Don’t Buy a Desktop PC With One of Intel’s Newest Processors—Here’s Why

We independently review everything we recommend. When you buy through our links, we may earn a commission. Learn more›

The Answer

Advice, staff picks, mythbusting, and more. Let us help you.

Photo: Intel

Published April 30, 2021

Share this post

Every year for the past decade, Intel has released a new generation of its Core processors. And every year, we’ve recommended that people buy the newest version they can get—if you’re paying hundreds or thousands of dollars for a computer, you should get one that will feel fast and run all the apps you use for as long as possible. But Intel’s 11th-generation Core processors are a little different, and there are some models we don’t think you should buy.

Specifically, the 11th-generation Core i5, i7, and i9 processors that will be available in many desktop computers in the next couple of months are difficult to recommend because they are only a little faster than the 10th-generation processors they replace, and because they run much hotter and use much more electricity than either those 10th-generation processors or competing AMD Ryzen chips do. Here’s what you need to know about the problems with these processors, what you should look for instead if you’re shopping for a desktop PC, and why, in contrast, we believe Intel’s 11th-generation laptop processors are safe to buy.

Hotter, more power-hungry desktops

To understand why these 11th-generation desktop processors are having problems, you need to know a little about how the processors in computers, tablets, phones, and game consoles get better over time. First, there’s the chip’s architecture, or how it has been designed—a processor is structured a bit like the blueprint of a house, with processor cores, cache memory, and blocks for playing 3D games or high-definition video files all laid out in a precise arrangement. And then there’s the manufacturing process, or how the chip is physically constructed in a chip maker’s factory.

This illustration is abstracted, but it’s more or less how a modern processor looks—it’s one solid chunk of silicon, with different pieces of the chip dedicated to different tasks. Illustration: Intel

Those two concepts are deeply intertwined. One way to make a processor faster is by adding more transistors to the design—a transistor is the basic building block of a computer processor, and the more of them you have, the more your processor can do. The transistor count of a typical desktop computer processor has increased from tens of thousands in the late ’70s to billions today. As you use the computer, those transistors are all being switched on and off constantly, which requires power, which in turn produces heat. So all else being equal, a processor design with more transistors requires more electricity to run and a bigger fan to cool.

But newer manufacturing processes make transistors smaller, which generally reduces the amount of power required to switch them on and off. That way, processor designers can add more transistors to make a processor architecture faster without worrying about making it physically larger or more power-hungry. If you’ve ever wondered why a MacBook Air you can buy today is faster, smaller, thinner, and lighter than a MacBook Air from a decade ago, that’s one major reason. ¹

So what happened to Intel’s latest desktop chips? Compared with the 10th-generation chips, the 11th-generation processors have an updated architecture but not a newer manufacturing process. This means that they can be faster sometimes, since Intel has added more transistors to their design. But each of those transistors requires the same amount of electricity as those in 10th-generation processors, and as a result, the 11th-generation processors run hotter and are more difficult to cool down. And because processors are designed to slow down (or “throttle”) when they get too hot to avoid burning themselves out, that increased heat can often cancel out whatever speed improvements Intel might have achieved by updating the processors’ architecture in the first place.

What you should buy instead

Intel’s 10th-generation desktop processors are still widely available, and they still perform reasonably well for most tasks, including gaming, professional photo and video editing, 3D modeling, and other tasks that benefit from a lot of processor power. And if you just need a basic desktop for editing documents and spreadsheets, browsing the web, and chatting on video calls, the 10th-generation Core i3 processor is an excellent value.²

We also like desktop processors from AMD, Intel’s biggest competitor in computer processors. Ryzen 5, Ryzen 7, and Ryzen 9 processors from the Ryzen 3000, 4000, and 5000 series are all as good as or better than Intel’s processors in both performance and power use (quite a bit better, once you start comparing Ryzen 7 and Ryzen 9 chips against the Intel Core i7 and i9 lineups). But AMD is a smaller company, and it has been a victim of its own success—AMD Ryzen systems are often harder to find and go out of stock more quickly than Intel PCs. Ryzen 5000 processors in particular are worth waiting for if you can get them, especially if you’re buying a PC for gaming or professional work such as video editing, coding, or designing 3D models. Just know that they have a reputation for being hard to find in an industry where currently everything is hard to find.

11th-generation laptop processors are good, actually

Our problems with Intel’s 11th-generation Core processors for desktops don’t extend to the company’s 11th-generation laptop processors, which are completely different chips despite sharing that Core name and generation. (Perhaps to distinguish between the two, 11th-generation Core laptops are often sold with Intel’s Evo branding instead, which literally downplays the “Core” by putting that word under the “Evo” in tiny print.) These chips offer a maximum of four processor cores, down from six or eight cores in the desktop chips, which means they aren’t as fast on some heavy-duty tasks like editing videos or playing games. But they generally offer a big step up in performance from their 10th-generation counterparts, they work great for everyday computing tasks such as browsing and editing documents and photos, and they offer excellent battery life in the laptops we’ve tested.

And although these processors were designed for laptops, they do appear in some desktops, mainly all-in-one PCs and mini desktop computers. We wouldn’t recommend those kinds of computers if you’re a professional photo editor or if you’re looking for a high-end gaming PC, but they’re great for everyday web browsing, video chatting, working from home, remote schooling, and less-intensive games like Fortnite (or older ones like Fallout 4).

If you’re shopping for a desktop and you need to be able to tell what kind of chip it has inside, you can look at Intel’s (admittedly bewildering) model numbers to distinguish the 11th-generation laptop chips from the desktop ones. The laptop chips have a four-digit model number followed by the letter G and another number, as in i5-1135G7 or i3-1115G4. The desktop chips have five-digit model numbers that are sometimes followed by one or two letters, as in i5-11400, i5-11600K, or i7-11700KF.

What if you need to buy an 11th-gen desktop anyway?

An ongoing shortage of silicon chips has made buying pretty much any piece of technology more difficult and expensive than it was even a few months ago, and that’s likely to be true throughout most of 2021. So what do you do if you need a desktop PC today, and one carrying an Intel 11th-generation chip is your only option?

If you’re in this position, the 11th-generation Core i5 processors are the least bad of the lot. They do use more power than 10th-generation i5 processors or AMD’s Ryzen processors, but they’re reasonably affordable, their six processor cores offer good-enough performance for graphics-intensive games, and they don’t emit so much heat that it will cause major problems in the long run. The 11th-generation Core i7 processors run hotter and use even more power, but their two extra processor cores do at least offer a noticeable speed increase for high-end video editing or 3D drafting apps—you wouldn’t notice the difference if you were just browsing or editing documents, though.

Regardless of the kind of work you’re doing, you should completely avoid the 11th-gen Core i9 models, which cost a lot more than the Core i7 versions and use more power without offering appreciably better performance. Early reviewers have even had problems with crashing and instability with the Core i9 processors (though we expect those issues to be fixed once Intel and the PC makers have had some time to work the bugs out).

Footnotes

1. Sometimes, chip makers choose to keep a processor’s architecture more or less the same while upgrading the manufacturing process. This results in a chip that performs the same but requires less power and heat and can be cooled with fewer fans or a smaller heat sink. This is why you often see new “slim” versions of video game consoles a few years after the originals were released.
Jump back.

2. There will be no 11th-generation Core i3 desktop processors; Intel will continue to sell 10th-generation Core i3 chips for budget systems.
Jump back.

Further reading

• The Best Laptops for Video and Photo Editing

  by Dave Gershgorn

  Photographers and video editors on the go need a powerful laptop with good battery life, and the 16-inch MacBook Pro is almost always the best tool for the job.

• You Shouldn’t Still Be Buying Intel-based Macs

  by Andrew Cunningham and Wirecutter Staff

  Apple announced it is transitioning new Macs from Intel processors to its own Apple silicon. We have advice for when—or if—you should buy a new Mac.

Wirecutter is the product recommendation service from The New York Times. Our journalists combine independent research with (occasionally) over-the-top testing so you can make quick and confident buying decisions. Whether it’s finding great products or discovering helpful advice, we’ll help you get it right (the first time).

• About Wirecutter
• Our team
• Staff demographics
• Jobs at Wirecutter
• Contact us
• How to pitch

• Deals
• Lists
• Blog
• Newsletters
• Make a Plan: Moving

Dismiss

Intel 11th Gen Core i5-11400F – 6 Cores & 12 Threads, 4.4 GHz Maximum Turbo Frequency, Dual-Channel DDR4-3200 Memory, 12MB Cache Memory, LGA 1200 Processor

Delivery options and delivery speeds may vary depending on the location.

Login to manage addresses

---

OR

Country

AfghanistanAlbaniaAlgeriaAmerican SamoaAndorraAngolaAnguillaAntigua And BarbudaArgentinaArmeniaArubaAustraliaAustriaAzerbaijanBahamasBahrainBangladeshBarbadosBelarusBelgiumBelizeBeninBermudaBhutanBoliviaBosnia And HerzegovinaBotswanaBrazilBritish Virgin IslandsBruneiBulgariaBurkina FasoBurundiCambodiaCameroonCanadaCanary Islands, TheCape VerdeCaribbean NetherlandsCayman IslandsChadChileChinaColombiaComorosCook IslandsCosta RicaCroatiaCubaCuracaoCyprusCzech RepublicDemocratic Republic Of The CongoDenmarkDjiboutiDominicaDominican RepublicEast TimorEcuadorEgyptEl SalvadorEquatorial GuineaEritreaEstoniaEthiopiaFalkland IslandsFaroe IslandsFijiFinlandFranceFrench GuianaFrench PolynesiaGabonGambiaGeorgiaGermanyGhanaGibraltarGreeceGreenlandGrenadaGuadeloupeGuamGuatemalaGuernseyGuineaGuinea-BissauGuyanaHaitiHondurasHong Kong, ChinaHungaryIcelandIndiaIndonesiaIrelandIsraelItalyIvory CoastJamaicaJapanJerseyJordanKazakhstanKenyaKiribatiKosovoKuwaitKyrgyzstanLaosLatviaLebanonLesothoLiberiaLiechtensteinLithuaniaLuxembourgMacao, ChinaMacedoniaMadagascarMalawiMalaysiaMaldivesMaltaMarshall IslandsMartiniqueMauritaniaMauritiusMayotteMexicoMicronesiaMoldovaMonacoMongoliaMontenegroMontserratMoroccoMozambiqueMyanmar [Burma]NamibiaNauruNepalNetherlandsNew CaledoniaNew ZealandNicaraguaNigerNigeriaNiueNorthern Mariana IslandsNorwayOmanPakistanPalauPanamaPapua New GuineaParaguayPeruPhilippinesPolandPortugalPuerto RicoQatarRepublic Of The CongoRomaniaRussiaRwandaRéunionSaint HelenaSaint Kitts And NevisSaint LuciaSaint Vincent And The GrenadinesSamoaSan MarinoSaudi ArabiaSenegalSerbiaSeychellesSierra LeoneSingaporeSlovakiaSloveniaSolomon IslandsSouth AfricaSouth KoreaSpainSri LankaSudanSurinameSwazilandSwedenSwitzerlandSyriaSão Tomé And PríncipeTaiwan, ChinaTajikistanTanzaniaThailandTogoTongaTrinidad And TobagoTunisiaTurkeyTurkmenistanTurks And Caicos IslandsTuvaluU. S. Virgin IslandsUgandaUkraineUnited Arab EmiratesUnited KingdomUnited StatesUruguayUzbekistanVanuatuVenezuelaVietnamZambiaZimbabwe

City

BerlinAach B TrierAach, HegauAachenAalenAarbergenAasbuettelAbbenrodeAbenbergAbensbergAbentheuerAbrahamAbsbergAbstattAbtsbessingenAbtsgmuendAbtsteinachAbtswindAbtweilerAchbergAchernAchimAchslachAchstettenAchtAchtelsbachAchterwehrAchtrupAckendorfAddebuellAdelbergAdelebsenAdelheidsdorfAdelmannsfeldenAdelschlagAdelsdorfAdelsheimAdelshofen, Kr FuerstenfeldbruckAdelshofen, MittelfrAdelsriedAdelzhausenAdenauAdenbachAdenbuettelAdendorfAderstedtAdlersteigeAdligstadtAdlkofenAdmannshagen-BargeshagenAdorf/Vogtl.AebtissinwischAerzenAffalterbachAffingAffinghausenAfflerAgathenburgAgethorstAglasterhausenAhamAhausAhausenAhlbeckAhlden (Aller)Ahlefeld-BistenseeAhlenAhlerstedtAhlsdorfAhlstaedtAhnatalAhnebyAhnsbeckAhnsenAholfingAholmingAhorn, BadenAhorn, Kr CoburgAhorntalAhrbrueckAhrensboekAhrensburgAhrensfeldeAhrenshagen-DaskowAhrenshoeftAhrenstedtAhrenvioelAhrenvioelfeldAicha Vorm WaldAichachAichelbergAichenAichhaldenAichstettenAichtalAichwaldAidenbachAidhausenAidlingenAiglsbachAilertchenAindling

Intel Core i5-11400H Review – Mobile Processor Test

Intel Core i5-11400H Review – Mobile Processor Test

The Intel Core i5-11400H is Intel’s new H-series processor, the successor to last year’s Core i5-10400H. The model is a direct competitor to the AMD Ryzen 5 5600H and is used in many high-performance and gaming laptop series.

Core i5-11400H is a Tiger Lake-h55 architecture processor. It comes with all the architectural updates: 10nm SuperFin process, improved cache sizes, clock speed and efficiency, and new features like PCIe 4.0 support. It has an integrated Xe GPU, albeit a much cut down one from the top Tiger Lake models.

Unlike the 8-core i7 and i9 models, the Intel Core i5-11400H has 6 cores, 12 threads, and 12MB L3 cache. The base clock is 2.7GHz at 45W TDP or 2.2GHz at 35W TDP. The maximum clock speed in Turbo mode is 4.5 GHz for a single core or 4.1 GHz when all 6 cores are loaded.

The Intel Core i5-11400H has integrated UHD graphics, but they are weak enough that laptops with this processor come with a discrete graphics card to provide adequate performance.

Tested using an XMG Core 17 laptop shipped with an Intel Core i5-11400H and the new NVIDIA GeForce RTX 3050 discrete graphics card for laptops running at 80W to 95W. It comes with 16GB of RAM running at 3200MHz and a 1024GB SSD. Testing was done with the default processor TDP set to 45W for 11th Gen Intel H processors.

Testing Intel Core i5-11400H

• Cinebench R20
• Handbrake
• Blender
• Code compilation
• MATLAB
• Microsoft Excel
• 7-Zip
• Adobe Acrobat PDF
• Adobe Photoshop
• Adobe Premiere
• Resident Evil 2
• Borderlands 3
• Rainbow Six Siege
• Cyberpunk 2077
• Hitman 3
• Core i5-11400H vs. Core i7-11800H
• Core i5-11400H vs. Ryzen 5 5600H
• Core i5-11400H vs. Ryzen 5 4600H
• Conclusion

Cinebench R20

The Core i5-11400H performs well in the multi-threaded Cinebench R20 test. Its performance is slightly lower than the latest generation Core i9-10980HK and better than AMD’s Ryzen 5 4600H, but it’s still slightly behind the newer Ryzen 5 5600H.

The i5-11400H ends up being 8% slower than the Ryzen 5 5600H and 19% slower than the i7-11800H, indicating that Intel’s eight-core processor is decently efficient despite the same power cap of 45 Tue

Since the Core i5-11400H is limited to 4.5 GHz per core, it’s not surprising that it performs weaker than other Tiger Lake processors in the single-threaded test, since this turbo frequency is the lowest. On top of that, the i5-11400H only has half the L3 cache compared to the i7-11800H, so overall performance is down 4% in this test. However, this is still enough for the i5-11400H to outperform the Ryzen 5 5600H by 5%.

Handbrake

In Handbrake, i5-11400H performance is similar to what was shown in Cinebench’s multi-threaded test. The i5-11400H is slightly faster than the Ryzen 5 4600H and slightly slower than the Core i9-10980HK. It is 5% behind the Ryzen 5 5600H and 13% behind the Intel Core i7-11800H.

Blender

Blender i5-11400H doesn’t work as well as Handbrake, it just matches Ryzen 5 4600H. We get 8% lower performance than the Ryzen 5 5600H and 17% lower performance than the Core i7-11800H. Therefore, the i5-11400H is not the best choice for such heavy multi-threaded tasks.

Compiling code

However, when compiling code, things are somewhat different. The i5-11400H is slightly ahead of the Ryzen 5 5600H here and is on par with the latest generation octa-core processors. Although this is a good result compared to the Core i5 and Ryzen 5, the performance is still 19% lower than the Core i7-11800H.

MATLAB

Matlab is another good result for the Core i5-11400H. It is only slightly behind the other models in the lineup and outperforms all Ryzen 5000 processors in this test.

Microsoft Excel

The Microsoft Excel Core i5-11400H suffers greatly from L3 cache shrinking from 24MB in higher end processors to just 12MB in this model. As a result, performance dropped by more than 30% compared to the i7-11800H. Although it is still competitive with the Ryzen 5 5600H and only 3% behind it.

7-Zip

7-Zip compression also suffers from the i5-11400H’s reduced cache size, combined with lower Turbo frequencies compared to other Intel processors. As a result, the performance is still better than the Ryzen 5 5600H, although the lead is only 4%, however, it cannot be compared with octa-core processors such as the i7-11800H or the Ryzen 7 5800H.

In decompression, the i5-11400H lags behind AMD Ryzen processors. This is a very strong test for AMD, so the performance of the processor from Intel is even slightly lower than the Ryzen 5 4600H, and 17% lower than the Ryzen 5 5600H. It’s also 27% weaker than the i7-11800H, which benefits from more cache and cores and higher clocks.

Adobe Acrobat PDF

When exporting Acrobat PDF, the Core i5-11400H performs very similarly to the Ryzen 5 5600H and Ryzen 7 5800H, we are talking about a few percent anyway. However, in this single-threaded test, performance is 6 percent slower than the 11800H.

Adobe Photoshop

In Adobe Photoshop, the Core i5-11400H delivers almost the same performance as the Ryzen 5 5600H.

Adobe Premiere

The Adobe Premiere Ryzen 5 5600H and Core i5-11400H work in roughly the same way when exporting.

Shadow of the Tomb Raider

In Shadow of the Tomb Raider, two laptops with RTX 3050 graphics showed similar performance, although the Intel Core i5-11400H model was slightly slower. The minimum FPS was 6% lower on the Core i5 laptop, but the average FPS was only 2% behind.

Resident Evil 2

In Resident Evil 2 again, there is not much difference between the i7-11800H and the i5-11400H. At best, we see a 4% difference when comparing minimum FPS.

Borderlands 3

In Borderlands 3, the laptop with the i5-11400H was 3% slower in average FPS and 6% slower in minimum FPS compared to the older model.

Rainbow Six Siege

In Rainbow Six Siege with an RTX 3050 graphics card, both processors produced the same result.

Cyberpunk 2077

Cyberpunk 2077 had a similar situation – there is no difference in performance between processors.

Hitman 3

Hitman 3 had the biggest FPS difference, with the i5-11400H being 6% slower on average FPS than the i7-11800H and 15% slower on minimum FPS.

Comparison of Intel Core i5-11400H and Intel Core i7-11800H

On average, the i5-11400H is less than 2% slower than the i7-11800H, so you won’t feel much difference in games.

However, in performance tests, the difference between Core i5-11400H and Core i7-11800H is higher.

In multi-threaded tasks, the Intel Core i5-11400H is over 15% slower due to fewer cores. The difference becomes especially strong in applications where the cache size is important, since the Core i5-11400H has 2 times less than the i7.

Intel Core i5-11400H vs. AMD Ryzen 5 5600H

The real battle is between the i5-11400H and the Ryzen 5 5600H, which are almost on par with each other. In longer, multi-threaded workloads, the AMD processor has an advantage, but the difference between them is less than 10%. In low-threaded tasks, the processors are close to each other.

Intel Core i5-11400H vs. AMD Ryzen 5 4600H

The Core i5-11400H is well ahead of AMD’s previous generation Ryzen 5 4600H.

Conclusion

Overall, the Intel Core i5-11400H did a great job and is probably the best mobile processor Intel has made in quite some time. This is a very competitive processor, on par with the AMD Ryzen 5 5600H.

The Ryzen 5 5600H has an advantage at 45W in multi-threaded workloads, but in most other workloads the i5-11400H is either close to or better than the 5600H. This advantage for Intel increases with more power, and the i5-11400H is able to close the gap.

The Intel Core i5-11400H is also a great platform for gaming as the performance difference compared to higher end processors is negligible.

In addition to performance, the Core i5-11400H has an excellent platform and feature set. Intel offers 20 PCIe 4.0 lanes, while AMD is stuck with PCIe 3.0, Intel also has Thunderbolt 4 support, which may be a key selling point for some.

The Core i5-11400H also has disadvantages. The built-in graphics card is too weak: almost two times inferior to that of the AMD Ryzen 5 5600H. Another disadvantage is the price. In general, laptops with an Intel processor are 5-10% more expensive than laptops with AMD.

Gaming laptops based on Intel Core i5

Intel Core i5-12400 test. Comparison with Core i5-11400 and Ryzen 5 5600X – i2HARD

Examining the performance of the Intel Core i5-12400 in comparison with competitors. Stock and overclocked tests, DDR4 and DDR5.

Many people remember the i5-11400 as a productive, but hot budget employee, which makes you think about choosing a motherboard because of its high power consumption. i5-12400, which came to replace, must provide not only 19% higher IPC, but also higher energy efficiency due to a smaller process technology. It doesn’t have energy-efficient cores like the 12600K has, and the frequency is slightly lower than its predecessor, but something tells us that it will be popular. It’s i5.

To see the progress over the generation, it would be logical to compare the new product with the i5-11400, and the Ryzen 5 5600X will fight from the red camp.

Contents:

• Test bench
• Stock tests
• Synthetic tests
• Stock cooling tests
• Tests in games
• Overclocking processors and memory
• Synthetic tests
• Cooling during acceleration
• Tests in games
• Conclusion

Test bench

To see the progress over the generation, it would be logical to compare the new product with the i5-11400, and the Ryzen 5 5600X will fight from the red camp.

• Processor #1: Intel Core i5-11400
• Processor #2: Intel Core i5-12400
• Processor #3: AMD Ryzen 5 5600X
• Processor #4: Intel Core i5-12600K
• Motherboard #1: ASUS ROG Maximus XIII Hero
• Motherboard #2: ASUS TUF Gaming Z690-Plus WIFI D4 9002 0
• Motherboard #3: ASUS ROG Maximus Z690 Hero
• Motherboard #4: ASRock B550 Taichi Razer Edition
• Video Card: Palit GeForce RTX 3080 Ti GameRock OC
• DDR4 RAM: G.SKILL Trident Z F4-3200C14D-32 GTZ 2×16 GB
• DDR5 RAM: T-FORCE DELTA RGB FF3D532G6200HC38ADC01 2×16 GB
• Cooling system: ARCTIC Liquid Freezer II-360
• Storage: Crucial MX500 2 TB
• Power supply: Deepcool DQ8 50-M-V2L
• Body: Open Stand
• Operating system: Windows 10

We usually use DDR5 as an additional option both in stock and overclocking, but this time we will touch on another important topic. Timings in XMP. Here is the same XMP, but on different platforms. See the difference?

For the i5-12400, ASUS TUF Gaming set almost all tertiary timings higher than the Hero for the 11400. And it’s not that the Hero is an overclocker motherboard, so this is expected, no. Even the simplest board for past generations will set similar timings. Checked. And not a specific TUF or BIOS is to blame for the inflated thirds. On ASUS Prime B660M-A, everything is the same.

So who is to blame? There are two possible answers to this question.

Either ASUS motherboards with new chipsets automatically set timings too high, or all motherboards of the new generation are guilty of this. Unfortunately, there are few people on the Internet who demonstrate XMP timings to the public, so I could not find other examples.

Based on all of the above, in stock we will consider the i5-12400 not only with its own XMP, but also with the identical XMP of the 11400. It will be useful to find out how important these timings are, and we will make sure if they are not like that everywhere.

Stock tests

Synthetic tests

AIDA64 tells us that we shouldn’t have bothered. The speed of writing, copying, as well as the delay, are identical. Unless the reading speed has grown, but only a little. But what has grown decently is the memory latency compared to the previous generation. The difference in 11 nanoseconds is hard not to notice. The 12400 came close to Ryzen in this respect.

In CPU-Z, the novelty outperforms its predecessor by 19% in the single-thread test, and with all the cores involved, the frequency of 12400 is lower, respectively, the gap is smaller.

In Cinebench R23, Ryzen lost. If before that it was breathing in the back of 12400 in the multi-stream test, now it is only slightly faster than 11400.

GeekBench 5 responded poorly to the new microarchitecture. Surely, a large latency had a hand in the result. At the same time, the tweaked XMP made itself felt, increasing the result by another 2.5%.

The effect of tertiary timings in Adobe Premiere Pro is especially noticeable. Just think, they gave almost the same increase as the new processor microarchitecture.

Stock cooling tests

If the boxed cooler coped with 12100 even in stress tests, the situation is different with 12400. For literally a minute, while the radiator warms up, this buzzer is able to divert almost 100 watts, but after this period, even 85 watts turn out to be an insurmountable obstacle. At the same time, a simple cooler with 4 heat pipes can easily cope with i5. And the water cooling system is redundant.

As for games, even with the FPS limit, due to which the processor consumes less than 60 watts, the temperature still stays around 80 ° C – not without the help of the video card, of course, splashing out its 200 watts. At the same time, do not forget about the feeling of the included hair dryer, forgotten in the system unit. In general, it is better to spend money on a better cooler.

An important issue is the behavior of the processor when using AVX-512 instructions. The already non-cold i5-11400 with the power consumption limits removed, when they were activated, requested almost 1.5 Volts for itself, after which it went into throttling. Fortunately, the algorithms in this case, sacrificing a couple of hundred MHz, achieve stable 90°C, not 100, but 12400 at the same time in complete peace.

Even Blender’s 11400 draws a hefty 1.3v, causing its power consumption to reach 140 watts. Due to the huge die area, the temperatures under the usual tower cooler remain normal, but the 5600X and i5-12400 allow you not to worry at all.

Tests in games

There were enough synthetics, it’s time to move on to games.

Call of Duty: Warzone, eSports settings, DLSS – ultra performance, not to be limited by the graphics card. As always, all tests were repeated many times and averaged. Recording was carried out with a capture card from a third-party computer. Something is not impressive at 12400, is it? With equal timings, it is a little faster, but without this refinement, it loses to its predecessor, which is generally nonsense. However, the statistics of rare and very rare events at 11400 are well behind. This game is very random in terms of these indicators, so let’s go watch the next one.

Cyberpunk, still old version 1.31. Graphics preset: ultra ray tracing, RT reflections off, DLSS – ultra performance, high crowd density. The situation is similar. With automatic timings, 12400 is worse than the opponent on the left in all respects, and with the same thirds as 11400, which gives an 8% increase, it is barely ahead. According to benchmarks, we saw a 20% increase, and given the greatly increased volumes of caches of different levels, this number should have been repeated in games. But we have reality. And in it, 12400 is worse than the i5 of the previous generation. At the same time, Ryzen keeps at the level of 11400, especially not inferior or superior to anyone.

Far Cry 6, ultra preset, DXR shadows and reflections on, HD textures off, FSR performance. It looks like you shouldn’t judge the novelty by two games. Even with higher timings, it is faster than 11400, and with the same thirds it becomes 4% faster. But wait. It turns out that these timings gave again more than the difference in generations. Even in total, the increase is around 7%, which is also small.

Shadow of Lara. Highest preset, resolution modifier – 20%. Ryzen, as usual, grabbed more FPS at the beginning of the scene due to its L3 cache, and when the camera drops to the market, it no longer stands out from the crowd. But he is not the only one. 12400 with automatic timings on minimum FPS is worse than 11400. 128 vs 131 FPS. If the thirds were identical, then yes, he would be faster, because they increase the FPS in this game by 6%. But again, we are not talking about how far ahead he is, but whether he is ahead at all. In terms of energy efficiency, everything is gorgeous. Energy consumption was reduced by a third. This is partly due to less stress, but still. And with the same TDP as the 5600X, it’s easier to remove heat from the 12400.

StarCraft II, all settings maxed out. Here you go. Another thing. Finally, we see a good increase. 13% on average FPS and a little less on rare and very rare events. The importance of tertiary timings is minimal here, it is even at the margin of error. StarCraft has always loved big caches and fast cores, which is what we’re seeing. Ryzen’s 32 MB L3 cache made itself felt, adjusting it under the 12400 sideways.

Total War Saga: Troy, ultra preset, crowd density and grass detail – extreme, resolution scaling and anti-aliasing – minimum. Unlike the first 4 games in comparison, Troy responds very cheerfully to the new Intel microarchitecture. So it was in the comparison of the 11th and 10th generation, so it is now. It is much closer to synthetic benchmarks than to gaming ones. Despite the higher latency, low-frequency memory and the same number of cores, the 12400 outperforms the i5 of the previous generation by 11%, and by less than 1 and 0.1% by all 15. At the same time, equalizing the tertiary timings gave only 4%, which is less than in the same Cyberpunk or Lara.

Well, where without a counter-strike? Minimum settings, maximum textures. Record growth in a generation. Almost 20%, but you can’t argue with Ryzen’s L3 cache, it outperformed everyone here and decently. Of course, we are talking about transcendental FPS. Fortunately or unfortunately, this difference is not felt by a normal person.

However, this is not the only game that responds well to L3 cache and fast cores. Many MOBA games have similar behavior.

CPU and memory overclocking

The 5600X is the same. The memory with our sample is stable at a maximum of 3800 MHz with the first timing of 14. There are rumors of a new revision, the overclocking potential of which is much higher, but we do not own one. The cores were overclocked using Curve Optimizer.

The

i5-11400, with minimal overclocking on the bus, took 3690 MHz from memory in Gear 1 mode with the first timing of 14. And even for such a frequency, sufficiently high IO and SA voltages were required for a stable start.

i5-12400, on the contrary, is limited by low voltage on SA in region 900 mv. It is not adjustable even on the Z chipset, and this was only enough for 3600 MHz in Gear 1 mode, which is even worse than 11400. Therefore, even memory with XMP 3600 MHz can be unstable on processors with a locked multiplier if a less successful sample comes across.

Well, what about without overclocking on the bus? We could not ignore such an opportunity on the i5-12400. 5 GHz turned out to be the maximum for our liquid cooling system. DDR5 on Hynix chips took 6250 MHz with the first timing of 28. Moreover, overclocking the i5 on the bus makes much more sense than i3.

Let’s add the i5-12600K to the comparison with almost identical overclocking. If we recall the test where we turned off its energy-efficient cores, then in most games there was no FPS loss, and in some places it even increased. Therefore, buying a 12400 with a more expensive motherboard and memory with the money saved may make sense.

Synthetic tests

AIDA64 recently had a micro update, but even the new version overestimates the performance by the BCLK overclocking percentage. With memory overclocking, the 12400 has closed the latency gap, but the difference is still around 8 ns, which makes it closer to Ryzen.

Naturally, for synthetics 12600K will be preferable. 4 power efficient cores give it a 28% boost in the CPU-Z multi-thread test.

Cinebench demonstrates the ability to parallelize no less than the previous benchmark. Overclocking cores by 25% in the i5-12400 gives the same increase, and Ryzen, due to the PBO setting, catches up with the stock result of 12400.

In Geekbench, overclocking on the bus in combination with DDR5 gives a 28% increase in the new product compared to its stock version on a tuned DDR4. Ryzen is not far behind, with the 11400 trailing by just over 10% when using all cores.

If earlier 12400 was equal in rendering speed to 5600X, now the second one turned out to be slightly faster. Overclocking the bus in combination with DDR5, of course, speeds up this business by another 28%, but the cost of such a system, you know, is disproportionately higher. And 12600K with DDR4 would still be faster. Judging by our past tests, he managed in 231 seconds.

Cooling during acceleration

There is something to add about cooling. As you can see, the 12600K with small cores disabled at the same frequency and voltage generates 25 watts more heat, but the temperature of the cores is 6°C lower. This can be partly justified by the fact that its cores are most likely scattered over a ten-core crystal, and in 12400 they are bunched together, which is why it is hot. But there is also a suspicion about using thermal paste instead of solder.

Tests in games

Back to games. Even with less memory overclocking and no bus overclocking, the 12400 outperforms its predecessor in COD: Warzeone. Only 6% on average FPS, but at least not behind. Overclocking on the bus and tuned DDR5 give it another 18% additional frames. However, Ryzen is not far behind this option – only 7%. As for the 12600K, it’s not far off. The average FPS is almost the same, and the random statistics of rare and very rare events in an online match are still higher on a ten-core.

Cyberpunk 2077. We have mentioned more than once the great memory hunger of the 12th generation compared to the previous ones. However, we do know the culprit – inflated tertiary timings in XMP. Now that we can compare with an alternative XMP identical to that of the 11400, is this statement still valid? And yes, it’s relevant. If 11400 accelerated by only 7% from overclocking, which is actually very little, then 12400 received a 14% increase from less overclocking, when compared with the XMP modified by us, or 24%, when compared with the native one. Ryzen increased by 19%.

Far Cry reacted differently. The 11400 got 11% of dual rank overclocking, the 12400 got 13% and 17% over different XMPs, and the 5600X got 13%. Very close results in contrast to Cyberpunk. And not such a big difference from the previous generation, if you look at it that way. In terms of FPS, we have an 8% advantage of 12400 over 11400. 5600X is between them, and overclocking on the bus somewhat leveled DDR5, reducing the gap from DDR4 to a small 9%. At the same time, 12600K on Windows 10, after a bunch of updates, still has problems in Far Cry. It serves as a reminder to us that the 12600K will somehow force you to upgrade to a new version of Windows.

In SOTTR, the i5-12400 received a 22% increase in overclocking compared to its XMP and 14% compared to the modified one. Completely different driving experience. Is not it? At the same time, 11400 received almost the same 13% gain. And considering that the frequency of its cores was increased by 2.5%, and the memory frequency is slightly higher in overclocking, it turns out that the 12th generation still consistently gets more from memory tuning. Ryzen, by the way, also got 14%, despite the fact that its large cache in Lara reduces the gain from overclocking. The 12600K with small cores off was faster in Lara than it was with them on, so the overclocked 12400 followed suit.

In StarCraft, DDR5 again downplays the gain from overclocking 12400 cores to 15% compared to the DDR4 variant. An interesting situation emerges. Games that are highly sensitive to the frequency of cores, as a rule, do not parallel well (Troy does not count), and memory bandwidth is not so important to them. Therefore, the current DDR5 performs worse in them than DDR4. At the same time, AAA projects treat it, if not well, then certainly not bad. However, the gain from overclocking the cores in such games does not give much. It turns out that due to the fact that overclocking 12400 over the bus is currently possible only on some boards with DDR5, such a combination cannot manifest itself in games in all its glory.

Even in Troy, which adores DDR5, the overclocked 12400 on the bus is ahead of its much more budgetary version by 27%. Pretty good, of course. But if you think about it – just one game, which is the exception that proves the rule. Yes, and 12600K in this game, you know, much more productive.

In Contra, Ryzen was unbeatable. Thanks to overclocking through Curve Optimizer, the frequency of its cores became 200 MHz higher, which allowed it to hold the lead, though only in terms of average FPS. The statistics of rare and very rare events are inferior even to 12400 with DDR4. We did not test the 12600K in this game when we disabled its cores, but now we are convinced that small cores are harmful in Contra.

Conclusion

What is the end result? And it turns out that in terms of FPS in stock, the new product is not far from 11400. And if we used only AAA projects in the test, then .. You understand. It is not entirely clear what kills the potential of Alder Lake so much. High memory latency? Too high timings or something else? This handbrake greatly hinders the fresh microarchitecture from showing itself. Recall the same AMD Ryzen. They have had a decent increase in games for the last two generations, and in software. Here, not always and not everywhere. Yes, on average, the 5600X came out faster, but you yourself understand that CS:GO had a very strong influence here, without it Ryzen would have dipped, and so would the 12400.

Prices at the moment are as follows: 12400F assembly is more expensive than 11400F assembly. Newer. A couple of thousand for the processor, a couple – for the motherboard. Total 4 thousand is money. And the FPS in half of the games is the same, if not worse. You can also take a fee on the H610 chipset, in some stores their cheapest representative costs 2 thousand cheaper than the cheapest B660, but it depends on you. We have already compared them, and for the 12400 with its low consumption, the difference will only be in memory overclocking. On the H610, the limit is 3200 MHz.

The new i5 has its own advantages, however. The processor is cooler, more energy efficient than its predecessor. In synthetics, the increase was noticeable, and this will affect everyday life. The same browser will be more responsive. So it makes sense to pay extra.

At the same time, the 5600X, which proved to be a cross between i5s in synthetics and approximately equal to 12400 in games, is quite expensive in our retail. Even if you take it from Aliexpress on sale with a promo code of thousands of commercials for 18, it still doesn’t look very profitable, if it’s not about upgrading from another Ryzen.

i5-12400 with overclocked DDR4 is 11% more cheerful than 11400. If you don’t drive the second bus, then this is another plus percent. If a more successful controller comes across and 12400 takes 3800 MHz, then this is another 2-3%. And 15% is a quite tangible increase without monitoring, worth the additional payment.

Overclocking on the bus and DDR5 gave an average of 19% more. And if not for Troy, such a bunch would be a drop faster than 12600K, at least in games. However, the 12600K has small cores that can take over the background load if you have a lot of it.