Intel 10th Gen Core i9-10900K and i5-10600K Review
Comet Lake-S Boosts Core Counts and Clocks
Intel’s 10th Gen Core “Comet Lake-S” processors are launching today, and with this generation Intel is aiming to improve their position in the desktop CPU space by increasing core counts for the third time in as many years.
Intel’s desktop strategy has necessarily shifted to an increase in core and thread counts to compete with AMD’s Ryzen parts, in stark contrast to the deprecated “tick tock” cadence of the past, in which every year saw either a die shrink or a new architecture.
Intel is, nonetheless, offering better price/performance across the board compared to 9th Gen parts, with the core/thread increases not accompanied by price hikes; as these 10th Gen CPUs are launching with the same list pricing as their less-capable 9th Gen counterparts.
As announced on April 30, here is Intel’s 10th Gen CPU lineup:
Retail pricing will naturally vary, and thus far we have seen pricing for the Core i9-10900K at $529.99, the Core i7-10700K at $409.99, and Core i5-10400 at $194.99.
Intel’s 10th Gen Core i9-10900K and Core i5-10600K
We received the 10-core, 20-thread Core i9-10900K (not to be confused with Core i9-10900X, a $799 HEDT part) and 6-core, 12-thread Core i5-10600K processors, and will cover just their performance in this launch review.
Intel 9th and 10th Gen “K” Series Processors | |||||||||
---|---|---|---|---|---|---|---|---|---|
Model | Cores / Threads | Base / Boost | TVB Frequency (Single/Multi) | Smart Cache | Memory Type | TDP | Launch Price | Launch Date | |
Core i9-10900K | 10 / 20 | 3.70 GHz / 5.20 GHz | 5.30 GHz / 4.90 GHz | 20 MB | DDR4 2933 | 125W | $488-$499 | Q2 2020 | |
Core i9-9900K | 8 / 16 | 3.60 GHz / 5.00 GHz | – | 16 MB | DDR4 2666 | 95W | $488-$499 | Q4 2018 | |
Core i7-10700K | 8 / 16 | 3.80 GHz / 5.10 GHz | – | 16 MB | DDR4 2933 | 125W | $374-$387 | Q2 2020 | |
Core i7-9700K | 8 / 8 | 3.60 GHz / 4.90 GHz | – | 12 MB | DDR4 2666 | 95W | $374-$385 | Q4 2018 | |
Core i5-10600K | 6 / 12 | 4.1 GHz / 4.8 GHz | – | 12 MB | DDR4 2666 | 125W | $262-$263 | Q2 2020 | |
Core i5-9600K | 6 / 6 | 3.70 GHz / 4.60 GHz | – | 9 MB | DDR4 2666 | 95W | $262-$263 | Q4 2018 |
There will be no architectural discussion as Intel has not released detailed information about these CPUs just yet (in a tradition that goes back to the Skylake launch), though we aren’t expecting too many surprises here.
Aside from the new core/thread configurations there are some notable changes with 10th Gen desktop parts compared to the 9th Gen, with the implementation of Thermal Velocity Boost for clocks of up to 5.3 GHz, official DDR4-2933 support (up from DDR4-2666 with 9th Gen), and the big one: a move to the LGA1200 socket.
Thermal Velocity Boost is an interesting item, which takes advantage of available thermal headroom to boost clocks beyond the normal Turbo Boost limits. As Intel explains:
“Intel Thermal Velocity Boost (Intel TVB) is a feature that opportunistically and automatically increases clock frequency above single-core and multi-core Intel Turbo Boost Technology frequencies based on how much the processor is operating below its maximum temperature and whether turbo power budget is available. The frequency gain and duration is dependent on the workload, capabilities of the processor and the processor cooling solution.”
The big difference that will be immediately apparent to anyone looking into one of these new processors is, of course, the LGA1200 socket. This launch becomes as much about the new platform as the CPU, as the two must be purchased together if you move up to 10th Gen.
Though identically sized, the Core i9-9900K has 1151 pads (left), and the Core i9-10900K has 1200 (right)
The LGA1200 socket arrives with the Z490 platform, with H570 and B460 on the horizon. Connectivity is a big focus for the 10th Gen, as Intel Wi-Fi 6 AX201 is integrated into the CPU, with Intel’s I225 2.5G Ethernet and Thunderbolt 3 also part of the 10th Gen platform.
We can explore the platform in greater depth when we get into our first Z490 motherboard reviews, so for now we will move on to the performance numbers.
Compute Benchmarks
Here we look at a handful of results using some common benchmarks, which have all been freshly re-run by yours truly over the past few weeks using the latest BIOS updates (including AGESA 1.0.0.5 for the Ryzen CPUs) and platform driver versions on Windows 10 1909 (build 18363.592). All results are the average of three runs.
PC Perspective CPU Test Platforms | |
---|---|
Motherboards | ASUS ROG MAXIMUS XII EXTREME (BIOS 0403) Gigabyte Z390 AORUS PRO (BIOS F12d) Gigabyte Z390 I AORUS PRO WIFI (BIOS F8c) ASUS Strix Z370-H Gaming (BIOS 2203) Gigabyte H270 Gaming 3 (BIOS F9d) MSI MEG X570 ACE (BIOS 7C35v192, AGESA 1.0.0.5) |
Memory | Crucial Ballistix LT 16GB (2x8GB) @ DDR4-2933 16-18-18-36, 1.35V Crucial Ballistix LT 16GB (2x8GB) @ DDR4-2666 16-18-18-36, 1.35V Corsair Dominator Platinum 16GB (2x8GB) @ DDR4-2400, 16-18-18-36, 1.35V Crucial Vengeance LPX 16GB (2x8GB) @ DDR4-2400, 16-18-18-36, 1.35V |
GPU | NVIDIA GeForce RTX 2080 FE |
Storage | Corsair Neutron Series XTi 480GB |
Power Supply | CORSAIR RM1000x 1000W |
Operating System | Windows 10 64-bit Version 1909, November 2019 Update (Build 18363.592) |
GPU Drivers | NVIDIA GeForce Game Ready 445.87 |
Every effort was made to keep these results as consistent and accurate as possible. Memory speeds were set to the maximum official spec, with AMD Ryzen processors at DDR4-3200 and Intel systems ranging from DDR4-2400 to DDR4-2933 depending on generation.
It was noted after testing had been completed that technically only the Core i9 and i7 series CPUs in the 10th Gen lineup officially support DDR4-2933, leaving Core i5 and below at DDR4-2666. Both the Core i9-10900K and Core i5-10600K were tested using the same memory settings, thus our i5-10600K was tested at DDR4-2933, and not DDR4-2666.
Blender 2.82a
Here we look at render times with both the classic BMW and larger Classroom test images, using the latest release of Blender (2.82a).
The Ryzen 9 3900X has a 2-core advantage over the i9-10900K here, which settles for second place. As to the i5-10600K, it performs as one would expect when considering that it, like the Coffee Lake i7-8700K, is a 6-core, 12-thread part. There are some clear performance benefits from this new 10th Gen 6c/12t Core i5 CPU, however, and it has a comfortable lead over the i7-8700K (though just behind the 8-core, non-HT i7-9700K) in this test.
Cinebench R20
What is there to say about Cinebench? AMD loves it, Intel hates it, and a lot of people use it. It’s here for your reference.
Intel does take back the crown in single-threaded performance here, something the R9 3900X had been able to boast about vs. the i9-9900K. But there is a world which exists outside of Cinebench (a world I only managed to escape from recently). Let us move on.
Geekbench 5
Using the latest version (5.1.0) of Geekbench we will take a look at overall single and multi-threaded results.
First, we see an impressive win for the i9-10900K here in the overall single-threaded score, which is 90 points higher than the second-place Core i9-9900K. The Ryzen 9 3900X is a close third, and the new Core i5-10600K finishes in a tie with the Core i7-9700K.
Moving on to the multi-core tests, the Ryzen 9 3900X is the winner by a comfortable margin, showcasing its core advantage in the face of the i9-10900K’s single-core performance lead in this test.
A closer look at both integer and floating point performance from these tests is provided for your reference in the charts below.
7-Zip 19.00 64-bit
The ubiquitous 7-zip offers a handy built-in benchmark that averages the performance across four iterations of its file compression and decompression tests, and it scales nicely with CPU thread count.
Here AMD enjoys an obvious advantage, and while 7-zip scales with core/thread count some of the differences could be due to AMD’s available memory bandwidth at DDR4-3200. It might be a more ‘fair’ test if all CPUs were paired with identical memory speeds, but the author doggedly insisted that max officially supported memory speeds be used on all platforms (even though he messed up and tested the Core i5-10600K at 2933 instead of 2666), though memory timings were identical in all cases, at least.
x264 HD Benchmark
While this is an older test it is at least the latest version of an older test. The x264 HD Benchmark (version 5.0.1, 64-bit mode) encodes a high quality 1080p test file using the apparently reviled 2-pass method that the kids despise. But it’s another very consistent command-line benchmark that takes full advantage of all available CPU threads.
The first chart shows first and second-pass FPS averages, with the second showing total average 2-pass encode time (all numbers are averaged from four 2-pass runs):
The first chart shows first and second-pass FPS averages, with the second pass of far greater importance here as it provides a significantly more challenging load for the CPU. Though the Intel Core i9-10900K easily wins pass 1, pass 2 goes to the Ryzen 9 3900X. Was it enough to win overall?
This second chart shows total average 2-pass encode time, and of course the only thing that really matters is the total encode time with this test, anyway. Here the 12-core Ryzen 9 3900X finished the job 32 seconds faster than the second-place Core i9-10900K. Once again the Core i5-10600K is offering better performance than the 6c/12t Core i7-8700K, but finishes behind the 6c/12t Ryzen 5 3600X.
Core i9-10900K Clock Speeds
I wondered if the Core i9-10900K was successfully hitting its potential 5.3 GHz single core, and 4.9 GHz all-core TVB frequencies, or if the Corsair H115i RGB Platinum 280 mm liquid CPU cooler used for our testing was enough to provide the needed thermal headroom.
We were able to quickly verify this with data logged using HWINFO64 during both a single-core and multi-core Cinebench R20 run.
With Windows configured with the High Performance Power Plan, single core frequencies with the Core i9-10900K ranged from 4.9 GHz to 5.1 GHz for the most part, in and out of the Cinebench run. There were several spikes up to 5.3 GHz single core, as the opportunistic nature of TVB is evident. It won’t hit 5.3 GHz all the time, but it certainly can.
Here the actual timing of the Cinebench R20 all-core run is clearly evident, as the Core i9-10900K spiked initially to 5.3 GHz (on core 8), before all cores settled to the full 4.9 GHz all-core TVB frequency for the duration of this short test. During a longer sustained all-core test we would probably be seeing lower frequencies, but I did this just to verify that Thermal Velocity Boost worked.
One final note here, as I manually set CPU performance in the UEFI setup to disable multi-core enhancement and “enforce all limits” on the CPU (as always). With high performance cooling and some tweaking, our deluxe ASUS ROG Maximus XII EXTREME board is assuredly capable of much more, but CPUs are always benchmarked in with settings to ensure stock performance.
Gaming Performance
As mentioned in our recent Ryzen 3 review, exploring the potential gaming performance differences between CPUs with a given GPU requires benchmarking at settings that intentionally place more of a bottleneck on the CPU. This is academic, but can be useful for certain applications.
We tested an NVIDIA GeForce RTX 2080 FE at 1920×1080 and less demanding settings, which is admittedly not a realistic scenario for enthusiast-level hardware unless you are a competitive gamer, or have a high refresh rate 1080p monitor, or some other scenario that favors FPS over all else.
Far Cry 5
Run at the “high” preset with standard textures, Far Cry 5 at 1920×1080 is not a particularly demanding title, but there is clear CPU scaling to be seen with FPS averages ranging from 106.3 to 151.7 with the same RTX 2080 FE card.
At the top of the heap is the new Intel Core i9-10900K, which flexes its gaming muscles here to the tune of a ~7% FPS increase over the second-place Core i9-9900K. That may not seem like much, but extracting an additional 10+ FPS with the same graphics card is impressive, and shows that even the i9-9900K could be a bottleneck – at this resolution/detail level, anyhow.
The i5-10600K is just behind the i7-9700K here (less than 1 FPS difference might as well be a tie), and Intel processors occupy the top six spots on the chart.
Metro Exodus
We ran Metro Exodus at the “normal” (medium) preset, as this is a much more demanding game than Far Cry 5. We see a similar story, however:
The Core i9-10900K is again on top and enjoys a nearly identical ~7% lead over the Core i9-9900K here, but unlike the Far Cry 5 test we now see its 10th Gen Core i5 counterpart finishing in second place, ahead of both the Core i9-9900K and Core i9-9700K.
DOTA 2
This is our lone ‘competitive gamer’ test, with a consistent benchmark being performed via this handy tool which uses the teamfight from match 3787212636. This was run at 1080p with the “fastest” preset settings and DX11.
In this result we see the Core i9-10900K leading by 48.4 FPS over the i9-9900K in our DOTA 2 test; a ~19% increase.
3DMark Time Spy
And finally, just for reference, the results from 3DMark Time Spy. These show significant scaling from the physics test, but not a big overall difference from the GPU tests.
Intel processors dominate the gaming charts in general at 1080p medium/high detail settings like these, but things tighten up at max settings and – of course – resolutions of 1440p and above, when the GPU becomes more of a bottleneck.
Power Consumption and Thermals
Measured power presented in the chart is total system draw, with the idle and load power recorded using a watts up? PRO power meter. For these tests a low-power GPU in a zero-RPM fan mode was used to help minimize impact on the test results, in this case an MSI GTX 1650 Gaming X. Loads were generated using a Cinebench R20 all-core test.
The Core i5-10600K and Core i9-10900K load numbers are, unsurprisingly, higher than the 9th Gen Core i5 and Core i9 processors on the chart. Pushing Intel’s 14nm process to higher clocks and with more cores and threads obviously results in higher power consumption.
Looking at the 10900K vs. the 9900K in particular, we see a rise from 210W to 278W (total system draw), which is an increase of about 32%. What does this power increase buy us? Looking back to the performance charts we see some large multi-core performance gains, with about 30% in Cinebench R20 and nearly 35% in Geekbench 5. On the low end a reduction in render time of 11-14% in Blender 2.82a, and reduction in x264 HD encode time of about 17%, was less impressive.
Still, Intel is marketing these “K” series parts to gamers, and we saw solid performance gains over the i9-9900K at 1080p, with a 10.3 FPS advantage in Far Cry 5 and 10.8 FPS in Metro Exodus representing a 7% increase with our RTX 2080 FE. Oh, and that 48.4 FPS over the i9-9900K in our DOTA 2 test.
Thermals – We Need More Data
Another aspect of this higher power consumption is that more heat needs to be dissipated. Intel did make some changes to the implementation of the integrated heat spreader (IHS) with the unlocked versions of these 10th Gen processors, with a Thin Die STIM (solder thermal interface material) and a thicker IHS.
Intel’s official guidance for cooling the i9-10900K is “Thermal Solution Specification: PCG 2015D”, which is 130W. This is the same as the Core i9-9900K, but clearly some additional cooling is needed to maximize headroom with this 10-core part (this is what happens when you reach HEDT power and performance levels on consumer desktop).
For this launch review we used a Corsair H115i RGB Platinum 280 mm liquid CPU cooler. To mitigate the impact of thermals on our results the pump and both fans were set to 100%. Under these conditions the thermals of the i9-10900K were well under the 100 C TJUNCTION, with the highest observed temps under 70 C in a 20-22 C room.
Between the high performance cooler we used, and the relatively short workloads of tests like Cinebench R20, proper thermal testing has not been done. Extended testing with more taxing workloads will help us understand the thermal limits of this design, and the processor is certainly a good candidate for some high-end air cooling tests, as well.
The Value Proposition
Pricing for a retail 10-core flagship Core i9-10900K is $529.99 ($488 is the official 1k list price), which is the current selling price for the 9th Gen Core i9-9900K. The 1k price of the Core i5-10600K was announced at $262, so retail pricing will likely be in the $269-$279 range for this part – not bad for a processor that competes with the ~$379 Core i7-9700K.
As these new processors use a new LGA socket, moving to a 10th Gen Core desktop CPU requires adoption of a new platform. This is never a popular choice, and of course the cost of a new motherboard will have to be factored in if you are planning to purchase any of these new 10th Gen parts.
It is notable 16 of the 45 available Z490 motherboards on Newegg are priced at $299 and above, which places Z490 at the HEDT (X299) level – at least at the high end. On the other hand, 18 of 45 boards are $200 or less. And at the entry level there are Z490 motherboards from $149.
Lower-cost platforms including H470 and B460 will also be available, though Z490 is what we have at launch.
Conclusion
That Intel would still be on 14nm for their mainstream desktop parts in 2020 was unfathomable when the company launched Skylake back in 2015, and obviously Intel is making the most of their situation right now. By targeting core counts, thread counts, and clock speeds, Intel is extracting every possible ounce of performance from this, shall we say, mature process technology.
Regardless of how we got here, Intel’s 10th Gen Core processors represent a clear and measurable improvement over the previous generation, without a price hike. In other words, price/performance has improved, and consumers benefit from the increased compute performance from these new configurations at their release prices, even if this is not what Intel originally had in mind for their 2020 desktop lineup way back in 2015.
But we are comparing Intel to Intel here. AMD’s Ryzen desktop CPUs are still an incredible price/performance value, and AMD still offers generally better multi-threaded compute performance – particularly at the high end – in our testing. A processor like the Ryzen 9 3900X is a better buy if you don’t need the highest single-thread performance, or if you only game at resolutions of 1440p and above.
Intel’s 10th Gen platform, on the other hand, which now includes integrated Wi-Fi 6 as well as 2.5 GbE and Thunderbolt 3, offers better connectivity and class-leading single-threaded performance. But Intel’s desktop processors still offer fewer CPU PCI Express lanes, and remain on PCIe Gen3 for another generation – though we understand this platform may support it with 11th Gen.
Did Intel deliver “the world’s fastest gaming processor” as promised? Well, the Core i9-10900K is the fastest processor we’ve ever tested at 1080p, though as this author has pointed out the real-world application of a $529 processor and $699+ graphics card to play at 1080p medium/high settings is somewhat limited.
Unquestionably, the Core i9-10900K is the CPU to buy if you are a competitive gamer looking to extract the highest possible FPS from your GPU. And for its part the Core i5-10600K is very impressive, offering nearly the performance of a Core i7-9700K for more than $100 less.
Review Disclosures
This disclosure statement covers the way the product being reviewed was obtained and the relationship between the product's manufacturer and PC Perspective.
How Product Was Obtained
The processors are on loan from Intel for the purpose of this review.
What Happens To Product After Review
The processors remains the property of Intel but will be on extended loan to PC Perspective for the purpose of future testing and product comparisons.
Company Involvement
Intel provided the product samples and technical brief to PC Perspective but had no control over the content of the review and was not consulted prior to publication.
PC Perspective Compensation
Neither PC Perspective nor any of its staff were paid or compensated in any way by Intel for this review.
Advertising Disclosure
Intel has not purchased advertising at PC Perspective during the past twelve months.
Affiliate Links
This article contains affiliate links to online retailers. PC Perspective may receive compensation for purchases made through those links.
Well, Intel can keep selling those as long as it can sell gaming numbers and as long as OEMs are not willing to push AMD systems in the market. Even after Zen 3 Intel will keep selling these CPUs because, as Renoir proved, there are many ways to make inferior CPUs look as the only option available.
It will be interesting to see how much of these numbers seen in this and all those other reviews, remain true with a cooling solution under $100 and a motherboard under $200 in a closed case of less than $100.
It seems like the design of “K” series unlocked parts do in fact have good thermal characteristics, in spite of the additional power draw. I always wait to read other reviews and see if my own findings are corroborated, and they were – but, again, with “K” series parts (the only CPUs with the thin die/thick IHS combo).
As far as the under $200 boards, it will be indeed interesting to see how the lower cost boards perform with these parts. But no CPU does well in a “closed case”. Airflow is important for a case, as always, and we’ve seen it make a comeback (as silly as that sounds) in recent months. Good thing there are plenty of cases at or below $100 with good airflow.
I missed that part, that only the K models have “the thin die/thick IHS combo” even having read your and other reviews.
Thanks for the rational conclusion:
” A processor like the Ryzen 9 3900X is a better buy if you don’t need the highest single-thread performance, or if you only game at resolutions of 1440p and above.”
“Did Intel deliver “the world’s fastest gaming processor” as promised? Well, the Core i9-10900K is the fastest processor we’ve ever tested at 1080p, though as this author has pointed out the real-world application of a $529 processor and $699+ graphics card to play at 1080p medium/high settings is somewhat limited.”
It gets a little old reading about how Intel ‘retains the gaming crown’ without the ‘ at 1080p’ caveat when, even I, someone that maintains a mid-tier gaming system, hasn’t gamed at 1080p in nearly 5 years now. Not to mention the additional costs that are required for power and cooling to keep these CPUs running anywhere near peak performance.
Regardless, Intel will sell as many of these 10th gen desktop CPUs at these prices as they can justify allocating mfg capacity to make to the Intel loyalists. Granted, this won’t attract any more pragmatic buyers to their camp, but it will keep those loyalists blithely touting Intel CPUs for one more product cycle.
You’ve got to give it to Intel – they’ve truly wrung every last bit of performance out of the 14nm process. It’ll be interesting to see what they can do with whatever their next process tech is, but until then I reckon its Ryzen all the way.
Excellent write-up Sebastian
This is a lot to like here including the price. the comparisons with the top AMD Ryzen CPU are very interesting.