How Helldivers 2 Async Compute Setting Effects FPS

preview_player
Показать описание
In this video I test Async Compute in multiple hardware configurations, like the Ryzen 7800X 3D, 5900X, 2600 CPU's and RTX 4090 and RX 7900XTX GPU's.

Music by: Karl Casey @WhiteBatAudio

Hardware Used:
Processors - R7 7800X 3D, R9 5900X, R5 2600
GPU's - Asus TUF RTX 4090, Sapphire Nitro+ RX 7900XTX

AM5 Build:
32GB 6200 mhz CL 32 DDR5 Memory
ASUS ROG Strix X670E - A Gaming WiFi Motherboard
360 AIO H150i Corsair Elite Capellix
ASUS TUF Gaming 1000 Watt 80 Plus Gold ATX 3.0 Fully Modular Power Supply
1TB Samsung 980 Pro NVME SSD Operating System Drive
2TB Samsung 980 Pro NVME SSD Game Drive
Lian Li 011 Dynamic EVO PC Case

AM4 Build:
ROG Strix B550-F Gaming WiFi Motherboard
32GB Trident Z Neo 3600mhz CL16 (8GB x 4)
1TB WD Black SN850 NVMe SSD (PCIe Gen 4)
1TB WD Black SN750 NVMe SSD (PCIe Gen 3)
EVGA 1000GT - 1000 Watt PSU
Lian Li Lancool 2 Mesh White (case)

Windows 11 Pro
Radeon Driver 24.4.1
GeForce Driver 552.22

00:00 Disclaimer
00:23 What is Async Compute
02:06 7800X 3D and RX 7900XTX
04:54 7800X 3D and RTX 4090
09:00 5900X and RX 7900XTX
11:06 5900X and RTX 4090
13:19 Ryzen 2600
16:29 Conclusion

#helldivers2 #amd #nvidia
Рекомендации по теме
Комментарии
Автор

Holy shit that post in 1:22 is so stupid. Async Compute is NOT meant to make the GPU perform "CPU math to help out with processing", it's supposed to speed up processing of compute shaders that can be executed asynchronously with classic shaders such as vertex shaders, geometry shaders, pixel shaders etc.

Reinhardovich
Автор

Turning async compute off gained me ~10 fps on my 7800x3d and 7900xtx.
Turning display mode to borderless fullscreen gained me an additional 10 fps. Exclusive fullscreen ironically loses fps in this game. I'm starting to notice a trend with this for DX12 games.

hydrosmusic
Автор

Honestly knowing Arrowhead implemented explosion physics that pulls you in instead of pushing you off, it could be the case that they simply reversed the option on Async-compute lol

rzkrdn
Автор

Thank you for the disclaimer at the beginning, much respect to you.

I'm playing with it off as it drops by frames by about 10fps when its turned on. I've had mixed experience with the feature in games. Either way, i've never noticed it either help or hurt that much.

5800x3D
6900XT

Damaged
Автор

I noticed that during certain atmospheric events or heavy action, I would see decreased framerate with async turned off. Seemed like having it on for me, sacrificed little bit of max fps for much more consistent fps.

btbarr
Автор

First off all thanks for seing our comments and reading our reply's. Second off all i was know something is wrong with ASYNC compute settings but i wasn't know the my CPU is holding me back. I have Ryzen 5 7600X with 6000mhz DDR5 and RX 7900 XT. This is mean i will get Ryzen 9000 or Ryzen 9000X3D chip for my next upgrade part. Your video's helping lots of people my friend. Thank you for your all effort.

NextGenCeoTV
Автор

I think the most interesting thing with the vidoe that I dont think that you really mentioned is that:

Nvidia clearly has a higher driver overhead performance cost on the cpu. In cpu bound scenarios, the 4090 is performing noticeably worse.

Very interesting.

TheIndulgers
Автор

Would be intresting to see if there is any diffrences with SAM on vs off

Whatever-el
Автор

hi my friend and thanks for video!!i read some bad commentaire !!dont mind about them... make ur video every day feel more confident !!all they start as and u will discover more things and get more expérience !!have start nice week and always glad to flow u and watch every video !!have fun

hocine
Автор

''Aaand we went from 50 fps to 50 fps.''

Sick gains bro 😂
On my 6800xt I saw a loss of around 4-6 fps so even on a mid-range setup it's pointless..

Thanks for sharing man!

constantinesoldatos
Автор

I have an extraordinarily CPU bound system as i save up for more upgrades. With a 6850K and 7900XT, Async Compute limits my maximum FPS which is achieved on the ship from 130 to 100, but raises my minimum FPS during matches from the high 30s to mid 50s during moments of extreme chaos.

GeneralLee
Автор

When I check the average FPS of my gaming sesions some days I get 100 FPS, other days 60 FPS, I have not changed my presets... I dont know what is going on, it might be that different planets have different performances?

ElPato_rof
Автор

Does Async Compute help with any kind of stutter, 1% and .1% lows?

JamesSmith-swnk
Автор

Glad I paired my 5900x with a 7900 xtx, instead of a 4090 i was considering. Saved money and a bit more FPS. At least in this scenario. I wonder if other games would be the same?

mikeshogo
Автор

could you test HAGS on/off would be greatly helpful thanks

morgan
Автор

should I trust amd's built in frame counter or msi afterburner? On msi software it shows my cpu usage 50-60% and gpu at 80-98%. When I use amd software it shows cpu 70-100% and gpu 70-90%. My render scale is set to native and all settings on ultra at 1080p. Useing a rx 6800xt and ryzen 5 7600.

toomas
Автор

Believe it or not, the Autodesk engine in use by HD2 is rather old. It would make sense if it isn’t designed with modern CPUs in mind.

The kind of load that is able to be done on the CPU wouldn’t show up on the GPU usage as much because it won’t be massively multithreaded and won’t use much of the main component of the gpu.

lodewykk
Автор

Latency is a factor, especially for NVIDIA. AMD has an advantage over NVIDIA in asynchronous. In general tho this asynchronous thing is having a real impact mainly for older hardware or specific combinations. I had a first gen threadripper and this setting sometimes boosted my fps by 50+% absolutely insane.

ViewTube_Emperor_of_Mankind
Автор

My 2 cents: your test is probably wrong. If this affects CPU usage, then CPU will work most when there is a lot of things happening on the screen (EG: doing math for what each enemy AI does, how rockets and lasers flow, where bullets go, etc). Those are the CPU intensive tasks in general (I may be wrong, I dont know specifics of AH's engine).
This means that testing a static setting is a poor test all in all for this setting, and it's probably not helping because, indeed the CPU is never the one bottlenecking.
This being said, I did not run any tests, it's just my guess.

TheShadow
Автор

1:11
I tested it again to see if it has changed with the last updates but when I turn it on I lose 20fps. On the ship I usually have 145-160 fps with async compute off, when I switch it on it's only 120-135 fps.
Ryzen 7 5800x3d bclk 104 CO-20 (all core boost 4628)
(The same with standard settings but slightly lower overall fps)
Graphics: Sapphire RX 6900xt Toxic limited edition.
RAM: 32GB Kingston Fury beast cl 16 3200MT's
Everyone should probably just test for themselves what brings more performance. I find your videos quite interesting and it's great that you also respond to your comments.
greetings from Germany

Gehirnsammler