Firestrike Physics Test Score Difference when changing GPU.

Toggle sidebar Toggle sidebar
C

coops1

Honorable
Jul 3, 2013
243
1
10,760
Not sure if this should be in CPU or GPU section but here we go.

So I bought a GTX 780 second hand off a friend and by happy circumstance ended up getting an R9 290 for free.

I decided to see which one would post higher firestrike scores, so put them each in my system with my overclocked FX 6300 and benchmarked each in turn. The graphics scores were similar at stock, however my physics score is consistently lower with the R9 290 (roughly 6800 with the 290, 7300 with the 780).

I was just wondering what could be causing this? My initial thought was PhysX would be heloping the 780, but I wasn't sure if PhysX helped the firestrike physics tests at all?

Cheers all.
 
Sort by date Sort by votes
M

mdocod

Distinguished
Aug 15, 2010
1,797
0
20,460
My guess is that whatever libraries/API are called on for the test are different with the nvidia drivers vs the AMD drivers. Whatever software resource is being called on, nvidias implementation is better optimized, and is getting more out of your CPU.

Ironically, it's not uncommon to see nvidia cards performing better on AMD CPUs, then AMD's own GPUs in compute intensive workloads.
 
Upvote 0 Downvote
C

coops1

Honorable
Jul 3, 2013
243
1
10,760
Turns out, the 290 draws significantly more power than the 780 - coupled with my CPU OC it was close to what my PSU could handle, which seems to be why my physics score dropped (dropping the 290 back to 900MHZ core with an undervolt essentially fixed the problem). Thought I would have been okay with a CX600m but ah well :/
 
Upvote 0 Downvote
M

mdocod

Distinguished
Aug 15, 2010
1,797
0
20,460
Curious... The PSU doesn't have any way of communicating with the computer and telling it that something is "wrong" or that the computer needs to slow down. Either the PSU provides enough clean power, or it doesn't. Granted, a CX600 is not adequate for such a configuration all overclocked, but that doesn't mean it should have effected the bench-marking results. I would keep digging and testing. Sounds to me like something else is amiss.
 
Upvote 0 Downvote
leeb2013

leeb2013

Honorable
Aug 5, 2013
1,967
0
12,460
Yeah still doesn't seem right. I pushed my cx600 to 720w with my i5 at 4.8GHz and two 7950s at 1200mhz, all it did eventually was shut down, but benchmarks weren't affected. I wouldn't expect your non-oc setup to hit 600w even.
 
Upvote 0 Downvote
A

AirForce101HD

Reputable
Aug 3, 2014
170
0
4,760
"3DMark Fire Strike Physics test benchmarks the hardware’s ability to run gameplay physics simulations on the CPU. The GPU load is kept as low as possible to ensure that only the CPU is stressed. The Bullet Open Source Physics Library is used as the physics library for the test.
The test has 32 simulated worlds. One thread per available CPU core is used to run simulations. All physics are computed on CPU with soft body vertex data updated to GPU each frame."
Which can be found here
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom