Intel Core Ultra 9 285K Review

At this time, we will lastly take the covers off Intel’s newest CPU technology, the “Core Extremely” collection, codenamed Arrow Lake. The Arrow Lake structure is new for desktop methods, utilizing a tile-based processor design, with the aim of shifting away from a big monolithic chip based mostly on the newest node and as a substitute splitting the CPU into blocks, or “tiles,” as they’re generally referred to.

This method permits Intel to develop the CPU cores and iGPU utilizing cutting-edge nodes, whereas different facets of the processor, akin to I/O interfaces, will be constructed utilizing an older, more cost effective node. This method is just like what AMD has been doing with Ryzen.

Since Arrow Lake represents such a radical change from the earlier Raptor Lake technology, we might spend a variety of time discussing the architectural and platform modifications, a few of which we have been protecting over time in anticipation for this launch. Nonetheless, since most of you’re right here for the benchmarks and precise efficiency outcomes, we’ll rapidly go over the flagship CPU we’re testing in the present day, contact on the brand new platform, after which dive into the graphs.

Meet the brand new Core Extremely 9 285K

The brand new flagship is the Core Extremely 9 285K. Sure, it isn’t the very best title, and sure, the ‘Extremely’ is pointless, however Intel is simply following the development – fashionable product names are usually unnecessarily lengthy and considerably uninspired, in order that field is checked.

The 285K options 8 P-Cores with 8 threads, as Hyper-Threading is not used. These P-Cores have a base frequency of three.7 GHz, boosting as much as 5.6 GHz, with a thermal velocity enhance of 5.7 GHz – leading to a 5% frequency discount in comparison with the 14900K.

There are additionally 16 E-Cores, every with 1 thread, as there is no such thing as a SMT assist right here. These E-Cores function at a base frequency of three.2 GHz, with a lift as much as 4.6 GHz, which is a 5% improve over the E-Cores within the 14900K.

In whole, there are 36 MB of L3 cache and 40 MB of L2 cache. Every P-Core will get 3 MB of L2 cache, whereas the E-Cores are grouped into clusters of 4, with every cluster sharing 4 MB of L2 cache. Lastly, the bottom TDP is 125W, with a max turbo TDP of 250W. Intel is pricing this at $590 per 1,000 items, so we count on an preliminary retail value of not less than $600.

It is also price noting that each one Ok-SKU fashions, that are the one ones introduced thus far, assist dual-channel DDR5-5600 UDIMM reminiscence or DDR5-6400 CUDIMM reminiscence. CUDIMM reminiscence features a small clock driver circuit immediately on the module, enabling extra exact timings required at greater reminiscence speeds.

All CPUs present 20 PCIe 5.0 lanes and 4 PCIe 4.0 lanes, together with an 8-lane Direct Media Interface 4.0 bus to the chipset. As ordinary, the Ok-suffix processors have an unlocked multiplier, making them preferrred for overclocking.

A New Socket

With Arrow Lake, Intel has moved away from the LGA1700 socket, switching to LGA1851. Regardless of the elevated pin rely, the brand new socket maintains the identical dimensions and cooler mounting gap spacing as LGA1700, making certain continued compatibility with current CPU coolers.

For testing, we used quite a lot of new motherboards, primarily the Asus ROG Maximus Z890 Hero and MSI MEG Z890 Unify-X. A considerable variety of BIOS updates have been wanted to realize the efficiency figures we’re about to current, so it is truthful to say the Arrow Lake assessment course of has been removed from easy.

Testing Notes

First, to be clear, we had very restricted time to organize this assessment, as Intel solely despatched us the assessment equipment three days earlier than the embargo lifted. Three days just isn’t sufficient time to run all of the assessments we would wish to, so some assessments can be lined in follow-up content material.

Since these CPUs can be out there for buy on October twenty fourth, we needed to have a assessment prepared to assist information your shopping for choices. Nonetheless, delays in receiving assessment samples turned out to be the least of our considerations, as Arrow Lake is presently a little bit of a scorching mess, a lot in order that we’re undecided the place to begin.

Let’s begin with Home windows 11. Lots of you understand that Home windows 11 24H2 was launched a number of weeks in the past, considerably bettering gaming efficiency for contemporary Ryzen processors in some circumstances. It even advantages thirteenth and 14th-gen Intel fashions. Nonetheless, when updating our information, we seen that Twelfth-gen Intel CPUs carried out noticeably worse on 24H2 in comparison with 23H2.

This problem wasn’t restricted to Twelfth-gen processors. We encountered critical efficiency and stability issues with Arrow Lake on 24H2. After a number of recent installs didn’t resolve the difficulty, we needed to revert to testing on 23H2. That is lower than preferrred, as we all know some video games carry out worse on the older model of Home windows 11.

Even with a recent 23H2 set up, we nonetheless skilled stability points with the 285K. Video games would often crash, as would sure purposes. The issue was worse when utilizing the DDR5-8200 reminiscence supplied for testing. This persevered even after quite a few BIOS revisions for 2 totally different motherboards.

We count on these teething points can be addressed rapidly, although we’re not sure why 24H2 precipitated so many issues. It is significantly odd as a result of Intel reportedly did all their benchmarking on 24H2 utilizing the Insider Preview construct from August and didn’t report any points. However, we have spoken with a number of reviewers who additionally encountered issues, even with 23H2, and board producers confirmed there are points with 24H2.

Take a look at System Specs

Earlier than we get into the benchmark information, listed here are the check specs and methods used. One factor to notice is that we benchmarked the 285K with each DDR5-7200 and DDR5-8200 reminiscence.

The 7200 CL34 reminiscence runs with a lot tighter timings in comparison with the DDR5-8200 CL40 reminiscence. If an software or sport is extra latency-sensitive, the 7200 reminiscence might carry out higher. Nonetheless, if bandwidth is the limiting issue, the 8200 equipment might need the benefit.

Now, let’s examine how the numbers shake out…

Productiveness Benchmarks

Cinebench Clock Conduct

First, let’s check out how the 285K behaves below load. For cooling, we’re utilizing the brand new MSI MAG Coreliquid I360, which was supplied in our assessment equipment and is particularly designed and optimized for the brand new Arrow Lake CPUs.

MSI has developed a novel bracket that shifts the chilly plate north, leading to a 3-degree discount in temperature because it higher targets the hotspot on these new CPUs. A pleasant function of this design is that the distinctive mounting bracket additionally supplies optimum protection for LGA1700, AM4, and even AM5 processors, making it a flexible, one-size-fits-all answer.

With the Coreliquid I360 put in, we loaded up the 285K with Cinebench and noticed a median clock frequency of 4.6 GHz on the E-Cores and 5.3 GHz on the P-Cores, all whereas staying inside the inventory 250W energy restrict. The CPU reached a peak core temperature of 84°C, which is properly under the 105°C TjMAX.

Cinebench 2024, Multi Core

Now for the benchmark graphs: the Cinebench multi-core efficiency appears very sturdy. We already know that E-Cores work properly in this sort of workload, based mostly on earlier fashions. The E-Cores within the 285K are based mostly on an up to date structure and run at barely greater clock speeds. Utilizing both DDR5-7200 or 8200 reminiscence, we noticed a rating of simply over 2,500 factors, making the 285K 14% sooner than the 14900K and seven% sooner than the 9950X.

Cinebench 2024, Single Core

After we take a look at single-core efficiency, the Arrow Lake P-Cores proceed to impress, with the 285K being 13% sooner than the 14900K and 6% sooner than the 9950X – one other nice end result.

Cinebench 2024 Energy Consumption

Subsequent, let’s focus on energy consumption. This a part of the assessment gave us some bother, however due to Steve from Players Nexus, we have been capable of catch a difficulty with the testing. The Asus ROG Maximus Z890 Hero has a novel design: as a substitute of delivering all of the CPU energy by the twin EPS12V rails, 4 of the vCore energy levels are related to the 24-pin ATX energy cable.

Initially, we have been puzzled as to why the 285K, with a 250W restrict, was drawing solely 196W. After Players Nexus alerted us to this, we confirmed that about 50-60W of energy was being equipped through the 24-pin ATX cable. This was later confirmed by Asus engineers.

After retesting with the MSI MEG Z890 Unify-X, we confirmed the proper EPS12V information, displaying the 285K drawing 258W throughout this check. Whereas that is an enchancment over the 14900K, it isn’t significantly spectacular, contemplating the 9950X was solely 7% slower however consumed 11% much less energy. Much more regarding is that the 7950X3D was simply 13% slower than the 285K however used an enormous 43% much less energy.

Nonetheless, in comparison with the extremely inefficient 14900K, the 285K represents a step in the correct course, consuming 17% much less energy whereas delivering 14% extra efficiency.

7-Zip File Supervisor Compression

In terms of compression efficiency, the 285K is stable, however not groundbreaking. It performs equally to the 14900K, 9950X, and 7950X. Whereas it isn’t an enormous leap ahead for a brand new technology, it is actually not a foul end result.

7-Zip File Supervisor Decompression

Nonetheless, decompression efficiency is kind of weak, largely as a result of removing of SMT. Consequently, when utilizing the identical frequency reminiscence, the 285K is 9% slower than the 14900K and a big 25% slower than the 9950X.

Blender Open Knowledge

The Blender Open Knowledge outcomes are good, although not distinctive, because the 285K matches the 9950X. Nonetheless, given the improved energy effectivity, that is nonetheless a constructive end result total.

Corona 10 Benchmark

Within the Corona 10 Benchmark, the 285K performs properly, although it is 9% slower than the 9950X, regardless of displaying a 17% efficiency improve over the 14900K.

Adobe Photoshop 2025

Regardless of the spectacular single-core scores seen in Cinebench, the 285K’s efficiency within the flippantly threaded Photoshop benchmark is disappointing. It is barely slower than the Core i5-12600K, making it one of many slowest CPUs we have examined on this benchmark, which is each puzzling and disappointing.

Adobe Premiere Professional 2024

Thankfully, efficiency in Adobe Premiere Professional is significantly better, although nonetheless barely behind the 14900K. This end result, whereas higher, continues to be underwhelming for a next-generation product.

Gaming Benchmarks

Let’s dive into the gaming benchmarks, beginning with Star Wars Jedi: Survivor. Ryzen processors acquired a notable efficiency enhance on this title with Home windows 11 24H2. Whereas the features for 14th-gen Intel processors have been smaller, they have been nonetheless current. Sadly, elements just like the 285K are presently round 5-10% slower on 24H2, so the 285K information introduced right here was gathered utilizing 23H2.

Star Wars Jedi: Survivor

Ignoring the X3D CPUs, that are blazing quick on this check, the Intel 285K delivered pretty mid-range efficiency, coming in 4% slower than the 14900K. Moreover, the lower-latency DDR5-7200 reminiscence was barely sooner than the 8200 reminiscence on this case.

The Final of Us Half 1

The 285K carried out impressively in The Final of Us Half 1, delivering efficiency akin to the 7800X3D and barely sooner than the 14900K. Nonetheless, with the identical 7200 reminiscence, we solely noticed a 2% enchancment, which is underwhelming for a next-gen product. Some may even name it a disappointment.

Cyberpunk 2077: Phantom Liberty

That is the place issues actually go off the rails. You may see evaluations the place the 285K performs equally to the 14900K in Cyberpunk 2077: Phantom Liberty, however these evaluations seemingly used the built-in benchmark or examined sections of the sport that are not very CPU-intensive. We’ve got proof for this, which we are going to clarify later.

In additional CPU-demanding sections of the sport, the 285K struggles considerably. We have spent appreciable time making an attempt to extract extra efficiency on this check, with the very best outcomes coming from lower-latency reminiscence, the place tightening timings helped probably the most. Nonetheless, regardless of testing motherboards from MSI, Gigabyte, Asrock, and Asus, and even performing a number of recent Home windows installs, the 285K remained not less than 20% slower than the 14900K, making it the slowest CPU on this check.

Hogwarts Legacy

Efficiency in Hogwarts Legacy is healthier, however nonetheless poor total. The 285K lagged behind the 14900K, making it slower than even the Ryzen 7 7700X. It is stunning to see how a lot Home windows 11 23H2 was holding again Zen 4 and Zen 5 processors, as we now commonly discover the 7700X matching the 14900K’s efficiency. Moreover, the Intel CPU is not working with out energy limits.

Assetto Corsa Competizione

In ACC, the 285K fell behind the 14900K by 5%, making it 14% slower than the 9700X and 32% slower than the 7950X3D, that means Arrow Lake just isn’t a super selection for this driving simulator.

Remnant II

Remnant II produced information nearer to what Intel had promised, with the 1% lows being considerably higher on the 285K in comparison with the 14900K. Nonetheless, regardless of matching Intel’s earlier flagship, the 285K nonetheless trailed behind the Zen 4 X3D elements.

Homeworld 3

One other sport the place the 285K’s efficiency falls to the underside of the chart is Homeworld 3. On this check, the 285K delivered Core i5-12600K-like efficiency, making it virtually 20% slower than the 14900K and 7950X3D, and not less than 23% slower than the 9700X.

A Plague Story: Requiem

Efficiency in A Plague Story: Requiem can be surprising. In case you’ve seen different evaluations displaying significantly better outcomes on this title, we’ll clarify the discrepancy later. For now, the 285K is underwhelming, permitting for simply over 120 fps on common, with 1% lows within the mid-70s. It is not less than 17% slower than the 14900K and 27% slower than the 7950X3D, once more delivering Core i5-12600K-like efficiency, making it even slower than the Ryzen 5 7600X.

Counter-Strike 2

In Counter-Strike 2, the outcomes align extra intently with Intel’s guarantees, with the 285K roughly matching the 14900K. Whereas it is good to see this sort of parity, it’s miles from the norm based mostly on our different assessments, and nonetheless an underwhelming end result total.

Starfield

Apparently, the 285K delivered its greatest efficiency in Starfield, narrowly beating the 14900K and even surpassing the 7950X3D. Nonetheless, utilizing the identical reminiscence, it was solely 2% sooner than the 14900K, which, whereas one of many higher outcomes we have seen, continues to be disappointing for a brand new technology.

Warhammer 40,000: Area Marine 2

Area Marine 2 ran wonderful on the 285K, although this new flagship solely supplied mid-tier efficiency, coming in slower than the 14700K and roughly matching the Ryzen 7 7700X. It is not a catastrophe, however removed from spectacular.

Hitman 3

The 285K carried out decently in Hitman 3, being one of many quickest CPUs on this title. Nonetheless, it was nonetheless barely slower than the 14900K – by simply 2% – so it is laborious to get too enthusiastic about this end result, as we’re taking a look at 14700K-like efficiency.

Watch Canine: Legion

For testing Watch Canine: Legion, we used the built-in benchmark, however averaged the info from six runs, because the preliminary outcomes are sometimes inflated. To precisely examine with this information, you may must take a median of six back-to-back runs.

Doing so reveals extraordinarily disappointing outcomes for the 285K, which solely matched the 14600K and 9600X with 162 fps on common. This made it 11% slower than the 14900K and 20% slower than the 7950X3D.

Star Wars Outlaws

Lastly, in Star Wars Outlaws, the lower-latency DDR5-7200 reminiscence supplied the very best outcomes, with 134 fps – the identical efficiency stage because the 12900K. And with that, we conclude our extremely disappointing gaming benchmarks.

Aspect-by-Aspect Knowledge

To recap, this is how the 14900K and 285K in contrast, each utilizing DDR5-7200 reminiscence. On common, the 285K was 6% slower throughout the 14 video games we examined, with important losses in Watch Canine: Legion, A Plague Story: Requiem, Homeworld 3, and Cyberpunk 2077: Phantom Liberty.

Gaming Energy Consumption

Arrow Lake makes important strides in energy effectivity, and whereas it is a lot improved over the 14900K, the outcomes nonetheless fall quick when in comparison with Ryzen processors.

In our Cyberpunk testing, we noticed a considerable 32% discount in energy utilization for the 285K in comparison with the 14900K. Nonetheless, it nonetheless consumed 28% extra energy than the 9950X, and efficiency for the 285K was poor on this title. Let’s now take a look at a sport the place the 285K carried out higher.

In The Final of Us Half 1, the 285K consumed 32% much less energy than the 14900K, which is spectacular given it was a number of % sooner. Sadly, when in comparison with the 7950X3D, the 285K used 71% extra energy for a similar stage of efficiency, which is disappointing. Whereas Intel has made enhancements in energy effectivity, lowering energy consumption by virtually 70W from the 14900K, it nonetheless lags behind AMD, particularly on the present efficiency ranges.

14 Recreation Common

Trying on the 14-game common, we see that the 285K is 5% slower than the 14900K, when calculating utilizing the geometric imply. Because of this, on common, it is barely slower than the 7700X and about on par with the 9600X, which is kind of stunning. It is also 14% slower than the 7950X3D.

Right here, we see simply how a lot Ryzen processors have benefited from the 24H2 replace, with the 7800X3D being 14% sooner than the 14900K, which is exceptional. After all, we are actually power-limiting the Core i9, besides, that margin is spectacular.

Price Per Body

Sometimes, our cost-per-frame evaluation for CPUs contains the price of the motherboard and reminiscence, however on this case, the costs for these parts are fairly comparable. Whereas this may occasionally scale back the general margins barely, the image is obvious: the 285K presents poor worth for avid gamers.

It supplies comparable worth to the 9950X, making it a lot worse than the 7950X3D, which might be the very best all-rounder in the meanwhile. It additionally performs worse than the outgoing 14900K, which now we have by no means really helpful.

Apparently, the 7800X3D, which has offered for as little as $340 this 12 months, nonetheless presents much better worth at $480. Even with current value will increase, the 7800X3D stays a robust contender by way of worth.

Home windows 24H2 Gaming Efficiency

As talked about earlier, Arrow Lake’s gaming efficiency was even worse on Home windows 11 24H2. Intel is conscious of this problem and has recommended enabling the ‘excessive efficiency’ or ‘excessive energy’ profile in Home windows for optimum efficiency on 24H2. Additionally they speculated that there is likely to be a difficulty with the ring bus clock on 24H2, suggesting a BIOS change by disabling the ‘CCF Auto GV’ setting.

We explored these solutions utilizing a restricted set of gaming benchmarks to see the impression. Moreover, we included a recent set up of Home windows 10, as many customers nonetheless depend on this OS and could also be curious in regards to the 285K’s efficiency on it.

Star Wars Jedi: Survivor

Trying on the 23H2 information, the 285K’s efficiency improved by 3% when utilizing Home windows 10, which is an effective signal. Nonetheless, it is seemingly that elements just like the 14900K would see an identical or presumably better efficiency uplift.

Utilizing the default 24H2 configuration with the balanced energy profile severely hindered the 285K, lowering the common body fee by 8% in comparison with 23H2, with the 1% lows dropping by 36%. Clearly, one thing is unsuitable right here. Switching to the ‘excessive efficiency’ energy plan improved the body fee to 150 fps on common, although nonetheless under what we noticed with 23H2, however a notable enchancment nonetheless.

Cyberpunk 2077: Phantom Liberty

In Cyberpunk 2077: Phantom Liberty, the 285K carried out greatest utilizing 23H2, with a slight body drop when utilizing Home windows 10, although the distinction was inside the margin of error.

Apparently, the 24H2 balanced energy plan carried out higher right here, because the body fee regressed by 3% when switching to the excessive efficiency profile. Adjusting the ‘CCF Auto GV’ setting had no impression, just like what we noticed in Star Wars Jedi: Survivor.

Homeworld 3

The Homeworld 3 information underscores simply how erratic the Arrow Lake testing has been, although the outcomes align with expectations. We all know Homeworld 3 performs poorly on 23H2, whereas 24H2 considerably boosts efficiency for Intel’s thirteenth and 14th-gen and AMD Ryzen processors. The truth that Home windows 10 carried out 20% higher than 23H2 is sensible on this context.

What would not make sense initially is the dearth of enchancment on 24H2 for the 285K – till you manually regulate the ability plan. After doing so, the efficiency matched the Core i7-14700K.

Sadly for Intel, now we have to decide on a single OS configuration for testing, and there appears to be no good choice the place every part works because it ought to, not less than not at the moment.

A Plague Story: Requiem

Efficiency in A Plague Story: Requiem was disappointing on the 285K, and switching to 24H2 did not resolve the difficulty. Oddly, Home windows 10 carried out poorly on this title, although this was a development we noticed in a number of of the 14 video games examined.

Counter-Strike 2

In Counter-Strike 2, the efficiency on 23H2 matched what we noticed with Home windows 10. Nonetheless, 24H2 was constantly slower, although switching to the excessive efficiency energy profile supplied a small enhance.

Intel APO Optimization

Lastly, it is vital to notice that each one the gaming benchmarks you simply noticed have been run with out Intel’s APO (Software Optimization) function enabled. This function was not enabled by default throughout our testing, however Intel has indicated that this can be addressed in future BIOS updates.

We did not concentrate on this for 2 causes. First, out of the 14 video games we examined, solely two – Counter-Strike 2 and Cyberpunk 2077 – assist APO. Second, out of curiosity, we went again and enabled APO in these video games and located solely a 1-2% change in efficiency, which falls inside the margin of error.

This implies that both APO has minimal impression on these titles, or it is not absolutely purposeful but. We’ll revisit this function in future content material to see if there are any important updates or enhancements.

What We Realized: Inconsistent Gaming Efficiency

So, there now we have it – a chaotic mixture of outcomes that usually do not appear to make a lot sense. Clearly, Intel has a big variety of points to resolve, and whereas we did not count on them to flawlessly execute a brand new platform and CPU structure launch, we had hoped for a way more polished product.

Relying on the {hardware} and software program configuration, stability will be a difficulty, although we count on this to be addressed pretty rapidly. Nonetheless, it is one thing price protecting in thoughts. We notified Intel in regards to the poor gaming efficiency, and whereas they’ve supplied a number of options, none have labored to a level that we discover passable.

Because it stands, gaming efficiency is inconsistent, to say the least. General, that is very disappointing for a next-generation flagship product. If Intel cannot make important enhancements, Arrow Lake might find yourself being a poor selection for gaming.

When it comes to productiveness, the 285K performs higher, nevertheless it’s nonetheless not distinctive. Whereas there are actually workloads the place it excels, there are numerous the place its efficiency is underwhelming in comparison with the 14900K and 9950X.

One space the place the 285K has massively improved over the 14900K is energy effectivity, which is nice information. Nonetheless, usually, it is nonetheless removed from the place it must be to really compete with AMD.

The worth can be problematic. It is anticipated to be round $600+, which is way too excessive. For a similar quantity, you will get the 9950X or the 7950X3D, and even the 7950X is cheaper at $510. Any a kind of AMD CPUs can be a more sensible choice in our opinion.

Even when Intel manages to completely resolve stability points, repair efficiency on Home windows 11 24H2, and handle the inconsistent gaming outcomes, the 285K would nonetheless solely warrant a price ticket of about $500. So, given all that, we’re clearly not recommending the Core Extremely 9 285K at the moment, and certain will not till these points are resolved.

As for overclocking, given all the present issues and the necessity to check efficiency throughout a number of variations of Home windows, we deprioritized overclocking for this preliminary assessment. It is one thing we’ll little doubt cowl in future content material, however for now, it is considerably irrelevant given the variety of points that must be addressed first.

Reminiscence overclocking is one other subject, and whereas the assessment kits got here with DDR5-8200 CUDIMM reminiscence, it would not make a lot sense as a result of excessive value and restricted efficiency features. We’re not sure of the precise price of this reminiscence, however we all know commonplace UDIMM kits are priced close to $300 for a 24GB equipment, so the CUDIMM kits will seemingly be much more costly.

Due to this fact, one thing like DDR5-7200 CL34, priced round $115, appears to be a extra cheap choice. As soon as the problems we have mentioned are addressed, we’ll look into reminiscence scaling to find out the optimum reminiscence assist for Arrow Lake.

Wrapping this assessment up, we expect there’s potential for Arrow Lake to get quite a bit higher because it seems below cooked at this stage. Whereas we strongly advocate not speeding out to purchase into Intel’s new platform proper now, we’re not prepared to jot down it off utterly simply but.

Purchasing Shortcuts:

• Intel Core Extremely 9 285K on Amazon (quickly)
• Intel Core Extremely 7 265K on Amazon (quickly)
• Intel Core Extremely 5 245K on Amazon (quickly)
• AMD Ryzen 9 9950X on Amazon
• AMD Ryzen 9 7950X3D on Amazon
• AMD Ryzen 7 9700X on Amazon
• AMD Ryzen 7 7800X3D on Amazon