, the power of the Core i5-14600K will be quite enough even in standard mode. Overclocking older 14th generation Core chips significantly increases the already considerable power consumption, and the resulting performance increase of 5–10% probably does not justify the additional effort and costs. So we are content with the standard mode, and ideally even superficially experiment with downvolting. Reducing the operating voltage will reduce the heating of the crystal.

We also pay attention to the choice of the Core i5-14600K, despite the slightly more affordable Core i5-14600KF. In general, for a gaming system, the presence of integrated graphics is not essential, but in light of potential power supply problems, it is worth considering the version with the iGPU activated. In such cases, you can use the PC in the most energy-saving mode, which will allow you to work longer from the charging station.

Among AMD chips for the optimal configuration, the best candidates are the Ryzen 7 7700 (8/16; 3.8/5.3 GHz, 32 MB L3) and the Ryzen 5 9600X (6/12; 3.9/5.4 GHz; 32 MB L3). Both chips have very similar price tags and their own set of advantages and features. Frankly, we continue to be happy with the opportunity to purchase an 8-core 16-thread Ryzen 7 7700 for only a little more than $200 (tray version). A very good offer that allows you to close the issue of a possible shortage of computing units even in the most demanding gaming projects. Even if it does not offer maximum fps, especially in Full HD mode, the functional formula 8/16 will certainly eliminate possible stutters not only in current but also in future games.

In turn, the 6-core Ryzen 5 9600X uses the Zen 5 computing architecture with increased performance per clock, and aggressively increases frequencies when not all cores are loaded. In certain projects, it is the increased IPC that is important, so often the new product has even better average fps than the 8-core of the previous generation on Zen 4. The asset of the Ryzen 5 9600X can also be attributed to better efficiency. Despite its relation to the X-series, the developers set the TDP for this model at 65 W. Because of this, when loaded on all cores, the CPU frequency is kept at a relatively low level so as not to exceed the set limit.

This approach caused some dissatisfaction among the community, so AMD had to react quickly, allowing the TDP to be increased to 105 W without losing the manufacturer's warranty. And the process of transitioning 65 W -> 105 W was simplified as much as possible - it is enough to change one parameter in the motherboard BIOS settings. After increasing the energy limit, the performance in multi-threaded tasks improves significantly, although in games the difference is minimal due to the variable nature of the load. And in this case, the Ryzen 5 9600X is initially actively accelerated.

If the Ryzen 7 7700 and Ryzen 5 9600X chips are used in standard modes, simpler cooler models can be used to cool the processors than in the case of the Core i5-14600K. So this is an opportunity to save $10–20.

Recently, at the same time as the Ryzen 5 5600F, AMD also introduced the Ryzen 5 9500F (6/12; 3.8/5.0 GHz; 32 MB L3) – the basic 6-core model of the Granite Ridge line for the AM5 platform. Although the recommended price has not yet been indicated, it will probably be the most affordable Ryzen 9000 series chip on the Zen 5 architecture. In addition to the more modest frequency formula, the Ryzen 5 9500F also has deactivated integrated graphics. We hope that the new model will be able to repeat the success of its famous predecessor – the Ryzen 5 7500F, and this time it will do without unpleasant surprises in the form of thermal paste instead of solder under the cover.

Video cards. With all the wealth of choices for optimal configurations, the GeForce RTX 5070 12 GB remains the most balanced solution for ~$650. From the NVIDIA range, the GeForce RTX 5060 Ti 16 GB models, despite the "correct" memory capacity, are somewhat below this category of platforms, and the cost of the GeForce RTX 5070 Ti 16 GB automatically excludes it from consideration for PCs of this class. As for the options from AMD, the Radeon RX 9060 XT 16 GB is a direct competitor to the GeForce RTX 5060 Ti 16 GB, so like the latter, it is a last resort in case of total savings. The Radeon RX 9070 16 GB would be very appropriate here, which even has a slight advantage over the GeForce RTX 5070, but the current retail price of ~$720+ "ruins" all good intentions to allocate it to the optimal configurations. It's even more disappointing that both models – the GeForce RTX 5070 12 GB and the Radeon RX 9070 16 GB – have the same recommended price for the US market ($549).

In general, to improve the situation in the middle segment, there is a great lack of additional alternatives that could be the same solutions from Intel. There is more and more evidence that the latter continues to work on older models of discrete graphics cards of the Battlemag family. At least, there is every reason to hope that we will see a solution on the BMG-G31 chip with 32 Xe clusters, a 256-bit bus, PCI-E 5.0 support and 16 GB of onboard memory. Potentially, with such equipment, the graphics card looks like a competitor to the GeForce RTX 5070 and Radeon RX 9070, but there are still no real deadlines for the appearance of the conditional "Intel ARC B770". Time will tell whether the recently announced cooperation with NVIDIA will affect the plans for future graphics cards. For now, the developers assure that they will continue working on their own graphics solutions.

Despite the popularity of the GeForce RTX 5070 12 GB, it is probably not worth emphasizing the main drawback of this model once again. Of course, we are talking about the memory capacity. 12 GB may not always be enough even for the 1440p mode, which is actually a priority for video cards of this class. NVIDIA developers understand this perfectly well, and they are going to correct the situation in the upcoming SUPER modification. According to preliminary data, the GeForce RTX 5070 SUPER will receive versions of the GB205 graphics processor with an increased number of processors. Thanks to the activation of two additional stream multiprocessors (SM), the number of processors will increase from 6144 to 6400, and the number of tensor and RT cores will also increase proportionally. But the most anticipated upgrade will be an increase in memory capacity from 12 GB to 18 GB. At the same time, the bus width will remain at the same level – 192 bits, as well as the bandwidth of 672 GB/s. A one and a half times increase in the local buffer will be possible thanks to GDDR7 chips with a capacity of 3 GB. Six such chips will allow you to get 18 GB, respectively. It is expected that the TGP of the video card will increase from 250 W to 275 W.

If the recommended price of the GeForce RTX 5070 SUPER 18 GB is at the current level for the basic GeForce RTX 5070 12 GB, it will be a kind of "imbo" option in its class. But we will probably find out about it no earlier than January CES 2026. It was expected that NVIDIA could offer SUPER versions on the eve of the Christmas holidays, but the developers decided not to rush to launch modified options. In some ways, they can be understood, especially if you look at the current distribution of the discrete graphics card market. NVIDIA's total dominance with a record share of 94% according to the results of Q2 2025 is a reason to slow down the pace of announcements. AMD needs to catch up more actively, and Intel should also accelerate with its productive solutions. The lack of changes also seems to be a result, but not when your total share is almost zero.

Progressive Gaming PC of the Month ($1800+)

Accessories for the “Progressive Gaming PC” from our partner HATOR

HATOR Quasar 3 Ultra 8K Wireless Gaming Mouse

HATOR Rockfall 3 TKL Wireless Gaming Keyboard

HATOR Hypergang 3 Wireless Gaming Headset

Processors. For a progressive gaming system on Intel, we started using the LGA1851 platform, equipping it with the Core Ultra 7 265K processor (8 + 12; 3.9/5.5 GHz + 3.3/4.6 GHz). Let us recall that the reason for this was a significant reduction in the cost of this model, which officially lost $100 in price. The current cost of a 20-core chip of ~$330 makes it quite possible to count on the Core Ultra 7 265K in the context of a powerful gaming platform.

Intel realizes that the chips for LGA1851 are far from ideal and it would be useful to "tweak" them to add attractiveness, especially in terms of gaming solutions. There is a positive in the fact that here we are dealing not just with a statement, but with a practical attempt to improve the situation. In addition to the Intel 200S Boost initiative with automatic overclocking of internal buses and memory modules, developers continue to improve Intel Application Optimization (APO) technology to optimize loads between cores of different types in games. More efficient use of productive processors (P-Cores) allows you to increase performance. However, the mechanism requires preliminary preparation of appropriate profiles for each specific game, so this process takes some time. Recently, the list of "prepared" projects was replenished with 15 more games. So the process is underway here, although the pace leaves much to be desired, and there are not many new products among the optimized projects.

Intel is also working on updating its chipset for LGA1851. As we mentioned earlier, radical changes are not expected here, although the manufacturer promises to offer the Arrow Lake-S Refresh line. It seems that the improvements will be limited to frequency adjustments. But if Intel manages to build an updated line, closing the positional gaps and offering objectively good price tags, this will help LGA1851 hold its own until the arrival of the future LGA1954 platform.

As for the timing of the appearance of Arrow Lake-S Refresh chips, then this is also just speculation. By the end of 2025, we are unlikely to see any changes here, most likely we are talking about the beginning of next year. The same CES 2026 exhibition may well become a platform for the announcement/launch of updated CPUs. Well, if we dream about Nova Lake-S for LGA1954, then Intel's new desktop platform should appear by the end of 2026. The manufacturer has high hopes for these chips and hopes to regain its positions in the segment of top consumer solutions. Meanwhile, from various sources, the number of confirmations that in the maximum configuration, future CPUs will offer up to 52 cores (16P+32E+4LP) is increasing.

Let us recall that AMD also intends to increase the number of cores on one CCD crystal from 8 to 12 for future processors based on the Zen 6 architecture, and, accordingly, to bring the total number of cores on models with a pair of CCDs to 24. So, there is a competition ahead in the number of computing units. Well, let such a "nuclear war" be better than another.

Returning to the present and the choice of an AMD processor for a progressive configuration, we load the platform with the most titled chip - Ryzen 7 9800X3D (8/16; 4.7/5.2 GHz; 96 MB L3). Expensive, but it knows its stuff. When it comes to games, expect another record-breaking performance. Still, the idea with an additional 3D V-Cache buffer "shot" and allowed AMD to reshape the gaming hardware market according to its own patterns. The exceptional efficiency of increasing L3 in very specific tasks, namely games, was recognized even by the main competitor, which is secretly already working on a similar solution of its own (bLLC, big Last Level Cache).

Even in the tray configuration, the Ryzen 7 9800X3D costs about $515. Although when buying a processor of this level, you probably still want to get the chip in retail packaging, not in a transparent blister, so it makes sense to pay a little extra. The desire to use the best is quite natural, but rational users will probably have a question about the justification of such expenses. Especially for 1440p and 4K modes, where the limited power of the video card does not allow to reveal the full potential of productive CPUs. However, it is worth recalling that the Ryzen 7 9800X3D received a second-generation 3D V-Cache, which does not hold back operating frequencies under high load, and even allows you to experiment with overclocking if desired.

However, if you have a 4K monitor and your budget clearly needs a rebalance towards a graphics card, the Ryzen 7 7800X3D (8/16; 4.2/5.0 GHz; 96 MB L3) for $350 or even the Ryzen 7 9700X (8/16; 3.8/5.5 GHz; 32 MB L3) for $300 will handle processor-based gaming challenges.

Video cards. Choosing a video card for advanced gaming configurations is now perhaps the easiest task. The starting models here are the GeForce RTX 5070 Ti 16 GB and Radeon RX 9070 XT 16 GB. Both demonstrate very similar performance on average, and the difference traditionally largely depends on the choice of the list of gaming projects. In some cases, a certain advantage will be on the side of the NVIDIA solution, in others - the current flagship of the AMD line will provide more fps. In general, very equal opponents with their own characteristics.

This time, as an "anchor" we use a video card from AMD - Radeon RX 9070 XT 16 GB. The basic models of this series are offered from $800. This is a very good price tag for its capabilities. The cost of the most affordable GeForce RTX 5070 Ti 16 GB starts at $950–1000. For this reason, they cannot be put side by side in the final table of recommended configurations.

Despite the significant difference in price, there are also arguments in favor of the more expensive GeForce RTX 5070 Ti 16 GB. If you expand the gaming pool as much as possible, there will be a few more cases of a certain advantage of the NVIDIA model. Also, the indicators after activating ray tracing look more predictable, especially when using resource-intensive modes with path tracing. Let's also take into account the larger number of projects with support for high-quality DLSS 4. Typically, the RTX 5070 Ti is also 20–40 W more economical in real games. Work tasks in which the GPU is used as an accelerator for calculations are often better optimized for NVIDIA video cards.

We recently looked at how the system works on the Ryzen 7 9800X3D + GeForce RTX 5070 Ti 16 GB while testing the updated graphics card for the BTF platform.

On the other hand, $200 is always $200, and if the priority is the gaming capabilities of the video card and there is no personal bias towards graphics from AMD, the Radeon RX 9070 XT 16 GB will be a very good opportunity to reduce the budget for the purchase of a powerful platform. In addition, the developers recently updated the drivers, adding support for FSR 4 to most projects with FSR 3.1 (DX12). There will be an opportunity to experiment with smart scaling and evaluate the qualitative changes and fps gain.

As for future graphics plans, we also expect SUPER variants from NVIDIA in this class of video cards. According to previous leaks, the GeForce RTX 5070 Ti SUPER will also use 3 GB chips, which will increase the total memory capacity from 16 GB to 24 GB. Similar transformations are also expected for the GeForce RTX 5080 SUPER. In addition, the effective frequencies of the GDDR7 chips will be further increased for the older model, so the memory bandwidth will increase from 960 GB/s to 1024 GB/s. Changes in the nominal frequency formulas of the GPU are also possible. Upgraded versions of the video cards will probably be presented at the same CES 2026, which will take place early next year.

AMD is already actively working on a more global update of its graphics solutions. Future processors based on the RDNA 5 architecture for desktop video cards have even received the code name Alpha Trion, which comes from the Transformers universe. It is expected that within the next line of video cards AMD will return to the category of graphic "superheavyweights" and will even be able to impose a fight on the conditional future GeForce RTX 6090. Current market indicators require extraordinary efforts from developers, but we hope that the company will be able to correct the situation. Healthy competition has never been superfluous.