Raw performance data vs "It runs just better"

Toggle sidebar Toggle sidebar
G

Gragiulo2000

Reputable
May 28, 2014
10
0
4,510
Hi,
i am here to ask a simple question.
Why sometimes processors like the FX-6300 from AMD performs equally if not better than for example the i5-4430 in raw benchmark data (see passmark website cpubenchmark for reference the list is daily updated), but everywere you see this processor listed behind other intels with the same benchmark. As i pointed out in the last article here in tom's Hardware.

It seems like the price tag is more important than the raw performance.
A lot of ppl here will answer "The processor x performs just better", and to me this is awful, because we are talking about processing power, we are talking about benchmarks and numbers, not dishes, that is the point.

Why do i have to pay twice the price with the SAME benchmark level?
What do we want from a processor? A coffee? A pizza? NO. Processing power.
 
Solution
photonboy


More specifically, most games don't use much more than TWO of the CPU's cores.

Recent Intel CPU's are better per-core than the AMD processors so at the SAME frequency when comparing just one core of each CPU the Intel might be as much as 40% better.

In more RECENT games like Battlefield 4 that can use more CPU cores we are starting to see roughly similar performance at times (i.e. FX-8350 8-core, vs i5-4670 4-core). So having more cores that are actually being USED by the game can make up for the relatively poorer performance of each core.

This trend will CONTINUE, but it only applies if the game was coded that way so if the game is already made and supports roughly TWO cores regardless of the CPU...
Sort by date Sort by votes
P

Pinhedd

Champion
Aug 7, 2007
10,342
20
56,965


Passmark and cpubenchmark are not real benchmarks. If you wish to see the true capabilities of each microprocessor, the best and most detailed benchmark suite is probably SiSoft Sandra. They have online rankings for each benchmark expressed as aggregates and individual for your searching pleasure.
 
Upvote 0 Downvote
bicycle_repair_man

bicycle_repair_man

Splendid
Jan 10, 2014
5,752
2
28,960
In benchmarking, the only way to get meaningful, comparable results is to keep every variable factor identical. In the real world, this simply isn't possible.

Also consider that hardware will only perform as well as software allows it to. All software has to be developed with an idea of the hardware that will underpin it, so some will naturally favour Intel over AMD. A benchmarking tool is software and therefore has this underpinning, so you could argue that the results will be inconsistent before you even start.

To me, benchmarking and raw figures are meaningless. What I want to know is how a component performs at the task I want it to do.

Good discussion thread.
 
Upvote 1 Downvote
photonboy

photonboy

Titan
Jun 20, 2006
21,567
13
85,015


More specifically, most games don't use much more than TWO of the CPU's cores.

Recent Intel CPU's are better per-core than the AMD processors so at the SAME frequency when comparing just one core of each CPU the Intel might be as much as 40% better.

In more RECENT games like Battlefield 4 that can use more CPU cores we are starting to see roughly similar performance at times (i.e. FX-8350 8-core, vs i5-4670 4-core). So having more cores that are actually being USED by the game can make up for the relatively poorer performance of each core.

This trend will CONTINUE, but it only applies if the game was coded that way so if the game is already made and supports roughly TWO cores regardless of the CPU that won't change.

Other:
DX12 (and Mantle/OpenGL) and newer game engines are not only able to use MORE CORES, but the coding is becoming more efficient so they will be able to do the same job with less processing power.

You probably heard about that with MANTLE benchmarks because the people who benefited most were those with a relatively weak CPU.

*So, future games will use MORE of the processing power that's currently not being used, and be more efficient at the same time which will significantly change how well a CPU like the FX-6300 performs compared to the i5-4430 for example.

A third awesome thing is that more of the current CPU code will be re-written to use the Graphics card which further reduces the CPU relative processing power needed. Neat! Physics is a great example as we have NVidia's PhysX which at times runs on the CPU or GPU. Speaking of which, we also may see graphics cards with on-board CPU's such as the "Denver" ARM CPU for upcoming NVidia cards which can alleviate some CPU tasks though probably needs to be specifically coded for.

*If we factor in everything, it's not unreasonable to think that a 2017 game that is well coded could use an FX-6300 with a graphics card that is 4X more powerful than what would presently pose a major bottleneck.
 
Upvote 0 Downvote
Solution
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom