Ryzen 9 3900X CPU Vs. Intel 8 Core i9-9900K Review And Comparison

Ryzen 9 3900X CPU Vs. Intel 8 Core i9-9900K Review

AMD has launched their new 12 core Ryzen 9 3900X CPU at a similar price point to Intel’s
8 core i9-9900K, so let’s see what the differences are and find out how they perform in games and applications, both at stock and while overclocked.

Specifications:

Let’s start with the specs. The major difference between the two is that the AMD 3900X has 12 cores and 24 threads, while the Intel 9900K is 8 cores 16 threads, so straight away we’re expecting the AMD chip to have an advantage in multicore workloads. The 3900X has a higher base clock, while the boost speed on the 9900K is higher, though these aren’t directly comparable due to architecture differences. While Intel is still at 14nm, Zen 2 makes use of 7nm, though these aren’t directly comparable either anyway. Otherwise the 3900X has a fair amount more cache compared to the 9900K. I’m comparing these two CPUs because they’re available for around the same price point. The 3900X is launching at 499, while at the time of recording Intel’s 9900K is going for 495, although the KF version without integrated graphics is available for 480, granted neither of Intel’s offerings includes a cooler. There was recently news that Intel may lower prices though, so check the description for updated prices. I’ll also, be comparing the 8 core 3700X in a future video, so make sure you’re subscribed for that one. Both CPUs were tested in the same system, however, I’ve had to change motherboards. For the AMD 3900X I’ve tested with the MSI X570 ACE motherboard and for the Intel 9900K I’ve used the MSI Z390 ACE motherboard. The rest of the components were otherwise the same, I’ve tested with 16gb of DDR4-3200 memory running in a dual-channel at CL14 and with an Nvidia RTX 2080 Ti to reduce GPU bottlenecks. Although the 3900X comes with a cooler included

in the box, I haven’t tested with it. I’ve used the same Fractal S36 AIO with Noctua NT-H1 paste for both CPUs so we can get an apple to apple comparison.

I’ve tested both CPUs at stock, and with the 9900K overclocked to 5.1GHz and 3900X overclocked to 4.3GHz, so we can see how overclocking helps improve performance with both, and I also, haven’t kneecapped the 9900K to a 95-watt TDP because in my opinion that doesn’t make sense in terms of practicality.
With that in mind we’ll first check out the differences in various applications, as well as power draw and thermals, followed by gaming tests at 1080p and 1440p resolutions afterward, then finish up by comparing performance per dollar.

Let’s start with Cinebench R20. I’ve got the overclocked results on the upper half of the graph while the stock results are in the lower half. As expected the 3900X is dominating in the multicore workload due to the 50% higher core count, while the single-core is much closer. At stock, the 3900X was slightly ahead when it came to single-core, however the score went down once it was overclocked to 4.3GHz on all cores. This is a trend we’ll see as we continue, basically the 3900X would boost higher than this in the single-core at stock, so by manually overclocking all 12 cores to this speed, we’re lowering single core performance.

While Cinebench R15 has been replaced by the newer R20 just covered, I wanted to also include the results of this one too. Again the 3900X is smashing the 9900K in the multicore score, though this time the 9900K was ahead in single-core performance, while again the 3900X saw a lower single-core result due to the 4.3GHz all-core overclock.

I’ve tested the Blender BMW and Classroom benchmarks, and as a test that works better with more cores, it’s another clear win for the 3900X. At stock, the 3900X is completing the BMW test 39% faster than the 9900K and 40% faster in the Classroom test. This gap closes a little once we overclock both CPUs, with the 3900X reduced a little to a 37.5% lead in the BMW test, while coming out 39% ahead for the Classroom test.

Handbrake was used to convert a 4K file to 1080p, and then a different 1080p file to 720p. This is another workload that benefits from more cores, so at stock settings we’re seeing the 3900X complete the 4K conversion 27% faster than the 9900K and 36% faster for the 1080p file. Again once overclocked the gap narrows in a bit, with the 3900X now 22% ahead for the 4K conversion and 34% faster for the 1080p file.

Adobe Premiere was used to export one of my laptop review videos at 1080p, and the results between the two CPUs were very close in this particular test. At stock, the 3900X was completing the task just a few seconds faster than the 9900K, only making it 2% faster. Once both CPUs are overclocked though the tables are turned, the 9900K is now just a couple of seconds faster than the 3900X.

Steve at Hardware Unboxed saw more than a 20% improvement with the 3900X when exporting a 4K video though, so this just goes to show I need to move away from my standard 1080p export test. Unfortunately, I couldn’t retest this as I was only able to borrow a 9900K for a couple of days, so I suggest checking Steve’s video out for more results.

I’ve also tested the warp stabilizer effect in Adobe Premiere, basically this processes a video file to smooth it out, and in this test, the 3900X was 9% faster than the 9900K at stock, and then 5% ahead with both overclocked. Again the 9900K is closing the gap once we overclock both CPUs.

I’ve used 7-Zip to test compression and decompression speeds, and this test saw the largest difference between these two chips out of all applications tested. With stock configurations the 3900X was performing compression operations 57% faster than the 9900K, and a massive 73% faster for decompression. Interestingly once overclocked the 3900X compression speed lowers slightly, now putting the 3900X a slightly lower 55% ahead of the 9900K while being 62% faster at decompression, again showing the 9900K seems to benefit more here once both are overclocked.

VeraCrypt was used to test AES encryption and decryption speeds and in this test the 9900K was ahead a fair amount both at stock and when overclocked. At stock the 9900K was around 29% faster, though its speed did drop a little once overclocked, while the 3900X saw a slight increase. The results from this test can vary quite a bit though, so each of these averages was taken from 10 runs.

The V-Ray benchmark is another that relies on core count to boost performance, and as a result the 3900X is 37% faster than the 9900K at stock, and then with both CPUs overclocked it’s down just slightly to a 36% lead.

The Corona benchmark is pretty similar, at the stock, the 3900X can complete the task 33% faster than the 9900K, and then 30% faster with both CPUs overclocked. We can see that the overclock only speeds up the 3900X by a couple of seconds here, while the 9900K gets a larger 5-second improvement.

These are the differences between the 3900X and 9900K CPUs in all of these applications, as we can see it depends on the specific workload. In most cases, though the 3900X is coming out ahead, as most of these tests are multi-core workloads. On average overall tests, the 3900X was almost 24% ahead of the 9900K, however, I have grouped in both multi core and lower single-core performance here too so I suppose it’s not a true average, just the average of all my results.

These are the results with both the 3900X and 9900K CPUs overclocked, so 4.3GHz on the 3900X and 5.1GHz on the 9900K. In most cases, the gap between them narrows a bit as we’re able to squeeze out more performance from the overclock on the 9900K, however overall tests the 3900X was still 21% faster on average.

I’ve also measured total system power draw from the wall while running the Blender benchmark. At stock the 3900X is using almost 10% more power, however, it’s worth remembering that it is also completing the task close to 40% faster than the 9900K. With both CPUs overclocked they were much closer together now, however, the 9900K was using more power, while the 3900X was completing the task 37.5% faster. In this specific workload, it would seem that AMD’s new 7nm architecture is helping out here.

These are the CPU temperatures with the same blender tests running. Both at stock and while overclocked the AMD chip was running a bit warmer, but again it’s also worth remembering that it’s got 50% more CPU cores and is completing this test around 40% faster, so based on these differences I think the temperature increase is justified.

Game Testing:

Let’s get into the gaming results next, I’ve tested these games at all setting levels at both 1080p and 1440p resolutions, and just as a reminder I’m also using a 2080 Ti to reduce GPU bottlenecks as much as I can. We’ll start with stock results, then look at overclocked results afterward.

Shadow of the Tomb Raider:

Shadow of the Tomb Raider was tested with the built-in benchmark. In all upcoming gaming graphs I’ve got Intel’s 9900K shown by the blue bars and AMD’s 3900X shown by the red bars. In this game, the 9900K was ahead at all setting levels, at highest settings this results in a 12% higher average frame rate. Stepping up to 1440p the results get much closer together at higher settings, presumably as we start getting more GPU bound at this resolution. There’s just a four FPS difference at highest settings now, resulting in 3.6% higher average FPS from the 9900K.

Assassin’s Creed Odyssey:

Assassin’s Creed Odyssey was also tested with the built-in benchmark and in this test AMD’s 3900X was ahead when it came to 1% lows, however for average frame rate the 9900K was still in front, with a 7% lead at maximum settings. At 1440p the 3900X is still ahead for 1% lows, though only at high settings and above now. In terms of average frame rate the 9900K was now 6% ahead of the 3900X, so again the gap lowers with higher resolution.

Battlefield 5:

Battlefield 5 was tested in campaign mode, I prefer testing this as I can more precisely perform the same test run with less variance, however, multiplayer does seem to be more CPU intensive. At 1080p there wasn’t too much of a difference in average FPS, however, the 1% low from the 9900K was always a fair amount in front. At ultra settings the 9900K was almost 8% higher than the 3900X in terms of average FPS. At 1440p the 1% lows get closer together at high and ultra settings, honestly basically margin of error territory there, however, the averages are still clearly higher from the 9900K, which is now scoring 7.5% higher average FPS at ultra settings.

Far Cry New Dawn:

Far Cry New Dawn was tested with the built-in benchmark, and at 1080p we’re seeing the largest improvement with the 9900K out of all games tested. At ultra settings the 9900K was 20% ahead of the 3900X both in terms of average FPS and 1% low. At 1440p we’re not losing too much FPS in this test, and the 9900K still has a clear win, with just a slightly lower 19% higher average frame rate over the 3900X at maximum settings now, and a 22% higher 1% low.

Watch Dogs 2:

Watch Dogs 2 is a game that loves CPU power, and at 1080p was another game where the 9900K was in front. At 1080p the 9900K was 9.6% higher in terms of average frame rate when compared against the 3900X, though I admit I couldn’t tell you the difference between the two while playing the game. At 1440p there’s not too much of a change at lower settings, however at 1440p the 9900K is now nearing an 11% lead, the gap between the two got larger at this resolution rather than smaller.

CS: GO:

CS: GO is a game that’s well known for performing better on Intel CPUs, so I just had to try it out. I’ve tested it with the Ulletical FPS benchmark, and the results weren’t really too different. There’s a bit of an improvement at lower settings with the 9900K, which to be fair is where a lot of people playing this game are at for max performance, however with all settings at maximum the results were the same, just a 0.3% higher average frame rate with the 9900K. At 1440p now the results are the same at minimum settings, while maximum has pretty much the same 1% low result, while the average frame rate from the 9900K is now 4.5% higher than the 3900X.

Rainbow Six Siege:

I’ve also tested out Rainbow Six Siege with the built-in benchmark. At 1080p with lower settings there wasn’t much of a difference, while at ultra settings this increases to the 9900K coming out 10% ahead in terms of average FPS and with a 14% higher 1% low. At 1440p the frame rates drop back quite a bit, though are still quite high for this test. At ultra settings, there’s now just a 5% higher average frame rate from the 9900K at this resolution.

Gaming Overview:

Out of all 7 games tested we’re looking at a 9.5% higher average frame rate with the 9900K at highest setting levels at 1080p. Some games like Far Cry New Dawn saw a massive improvement with the 9900K, while others like CS: GO saw next to no change, it varies by game. At 1440p as we become more GPU bound the difference lowers slightly to an 8.1% higher average FPS with the 9900K. The only game out of these tests where the 3900X was coming out ahead was for the 1% low in Assassin’s Creed Odyssey.

That’s stock settings, so what about with both CPUs overclocked? As testing every single setting level takes a long time I’ve just picked one set to test overclocked results at 1080p. I’ve also tested the 3900X with Precision boost overdrive enabled, noted as PBO onwards, which essentially boosts power limits. Speeds of each core are also controlled for us here, so in general, we should see better single-core performance when compared with our manual 4.3GHz all-core overclock.

Assassin’s Creed Odyssey:

Assassin’s Creed Odyssey got a nice improvement to the 1% low result with the overclocked 9900K, however, it was still behind what the 3900X was able to offer. While manually overclocked to 4.3GHz the 3900X ends up performing worse, however with PBO enabled it was able to get a nice boost and close the gap quite well.

Battlefield 5:

Battlefield 5 saw very nice improvements from either CPU once overclocked, especially in terms of 1% lows, while PBO was also able to give the 3900X an even higher average frame rate. The overclocked 9900K has a 13% higher 1% low compared to the 3900X with PBO, though in terms of average frames per second, it’s just 4.6% ahead.

Shadow of the Tomb Raider:

Shadow of the Tomb Raider saw a larger improvement with the overclock on the 9900K, though this only boosted average FPS by less than 2%, while the overclock on the 3900X gave a smaller 0.7% higher average frame rate in this test, and no difference between PBO and the manual overclock.

Watch Dogs 2:

Watch Dogs 2 saw the 9900K get a 6.4% higher average frame rate once overclocked, while the 3900X had a smaller 4.7% boost with our manual 4.3GHz overclock, rising slightly to a 5.3% boost with PBO, though honestly as a game I can play with a stable 30 FPS you’re not practically going to notice a difference while playing this.

CS: GO:

In CS: GO we’re seeing a much larger improvement with the overclock on the 9900K, while the 4.3GHz all-core overclocks on the 3900X is performing worse when compared to stock. This seems to be another title were capping the clock speed of the 3900X to 4.3GHz actually ends up with worse off results as single-core speeds can’t boost up higher anymore, and this is why we’re able to get a little extra with PBO.

Rainbow Six Siege:

Rainbow Six Siege didn’t see much of a change once applying the overclocks. At stock settings, the 9900K is about 11% faster in average FPS compared to the stock 3900X, however, with the 9900K overclocked to 5.1GHz and 3900X with PBO enabled this lowers to a 7% lead, so despite not catching up PBO is certainly narrowing the gap.

These are the differences between these games at stock, shown by the purple bars, and with best case modifications, shown by the red bars, so the 5.1GHz overclock on the 9900K and the 3900X with PBO enabled. With PBO enabled we saw equal or better performance in games from the 3900X compared to our manual overclock, so it appears to make more sense to just turn that on. The gap between the two CPUs narrows with the modifications in place in games like rainbow six, battlefield 5 and assassin’s creed, while CS: GO saw a massive improvement from the 9900k’s overclock.

When it comes down to it I don’t think we’re seeing that big of a difference in gaming performance at stock. To be fair I have only tested 7 games here, however from what I’m seeing you can get decent gaming performance with both, I honestly can’t say I personally was able to notice the extra FPS from the 9900K while testing. Despite this though, it was the clear winner in the games tested. The 3900X was only coming out ahead in the 1% lows for Assassin’s Creed Odyssey.

If you only care about gaming then at first glance it appears that the 9900K is the better choice, however, we need to factor in overall costs. These are the dollar per frame values at 1080p averaged out over all 7 games tested at maximum settings. I’ve got the 3900X shown up the top in red, with a few different options for the 9900K in blue. To use the 9900K you have to buy a cooler for it, making the two bars towards the bottom essentially useless, you can’t use the CPU without a cooler.

Decent coolers like the Dark Rock 4 from Be Quiet are around 75, it doesn’t make much sense to buy a top of the line processor and cheap out there, however for argument's sake I’ve also included the results with a cheaper 50 cooler too. I’ve also added in the 9900KF, which is the same CPU but with the integrated graphics disabled, so you can save some money with that one. Even if we factor in a 50 cooler the 3900X is still ahead of the 9900K, and only just a little more than the KF. Using a proper 75 cooler with the 9900K though and the 3900X is further ahead in terms of cost per frame.

Here are the results with both CPUs overclocked. As we saw the 9900K overclocked better than the 3900X, which now narrows the gap between them. The 3900X is extremely close now to the 9900K, and with PBO enabled the 3900X is slightly ahead of even the 9900KF with good a cooler. It’s worth considering that while overclocked to 5.1GHz you’re definitely going to want to go with a higher-end cooler, so with these results, I wouldn’t personally even consider a cheaper option.

It’s not all about gaming though, as we saw earlier the extra four cores with the 3900X helps it come out ahead in multi-core workloads. These are the cost per frame values while exporting a 4K video file to 1080p with Handbrake, and I’ve chosen this particular test as the 3900X was 27% faster than the 9900K, which was close to the average 24% the improvement we saw out of all application tests, plus it’s a very real-world workload.

At stock, the 3900X is offering significantly better value compared to the 9900K, and if we were looking at a different result like Blender than the 3900X would be even further ahead, it depends on your specific workload.

Once both CPUs are overclocked we see the 9900K results move in a bit closer, as the 3900X overclock saw far less improvement in this particular test. Despite this though, the 3900X is still far better value for a multicore workload like this. Again as these are overclocked results I wouldn’t even consider the cheaper cooler as an option for the 9900K. I’ll also acknowledge that the cooler I’m using with the 3900X is most likely better than the stock cooler, so with the stock cooler we may see slightly different results.

The 8 cores on offer from the 9900K are likely going to be enough for gaming for the foreseeable future, however, if you’re someone that also uses your machine for productivity tasks, as we saw earlier the 50% higher core count can help out quite a lot more there. Speaking of the future, we already know the 16 core 3950X is coming out in September 2019 which will be supported by the same AM4 socket. At this time it’s not clear if current Intel motherboards will be able to push beyond the 9900K.

If you only care about gaming and want the best performance in terms of raw FPS, then you’re of course better off getting the 9900K. If you overclock it and even use a the decent cooler the value proposition between the two in terms of dollar per frame is similar in games. If you’re also or otherwise instead using your system for other productivity tasks where the extra CPU cores are more beneficial though, then the 3900X is offering an excellent all-round solution for the price.

For me, the frame rates I was getting from both CPUs were high enough and the games were playing well with either. I’d be more than happy to sacrifice on average less than 10% FPS in gaming and take the extra cores for other tasks, but that’s just me. It’s also possible, even likely, that over time we’ll start seeing the 3900X catch up to the 9900K in terms of gaming performance, as future titles and game updates continue to make better use of multicore CPUs, so perhaps a revisit in a year or two from now would be worthwhile.

Let me know which CPU you’d pick and why down in the comments, Intel’s i9-9900K or AMD’s Ryzen 9 3900X? I’m interested to hear which you’d go for. I’ve also got the 3700X here for testing, so if you’re new to the channel you’re going to want to get subscribed for my future CPU comparisons.