• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Fallout 4 CPU benchmark thread(need some Zen3 and Zen4 results!)

I wonder what OCed 12700K and 12900K results will be like. They should be a lot faster cause of bigger cache.
aX0c027

whkNc9o
Unfortunately, nobody with those CPUs has run the test. Neither here nor on the original AT thread.

Anyway, updated scores including a user with a 5800X3D over on AT who has faster RAM than fish:
lYlZcQ7.png
 
The Creation engine truly is utter ****. I really hope Creation 2 turns out some big improvements.
Yep, no arguing there.
These benches here mainly deal with draw calls. The AI and other script mainly run on one thread too.

Engines get harder and harder all the time, for instance CD Projekt have stopped developing REDengine and are going UE5.

What I'd really like to see is an AI framework. A versatile and extendable one - don't want every game's NPC to behave the same - and possible with custom AI hardware in a future console generation.
 
When I get mine, I'll test mem and IPC impact separately to see what's going on.

AMD sadly doesn't have something nice like vtune to do pipeline analysis so it'll be a bit of trial and error to see where the X3D bottleneck.
 
Was bored so I tried it on my laptop, 11900H (24MB L3 Cache).

~4800Mhz CPU 1.23v 4.4ghz cache
16GBx2 3200MHZ C16-17-17-33-1T 1.2v Gear 2
RTX 3080 16GB Laptop GPU

S1 11711 73 FPS
S2 7945 88.1 FPS

AVG 80.5 FPS

Windows 10 21H2. Can probably get slightly higher with a fresh install of win10 as that maxes out memory performance.

AMD users should be using BGS (Bank Group Swap) Alt Mode and should also disable GDM (Gear Down Mode) in addition to optimizing subtimings.
12600KF (5.125GHz, 100.5BLCK x 51, x40 ECores, x43 Ring, 2x8GB DDR4 3920MHz 15-15-15-32 Gear 1 CR1 1.55V Samsung B-Die), RTX 3080, Win11
Save1 (Corvega): 11693 - 96.8 FPS
Save2 (City): 8029 - 123.4 FPS
AVG: 110.1 FPS

https://imgur.com/aX0c027
https://imgur.com/whkNc9o

I wonder what OCed 12700K and 12900K results will be like. They should be a lot faster cause of bigger cache.
aX0c027

whkNc9o

Thank you very much for the results!

@KompuKare BTW,if you look at the Ryzen 7 5800X3D results with 3200C16 results the AT numbers look higher - I wonder whether that is because the 64GB kit has weaker subtimings or whether the GTX1080TI seems to have less of a driver overhead??

The Corvega results are much lower with the RTX3080 but have higher drawcalls - the Diamond City results have less drawcalls but probably a higher immediately NPC AI load.

The Creation engine truly is utter ****. I really hope Creation 2 turns out some big improvements.

It's an ancient engine which has had bandaids put over it's Gamebryo engine roots.
 
Last edited:
@KompuKare BTW,if you look at the Ryzen 7 5800X3D results with 3200C16 results the AT numbers look higher - I wonder whether that is because the 64GB kit has weaker subtimings or whether the GTX1080TI seems to have less of a driver overhead??
To keep Ampere fed is supposed to require more CPU power, however, for this ultra low detail 480P benches the GPU shouldn't matter that much.
I suspect it's because fish is running 4x16GB and Madcap over on AT is possible 2x16? S2 is almost the same but S1 isn't. So raw draw call vs draw calls with more NPCs?
 
Last edited:
Ok, so memory overclocking was a bust with this kit and CPU, but i get different result depending on which monitor it runs on. On my second screen i get 123.9(7980) and 104.1 (11681) which is similar for one result but a lot higher for the other.
 
Just to copy my post over from Hexus :

Ryzen 7 5700G - no graphics card - 32GB DDR4 3200Mhz

Corvega=43.7 FPS (Drawcalls=11667)
Diamond City= 57.3FPS (Drawcalls=7975)

4GB ram allocated to the CPU for graphics.
 
Just to copy my post over from Hexus :

Ryzen 7 5700G - no graphics card - 32GB DDR4 3200Mhz

Corvega=43.7 FPS (Drawcalls=11667)
Diamond City= 57.3FPS (Drawcalls=7975)

4GB ram allocated to the CPU for graphics.

Surprisingly close to the Ryzen 5 4650G with a GTX1080TI and faster than a Core i5 3570K with an RX470 4GB! It shows you how CPU limited Fallout 4 is!
 
Yeah it's clearly a mem tune bump as the difference between my two 12600k is just me tuning every secondary and tertiary timing (if I recall).

I'm also curious why we all top out around 113fps max on test 1. If my x3d ever shows, I'll try digging into it.
 
Not got the game yet and I'm still tuning and dialling in my final memory overclocks, but I might yet give this a crack in time.
Cool the more the merrier!

Memory clocks and subtimings is really important for this game. I think CAT said that the base game should do for the benchmarks (at least until someone makes another Fallout4 bench thread using some Automatron, Contraptions Workshop save), so might be worth waiting until it's down to £3.50 or so (IsThereAnyDeal says it currently is £7, although there are cheaper sources).

Somewhat surprised nobody with a 12900K has benched this yet. Would be interesting to see if the extra cache makes any difference, or if the difference versus the 12600K is simply too little.
 
Just got my 5800x3d so i thought i'd contribute since i've been lurking here for a while

Corvega 11706 calls, 78fps
Diamond City 8020 calls, 95fps

Specs are:
5800x3d
5700XT
32GB 3600mhz CL14
win 10

guess the red GPU holds back the x3d still
 
Just got my 5800x3d so i thought i'd contribute since i've been lurking here for a while

Corvega 11706 calls, 78fps
Diamond City 8020 calls, 95fps

Specs are:
5800x3d
5700XT
32GB 3600mhz CL14
win 10

guess the red GPU holds back the x3d still
Well, most of our scores are with Nvidia but my own scores weren't that much better with the RTX 3050 vs the RX 470.

However, for some reason yours is the worst 5800X3D score yet:
U6pimCY.png
so unsure what is going on.
Maybe we can get LtMatt to test with both the 6900XT and the 3090, or some others may step forward.
 
Back
Top Bottom