• 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.

GTA5 CPU Benchmarks (I3 beats FX yet again)

Martini1991 said:
Just driving through the dark in the same street as per the other 2 shots.
Anyway, MSI AB started working again, and then CPU usage started going up as it became lighter.

Hilariously frame rate's the same.

Lol_1.png


That CPU 1 core went back down quickly too.
Click to expand...

Different parts of a game, even different views on the same part...

You can be looking one way; the i3 and i5 is better, turn 30 Degrees and the FX-8350 is better.

The difference between Draw Calls and multi-threaded Calc is constantly changing as you move around.

This is why we get cherry picking in this room where one review says the i3 or i5 is better and another review says the FX-83## is better... they are all correct, the results are correct for what they did to get those results.

Really, unless its consistent built in benchmark its all just a nonsense.

PS: Core Thread 1 may well still be the Draw Call thread and probably is, but it can and does move around.
 
Last edited:
I deactivated the core 1...
And I'm driving through the same place.

I'll put up 2 little crappy videos I did doing a little run through the same street, obviously I'm just going forward as I haven't got 4 arms :p
 
Last edited:
Martini1991 said:
But it's evident there's no bottleneck from that side now though.

Disabling a thread does sometimes hilariously improve threading in games.
Click to expand...


Its not because you disabled a thread its because you moved to a different part of the game, i thought i explained this?
 
humbug said:
Its not because you disabled a thread its because you moved to a different part of the game, i thought i explained this?
Click to expand...

So, if I take an image in exactly the same exact place it'll change your opinion? A static image is easy to reproduce but it's meaningless for performance. Gameplay is dynamic.

Two run throughs of the same street is far more meaningfully than a perfectly reproduced static image.

GTA 5 does have a built in Benchmark though.

But some games do get improved performance by disabling a thread, AC4/AC3 for example.
 
Last edited:
Martini1991 said:
So, if I take an image in exactly the same exact place it'll change your opinion? A static image is easy to reproduce but it's meaningless for performance. Gameplay is dynamic.

Two run throughs of the same street is far more meaningfully than a perfectly reproduced static image.
Click to expand...
Yes Game play is dynamic.

What are you proposing to do, stand on the same spot looking at the same space one with 4 cores running and one with 3?
 
humbug said:
I don't have the game, in what way is it not great?
Click to expand...

In the same way Heaven isn't exactly a good indication of game performance.
It's only 2 sections that have sort of actual in game situation.

I didn't monitor the core usage as I was in the shower, but

All cores enabled ;

Frames Per Second (Higher is better) Min, Max, Avg
Pass 0, 40.051174, 87.333397, 62.145885
Pass 1, 22.125916, 77.355820, 46.961136
Pass 2, 32.510269, 108.586182, 58.472534
Pass 3, 40.985275, 91.194420, 63.887218
Pass 4, 5.746800, 93.130074, 58.233692

Time in milliseconds(ms). (Lower is better). Min, Max, Avg
Pass 0, 11.450374, 24.968058, 16.091169
Pass 1, 12.927276, 45.195869, 21.294205
Pass 2, 9.209275, 30.759512, 17.102047
Pass 3, 10.965583, 24.399006, 15.652583
Pass 4, 10.737670, 174.009872, 17.172190

1 Core disabled ;

Frames Per Second (Higher is better) Min, Max, Avg
Pass 0, 38.686230, 83.222961, 61.426086
Pass 1, 28.076189, 83.111664, 48.197716
Pass 2, 32.056263, 102.731117, 59.801090
Pass 3, 42.690437, 103.075005, 66.346680
Pass 4, 16.749567, 102.098961, 58.704788

Time in milliseconds(ms). (Lower is better). Min, Max, Avg
Pass 0, 12.015915, 25.848991, 16.279728
Pass 1, 12.032005, 35.617371, 20.747871
Pass 2, 9.734149, 31.195152, 16.722103
Pass 3, 9.701674, 23.424450, 15.072345
Pass 4, 9.794419, 59.703037, 17.034386

Overall the performance was actually better with a core disabled (It's far from the first game that sees performance benefits from a core disabled). I'm pretty sure it just helps the load balancing on the cores. Either way, I'm confident there's no API bottleneck. Even though the benchmark figures don't actually indicate that.

No matter which way you split it, you'd never expect performance to stay the same/get better if I'm taking away 25% of the physical cores in a game that can use them.

If I ran the game at the lowest settings, that'd be quite interesting to test.

I'm about to go see Avengers, then hit the town. Tomorrow I might run the benchmark on the lowest settings with a core disabled.
 
Last edited:
Th0nt said:
At least we are testing the game and not relying on some funded reviewer that runs skewed tests.
Click to expand...

Not that it helps, me and Humbug both don't agree on what Bottleneck is what :p
And based on me being able to make one of my cores hits 93% means I'm not confident with the FX's in GTA, because the FX would have maxed out far quicker.
 
As long as we know what resolution and clock speed ball park it is to run 60+ fps then we have a mark to work from. I dont trust the reviews especially when posted from dave2***.
 
The sites he used are sites that AMD users have used to claim good performance in previous titles.

To me, I don't like graph wars. Rarely post benchmarks.
 
This is precisely why I would take screenshots on here from other users over graph wars. More people helping the better.
 
We need to come up with a clear way of identifying the limiting factor or "bottleneck" if we must call it that. It will take someone smarter than me for that though.

Martini, since you seem to think your GPU is the limit, could you not try and downclock your CPU to show when the CPU is the limiting factor? Not sure how much use it will be, but I'm interested to know (Would do it myself but still downloading)


Also since you have shown that disabling a thread seems to have little affect, how many threads do you need to disable until it does.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom