Upon release a few media outlets scrambled to the benchmarks and it quickly became apparent that the game was extremely CPU intensive and not in the way that ARMA 3 or Planet Coaster are. Rather than just taxing a few threads heavily the game eats up cores, lots of cores. There are reports of it using up a Threadripper 1920X for example. As a result modern quad-cores were getting slayed and there were reports that the game was simply unplayable on the Core i5-7600K. And well, that peaked our interest. We’ve already seen the 7600K and other quad-core CPUs struggling in big 64-player Battlefield 1 battles and while the performance isn’t always ideal, it’s certainly playable in our opinion. Anyway I believe it’s the Computer Base results that have have caused most of the excitement as they showed the Core i5-7600K getting trampled by the Ryzen 5 1500X as it only managed to match the Ryzen 3 1300X. However the 1% low results were the most shocking as the i5-7600K dipped down to 41 fps making it slightly slower than the 1300X. Honestly I’m not sure how that’s possible, but let’s ignore that for a moment.
Essentially the 7600K was almost 30% slower than the Ryzen 5 1600X and almost 40% slower than the 7700K. Countering this information though, were results published by GameGPU around 3 days earlier showing the 7600K never dropping below 60 fps and consequently beating the Ryzen 5 1400 while easily beating the Ryzen 3 CPUs. The higher core count Ryzen parts still do very well in their test but the 7600K hardly looks pathetic. Looking at where each media outlet tested, as best as I can tell GameGPU actually tested a more demanding section of the game, which adds to the confusion. Therefore we decided to have a look and see if we could work out what was going on. For measuring CPU performance we’ve found getting on your steed, in this case a camel, and having a trot through the open-world city of Siwa (seewa) is the way to go, there’s generally loads of non player characters here that give the CPU a hard time. So using the GeForce GTX 1080 Ti and three quality presets, let’s have a look at some of the results in graph form before checking out some gameplay.
The Benchmarks
Let’s start with the Ultra High quality preset, we are using a GTX 1080 Ti after all. Here the 8th-gen Core series is limited to just over 90 fps on average with a 1% low result of 71 fps. The 7700K roughly matched the average but was 7% slower for the minimum as it’s often close to being maxed out. Then we see a pretty large drop for the 7600K which bottomed out at 52 fps making it 21% slower than the 7700K and 27% slower than the new Core i5-8400. This also means it’s a tad slower than the Ryzen 7 1800X and Ryzen 5 1600X, though faster than the R5 1500X.
We would just like to point out that all CPUs were tested using the same DDR4-3200 CL14 memory. I’ll also get to it in a moment with some gameplay footage, but the 7600K as well as the 1500X both provided highly playable performance in busy sections of the game.
Reducing the quality preset to high provides a decent performance uplift and the Core i5-7600K in particular does much better. Whereas the minimum frame rate for the 1500X was increased by 17% the 7600K saw a massive 27% increase to 66 fps and was now able to deliver very smooth performance. The 7700K was also able to max out the GTX 1080 Ti alongside the 8th-gen Core processors.
Dropping down to the medium preset only boosted the minimum frame rates of the Core i5-7600K and Ryzen CPUs by a small margin. Interestingly though, it did boost the minimum and average frame rates of the Core i7-7700K as well as the 8th-gen CPUs quite a bit, so that does suggest the Ryzen and 7600K processors are creating a system bottleneck.
Here is a quick look at how the 7700K, 7600K and R5 1500X scaled using the various quality presets. We see strong and consistent gains as the GTX 1080 Ti is afforded more breathing room with the 7700K. The 7600K is near maxed out using High, while the R5 1500X sees virtually no improvement from High to Medium.
Game Footage
We’ve seen the graphs and I know there’ll been those screaming that bar graphs don’t tell the full story, the 7600K would have been a stuttery mess. Well, in an effort to head off that noise here is a look at some gameplay footage recorded with an external capture device. All footage was recorded at 1080p using the ultra high quality preset, so worst case scenario then. First let’s compare the 6-core/6-thread Core i5-8600K to the 6-core/12-thread Ryzen 5 1600X. Here you can see that all 6-threads of the 8600K are heavily utilized and at times the processors are completely maxed out, yet frame rates remain high and more importantly consistent. The game was jitter free in my testing. The CPU isn’t overclocked, multi-core enhancement isn’t enabled and we have 16GBs of DDR4-3200 memory installed. Something key to note here is the GPU utilization which is locked pretty much at 97% on the 8600K system. Now it we look at the R5 1600X, the GPU utilization is mixed in with the CPU threads so sorry about that, we can see that GPU utilization is usually around 80% but does fluctuate quite a lot and at times dropped as low as 53%. This is interesting as CPU utilization almost never cracked 90% and was often around 80%. Despite this due to the much lower GPU utilization the Ryzen CPU was overall much slower. Okay so time for the quad-cores, the 4-thread 7600K vs. the 8-thread 1500X. Both CPUs were utilized heavily, the 7600K was just about constantly maxed out while the 1500X sat around 95% for the most part. GPU utilization was similar at around 70 - 80% but at times did jump up much higher on the Core i5 processors which is ultimately what gave it the edge. Under heavy load they were much the same but when the CPU was afforded a little breathing room the 7600K ran away a little. The key thing to note here is that even with a lot of NPCs around both CPUs provided a smooth and playable experience. For those wondering dropping the memory speed on the 7600K configuration to DDR4-2400, from 3200, reduced the frame rate by 5 - 10 fps. So a decent reduction there but not massive and performance was still playable.
Putting It All Together
This wasn’t a test we planned for this week. In fact, we have something much bigger in the works which you’ll see in the next day or two, so consider this a filler. The main goal here was to see how lower end CPUs, not low end, “lower end” CPUs handle themselves in Assassin’s Creed Origins. In particular we were really keen to see if the Core i5-7600K would get crushed in a game we can accurately benchmark, but that wasn’t to be. We know the 7600K struggles in 64-player Battlefield 1 battles, but sadly it’s absolutely impossible to accurately benchmark a multiplayer title. There’s no point walking around in an empty map or avoiding the action, that defeats the purpose entirely.
Anyway, getting back on track, the 7600K played Assassin’s Creed Origins just fine and the section of the game we tested saw zero frame hitches. We’re keen to explore the game more, but for now we are satisfied with the testing done. It was disappointing to find that the Ryzen 7 1800X and R5 1600X couldn’t hold a candle to the Core i5-8400 in spite of the rumors circulating the Internet. As many of you know we really like the Ryzen 5 1600 as a balanced/best value offering here at TechSpot and while it did lay waste to the 7600K, Intel’s rapidly aging quad-core still provided playable performance. Meanwhile the HT enabled Core i7-7700K was able to keep the 1600X in its rear view mirrors. Shopping Shortcuts: There’s certainly more testing to be done but for now although Assassin’s Creed Origins looks to be very heavy on the CPU, the results and performance trends aren’t that much different to other recently released titles such as Project Cars 2, just to cite an example.
