[Part 3] Measuring CPU Draw Call Performance in Fallout 4

Page 7 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

KompuKare

Golden Member
Jul 28, 2009
1,013
924
136
I realise this post is over 6 months old but I'm up for the Oblivion testing. What do I need to get going here if your still interested?
Btw, what is that overlay software in your screenshot? apologise for the nOOb question!
The overlay is just the ENB config screen. If you follow the instructions @USER8000 linked to, you should get similar output.
 

USER8000

Golden Member
Jun 23, 2012
1,542
780
136
The overlay is just the ENB config screen. If you follow the instructions @USER8000 linked to, you should get similar output.

I linked to the Oblivion testing thread. Here are the Fallout 4 test instructions:
 

BoredErica

Member
Apr 26, 2022
41
64
51
Dang, RPL posters are fast. I guess it makes sense, but I have to move from AMD to Intel, and the ram timings/IMC differ, and I've been having install hell (bios flashback is a scam on my board). It's hard getting the many timings tightened without taking a ton of my time (sitting here all day while my computer is busy) or getting results a month from now (just test 1-2 timing changes each night when I go to sleep). I don't wanna post FPS w/ a potentially unstable setup or come back 2 weeks later with fully tuned ram and a different submission.

Can you check the L3 cache hit rates on your new CPU?
I'm not sure if I can. The utility has better support for zen2/3 than Intel CPUs. I thought I saw capframex on Twitter post l3 hit rates on ADL though but I have RPL. Also the L3 hit rates are hard to compare across different architectures (like ADL vs zen3). If people want to nudge Clamchowder on Twitter to work on it, I wonder if it helps loool.

halgari from my chart still has the 7950x and I'm sure he has FO4 on his computer but... I don't want to push him to do more work than I already made him do. xD My best friend might upgrade to Zen5 3d, and if they upgrade they'll definitely be willing to test Oblivion/FO4 for me. Currently they just run 5900x so it's basically my 5600x + 3% higher clocks.

I wonder what tuned (no trash ddr5 mobo auto rules) DDR5 7800 and up can do for FO4. Alas, these days I'm more about the value buys. :p
 
Last edited:

jpwolf23

Junior Member
Nov 24, 2022
12
1
41
Hi, new here, I was wondering what current cpu and RAM would go well with 1080 ti that would give 60+ fps because my current system (4790k, Asus Maximus Hero VII and DDR3 1600MHz 16gb RAM) is having the occasional ctd's (no warning) and lots of stuttering and extremely low fps down to 1 fps (like having slide pictures shown instead) where I have to close the game via Task Manager. PS, I do have mods (Fallout 4 MOFAM guide) running. I also play the other Bethesda Games like Morrowind, Oblivion, Skyrim, Fallout 3 and Fallout New Vegas.
 

KompuKare

Golden Member
Jul 28, 2009
1,013
924
136
Hi, new here, I was wondering what current cpu and RAM would go well with 1080 ti that would give 60+ fps because my current system (4790k, Asus Maximus Hero VII and DDR3 1600MHz 16gb RAM) is having the occasional ctd's (no warning) and lots of stuttering and extremely low fps down to 1 fps (like having slide pictures shown instead) where I have to close the game via Task Manager. PS, I do have mods (Fallout 4 MOFAM guide) running. I also play the other Bethesda Games like Morrowind, Oblivion, Skyrim, Fallout 3 and Fallout New Vegas.
Well, the FO4 leaderboard chart looks like this ATM:
VOMQnGu.png

And I guess any of those CPUs should give you some indication. For 5800X3D, the CPU is pretty much everything, for the others memory speed tuning matters a lot too!
 
  • Like
Reactions: Terka

jpwolf23

Junior Member
Nov 24, 2022
12
1
41
Well, the FO4 leaderboard chart looks like this ATM:
VOMQnGu.png

And I guess any of those CPUs should give you some indication. For 5800X3D, the CPU is pretty much everything, for the others memory speed tuning matters a lot too!
This is the chart that I found when I was looking for a recent benchmark for Fallout 4. Plus, it also gave additional information on CPU's, RAM and GPU on what they are using. But I had no idea that memory speed tuning mattered until I looked closer on the RAM settings. Thank you.
 

KompuKare

Golden Member
Jul 28, 2009
1,013
924
136
Some more results from the Overclockers UK thread.

E5B04hW.png
Thanks for posting them here. The reason I didn't post here (yet) is because....

... We are waiting for Zen4 3D results...

... But the obvious thing is: we might as well beg here too!

So any Zen4 3D owners out there willing to test this? The top 5 only have about 2% between them so a 7950X 3D should easily beat them?
 

Det0x

Golden Member
Sep 11, 2014
1,028
2,953
136
Thanks for posting them here. The reason I didn't post here (yet) is because....

... We are waiting for Zen4 3D results...

... But the obvious thing is: we might as well beg here too!

So any Zen4 3D owners out there willing to test this? The top 5 only have about 2% between them so a 7950X 3D should easily beat them?
Kinda big benchmark to download :D
1678994478899.png
 

Det0x

Golden Member
Sep 11, 2014
1,028
2,953
136
User: domdtxdissar/Det0x
7950x3d PBO CO
6600MT/s tight timings, 2166mhz FCLK --> same memory settings as shown here
SMT disabled
3090

CCD1 disabled, 106 baseclock
Corvega- Draw Calls: 7999 @ 178.6 FPS
Diamond City- Draw Calls: 11701 @ 141.9 FPS
1678999810080.png

1678999825072.png
 
Last edited:

KompuKare

Golden Member
Jul 28, 2009
1,013
924
136
User: domdtxdissar/Det0x
7950x3d PBO CO
6600MT/s tight timings, 2166mhz FCLK --> same memory settings as shown here
SMT disabled
3090

CCD1 disabled, 106 baseclock
Corvega- Draw Calls: 7999 @ 178.6 FPS
Diamond City- Draw Calls: 11701 @ 141.9 FPS
View attachment 78265

View attachment 78266
There you go: way at the top:
ZP2rkfK.png

Not only that but a huge leap from your previous 7950X with the same setup.
Or summaries vs your 7950X score and the top scored 13600K
dBmK7Bm.png


Fallout 4 really loves Zen4 3D.
 

zoko

Junior Member
Oct 15, 2022
11
38
51
Got multiple runs and step up with the 7950x3d and simulating the 7800x3d by disabling ccd1 from the 7950x3d.

Ram configuration:
6000 MT/s Tuned :- photo_2023-03-17_17-26-12.jpg
6200MT/s Tuned :-photo_2023-03-17_17-26-09.jpg

First run

User: zoko
7950x3d PBO + CO
6000MT/s tune timings
SMT on
3080 ti

Corvega- Draw Calls: 7996 @ 150.9 FPS
Diamond City- Draw Calls: 12017 @ 124.1 FPS
1679058717046.png1679059148626.png

Run two:

7950x3d PBO + CO
6000MT/s XMP
SMT on
3080 ti

Corvega- Draw Calls: 8021 @ 147.4 FPS
Diamond City- Draw Calls: 12102 @ 120.3 FPS
1679059081716.png1679059099678.png

Then I turned the CCD1 off to simulate the 7800x3d

User: zoko
7800x3d (Simulated) PBO + CO
6000MT/s tune timings
SMT on
3080 ti

Corvega- Draw Calls: 8013 @ 152.3 FPS
Diamond City- Draw Calls: 12010 @ 127.2 FPS
photo_2023-03-17_17-21-05.jpgphoto_2023-03-17_17-21-08.jpg


Finally, I tried pushing the ram more.

7800x3d (Simulated) PBO + CO
6200MT/s tune timings
980 ti

Corvega- Draw Calls: 8048 @ 161.3 FPS
Diamond City- Draw Calls: 11495 @ 118.9 FPS
photo_2023-03-17_17-24-21.jpgphoto_2023-03-17_17-24-23.jpg
 

KompuKare

Golden Member
Jul 28, 2009
1,013
924
136
Got multiple runs and step up with the 7950x3d and simulating the 7800x3d by disabling ccd1 from the 7950x3d.
Thanks for those:
77f2aot.png

quite a spread in your results, but for FO4 it seems the 7800X3D will be best as both your simulation and @Det0x's CCD0 only run were ahead.

EDIT: Updated zoko's 6200/mt results as they should have had the 980 Ti as the GPU.
 
Last edited:

lopri

Elite Member
Jul 27, 2002
13,209
594
126
What does this test, CPU, GPU or memory? I never got a chance to run this one and I am not sure if my 3060 Ti is up to snuff.

Rephrase the question: Does GPU matter in this test?
 

USER8000

Golden Member
Jun 23, 2012
1,542
780
136
Thanks for those:
53hdUal.png

quite a spread in your results, but for FO4 it seems the 7800X3D will be best as both your simulation and @Det0x's CCD0 only run were ahead.

EDIT: Updated zoko's 6200/mt results as they should have had the 980 Ti as the GPU.

SMT off makes such a big difference? Zoko also tested with SMT on. It was Det0x that turned off SMT.

Got multiple runs and step up with the 7950x3d and simulating the 7800x3d by disabling ccd1 from the 7950x3d.

Ram configuration:
6000 MT/s Tuned :- View attachment 78309
6200MT/s Tuned :-View attachment 78308

First run

User: zoko
7950x3d PBO + CO
6000MT/s tune timings
SMT on
3080 ti

Corvega- Draw Calls: 7996 @ 150.9 FPS
Diamond City- Draw Calls: 12017 @ 124.1 FPS
View attachment 78295View attachment 78302

Run two:

7950x3d PBO + CO
6000MT/s XMP
SMT on
3080 ti

Corvega- Draw Calls: 8021 @ 147.4 FPS
Diamond City- Draw Calls: 12102 @ 120.3 FPS
View attachment 78300View attachment 78301

Then I turned the CCD1 off to simulate the 7800x3d

User: zoko
7800x3d (Simulated) PBO + CO
6000MT/s tune timings
SMT on
3080 ti

Corvega- Draw Calls: 8013 @ 152.3 FPS
Diamond City- Draw Calls: 12010 @ 127.2 FPS
View attachment 78304View attachment 78305


Finally, I tried pushing the ram more.

7800x3d (Simulated) PBO + CO
6200MT/s tune timings
980 ti

Corvega- Draw Calls: 8048 @ 161.3 FPS
Diamond City- Draw Calls: 11495 @ 118.9 FPS
View attachment 78306View attachment 78307

Can you test with SMT off please? I am wondering if the other Ryzen 9 7950X3D score being higher than yours is because of the faster memory speed or the SMT being switched off.
 
Last edited:

zoko

Junior Member
Oct 15, 2022
11
38
51
SMT off makes such a big difference? Zoko also tested with SMT on. It was Det0x that turned off SMT.



Can you test with SMT off please? I am wondering if the other Ryzen 9 7950X3D score being higher than yours is because of the faster memory speed or the SMT being switched off.
I did test it with SMT off, didn't change my result a lot, also remember that Det0X is using BCLK 106, which I didn't and couldn't do since my motherboard doesn't support ECLK. (don't want to risk killing my nvme)
Difference between SMT off and on close to zero or margin of error. Here you go:-

7950x3d PBO + CO
6000MT/s XMP
SMT off
3080 ti

SMT OFF (2).jpgSMT OFF.jpg

Edit:-
BLCK/eCLK allows you to boost the v-cache ccd by an extra 200-250mhz, but you need to make sure your motherboard support eCLK, otherwise using bclk may kill your nvme as it applies it to the PCIe slots.
 
Last edited:

MajinCry

Platinum Member
Jul 28, 2015
2,495
571
136
What does this test, CPU, GPU or memory? I never got a chance to run this one and I am not sure if my 3060 Ti is up to snuff.

Rephrase the question: Does GPU matter in this test?

Draw calls are an entirely CPU based task. Memory can have an effect when the cache is full, as instructions and small pieces of data are chucked into memory. Which is why we see 5800x3D and the 7000x 3D CPUs doing so well. Just use the included settings, and your GPU won't affect much at all (unless it's pegged at 100% ofc).

Those 7950x3D results make me almost feel remorseful buying a 5800x3D. Then I remember the prices, and how the 5800X3D was already way out of my comfort zone.
 

KompuKare

Golden Member
Jul 28, 2009
1,013
924
136
Slight update:
Rober896r1 over on OCUK has come back with a tuned 13600KF:
47av05K.png
Pretty impressive especially considering that is "only" DDR4 albeit quite tuned. With the premium which Zen4D currently commands, that Raptor-Lake 13600KF is "almost" good value - even it only has 6P cores.

Dying to see a 7950x3d with a 4090 !!
Well, since we run this at a crazily low resolution and low settings, GPU barely makes a difference. Nvidia seem to have a bit lower driver overhead (Fallout is DX11 - for DX12 the driver overheads are swapped with team red ahead of team green). Even older cards like the 980 Ti see to be on the same or similr driver code path
 
  • Like
Reactions: USER8000