TechspotTom Clancy's The Division CPU Benchmarks

[ShintaiDK]

Isn't a pre scripted benchmark kind of like watching a movie where the graphics have to be rendered but that's it?

Pretty much. Else we could conclude that DX12 isn't needed and there are almost endless CPU power untapped for all CPUs

 

[Sweepr]

We've had GameGPU and PCLab results for days now, I bet you've seen it but didn't post for some mysterious reason. D:

 

[SlowSpyder]

Wouldn't it be nice with benchmarks from actual ingame play, instead of a super low CPU demand prescripted benchmark whos sole purpose is to test the GPU when you want to determine CPU influence?

Do you think its a valid benchmark if my 6700K at 800Mhz will show 60FPS+?

Is the scene different in the in game test compared to the benchmark? Couldn't that explain the FPS difference? I don't know how in-game benches are coded, if AI is active or if it's totally scripted.

 

[ShintaiDK]

Is the scene different in the in game than benchmark? Couldn't that explain the FPS difference? I don't know how in-game benches are coded, if AI is active or if it's totally scripted.

Some benchmarks are 100% scripted. No AI, no input etc.

So graphics wise they more or less fit. CPU wise they are day and night in difference. You could almost call it an API test.

 

[SlowSpyder]

Some benchmarks are 100% scripted. No AI, no input etc.

So graphics wise they more or less fit. CPU wise they are day and night in difference. You could almost call it an API test.

I guess we'd need to know which benches are scripted and which ones more realistically represent actual game play. Otherwise, yea, I agree for the most part. I'm all about the actual real life experience vs. benches anyway.

 

[ShintaiDK]

I guess we'd need to know which benches are scripted and which ones more realistically represent actual game play. Otherwise, yea, I agree for the most part. I'm all about the actual real life experience vs. benches anyway.

When you see minimalistic to no difference when you go from 4.5Ghz on i7/FX to 2.5Ghz. Then the red light should turn on.

Kabini with a GTX680 could also pull 53 FPS in the previous Tomb raider benchmark. Remember a 800Mhz Haswell did 74 FPS in the same benchmark and 22 FPS in actual gameplay.

So if you trusted that benchmark, you would think Kabini would be more than good to play it with a discrete GPU. But you end up with 15 FPS or so gameplay.

 

[Erenhardt]

Well, the benchmark run is a workload of some kind. If one CPU is faster than the other, it mean just that, no matter how you slice it.

Question if it is representative of the actual workload during gameplay is valid but it doesn't change the results.

 

[Thala]

When you see minimalistic to no difference when you go from 4.5Ghz on i7/FX to 2.5Ghz. Then the red light should turn on.

Could be GPU bound. With other words you just don't know...
Typically you have the physics engine running even if the benchmark is scripted. The only thing typically missing is the AI engine because...well that part is scripted
Of course if you add AI calculations, you are GPU bound later with respect to the CPU performance scale.

 

[ShintaiDK]

Could be GPU bound. With other words you just don't know...
Typically you have the physics engine running even if the benchmark is scripted. The only thing typically missing is the AI engine because...well that part is scripted

What on earth had you imagined it would be if not GPU bound if CPU speed didn't matter?

 

[TheELF]

I don't think it's called GPU bound if you intentionally stress only the GPU, it's a GPU benchmark made to show the limits of the GPU.

 

[frozentundra123456]

Well then, that particular benchmark is gpu bound.