Good way of GPU PVP

[NLTech]

BTW the PVP means Price VS Performance. 

But, I came up with a way to see how much power you get for your dollar. 

For years GFLOPS and TFLOPS have been used to rate GPUs, so I calculated the current AMD and NVidia's lineups and then divided the TFLOPS/GFlops by their price. Here ya go. 

The .txt has all links, performance ratings, etc and the .ods has all info. 

TFlops_Comparisons.txt

Untitled 1.ods

[DocSwag]

You can't compare tflops across architectures tho 

[Glennieboyyy007]

LTT made a video about this and a massive excel sheet with fps per dollar for 1080p 1440p and 4k

[Bananasplit_00]

tflops across architectures cant really be compared, within an architecture it can be a decent messurement of performance but what you should use is benchmarks if you want performance nummbers

[DocSwag]

4 minutes ago, Bananasplit_00 said:

within an architecture it can be a decent messurement of performance

Even within architectures it's not the best comparison metric. A 1080 has about double the TFlops of a 1060, but performance is considerably less than double.

(happens because performance doesn't linearly scale with cores)

[ARikozuM]

What they said.

 

35 minutes ago, DocSwag said:

You can't compare tflops across architectures tho 

I will compare my 4 in-line to a V6 though... Convertible wins.

 

Especially when drifting*. 

 

*Don't try this at home and leave it to professionals. 

[ARikozuM]

1 minute ago, DocSwag said:

 

happens because performance doesn't linearly scale with cores)

See Amdahl's Law.

[Bananasplit_00]

1 minute ago, DocSwag said:

Even within architectures it's not the best comparison metric. A 1080 has about double the TFlops of a 1060, but performance is considerably less than double.

 

(happens because performance doesn't linearly scale with cores)

yah, but the bigger number performs better basically, its not a great messure either way but you can use it