The Ryzen 7 3700X and the Core i7 9700K battle for the high-end share of the CPU market in one of the most processor-competitive periods the last 20 years has ever seen. At launch, they were both priced around the same, $329 and $359 for the 3700X and 9700K respectively but current retail prices favor the 3700X a lot; while the 3700X has gone down in price and can be had for as little as $290, the 9700K is still expensive at around $370. It can really be a tough choice getting one or the other; it mostly depends on what you’re doing with your system the majority of the time. Being addressed to the consumer market, both 8 core CPUs have enough horsepower to handle any daily task with great consistency and performance be it a gaming environment or a work/production effort.
These two CPUs compete in a battle of the giants which most likely doesn’t need any introduction; but we’ll do it anyways if you’ve been out of the cycle and need to learn more. Out of the things worth noting is that AMD’s processor is on a better and much newer process than Intel’s 14nm; TSMC’s 7nm has allowed Ryzen to stretch its legs in both performance and power consumption like never before. For example, this is the first time in the recent technology timeline that AMD surpassed Intel in IPC (Instructions per clock). Now, that’s not the whole performance story because it comes with a few implications that will be talked about if you stay along.
At the heart of the Core i7-9700K is the 8-core Coffee Lake Refresh silicon built on the same 14 nm++ process as the 8th generation Core processors; this CPU, unlike its 8 core brother, the 9900K doesn’t feature Hyper Threading, HT being exclusive to the i9. The die is laid out in the same essential way as Skylake, with two rows of CPU cores sandwiching the ring bus, each core with a dedicated L2 cache and 2 MB slice of the L3 cache. Each core can address the entire 16 MB of L3 cache on the die. Intel is starting to reach the core-count barrier beyond which the ringbus has to be junked in favour of Mesh Interconnect tiles, or it will suffer the detrimental effects of ringbus latencies. AMD’s alternative Mesh Interconnect, called Infinity Fabric is used by Ryzen processors in order to simplify access to processor cores while being very cost-effective when upping the core count. Infinity Fabric’s main disadvantage is the core to core latency it induces and Intel will soon be forced to switch to a similar architecture in order to compete with AMD; it will be very interesting to see how this will play out.
Intel has also improved the integrated memory controller to now support up to 128 GB of dual-channel DDR4 memory. For most users this is meaningless, but if you plan on building a future-proof machine that will last the next seven years or so, you never know when you’ll need more than 64 GB of RAM.
AMD’s 3rd generation Ryzen 7 3700X processor uses the Zen 2 micro architecture. For the past 15 or so years, Intel dominated AMD at IPC, while AMD attempted to remain competitive by upping the core count at a much higher rate than Intel at any given price point for competitive multi-threaded performance. Today’s software environment and game engines are coded to take advantage of multi-core load distribution and thus, even games are increasingly multi-threaded. With Zen 2, AMD set itself an ambitious 13% IPC uplift target to catch up or overtake Intel’s latest Coffee Lake micro architecture at single-core performance; and they matched it. AMD didn’t stop there and even increased core counts for the platform at higher price points. The 3rd generation Ryzen family even includes a 16-core processor, which is a tremendous core count for the mainstream-desktop platform.
Single & Multicore performance
Continuing the talk about performance, we can split this category into two main parts; single core and multicore. Both CPUs excel at them and are marketed to be high-end desktop processors that can do it all without breaking a sweat and that includes gaming, streaming, editing, rendering and other workloads, sometimes running at the same time! In reality, there’s a few key differences here that separate the two products.
Single core performance for the two CPUs is similar even though AMD’s Ryzen 3700X has a better IPC (instructions per clock) metric. This is because Intel’s 9700K is on a more mature process that can sustain higher clock speeds. Because single core performance is a junction between IPC and clock speed, both processors perform fairly similar in lightly threaded tasks. An obvious example is the following. Think about two people that each work in the same environment and are paid over a period of 30 days; one is getting paid $5 for 6 times a month and the other is getting paid $6 for 5 times a month. You probably guessed it right, the $5 guy is the Intel 9700K and the $6 guy is the Ryzen 3700X and they get paid the same amount at the end of the month; even if the 9700K is paid less on a paycheck, it compensates by more frequent paychecks – 6 times a month vs 5 times a month – hence why clock speed is also called frequency.
Multi core performance is very important as well for a huge range of applications and workloads, including the all popular gaming. While single core performance will always translate into increased performance, multicore is a slightly different beast; there are cases like well threaded applications that will efficiently use every ounce of core and thread your CPU has to offer and applications like game engines that will only need just a few to run well. This is why the majority of games will scale amazingly in performance to 4 or 6 cores but if you provide them with more they will simply just ignore them. The sweet-spot less than 5 years ago was the 4 core 8 thread CPU because frankly, it was more than enough for the majority of people. Today, the standard midrange performance for the majority of consumers are the 6 cores CPUs; 8 is already considered high-end but the sales values and application requirements in the past few years has showed a trend that people need more cores and threads for their daily computing adventures. This is why the 9700K and the 3700X are such great CPUs; they provide everything that 6 cores already do and also come with an additional 2 core ‘safety-card’ for the upcoming years’ performance requirements.
Tweaking and overclocking
Both Intel Core i7 9700K and AMD Ryzen 7 3700X come with an unlocked clock multiplier meaning that they can be manually tweaked by the user on compatible motherboards. While 9700K overclocking is enabled only on high end Z370 and Z390 motherboards, the 3700X is more consumer friendly and even their midrange B series motherboards permit it and that includes B450, B550 and even the older B350; how’s that for backwards compatibility?
The overclocking experience with the Core i9-9700K is more straight forward than you might initially think; great motherboard features and the intuitive BIOS interface let you adjust voltages, clocks and LLC (load line calibration) without much fuss. With air cooling, overclocking is generally held back by temperatures despite the soldered IHS. We can set a voltage offset of +50 mV, on air while staying outside of thermal throttling, which resulted in a stable clock of 5.0 GHz on all cores; 5.1 GHz is a little bit unstable with most samples, and we couldn’t bump the voltage any further without running into thermal throttling. Liquid cooling seems to be a must if you want to take this CPU over the 5 GHz threshold but it also depends on silicon lottery; you might get lucky and get a better overclocker than average.
Using a 240 mm all-in-one watercooling unit, temperatures are no longer a problem, and we could dial up the voltage +100 more than we did on air cooling. We picked +150 mV for this scenario as it’s a reasonable increase that won’t harm the lifetime of your processor; the stock Intel Core I7 9700K sits at a default voltage of 1.25V. On water, you can reach a completely stable 5.1 GHz on all cores even with +100 mV. You might be able to boot fine at 5.2 GHz, but the system has a tendency to crash if pushed with 100% load. It’s a little bit surprising but the i7-9700K reaches its maximum clocks so very similar to the i9-9900K. The silicon lottery plays a big role here, too.
On the other hand, AMD Ryzen 7 3700X overclocking was even easier as we could increase the voltage up to 1.4 V before we started getting high temperatures on air. At that voltage setting, our maximum manual overclock was 4.325 GHz on all cores — lower than the maximum boost frequency of the processor (4.4 GHz). The manual overclock can only shine in specific applications that fully load all cores, in which the processor at stock would dial down its clocks automatically; essentially, if you’re doing both gaming and heavy work with your 3700X, it’s generally better to leave it at stock thanks to the CPU’s built-in algorithm that intelligently boosts cores based on the type of load. This way, some cores may even sustain 4.4GHz and the performance will be better overall.
AMD’s superior 7nm process plays a huge role in the efficiency battle between these two CPUs. While the 9700K consumes ~180W during demanding multi-threaded tasks, the 3700X doesn’t go above ~135W and the latter even has SMT (Hyper-Threading) enabled. When we’re overclocking both CPUs, the Ryzen to 4.325GHz all core and the 9700K to 5.1GHz, power consumption remains in check for the 3700X at around 160W while the 9700K skyrockets to a whopping 235W. This is important if power consumption during multi-threaded tasks matters for you.
Even though these CPUs hit different maximum clock speeds, 4.4 for the 3700X and 5.0 for the 9700K, single-core performance is almost identical, the 3700X scoring 500 in Cinebench R20 Single-Core test and the 9700K only 2 points ahead at 502. This is due to what we talked about at the beginning of the article, IPC matters and clock speed is not the full story.
|CPU Name||Game Title||Graphics Quality||Resolution||Average FPS||1% Low FPS||0.1% Low FPS|
|AMD R9 3700X||Battlefield V||Ultra||1080p||156||113||102|
|Intel i7 9700K||Battlefield V||Ultra||1080p||170||126||98|
|AMD R9 3700X||Gears Tactics||Ultra||1080p||125||92||86|
|Intel i7 9700K||Gears Tactics||Ultra||1080p||133||102||91|
|AMD R9 3700X||Tom Clancy’s Rainbow Six Siege||Ultra||1080p||266||227||179|
|Intel i7 9700K||Tom Clancy’s Rainbow Six Siege||Ultra||1080p||273||213||170|
|AMD R9 3700X||Ghost Recon Breakpoint||Ultra||1080p||131||108||92|
|Intel i7 9700K||Ghost Recon Breakpoint||Ultra||1080p||137||112||92|
|AMD R9 3700X||Shadow of the Tomb Raider||Ultra||1080p||124||99||91|
|Intel i7 9700K||Shadow of the Tomb Raider||Ultra||1080p||128||98||87|
Intel is still ahead in the gaming department but the margin is not big enough to justify the big price premium and the lack of hyper threading. The small gaming performance advantage perishes before the stellar multi thread performance of the 3700X that is one class over the 9700K, equaling the Core i9 9900K.
Many would ask why is the 3700X slower in gaming than the 9700K if the single-core performance is on par and multi-core is a lot better and the answer lies in the study of latency. Intel’s ring bus architecture is still better for gaming thanks to its very low inter-core latency. The Ryzen 7 3700X features two 4 core chiplets working together on an Infinity Fabric surface that induces some latency; while the gaming 1% lows and 0.1% lows are still great, average FPS takes a small hit due to the increased core to core latency. AMD is set to eliminate this detriment next generation by switching to the use of 8 core chiplets which will avoid the chiplet to chiplet latency.
One good example is the newly released Ryzen 3 3300X which is a 4 core 8 threads CPU that has only one chiplet (unlike all the other Zen 2 CPUs); benchmarking it against other 4 core 8 threads CPUs proves its superiority in Single Core performance also translates to gaming. The 3300X is 5 to 10% faster in gaming than even the newest Intel Core i3 10100 which is also a 4 core 8 thread processor. We’re looking forward to AMD translating this positive change to their new range of CPUs.
Work-wise, you really can’t go wrong with the Ryzen 7 3700X when considering it against the i7 9700K. The increase in multi core speed the extra 8 threads bring to the table is no slouch and will allow you to finish tasks considerably faster than the i7 9700K. Don’t get us wrong, the Intel CPU will still perform well in all the productivity tasks because it has 8 strong cores but its performance pales when compared to the mighty 3700X.
The following chart will compare some of the most popular productivity use cases among PC users.
|Application||Intel Core i7 9700K score||Ryzen 7 3700X score|
|Cinebench R20 Single (score)||502||500 (0.4% slower)|
|Cinebench R20 Multi Core (score)||3712||4824 (30% faster)|
|Adobe Premiere PRO 4K Export (export time)||514s||412 (20% faster)|
|Blender Open Data (export time)||1296||972 (25% faster)|
|V-Ray Benchmark Version 4 (score)||10507||13494 (28% faster)|
|Corona 1.3 Benchmark (render time)||146||107 (27% faster)|
We can start seeing a trend; the Core i7 9700K can’t keep up with the Ryzen 7 3700X in well-threaded applications and the advantage lies somewhere around 26%.
How do they stack up against their newest competitor, the i7 10700K
Intel just released the Core i7 10700K as an 8-core, 16-thread CPU, same as the previous generation Core i9 9900K; this means Hyper Threading is back to i7 and these are great news for all consumers. Whereas the previous Coffee Lake part, the i7 9700K, featured a base clock of 3.6 GHz, the 10700K won’t drop below 3.8 GHz; as a result, this also increases the TDP 125 W. The boost clock frequency has also been increased by 100 MHz, which sees the 10700K hit 5.1 GHz. L3 cache also went through some changes as the new i7 part now sporting a 16 MB capacity, 4 MB higher than the previous generation part. Memory bandwidth is also up around 10% and Intel tweaked the integrated memory controller to a base DDR4 2933 instead of i7 9700K’s DDR4 2666.
In simpler words, coming in at a MSRP of $375 and current retail price of around $400, the i7 10700K is a i9 9900K in disguise but 25% cheaper and better, a good progress from last year’s i7 9700K. How does this translate into performance? Well, when compared to the i7 9700K, the new processor provides a 2% increase in single core speeds due to the increased boost clock / overclocking ceiling and a whopping 25% increase in multi core productivity applications due to the processor having 8 additional threads above the 9700K.
These improvements push the gaming performance of the i7 10700K 2-3% above the i7 9700K, a small and unnoticeable difference but the real highlight here is the increased productivity performance; the 10700K is a much better editing/rendering/streaming processor. If you’re currently on the 9700K, save your money because you won’t get noticeable improvements with the 10700K. Sadly, the 10700K is also more power hungry than the 9700K, consuming around 20% more and this is due to the addition of Hyper Threading.
But we also remember from what we’ve talked earlier that the i7 9700K trailed the Ryzen 7 3700X by around 26% in productivity related tasks. This changes with the release of the 10700K, the new processor being on par with the 3700X on core count and thread count. The single core performance is 2.5% better with the i7 10700K and multi core is now on par and in some cases even better than the Ryzen 7 3700X. Now Intel brings to the table a much closer battle between the two than they did with the previous generation i7 processor.
Is the 10700K a good value then? Well, you’ll have to decide for yourself if the marginally better performance is worth the big price premium over the Ryzen 7 3700X. The 10700K performs around 10% better in gaming and is on par with the Ryzen 7 3700X in productivity applications like Blender, After Effects, Premiere Pro, all while being a whopping 30% more expensive and that is not factoring in the extra cost of a new Z490 motherboard and a high end air/water cooler. If we were to compare prices, you will need to spend a minimum of around $650 for a 10700K/Motherboard/Cooler bundle and just $410 for a good 3700X & Motherboard bundle. The price difference is there and you are the only one who can decide if its a reasonable price jump for the performance.
We hope we managed to paint a full picture of how these two high end processors perform in a variety of applications, including gaming and productivity. The 3700X pulls ahead in the majority of metrics thanks to its 8 extra threads that the i7 9700K is just crippled without and cannot make up the advantage through better core to core latency or higher clock speed. The Intel CPU though can slightly pull ahead during gaming and will bless users with a slightly higher average framerate when compared to the 3700X – the average difference is around 5% in favor of the 9700K.
If you are looking for a do it all CPU that puts stellar value & performance on the table in gaming, editing, rendering and general productivity situations, look no further, the Ryzen 7 3700X is the processor for you. On paper you would think the 3700X should be the more expensive CPU (its productivity performance is almost on par with the much more expensive Core i9 9900K) but the AMD part also manages to be 80$ cheaper than the 9700K and come with a suitable box cooler, the Wraith Prism. If you factor in the cost of a good aftermarket cooler and a high-end Z370/Z390 motherboard for the 9700K, the cost difference between the two processor bundles comes around $200 to $250 in favor of the Ryzen 3700X.
Otherwise, if you are looking to squeeze even the last drop of gaming performance out of your GPU and still be able to do productivity tasks decently, the 9700K may just be your next CPU. Just make sure you are pairing it with a BIOS updated Z370 motherboard or a new Z390 so you can make use of the high overclocking ceiling of the Core i7 9700K. A good air cooler is a must in order to let the processor stretch its legs; tower coolers like the Noctua NH D14 or the be Quiet! Dark Rock Pro 4 should do the trick nicely!