How does Soulframe perform?
I finally got access to Soulframe preludes, I’m a long-time Warframe player, so I’ve been waiting ages for my access code! I have two graphics cards to benchmark here, the RTX 3070 and the RX 580 2048SP. So, a medium/high-end graphics card and an entry-level one. I’ll also be doing visual comparisons with low, medium, high and very high. I’ll also test 1080p v 1440p as well as DLSS and FSR for the RX 580.
Corrections:
- 1:11 I would eliminate all the enemies I could see before I would start benchmarking. This means that all mobs would be gone on the path I took.
- 2:32 The RM750x is a 750w PSU.
- 3:11 FSR was on Performance, NOT Auto.
Settings:
- All Benchmarks tested on a 3-run average.
- CPU and GPU are stock, with no overclock.
- Games tested in Fullscreen.
- Vsync is off.
- Nvidia Low Latency is Off.
- Frame rate uncapped.
- Ray Tracing is Off.
- REBar is On for the RTX 3070.
- SAM is Off for the RX 580.
- Nvidia Driver Tested on: 561.09.
- AMD Driver Tested on: 24.9.1.
- Latest Windows 10 Update 22H2.
- SoulFrame tested on DX11.
- All benchmarks done during ‘Day Cycle’
- DLSS is set on Quality.
- FSR is set on Performance.
- Benchmarks for RTX 3070 Patch: Hotfix 3.
-
-
-
- RX 580 Patch: Hotfix 4.
-
-
-
- OS: Windows 10
- Nvidia Driver: 566.03
- AMD Driver: 24.9.1
Test system(my system):
CPU: AMD Ryzen 5 5600 Stock
AIO: Arctic Liquid Freezer III 360mm
GPU: Zotac RTX 3070 Twin Edge LHR 8GB – Stock
-
- MLLSE RX 580 2048SP 8GB – Stock
SSD: Samsung 860 Evo 500GB (OS), Samsung 870 Evo 1TB(Games)
Ram: 32GB (4x 8gb) Patriot Viper CL18 DDR4 4000 (CL19-19-19-39)
Ram Cooler: Jonsbo NF-1 (Only used with Patriot Viper Kit)
Motherboard: Gigabyte Aorus Master B550 PSU: Corsair RM750x Gold 750w
Case: Corsair 5000D Airflow
Power supply: Corsair RM750x 750W (Gold)
Fans: 3x Inwin Sirius Loop ASL120 120mm (Intake), 3x Fractal Design Prisma AL-12 120mm (Intake), 1x Silverstone Air Blazer 120R 120mm (Exhaust), 2x Random Corsair 120mm fans (Intake on bottom).
Test system(my system):
Here is the video of benchmarking Soul Frame!:
Timestamps:
0:00 – 0:44 Intro and Structure of Video
0:45 – 2:29 Benchmarking Location?
2:30 – 3:57 System Specs
3:58 – 5:27 Benchmark – Low, Med, High
5:28 – 6:09 Low -Visual Comparison
6:10 – 6:47 Med – Visual Comparison
6:48 – 7:22 High – Visual Comparison
7:23 – 8:07 Visuals – Low v Med v High
8:08 – 9:06 Benchmark – Very High 1080p DLSS
9:07 – 10:12 Benchmark – Very High 1440p DLSS
10:13 – 11:07 Benchmark – Very High 1080p V 1440p, DLSS v No DLSS
11:08 – 12:17 Very High 1080p – Visual Comparison
12:18 – 14:13 Benchmark – RX 580 1080p
14:14 – 15:22 Custom FSR – Visual Comparison
15:23 – 16:52 Visuals – Low v Med v High v Very High v Custom FSR
16:53 – 17:42 Power Consumption L, M, H – RTX 3070
17:43 – 18:31 Power Consumption – 1080p v 1440p, DLSS v No DLSS
18:32 – 19:21 Power Consumption – RX 580
19:22 Final Thoughts
Graphics Quality Settings:
Low, Medium, and High are Presets.
Very High, Ludicrous Settings:
Custom FSR:
Benchmarks:
Low, Medium, High, 1080p – RTX 3070:
First up we have the RTX 3070, I’ll run through the numbers first and then show you some gameplay. As you can see here, all three presets low, medium and high performed very well on the RTX 3070 with over 300fps on the low settings. Frametimes were decent in all three graphics settings, but the high settings did have a few hiccups when I was doing the benchmark.
The low settings average FPS was 116% faster than medium and 207% faster than high settings. The 1% lows were also really good for 1% lows at 181.27fps.
The medium settings was 42% faster than high for average fps. The medium settings had an ok 1% low at 94.20, which was 30% faster than the high settings.
The high settings were the lowest performing but still provided a more than playable and enjoyable experience at 105.53 average fps, which was 30% slower than medium. The 1% lows were great as well at 72.23fps on high settings.
Very happy with this result showing that this game isn’t limited by the 8gb vram on the RTX 3070 too much.
1080p v 1440p and DLSS v No DLSS, Very High ‘Ludicrous’ Settings – RTX 3070:
At 1080p we achieved a more than playable 86.53fps with dlss off and 65.10fps for 1% lows.
When DLSS was on for 1080p the average FPS was 55% increase in performance over the non dlss and the 1% lows improved being 81.47 on dlss and 65.10 on non dlss which is a 22% difference. So not a bad improvement overall on DLSS. I’ll show you the visual quality difference at 1080p right after the 1440p results.
For a playable game I think the RTX 3070 is probably reaching its limits in 1440p with the higher settings. As you can see here at 1440p, no DLSS, we got 58.73fps, which was playable, and 46fps for 1% lows. While I would have liked something smoother than just 60fps, visually, the game looks really good at 1440p with these higher settings.
DLSS is here to fix framerates at 1440p a bit with 90.93fps which is a 55% increase in performance over non dlss 1440p and for 1% lows a 50% improvement going from 46fps on non dlss to 68.77fps on dlss 1% lows.
There was a solid improvement with DLSS here, and I think there wasn’t too much of a hit in visuals for the game.
Comparing the two resolutions now taking the non-DLSS average FPS 1080p was 47% faster than 1440p at average fps and 42% faster on 1% lows.
For the DLSS comparison, 1080p was 48% faster than 1440p at average fps, and 1% lows at 1080p were only 18% faster than 1440p. Pretty much the same performance decrease on average fps from 1080p to 1440p but 1% lows was a much smaller margin, I’d like to try this out with a better cpu to see if there is any difference in 1% lows in the future.
Low, Medium, High, Custom FSR, 1080p – RX 580:
Now to our lower end or entry 1080p card the RX 580 2048SP. Low, Medium and High settings did not use FSR upscaling and only the custom settings used FSR. This card performed well at 1080p with the low settings having 124.24fps on average and 1% lows being 94.90 which are both the highest for the RX 580. While visually not the best looking the game was more than smooth on low settings.
Next was medium which had a 70% decrease in performance on average fps compared to the low settings. 1% lows did alright at medium settings at 31.67fps considering that average fps was on 37.37fps.
At high settings the game was just not that nice. It was a bit jarring at 23fps and didn’t feel playable to myself. The 1% lows achieved 17.43fps. The RX 580 couldn’t really handle the high settings.
So what can we do? The low settings run really well, but they don’t look great. Here comes the saviour FSR upscaling. At 1080p with these custom settings we got 59.83fps which is pretty decent and a 60% increase in performance compared to average fps on medium settings. Even 1% lows were 50.77fps which made the custom settings feel great. There weren’t any problems that I could see with the frame times in using FSR, and the custom settings, and maintaining nearly 60fps constantly felt great to benchmark.
I didn’t test 1440p as I didn’t see the point of it on this card as we were already struggling with the custom settings to make the game look decent and run well.
Overall, I’m pretty happy that this game is still playable with lower-end cards.
Visual Comparisons: (The video is better for direct comparisons :D). MORE VIDEO VISUAL COMPARISONS WILL BE UPLOADED LATER 🙂
Low, Medium, High – 1080p:
On the low settings we can see that the textures are not as smooth anymore as well as lighting not being as accurate compared to the high or medium. If you take a look at the ground as well the stones seem to be very flat. The shadows that are present in the medium or high settings are gone as well giving brighter image in some areas. The cape on the back of the character did look slightly worse than the medium settings being more jarring and less smooth. There is also a lack of sunrays which I personally like for the ambience of the game when you’re walking around. Some of the finer details are gone from the textures which is an easy way to summarise the low settings.
Next is the medium settings. The textures here are a lot smoother than low with only slight loss on detail from the high settings here. The lighting is also a lot nicer compared low settings with light rays coming through just like the high settings. The characters cape looked pretty similar at medium and high settings. Shadows are actually apparent compared to the low settings showing a little bit more dept but not to the extent of the high settings. I felt that the textures were a nice increase in quality and smoothness over the low settings if you have a look over at the rocks by the waterfall. Not bad for the middle-ground settings.
On the high settings I got the impression that it felt a lot cleaner or crispy compared to the medium settings. There are finer details such as the ball object on the statue, there is even a little bit of shine on it as well. Lighting on the foliage like the trees look a bit more realistic as well compared to the medium preset. The characters texture with the cape on the back seems similar on both the high and medium preset, I can’t see too much of difference here.
While the medium settings are great, generally, I think that the high settings feel a bit more immersive and have a bit more in-depth compared to the medium settings preset.
Comparing the three runs side by side you can see the visual difference between the three texture quality settings to get a bit more of an indication of what all three presets offer. Personally I think that the game looks visually really good with the higher quality settings. If your graphics card can handle it I don’t really see any reason why you would need an extremely high framerate in this type of game considering its souls like. Playing something like Sekiro on the PS5 for example feels good to myself at 60fps and I don’t think that’s an exception for Soulframe as well. But at the same time I don’t doubt for a second that more fps you have or better 1% lows will offer a more fluid in game experience.
DLSS v No DLSS, Very High ‘Ludicrous’ Settings – RTX 3070:
In terms of the Very high or ludicrous settings, there are only a few differences between it and the high preset.
The fog is a bit more detailed, and there isn’t too much else, which is different, as you can see here. Both are quite crisp in terms of the quality and the texture quality seems to be the same on both. I could argue in the particles like in the fire the very high settings look a bit better but it’s a bit of stretch to be honest.
DLSS v Non DLSS
Now, to be honest, I couldn’t really tell too much of the difference between the DLSS and non-DLSS versions, but they are subtle when turning the camera around. When walking around there are some areas where the texture clarity isn’t too good because of the upscaling but it was a bit difficult to tell during the benchmarks and reviewing the footage right now. The truth here is that 1080p or 1440p DLSS is more than fine to turn on for this title and the visual fidelity loss isn’t too much in my opinion, so you can achieve a decent smooth fps while maintaining a relatively high graphics quality with the very high settings on the RTX 3070.
Custom FSR, 1080p – RX 580:
Having a look at the graphics quality for the FSR custom graphics settings at 1080p for the RX 580, visually it looks pretty decent.
The textures aren’t always the best like if you look at the waterfall rocks compared to even the medium settings they aren’t as smooth, they do look like the low settings rocks. But in terms of the light rays that are coming through and the shiny textures as well as some decent shadows I think this makes up for the lack of texture in some areas. If I put these side by side which one is the FSR custom quality and the other is the medium settings. Have a look which one is which?
It’s a little bit hard to tell the one on the right is the FSR custom quality and the one on the left is the medium. I think this gives some perspective into what graphics settings you actually need in terms of sacrificing performance for minor visual gain. It also proves that low end cards still have a place at 1080p with FSR and a bit of lighting or shadow quality. Not a bad result for the RX 580 from what I’ve seen.
Statue Visual Comparison – 1080p (Hover for names):
Statue Visual Comparison taken from my camera Fujifilm XS10 – 1080p (Hover for names):
Upstream Visual Comparison – 1080p (Hover for names):
Downstream Visual Comparison – 1080p (Hover for names):
Power Consumption and GPU Usage:
For the RTX 3070 at 1080p on the wattage, there was only a slight change from the low settings at 195.9w where, as the medium and high settings were at 219w. Again, similarly, on GPU load, the low settings were only 75% compared to the higher 98% on medium and 99.5% at high. Whereas on the GPU memory usage on average 73% of the 8gb was used at high settings compared to the 50.9% at medium and only 36.3% at low. This makes sense as the GPU memory usage has a slight correlation to the performance we got for low, medium, and high settings at 1080p.
At 1080p v 1440p on the very high settings, wattage and GPU core load were both pretty much the same whether it was 1080p or 1440p or if dlss was on or off. But in GPU memory usage the highest came from 1440p very high no dlss at 83.8% due to the higher detail and rendering out in 1440p. Where as 1080p with the same very high settings had 76.1% memory usage, a slight decrease.
When dlss is enabled 1440p got 77.3% memory usage and 1080p got 67.9% memory usage which were both lower than their non-dlss counter-parts.
A similar thing happens with the RX 580 on GPU core Load and Wattage, with low, medium, and high being similar again. The difference came from the amount of memory used. On the RX 580 the high setting had 73.1% memory usage compared to 58.7% at medium and 38.2% at low. While the memory usage was higher at high settings, the benchmark shows us that the GPU couldn’t keep up, and the frames rendered were quite low. The custom FSR result had a similar GPU memory usage to the high preset at 72.5%, but yet the game was much more playable. Its great to see that FSR can sort of breathe a bit of life into these older entry level cards for more graphically demanding games without sacrificing too much visual quality.
Final Thoughts:
- DLSS and FSR are more than fine to use for this game. In my opinion there wasn’t too much visual loss if you want more FPS.
- 8GB VRAM isn’t an issue if the graphics card is fast enough such as the RTX 3070.
- Entry level 1080p cards may struggle like the RX 580 and need some tuning in the settings.
- Soul Frame isn’t that CPU intensive yet even at 1440p. There are usually high levels of GPU load from this game.
- I’d like to see how this game evolves and it looks like an awesome game to benchmark going forward! 😀
- Thanks DE for another cool game, and cheers from another Warframe fan. I hope the game takes off well! 🙂
YSTech: https://ystech.org
YSTech Article: https://ystech.org/soulframe-benchmarked
-YS