Ending score - Junkyard wars Episode 7

[FatherChaos]

I have no idea why they equated the best score in each category to 100.

 

I feel like that creates a bias/handicap and makes the scores closer than they really are.. In reality i feel they should just treat each benchmark as points, then  add all the points together and divide by the respective cost of each computer. Linus got 18.16 points per dollar, and Luke got 16.91 point per dollar. Linus won in my mind.

 

The way it was done was weird. There is no reason for your score to down or up if the competition does worse/better.

What do you guys think? or am i just outta my mind.

[Spotty]

Top score = 100% (100pts)
Runner up score = their score / Top Score = Their percentage of performance compared to the other person

It's the easiest way to normalise the different scores between different benchmarks. The Cinebench scores were in the hundreds, but the 3DMark scores were in the thousands with huge difference between the systems of thousands of points between each system. If you just took the raw points from each test and added them up to get a final total score, the 3Dmark score would effectively render the Cinebench score worthless.

Here's an example.

Quote

If I score 500pts in Cinebench, and a competitor scores 300pts in cinebench. I get 100pts and they get 60pts because their score is 60% of my score. (300/500 = 0.6)
If I score 10,000pts in 3DMark, and a competitor scores 15,000pts in 3DMark. The competitor gets 100pts and I get 66.6pts because my score was 66.6% of their score. (10000/15000 = 0.6666...)

If I get 120FPS in games, and the competitor gets 150FPS. The competitor gets 100pts and I get 80pts because I got 80% of their score. (120/150 = 0.8)

My Score = 100 + 66.6 + 80 = 246.6
Competitor score = 60 + 100 + 100 = 260

 

 

5 minutes ago, FatherChaos said:

There is no reason for your score to down or up if the competition does worse/better.

Well, it was a direct competition between only those two systems to see who got the best (bang for buck) PC, so it entirely makes sense to score them against each other.

[lewdicrous]

25 minutes ago, FatherChaos said:

I feel like that creates a bias/handicap and makes the scores closer than they really are..

How so?

25 minutes ago, FatherChaos said:

In reality i feel they should just treat each benchmark as points, then  add all the points together and divide by the respective cost of each computer. Linus got 18.16 points per dollar, and Luke got 16.91 point per dollar.

Woudn't that be inaccurate?

[Mutoh]

In my mind, Linus wins just for not cheating this time.

[Spotty]

27 minutes ago, Mutoh said:

In my mind, Linus wins just for not cheating on camera this time.

ftfy

[FatherChaos]

I get the normalization , but it has an error in my eyes. The problem with it is that is 3dmark and fps in fortnite are pretty co dependent benchmarks if you get a high 3dmark score your fps will be higher in games(in this case fortnite) without a doubt , but here is the outlier since they have set a ratio of lukes fps to linus's fps in this case (67.87:147.87) and the 3dmark of (5216:13519) this suggest that if luke spent the same amount of money which  as Linus that he could have a lower 3dmark score and higher fps. Which is non sense.

 

 

[FatherChaos]

im good guys thank for helping me out!

 

[ChaseB]

8 hours ago, Mutoh said:

In my mind, Linus wins just for not cheating this time.

Ture. Luke went to Best Buy. Anyone can do that. Linus not only dug around and tried but also did it without cheating.

[Spotty]

1 hour ago, ChaseB said:

Ture. Luke went to Best Buy. Anyone can do that. 

He also bought a SSD, power supply, and graphics card to upgrade the system - as well as finding a person to purchase his unneeded parts to recoup some cost. It's not as if he just walked in to the first computer store he saw, bought a system off the shelf and then went to the pub for the next few days.

[ChaseB]

5 hours ago, Spotty said:

He also bought a SSD, power supply, and graphics card to upgrade the system - as well as finding a person to purchase his unneeded parts to recoup some cost. It's not as if he just walked in to the first computer store he saw, bought a system off the shelf and then went to the pub for the next few days.

True but I feel that Linus really worked for his system

[FatherChaos]

1 hour ago, ChaseB said:

True but I feel that Linus really worked for his system

Yeah I think he did, the stats are similar I believe you would be crazy to get the 330. I feel that A10 was weak and came down to the fortnite FPS. It seems like it was the weakest control of all of them.

[dalekphalm]

15 hours ago, FatherChaos said:

I get the normalization , but it has an error in my eyes. The problem with it is that is 3dmark and fps in fortnite are pretty co dependent benchmarks if you get a high 3dmark score your fps will be higher in games(in this case fortnite) without a doubt , but here is the outlier since they have set a ratio of lukes fps to linus's fps in this case (67.87:147.87) and the 3dmark of (5216:13519) this suggest that if luke spent the same amount of money which  as Linus that he could have a lower 3dmark score and higher fps. Which is non sense.

 

 

While Cinebench and FPS in games both require GPU power, they are not linked in a 1:1 ratio.

 

FPS has many variables - and while GPU is the strongest variable, it's not the only one. Having bottlenecks can severely hamper your FPS, even if you have a GTX 1080 Ti.

 

Also Cinebench is a more "all around" benchmark that tries to simulate content creation. It's based off of the developers 3D animation software, called Cinema 4D.

 

So while content creation like Animation, and gaming FPS might be similar, there are distinct differences that different hardware will handle differently.

 

You should not treat game FPS the same as a Cinebench score, because they are not the same thing.

 

If you think they should have used different tests, that's fine. But they chose 3DMark, Cinebench, and Fortnite FPS. So with those 3 tests, they measured and ranked them exactly as I would have: Normalize the score in each category and then add up the normalized scores from each test to get a total.

[FatherChaos]

19 minutes ago, dalekphalm said:

While Cinebench and FPS in games both require GPU power, they are not linked in a 1:1 ratio.

 

FPS has many variables - and while GPU is the strongest variable, it's not the only one. Having bottlenecks can severely hamper your FPS, even if you have a GTX 1080 Ti.

 

Also Cinebench is a more "all around" benchmark that tries to simulate content creation. It's based off of the developers 3D animation software, called Cinema 4D.

 

So while content creation like Animation, and gaming FPS might be similar, there are distinct differences that different hardware will handle differently.

 

You should not treat game FPS the same as a Cinebench score, because they are not the same thing.

 

If you think they should have used different tests, that's fine. But they chose 3DMark, Cinebench, and Fortnite FPS. So with those 3 tests, they measured and ranked them exactly as I would have: Normalize the score in each category and then add up the normalized scores from each test to get a total.

Yeah, I See that that. The Cinebench only benched the processor though they didn't test open Gl. FPS doesnt really stress the CPU as a dual core is all really all you need to get good frame rate, and that 950 that Luke used was capping out due to its own limitation, but not due to the processor bottle necking. Cinebench does the cpu, and 3d mark does both cpu and gpu. I believe it should of been weighted like 50/100/50(Cinebench/3dmark/Fortnite fps). I believe 3dmark is the best way to see if your system is copacetic.

[dalekphalm]

25 minutes ago, FatherChaos said:

Yeah, I See that that. The Cinebench only benched the processor though they didn't test open Gl. FPS doesnt really stress the CPU as a dual core is all really all you need to get good frame rate, and that 950 that Luke used was capping out due to its own limitation, but not due to the processor bottle necking. Cinebench does the cpu, and 3d mark does both cpu and gpu. I believe it should of been weighted like 50/100/50(Cinebench/3dmark/Fortnite fps). I believe 3dmark is the best way to see if your system is copacetic.

You're simplifying things greatly.

 

Saying that CPU doesn't matter, and that you only need a dual core to play games is simply not correct.

 

Under some resolutions, with some games, sure - the GPU will max out and bottleneck the system before the CPU. But that largely depends on resolution being used.

 

There are many games - mostly modern AAA games - that will benefit from a stronger CPU.

 

With Fortnite specifically? Perhaps it doesn't require a very strong CPU, but let's not generalize when we don't need to.

[Spotty]

1 hour ago, FatherChaos said:

Yeah, I See that that. The Cinebench only benched the processor though they didn't test open Gl. FPS doesnt really stress the CPU as a dual core is all really all you need to get good frame rate, and that 950 that Luke used was capping out due to its own limitation, but not due to the processor bottle necking. Cinebench does the cpu, and 3d mark does both cpu and gpu. I believe it should of been weighted like 50/100/50(Cinebench/3dmark/Fortnite fps). I believe 3dmark is the best way to see if your system is copacetic.

When you're talking about gaming rigs you have to also consider other tasks other than just gaming. Stuff like browsing websites with lots of tabs open or watching twitch or youtube videos can stress a weak CPU. It's important to consider CPU performance as well and not just focus purely on GPU performance.

[FatherChaos]

2 minutes ago, Spotty said:

When you're talking about gaming rigs you have to also consider other tasks other than just gaming. Stuff like browsing websites with lots of tabs open or watching twitch or youtube videos can stress a weak CPU. It's important to consider CPU performance as well and not just focus purely on GPU performance.

That is what i'm saying, a single thread application(video game) is not going to test a cpu. How many times do I have to say that. Im telling you that a DX11 game is not going to dictate the true performance of the cpu. Which is why you do Cinebench that is all CPU. A games FPS does not push a cpu to the limit, it is a GPU test there is no strain on a processor. 3D Mark Does CPU,GPU,RAM and should be weight more heavily. 3dmark is the comprehensive final exam. While Cinebench and FPS count is the homework.

53 minutes ago, dalekphalm said:

You're simplifying things greatly.

 

Saying that CPU doesn't matter, and that you only need a dual core to play games is simply not correct.

 

Under some resolutions, with some games, sure - the GPU will max out and bottleneck the system before the CPU. But that largely depends on resolution being used.

 

There are many games - mostly modern AAA games - that will benefit from a stronger CPU.

 

With Fortnite specifically? Perhaps it doesn't require a very strong CPU, but let's not generalize when we don't need to.

Never said a CPU doesn't matter. A Game is not going to stress the CPU.

 

Since you think I am Simplifying things greatly, let me reiterate and be more specific. A FPS Count in a DX11 game is not going to test the full throughput of a CPU that has  2 or more cores, therefore an FPS score should not be weighted equally to something that test both CPU and GPU to max throughput.

 

 

DX 11 doesnt use multiple Cores IT doesn't stress the CPU as long as you have a decent dual core it doesn't matter.

 

DX 12 uses multiple cores and hyper threading can stress the cpu more but still not a great show cpu potential

hand full of games use DX12.

 

Fortnite is a dx11 game.

 

 

 

 

 

Oh look a dual core performing the same as a quad core in video games, how weird.

 

 

[FatherChaos]

but yea sure if you go get an athlon 2  from 2007 and then a 1080 ti and tr to run it in 8k its going to explode.

[DrMacintosh]

#TeamLuke  

[FatherChaos]

1 minute ago, DrMacintosh said:

#TeamLuke  

lmfao I was really hope Linus would get it.