Having reviewed Ryzen 5000 12-core and 16-core models, today we're testing the Ryzen 7 5800X, AMD's latest 8-core CPU. So far we've been impressed by the Ryzen 9 5900X and 5950X, so let's continue to work our way down the product stack.
Probably this won't be the best way to start the review, but we've got to say the 5800X looks like the least compelling processor in the new series. There's certainly nothing wrong with an 8-core, 16-thread processor clocked at up to 4.7 GHz with high IPC. In fact, you could argue that's the sweet spot for gamers. The problem is price. At $450, it's just $100 cheaper than the 5900X, and while that nets you almost 20% in savings, you're also getting 33% fewer cores.
Whereas the 5900X costs $46 per core, the 5800X costs 22% more at $56 per core – and even 12% more than the 5950X flagship. If you're wondering why AMD has positioned the 5800X so poorly, the answer is simple: it doesn't make sense for them to sell it any cheaper.
As you're likely aware, Ryzen processors consist of multiple chiplets or smaller dies, rather than a single monolithic die like what we see with Intel processors. In the case of Zen 3, a CCD or 'Core Complex Die' packs 8 cores, so a CPU like the 5950X features two CCDs with all cores enabled for an 8+8 core configuration. Then there's a third die, called the I/O die which houses the dual-channel DDR4 memory controller, PCIe 4.0 root-complex, and a number of SoC features such as SATA and USB ports.
Making up a CPU like the 5950X are two CCDs and a single I/O die, and the same configuration is used by the 5900X, but the 12-core version doesn't require fully working CCDs, rather detective silicon with one or two failed cores can be used. Each CCD only has 6 of the possible 8 cores enabled, resulting in a 6+6 core configuration. That makes these 6-core CCDs less valuable as they can't be used in the more expensive 5950X. So you see where this is going...
The Ryzen 7 5800X and R5 5600X only use a single CCD, but the 5800X requires top shelf silicon with all cores enabled whereas the 6-core 5600X receives the same lower binned silicon used by the 5900X.
If the 5800X was to come in at roughly the same cost per core as the 5900X, it would cost just $370. But for the sake of this comparison, let's bump it up to $400 to price match the 3800X, which is also half the price of the 5950X. At that price, AMD is making less profit on the highest quality silicon, so rather than sell it at a lower margin in a $400 5800X, they're better off saving it for the $800 5950X, or alternatively selling the 5800X at an increased price, which is what they've been forced to do. It appears as though these 8-core chiplets are very valuable to AMD and we suspect they're also wanting to save as many as possible for Epyc 3 server processors, where the margins are even greater.
All of this leaves the 5800X in an unfortunate position where it ends up costing more per core than any other Ryzen 5000 series processor. For that reason we feel most will either end up being upsold to the 5900X, or opt to save some money and go for the cheaper 6-core 5600X.
Either way, it's time to test and then we'll re-evaluate the value of the 5800X towards the end of the review. For testing the AMD CPUs we're using the MSI X570 Godlike with four 8GB G.Skill TridentZ DDR4-3200 CL14 memory modules for a 32GB capacity and then cooling all test systems is the Corsair iCUE H150i Elite Capellix AIO. All productivity testing was performed using a GeForce RTX 2080 Ti, but for the gaming benchmarks we used the top-of-the-line RTX 3090 GPU.
Benchmarks
As usual we'll start with Cinebench R20. The 5800X scored 5982 pts in the multi-core test and that meant it was 22% faster than the 3700X, a serious performance uplift and it's what we've come to expect from Zen 3. We didn't bother retesting the 3800X as it's really no faster than the 3700X. In this test we're talking about a 3% performance boost.
When compared to the 10700K, the new AMD CPU is 20% faster even though both chips pack 8 cores with 16 threads. In fact, the 5800X is just 7% slower than the 10900K which features 25% more cores.
When testing with Cinebench we measured the 5800X's clocks. For the multi-core test where all cores are heavily loaded, the 5800X clocks at around 4.55 GHz which is well above the advertised 3.8 GHz base clock frequency. AMD also advertises a max boost clock frequency of 4.7 GHz and this should be achieved in single core or lightly threaded workloads. In the Cinebench single core test the 5800X typically operated at 4.85 GHz, so that's 150 MHz over the advertised spec.
The 7-zip compression performance is excellent. The 5800X even managed to edge out the 10900K as it was 24% faster than the 3700X and 22% faster than the 10700K. In other words, the 5800X is now the fastest 8-core desktop CPU by a country mile.
Decompression performance is also good, though the margins do shrink a little. Compared to the 3700X, the 5800X is 19% faster and it's also 31% faster than the 10700K, so a brutal 8-core smackdown there.
AES performance has improved by 12% over the 3700X and that made the 5800X a whopping 54% faster than the 10700K.
We're looking at a 16% performance uplift over the 3700X in Blender making the 5800X 12% faster than the Core i7-10700K. It was also 16% slower than the 10900K and 31% slower than the 5900X, which makes sense given it packs 33% fewer cores.
As we saw with the 5950X and 5900X, the gains for Zen 3 in V-Ray are extreme. The 5800X was 30% faster than the 3700X and not a great deal slower than the 10900K.
Performance gains in the Corona benchmark have also been strong and we're seeing a 26% performance uplift for the 5800X over the 3700X. Compared to the Intel competition we're talking about an 11% boost over the 10700K.
Code compilation performance gains are smaller, though a 12% performance uplift is nothing to sneeze at. It means AMD is 15% ahead when comparing the 5800X and 10700K, too.
In DaVinci Resolve Studio 16 the CPU doesn't have a huge impact on performance, but we're still looking at a 6% increase for the 5800X over the 3700X.
It's a similar story in Premiere Pro. The 5800X is beating the 3700X and 10700K by a 14% margin, while it was just 2% slower than the 10900K.
Photoshop relies heavily on single core performance and with Zen 3's massive improvement on this aspect, the 5800X is 27% faster than the 3700X. Moreover, because the 5800X, 5900X and 5950X all feature similar single core performance, they deliver comparable results in this benchmark.
After Effects is another program that relies heavily on single core performance and as a result the 5800X is able to beat the 3700X by an impressive 23% margin. It was also 12% faster than Intel's best mainstream desktop CPU and 26% faster than the 10700K.
Power Consumption
The Ryzen 7 5800X uses ~20 more watts than the 3700X, though that's not entirely surprising as the 3700X is in that sweet spot for efficiency. The 3800X, for example, consumes 20 watts more, which would make the 5800X and 3800X fairly even in terms of power usage.
It's also worth noting that the 5800X was 16% faster than the 3700X in this test, so a 13% increase in total system power is still very good in terms of efficiency. It's also 30 watts less than Intel's 10700K.
In terms of temperatures, we ran a quick comparison between the Ryzen 7 3800X and 5800X tested in a 21C room using Corsair's iCUE H150i AIO, MSI X570 Godlike, installed inside the Corsair Obsidian 500D. We saw the 5800X running 3 to 5 degrees hotter than the 3800X, which is pretty good given it's clocking just over 400 MHz higher at a similar voltage.
Gaming Performance
For our gaming tests we used Nvidia's RTX 3090. Checking out Far Cry New Dawn first, we find that in this latency-sensitive title the 5800X is a whisker faster than the 5900X and 5950X, basically it matched the 10700K making it just 3% slower than the 10900K.
That's an 18% performance uplift from the 3700X, which can be seen trailing even the Core i5-10400, at least when comparing average frame rates.
Performance in Tom Clancy's Rainbow Six Siege is strong. The 5800X matches more expensive Zen 3 processors with 508 fps on average. That made it 2% faster than the 10700K and 19% faster than the 3700X, another big generational performance leap.
We're also looking at a 16% performance uplift in Watch Dogs Legion, going from 99 fps up to 155 fps. The 5800X matched the higher-end 12 and 16-core models and both the Intel 10700K and 10900K.
The 5800X was blazing fast in F1 2020, beating its predecessor by a 20% margin and even edging out the 10700K by 5%. Basically it matched the 5900X and 5950X to come in just 3% behind the 10900K.
As seen previously with the 5900X and 5950X, Zen 3 offers solid performance in Horizon Zero Dawn and the 5800X drops just a few frames when compared to higher-end models. That meant it was 3% faster than the 10700K, matching the Core i7's level of performance and this was possible due to an 18% performance boost over the 3700X.
The gains seen in Borderlands 3 are less impressive, here the 5800X was 7% faster than the 3700x and that meant it was a little slower than competing Intel parts. It was just 4% slower than the 10700K though, which we'd generally deem a tie.
Where we don't see a tie is in Death Stranding, here the 5800X was 12% faster than the 10700K and you'd be hard pressed to make up that margin by overclocking the Core i7 part. The 5800X is also 33% faster than the 3700X which is great, though having now looked at the 5900X and 5950X, we're starting to get spoiled to seeing margins like this.
The 5800X managed to match the 10900K in Shadow of the Tomb Raider with 165 fps on average and that made it 28% faster than the 3700X and 9% faster than the 10700K, while it was just 4% slower than the higher-end Zen 3 base models.
Interestingly, the 5800X isn't as impressive as the 5900X and 5950X in Hitman 2, pushing the RTX 3090 to 151 fps at 1080p, placing it on par with the 10900K. Certainly not a bad result but we also weren't expecting it to be 7% slower than the 5900X.
The 5800X performed well in Star Wars: Squadrons, basically matching the higher core count parts to match the 10900K and beat the 10700K by an 8% margin. We're also looking at a big 27% performance uplift over the 3700X.
Last up we have Serious Sam 4 where the 5800X is seen matching the 5900X and 5950X with performance that placed it 11% ahead of the 10900K and a whopping 27% ahead of the 10700K. Moreover, we're looking at an insane 51% performance increase over the 3700X.
Now, if we average the gaming performance seen across all our game sample, we find that the 5800X is roughly on par with the 10900K and for the most part isn't much slower than the 5900X and 5950X. It also comes out 6% ahead of the Core i7-10700K and 23% ahead of the older 3700X.
Overclocking
We were able to hit an all-core 4.7 GHz overclock with our Ryzen 7 5800X using 1.375v, so the same frequency achieved by the 5950X. In the case of the 5800X though, this only boosted multi-core performance by 5% as the all-core frequency was increased from 4.5 GHz to 4.7 GHz.
This overclock reduces single core performance as we're now operating 150 MHz slower and as a result the Cinebench R20 performance dropped by 3%.
We're also only seeing a 4% performance uplift in Blender, so it hardly seems worth the trouble, especially given the decline in single thread performance.
The upshot being that we're only looking at a 3% increase in power consumption, so that won't really affect thermals.
The hit to gaming performance isn't huge, as we saw in the single core Cinebench R20 test, we're looking at a low single digit percentage drop, in Rainbow Six Siege we're talking about a 4% performance loss.
Performance in Far Cry New Dawn though goes unaffected, we're looking at the same 132 fps on average. Still although we didn't see a performance decline here, it seems quite clear that overclocking is pointless for this CPU.
Price to Performance
Unfortunately the massive price hike does mean that the 5800X is worse value than the previous generation 8-core parts such as the 3700X. You're paying a 12% price premium, though that does place it on par with the 10700K despite being much worse than the 5900X.
It's a similar story in Premiere Pro, the 3700X is better value and the 5800X is even edged out by the 10700K. At this price point, you're actually better off with the Ryzen 9 3900X.
Then for gaming the 5800X comes out costing slightly more per frame than the 3700X despite the massive 28% performance uplift. It also loses out to the Core i7-10700K in terms of value, though it does make out better than the 5900X as most games can't utilize the extra cores.
Looking at the value equation across all games tested, it's clear that while faster, Zen 3 isn't as good as Zen 2 in terms of value. The 5800X is edged out by the 3700X and even Intel's 10700K presents itself as a better value option at the more affordable $380 price point.
What We Learned
Based on what we've already seen from Ryzen 9 parts and how AMD has positioned its 8-core variant this generation, it's not surprising to learn that the Ryzen 7 5800X is a bit disappointing. In our opinion, AMD has unwisely afforded Intel some breathing room here, as the 5800X isn't the obvious choice for those looking to spend up to $450.
At least for gamers, the much cheaper Core i7-10700K offers more value. Yes, it's a little slower, but at $380 it's the better deal. The only weakness on Intel's platform is the lack of PCIe 4.0 support, but not everyone is planning to take advantage of that right away. Buyers may be faced with a situation where those looking to spend ~$400 are better off saving a few bucks by going with Intel, or alternatively they will need to dig deeper and come up with another $100 to land the Ryzen 9 5900X.
It's the same story for productivity, where the Ryzen 9 5900X offers a lot more bang for your buck while the Core i7-10700K is fairly similar in terms of value. Compare that to the match up we see between the 5900X and 10900K and it's hard not to walk away disappointed with the 8-core model. The 5900X dominates the 10900K in everything: performance, substantially better efficiency, and a better feature set, all for about the same price.
Sure, if we put pricing aside for a minute, the Ryzen 7 5800X remains a phenomenal leap forward from AMD's previous-gen 8-core processors. A 23% gaming performance uplift on average is amazing, and as we said earlier, AMD now offers the world's fastest 8-core desktop CPU.
But at $450, we'd recommend skipping the R7 5800X. You can either spend less on the 10700K or maybe the 5600X will be what you need. Alternatively, you could spend more to get the 5900X, or keep your eye open for a good deal on a 3rd-gen Ryzen part.