The new rendering king?
And there you have it folks, one crisp, clean, overly expensive, armoured-to-the-hilt Intel Comet Lake build, complete with 3TB of storage, 32GB of DDR4 @ 3600MHz, and 10 of Intel’s finest 14nm++ cores.
On the whole the build experience was very enjoyable; working in the Corsair iCUE 465X, despite first impressions, was a smooth ride with no major complications. Even afterwards, when we had to do a bit of troubleshooting to diagnose some suspect memory problems (reseat the CPU, it’s always reseat the CPU), the case provided us with no issues whatsoever.
And with that front criss-cross fan filter out of the way, this thing just looks killer. Corsair, if you’re reading this, you need to pump one of these out in an ITX form factor for sure – 240mm AIO support in the front, 2 x 120mm fans in the roof, and a vertical GPU mount would be a fantastic case. For a mid-tower though? It’s not half bad either.
Taking on the Ryzen
Okay, enough, let’s get down to the brass tacks here. The big question we want to know is whether Intel’s latest Comet Lake flagship can keep up with AMD’s Ryzen 9 3900X. It’s a tough competition. On the one hand, you have our Ryzen competitor with 12 cores, 24 threads, a whopping 68MB of cache, and a fairly solid 4.6GHz max turbo speed (or 4.2-4.3GHz across all cores), combined with a fairly decent stock cooler, available at just $839. On the other side, Intel’s Core i9-10900K comes with 10 cores, 20 threads, 20MB of “Smart Cache,” and a phenomenally impressive 5.3GHz max clock speed (which again is around 5.0-5.1GHz across all cores), all for $999. You don’t get a cooler with it – but to be fair, even if you did, if you were to buy either of these chips, we would wholeheartedly recommend that you pick up an after-market cooler anyway. So ultimately then, Intel is down on thread count and cache, but higher on the pure grunt that is single-core IPC and clock speed.
We should explain our testing methodology here too. Our comparative system, outside of storage, is very similar. It’s actually the 4K Gaming Monster we did last year, but with different memory and one of the RTX 2080 Supers disabled in all of our benchmark tests. Yes, it’s liquid cooled, allowing GPU boost and AMD’s PBO to work their magic, but on the other hand the GPU follows a basic PCB power design and has lower advertised clock speeds than our Zotac RTX 2080 Super in this build.
On top of that, we’re also bringing in a few additional CPU tests to give you a better idea of which platform performs best. There is, of course, more synthetic tests courtesy of Cinebench, TechARP x264, and Fry Render, and additionally we’ve added two real-world tests, thanks to Adobe. Yes, we’ll be putting both processors through their paces by getting each system to render a bespoke five-minute 4K video in Adobe’s Premiere Pro CC, using the H.264 codec and the YouTube 4K preset.
Furthermore, we’ve also put together a short-but-sweet Adobe
After Effects video intro, which again we are going to get both machines to export in 4K, but this time under the AVI file format.
The Results
Boy did we get some interesting figures from this. Straight out of the gate in our synthetic tests the AMD comparison system hammered Intel in Cinebench R15’s multi-core tests, and we’ve got to admit, at 192 points, its single-core performance was impressive as well, albeit still not a patch on the Core i9’s 225. It also beat the Core i9-10900K in TechARP’s x264 benchmark, too. However, this is where things get interesting: In Fry Render Intel won the competition by a single second. That doesn’t sound like much, but this is a rendering test, and Intel is down by four threads.
Moving on to the memory tests, and things fall where we’d expect them to. Latency-wise, Intel always edges out AMD, quite substantially. In part we believe this is down to the fact that AMD’s core complexes (each one housing four cores) have to communicate with each other and the memory when processing requests, and because of that there’s a small delay added on to the latency. It’s an architectural difference that’s difficult to get past. With Intel and its monolithic design, latency remains consistently lower, and has done since the days of Skylake.
Interestingly, despite both platforms supporting dualchannel DDR4, and both systems using 3,600MHz kits with a C18 latency, the Intel system falls behind in pure memory bandwidth, coming in almost 4,000MB/s lower than its AMD rival. We can only put this down to
“Okay enough, let’s get down to the brass tacks here. The big question we want to know is whether Intel’s latest Comet Lake flagship can keep up with AMD’s Ryzen 9 3900X. It’s a tough competition. ”
early teething issues with the motherboard BIOSes themselves, as the results should be very similar.
And now we get to the realworld tests, and this is where things really get shaken up a bit. Let’s keep this brief. 4K Premiere Pro video export? AMD, nine seconds faster. 4K After Effects video render? Intel, one second faster. It’s genuinely impressive.
Despite the extra cores, the extra cache, the synthetic tests showing that AMD had the advantage when it came to memory bandwidth, Intel’s single core (and likely Adobe optimisation) managed to narrow the gap significantly.
It’s nothing short of awesome, especially when that processor is effectively four threads down on a fight like that.
Wrap-up
So in conclusion, it’s a tough one to call. In real-world scenarios, the two processors are neck and neck. Seeing them that close to one another despite the differences is just remarkable, especially given how long Intel’s been on that 14nm manufacturing process. However, it’s still hard to deny just how much AMD has the advantage in terms of cost to performance. For most of those challenges, it was still either on par or beating a CPU from the competition that costs $160 more than it does. And on top of that, it supports PCIe 4.0 storage as well, on a very mature platform. And as you can see from those results, the differences as far as storage goes are staggering. If you’re moving big files around on a day-to-day basis – which let’s face it, is very common with videoediting – AMD’s offering is still by far the better value choice.
That said, you have to admire Intel for what it’s managed to achieve with this. A lot of engineering work has gone into making sure that its 10th generation is still competitive, despite the fact it’s based on a five-year-old architecture and a six-year-old manufacturing process.
It’s a tough one to call. For those of you who do need a fast CPU for work like this, you’re not going to be disappointed picking one of these up, but you’ve got to know that that comes with an extra $160 cost at a minimum, and you lose out on the storage. Yes, you gain in gaming, but that’s not for everyone. The smart choice is AMD, for sure, every time, but there’s no denying that the Core i9-10900K is still one killer chip.