Oblivion CPU Benchmark Thread

[BoredErica]

Inspired by the Fallout 4 CPU benchmark thread here is another one for Oblivion. The reason why I want an Oblivion test is because the engine in Oblivion is very different from FO4 and Oblivion struggles even more w/ CPU perf. In general I find Fallout 3/NV's performance is more similar to Oblivion, and Skyrim Special Edition is closer to FO4, at least when pushed hard. For example, in FO4 the CPU intensive areas tend to cause low L3 hit rates (as measured by Clamchowder's utility) on 5600x (as low as 49%). Similar story if I go to College of Winterhold in SSE fully modded with Splashes of Storms and splashes at 48 iterations (52%). 5800x3d saw a massive improvement in performance in FO4 which makes sense. In Oblivion the L3 hit rates were already okay (~70%), so it did not show massive gains.

1. Ensure latest version of Oblivion is not running any mods, including OBSE or DXVK. (Official DLCs are okay and encouraged.)
2. Replace Oblivion.ini with ini found in this link, and move the save file included to save directory. Keep the read-only flag on the ini. https://mega.nz/file/C1pynYKL#PcJqg3xUrNdaJWKwfpwl59HHiNGmjwKp29D4ldRagNw
3. Get FPS value 3-4 seconds after loading in. Try not to wait too long after that or take FPS before 3 seconds.
4. With just 1 mod (which has no dependencies), ImpeReal City - Unique Districts, FPS can be pushed down to 38FPS already on my CPU. DL is only 150kb. Get it here. (First or second DL, doesn't matter.)

Pictures are appreciated.

If Oblivion Launcher starts and says it is changing graphical settings, it should attempt to do so and fail silently since the ini is read-only. If you dunno where Oblivion ini goes because you don't use a mod manager, by default it belongs in My Documents/My Games/Oblivion.

In my personal game, the stuttering is a lot more annoying but would require an Autohotkey script + CapframeX to test correctly. (I actually have such a script but it requires more set up to test). Worst case I can get in my fully modded game standing still is 23.5fps vs my 5600x score of 37fps modded in this test, and 50fps vanilla. I also have an extreme ini that further pushes FPS down, but I'm not sure it's relevant for now because no CPU can run it well.



Username:
CPU and frequency:
Ram: (If you've manually tuned most or all of the subtimings please note this.)
GPU:

 

[BoredErica]

I dug out my original Oblivion DVD, applied the most recent patch, and got 55fps unmodded, 41fps modded. I have a 5950X, 2x32GB DDR4 3600 C18, Radeon VII.

Thanks for being the first submission on OCUK for this. Very strong for a non-3d zen3 part. Do you know what frequency the CPU was running at, and whether you manually tuned the subtimings of your ram?



My 13600k OC is not finalized yet, since I can't copy paste ram settings from 5600x to 13600k due to different IMC.

 

[BoredErica]

I ran it again using HWiNFO64 to record CPU speed. This time there were some background tasks running, so there are probably more cores running at lower speed than there were when I recorded 55/41fps, and indeed I got ~1fps less this time. It's the best I can do right now though:

So, it looks like there are three cores running at 4.85GHz at any given moment.

I have done a little bit of tuning of RAM timings, however it's ECC RAM, so 1) they never use the fastest chips for this stuff, and 2) there's a few percent hit to memory bandwidth from the ECC. Here's what CPU-Z says the primary timings are (I think I ended up leaving the secondary timings at the default slow JEDEC values):

Ahh thanks. Congrats on the nice score. My plan for ram is to focus on speed first, and then note if user says they manually tuned everything (binary yes or no).

 

[BoredErica]

Unmodded:

3090FE

Thanks for the submission.

 

[BoredErica]

Not sure if I am doing this right but it says video card can't be detected and sets settings to medium?

The first FPS figures are from the unmodded game,with no unofficial Oblivion patch installed,etc. The second FSP figures are with the mod in the OP installed and the unofficial Oblivion patch installed.

Systems specifications:
Ryzen 7 5700X. Three,sometimes four cores were running at 4.65GHZ so the Gamebryo Engine seems to generally use less cores than the Creation Engine.
32GB(2X16GB) 3600CL16 DDR4
Seagate FireCuda 520 2TB(game install drive)
RTX3060TI FE(Nvidia 516.59 drivers)

Oblivion 1.2.0416(GOG version).

FRAPS used for FPS measurements.

Unmodded

Average of three runs:55FPS

Modded

Average of thee runs:39FPS

55 unmodded.

5800X 4.8ghz fixed OC.

RTX 3070.
32GB 3733mhz memory at 3800mhz.

Thanks for the submissions, updated.
When you load Oblivion through launcher (you can go directly to game start w/ mod manager), the launcher will try to update settings but fail because the ini is marked as read only. If ini was incorrectly installed, FPS will be way too high.

My understanding of Oblivion is it mostly just runs on 1 core (but OS might split it across multiple cores to manage heat). It's not as memory sensitive vs FO4. Also far more stutter prone than FO4. A really different beast to be sure... but we'd expect this of course (2006 vs 2015 game, dx9 vs dx11).

FO4 Vanilla Corvega: 104
FO4 Vanilla Faneuil Hall: 85
FO4 Vanilla settlement: 60
FO4 modded Faneuil Hall: 81
FO4 modded settlement: 55
Oblivion Vanilla: 75
Oblivion Modded (for this chart): 55.5
Oblivion My Personal Modded setup (same ini as this test): 35
Oblivion My Personal Modded Setup + Ini adjustments to push light/specularity to very far render distance: 28

 

[BoredErica]

The Gamebryo Engine in Oblivion was the basis of the Creation Engine used in Skyrim AFAIK.

My modded Fallout 4 install has a few 100 mods and large settlements,so performance drops much more significantly over a stock install. It seems both the graphics rendering thread and NPC AI thread are the limiting factor here.

Yes, but so much time has passed and so many changes done, the way the engines work under the hood has changed significantly. Well, they are still both hard on the CPU though. Some things never change. BTW, this is a little OT but, do you use mods that reduce FPS a lot more than my modded FO4 setup? If so, do you know which mods are really heavy?

Speaking of differences, DXVK can be a big perf boost to some dx9 games, like FNV or Oblivion but isn't an auto free perf gain in FO4. In my testing, FNV straight up got better perf with DXVK, but in Oblivion on my old 5600x I had an issue where DXVK made stuttering worse while increasing average FPS. I just assumed it was bugged for Oblivion. Testing with 13600kf just now, and... it works as expected?

This test was done with an AHK script for consistency, where my character noclips through Anvil docks at 255 speed/100 athletics to test stuttering. It's modded and running ini for my normal playthrough. For 5600x 30 passes were done, for 13600kf 12passes (I can do all 30, I justed wanted to share what just happened.)

5600x worst frame time: ~100ms
1300kf: 70ms
13600kf + DXVK: 40ms

I will do follow up tests to confirm it works fine randomly even though I never updated DXVK and only my CPU. I have to ensure no performance degradation occurs because this feels like a fever dream.

 

[BoredErica]

13600kf OC is stabilized. This is my daily driver settings. I can push for 5.8/5.9ghz but I rather play it safer.

Extra data:

An aspect of perf I think is underrated, loading times. Especially with Bethesda games crashing, I want to get back in the game ASAP.

I've repeated this many times to people I talk to: CPU performance often makes larger difference than SSD speed for game load times, AND improves FPS. A drive can be as fast as it wants, there is still overhead for IO requests and CPU still has to do something with the data it gets. While Skyrim runs very smoothly, starting the game up takes a looong time. Starting near vanilla game quickly matters because I intend to do Skyrim texture testing which requires loading the game tens of thousands of times with a script. The loading time decrease here is so much it looks wrong, but seems correct from my testing. It just takes way less time to load in Skyrim. This is why I lean towards RPL over 5800x3d: RPL is faster in Oblivion because large L3 is far less important than in FO4, and the single thread perf outside of L3 is far lower. 13600kf has like 42% lead in that department. So until I get evidence to contrary, I believe RPL will be superior to 5800x3d in game loading time. OFC, 7800x3d will not be behind by much, so it looks a lot more promising. Time will tell. Do I go for Arrow Lake or Zen5 3d? Too early to tell.

Let's look at FPS figures. Massive gains in performance. DXVK on my 5600x increased avg fps but increased stutters. For some reason that problem might just be gone on my 13600kf. I didn't change Oblivion or anything, just the CPU. So... I dunno what's going on. Followup tests must be done to ensure there is no performance degradation.

It ends up being like this, in terms of perf increase. Note that the FO4 figures is higher than my score in FO4 benchmarking thread. I have my own tests I prefer to use in evaluating perf for my own personal game. Massive gains in perf, especially for an upgrade that ended up costing me <$300.

I'm starting to think about 240hz SSE gaming, but FO4/Oblivion are still way behind.

 

[BoredErica]

So, conclusion: is it worth it for me to dig out the Oblivion Wabbajack modlist, Heartland, again try it again with DXVK?

The last time I tried it (with the stock renderer) it wasn't that smooth. I have gone crazy with the RAM but even 64GB is not enough to cache a whole modlist (well maybe Morrowind but I cannot live with the blocky models). Playability-wise I do prefer SSE though even if the game mechanics has been downhill since Morrowind (or Daggerfall, spent far too long making totally custom classes in there exchanging weakness for something for extra points elsewhere).

DXVK in Oblivion is still being weird for me, particularly in conjunction w/ RTSS. Full screen RTSS overlay doesn't work unless I set custom Custom Direct3d Support, but then the game sometimes crashes when I start it. But I never had these issues w/ FNV. In Oblivion I'm not so sure, but it wouldn't surprise me at all if loading times are improved more by having a way faster CPU vs a ram disk. I think hardware reviewers have been getting it wrong with testing loading times only w/ SSDs.

I kind of enjoy Oblivion more than SSE, but the engine in Oblivion is so old. Performs poorly, crashes more often, modding community moved on. I took RPL out for a spin, ran across Skyrim at 7x normal running speed, and only got 1 stutter from Solitude to Whiterun.

Not too long from now there will be used 7800x3d sold cheaply as Lisa Su tries to sell new ones on fire sale to promote zen5. People will see just how massive the latest cpus are vs ones from 2yr ago. 80% gain in fo4/sse, 50%+ in Oblivion, way shorter Skyrim startup times. 5600x -> 13600k is bigger boost for me than 2012-2020 combined. It's insane. And we might be in the same situation again 2yr from now again to an extent w/ ARL/Zen5 3d.

People used to say Oblivion can't run faster because dumb engine, it's "engine limited". No, it's CPU limited and your CPU hasn't gotten faster in ages. And now I've shown serious diminishing returns IMO hasn't set in for loading times either.

 

[BoredErica]

Hey again

Do you have any idea what happened to the benchmark thread over at Anandtech forums ?
I submitted a run and suddenly the thread is gone/deleted ? WUT

Guess i can post it here instead then...

7950x with highly tuned memory settings @ 6400MT/s
(showing aida64 memorybench to give you an idea)

Newest game version from steam and a 3090


unmodded = 90fps
modded = 65fps

imgur.com

Discover the magic of the internet at Imgur, a community powered entertainment destination. Lift your spirits with funny jokes, trending memes, entertaining gifs, inspiring stories, viral videos, and so much more from users.

imgur.com

Recorded a short run to make sure correct settings: (ini was read only)

Oblivion benchmark

90fps

www.youtube.com

I'm not sure, I don't use Anandtech or OCUK that much TBH. I know Anandtech has a lot of safeguards to protect against spammers. I keep getting flagged for manual moderator review because the forum thinks I'm spamming lol.
Also, now is time for me to update the thread here in OCUK.

We do have det0x's 7950x result.


Very strong in general but also vs halgari's 7950x.

 

[BoredErica]

Vishvadeva tested same cpu w/ different GPU (Nvidia -> AMD) and FPS went from 40 -> 36.

 

[BoredErica]

Threaripper result.

 

[BoredErica]

In the popular player house mod Leveler's Tower for SSE, I made a mod few years ago that added occlusion planes for SE version of the house. I tested performance with my old 7600k (GPU was <90% utilization, same ram as now). FPS was 171 without my mod, 200 with. Now with my 13600kf I get 530fps. So x3 the FPS.

I've long since sold the 7600k, so it's not around for benchies anymore. Thought it'd be fun to share.

 

[BoredErica]

Two new 5800x3d results.