Ryzen 5 3500X versus i5-9400F in World of Warcraft
AMD or Intel?, many or just few cores? What is better for World of Warcraft? To answer that question I decided to benchmark Ryzen R5 3500X and Intel i5-9400F in a set of WoW scenarios.
R5 3500X versus i5-9400F
Intel i5-9400F is a well known CPU priced in Europe at around 155 EUR. The R5 3500X is a Ryzen CPU available on the Asian markets and usually not available in other regions - where R5 3600 reigns supreme. A tray version from Aliexpress can cost around 110 EUR - if you find any available as AMD isn't keen on having sales of this model outside intended markets. The price may be higher with shipping and other costs of ordering from China depending on your location. I got one and used in this benchmark to have two factory 6C/6T CPUs from both companies. More commonly available R5 3600 will likely be bit more expensive than 9400F (discounts may vary this).
| i5-9400F | R5 3500X | |
|---|---|---|
| Core count | 6C / 6T | 6C / 6T |
| Batch | Microcode revision 906EA B4 | BF 1928SUS Made between 8-14 July 2019, silicon wafer from Saratoga, ATMP in Suzhou, China |
| Base clock | 2900MHz | 3600MHz |
| All core turbo | 3900MHz | 4100MHz |
| Single core turbo | 4100MHz | 4100MHz |
| Rated RAM frequency | 2666MHz | 3200MHz |
| Max memory bandwidth | 39,74 GiB/s | 47,68 GiB/s |
| PCIe lines | 16 PCIe 3.0 | 20 PCIe 4.0 |
CPUs are quite similar to each other. 9400F has lower base clock but can boost to same 4100MHz in single threaded loads. At multicore it's bit lower. AMD product can reach 4100MHz on all cores but it frequency can also vary based on thermal headroom.
Both are six core and six thread and should perform quite similarly. Intel will have lower inter core latency where as AMD will have IPC, memory frequency and bandwidth advantage and those differences may play some role in specific workloads. Reason why i9-9900K is still top performer in some games is due to that lower latency. Also AMD Ryzen performance is best with high frequency RAM where as Intel is less reliant on this and may perform well with cheaper memory kit.
3500X is very similar to popular R5 3600. The difference is that it lacks hyper-threading which is a downside in some productivity apps while games are usually not affected by it. A Czech website pctuning.tyden.cz benchmarked R5 3500X from same batch, Hardware Unboxed, Gamers Nexus and other content creators put the R5 3500X to the test against all other modern CPUs including R5 3600 and i5-9400F so you can check them out if you want games-wide performance numbers.
Test platforms
| i5-9400F | R5 3500X | |
|---|---|---|
| RAM |
2 x 8 GB G.Skill Ripjaws V (F4-3200C15D Samsung B-die) 2666MHz for Intel and 3200MHz for AMD |
|
| Cooler | Spartan 3 Pro HE1024 Ceramique 2 thermal compound |
|
| Storage | 512GB Apacer AS350 | |
| GPU | GTX 1070, driver: 442.19 | |
| Motherboard | ASRock H310CM-DVS | Gigabyte B450M DS3H |
| BIOS version | 4.20 | F50 |
| WoW version | 8.3.0 (33369) from 13 Feb 2020 | |
| WoW settings | Mode 7, 1080p, FPS limiter off; DX12 mode; no addons | |
Methodology
I used MSI Afterburner to record average, 1% low and 0.1% low FPS for given amount of time in various scenes/scenarios within World of Warcraft. In more random ones (players presence) multiple recordings have been made.
The 1% and 0.1% low FPS is a metric indicating fluidity of the game. Even if you have 60 or 90 FPS on average very low FPS for 1% or 0.1% can indicate that the gameplay isn't fully fluid. With like 30FPS at 1% lows you may notice that some animations or rotating your camera has some stuttering to it.
Test Results - R5 3500X versus i9-9400F
Boralus flyby is done by keeping Boralus in the field of view when flying via flight paths on the side of the city. There is a lot of structures here to render. Both CPUs managed to get very similar scores with Intel one having few more extra frames.
In the Chamber of heart we constantly have players but the area is small and relatively simple. This benchmark was done by looking at the chamber from the side (having all player in the field of view). Results are very similar to the Boralus flyby although Intel has better 1% and 0.1% lows here.
Dalaran (Legion) can load for quite a while if you are not on a SSD as well has higher performance requirements in some areas of the city. This test was done via flying multiple times around the main street of Dalaran. Some areas will have higher or lower FPS and via multiple circles this is the average result. Also players presence adds bit of random factor to the scores.
In this test we can see how 1% and 0.1% low FPS is much lower than the average one. The FPS is not low enough, but it's close to become noticeable in some animations (like camera movement). Ryzen got better 1% and 0.1% lows while Intel got few FPS more on average. With the players random factor you can call it a tie.
Dazar'alor, looking at the seal gives some structures to render as well as lots of players moving in and out. There is some randomness to this benchmark but still fluidity in such scenarios is needed to have good gaming experience. Intel wins this one, although some level of randomness is present in those results.
Karazan the TBC raid can be used to somewhat simulate a raid/dungeon encounter. It's old content but we have access there to large quantities of mobs (4 rooms) than when pulled generate results very similar to what current raids/M+ do (aside of edge-case scenarios of some graphics effects etc.).
Both CPUs got pretty much same score with Ryzen pulling ahead barely.
Out of combat Karazan can be a reference on how big of an impact to the performance was that combat scenario. Just kill all the mobs, loot them and after a moment they will despawn. At that point the FPS should be much higher.
In this test we see Ryzen 5 3500X pulling ahead. There can be some variability here with camera angles being slightly different, although I tried to place it roughly with the same field of view.
Stonard is a place you can visit before Karazan, and is an example of an old zone with very simple graphics and no players. Here you should see max framerate of all places. AMD is pulling ahead in this benchmark as well.
Benchmarks summary
As we can see both CPUs performed nearly identical with maybe slight inclination for i5-9400F. If you are thinking about buying or building a PC for WoW and other games then if you are on a budget you have 9400F or Ryzen R5 3600 as a good but not cheapest options. Hypertherading on the 3600 should not affect WoW performance but it can affect non-gaming tasks or help with multiple apps running (like WoW, Discord, browser and alike or even streaming).
For purely or mainly WoW those CPUs are a bit of an overkill (unless you really want high fidelity high framerate) so if you really have to go cheaper you can look at cheaper quad core high frequency Intel Skylake CPUs or maybe Ryzen 1600 AF if available (it's a Ryzen+ chip at a very low price).
Going higher at 9700K or 9900K or stronger Ryzen 7 and Ryzen 9 is an option but it has to be paired with matching GPU, RAM and storage. That's a full blown gaming/productivity system and I'll pass on going further into this in this article.
Test Results - i5-9400F at 6, 4 and 2-core configurations
Using BIOS option I limited i5-9400F to run at lower amount of cores. This can be used to showcase how performance scales with amount of cores for Intel system. For AMD it could look similarly although there can be differences on how the OS and the CPU itself manages and distributes uneven load on cores.
Boralus has a slight decrease at 4C and bigger at 2C.
In the Chamber, being a small space even the 2C configuration holds good average framerate but 0.1% lows already show it weaknesses.
Dalaran has a steady decrease in framerates. Even though average is high you can start seeing stuttering on lower core count systems.
Dazar'alor shows significant FPS changes when going to lower core configurations. There is a lot of structures there plus players and NPCs. A lot of draw calls and tasks that can be spread across cores. Just note that this spot has some randomness to it.
Stonard is a simple spot... and weirdly enough 4C configuration got best results here (few recordings). This could potentially indicate that overhead of spreading the load across more cores is higher than workload needed to render such simple area (compare this to Dazar'alor).
Karazan saw only slight performance drop on the 2C configuration. 4C and 6C are nearly identical.
With empty Karazan the 4C got bit better results than 6C so there can be something with rendering old simple areas that doesn't benefit multi-core configurations.
As you can see it's good to have at least a quad core Intel CPU to handle WoW. For best performance in modern areas you already should be looking at good 6 or 8 core CPUs (or more if you have usages for such CPUs outside of gaming).
Ryzen on lowered core count
Ryzen and Threadripper CPUs can be configured to work with some cores disabled. As the cores are grouped into CCX (core complexes) not every core will be equal - disabling one 4-core CCX is different than like disabling 2 cores on two CCX each. This can be done via Ryzenmaster application but should not be done without a clear reason. For the most part it's intended for Threadrippers and in edge scenarios for Ryzen CPUs when an application can't work with large core count or when with some Threadrippers you want to avoid running NUMA nodes.
With 3 cores per CCX in 3500X I was able to disable 1 core per CCX. This had a significant effect on performance. Even when average FPS was comparable the 1% and 0.1% lows could be really low. Out of combat Kazaran had 133,5 FPS average but 16,8 FPS at 0,1% low. Stonard was nearly identical on all metrics so the stuff is more complex. In some areas stuttering was easily noticeable in such configuration.
Conclusions
It turns out that AMD and Intel are quite even and there is no always best part to pick. 9400F and 3500X (3600) are a good mid-range picks. R5 1600 AF (the AF is important) could be a value option. If price is equally low you can look at Intel quad core parts from Skylake or newer series. If the part has high clocks (or if it's an unlocked K-part) it could be better than R5 1600 AF Ryzen+ for WoW (although in other games or tasks core advantage could become a thing). Quad core i3-9100F is weaker than 1600 AF in modern games and productivity apps.
As of writing this article second gen Ryzen is starting to see some discounts, while Intel is preparing yet another 14nm refresh and price drops - some bargains may show up near you. Later this year we will see third gen Ryzen (4000 series) which should be even better than current ones. Plus rumors of new Nvidia GPUs. Do you wait or do you buy now? This is the question.
It's also good to keep the game on a SSD. Right now the prices are quite low and even a cheap but decent SATA SSD will do. For Windows + WoW I would recommend at least 240GB+. You need some sizeable free space for the SSD to perform. For more you can check 512GB and 1TB SSDs.
As for RAM 8GB would do, although 16GB is a standard right now. Ryzen needs fast RAM to perform as advertised so with 3000-series Ryzens you would be looking at 3200MHz-3600MHz RAM. With 9000-series Intel you may be fine with 2660MHz or at best bit more (overclocked RAM frequency support depends on the motherboard). Just note that after installing new RAM you have to go into BIOS and select the XMP profile to have the desired frequency and not base 2133MHz.
If you have problems with WoW performance you can check my debugging tips in my previous article. And if you want to see WoW benchmarks on some older hardware (multiple GPUs and CPUs) you can check the previous benchmark article.
Comment article