Posts: 17
Threads: 6
Joined: May 2014
05-28-2014, 10:49 AM
(This post was last modified: 05-28-2014, 10:51 AM by tethys.)
Can we trust this chart?
http://www.golubev.com/gpuest.htm
Because I saw that the smaller 5850 is faster than the 6850:
Radeon HD 6850--> 960 775 744.00 2745M 744M 14171 353932 43237
Radeon HD 5850--> 1440 725 1044.00 3852M 1044M 19886 496646 60672
Posts: 2,936
Threads: 12
Joined: May 2012
they are estimations, not benchmarks. they are to give you a rough idea of a card's performance.
but yes, a 5850 should be faster than a 6850. 5850 has 1440 stream processors @ 725 Mhz vs 6850's 960 stream processors @ 775 Mhz. 2088 GFLOPS vs 1488 GFLOPS.
just because it's a generation older doesn't mean it's smaller.
Posts: 17
Threads: 6
Joined: May 2014
Great, thank you!
Maybe I should buy two 5850 then.
I don't know though if their powers add up literally like 2 x 2088 GFLOPS = 4166?
Posts: 2,936
Threads: 12
Joined: May 2012
perhaps part of your confusion is how the model number convention changed between the 5xxx and 6xxx line.
prior to the 6xxx product line, the x8xx notation was the top of the line single GPU products, and x9xx was reserved for the dual GPU products. xx70 was used to denote the top-performing card in each product line.
starting with the 6xxx product line, x9xx became the top of the line for both single and dual GPU products, with xx70 being used to denote the top-performing single GPU card and xx90 used to denote the top-performing dual GPU card.
so a 5850 is more comparable to a 6950.
Posts: 2,936
Threads: 12
Joined: May 2012
performance scales linearly, yes. two GPUs will be exactly twice as fast as one GPU.
but note that GFLOPS aren't directly relevant for hash cracking since hash algorithms use integer math, not floating point math. but they do provide a decent reference point for comparing performance between similar GPUs.
Posts: 17
Threads: 6
Joined: May 2014
Thank you Epixoip, very informative.