Is this site reliable for "general" CPU comparison across CPU generations?

[Turbonium]

I was curious to see how fast/slow a given generation of CPUs is to another generation, and I found this site.

How reliable is it? Not in terms of the numbers (as I know they shouldn't be taken literally or linearly etc.), but in terms of the site's reputation?

Btw I tried searching the forum for something related to this, and didn't find anything.

 

[ShintaiDK]

Utterly useless site.

 

[Sable]

http://www.anandtech.com/bench/CPU/2

And for laptops

http://www.notebookcheck.net/Mobile-Processors-Benchmarklist.2436.0.html

 

[sm625]

lol @ the useless site comment. Passmark remains the best single unit benchmark. If you had to give one number to most accurately describe a chip's performance, yes it would be its passmark rating. It is very accurate for a single number. Take 2 different CPUs that score 4000 and you will find that they perform very similarly overall. Same goes with two chips that score 1000. Notice for example that an AMD E2-2000 and a Pentium D 3.73GHz both score the same.

 

[Enigmoid]

lol @ the useless site comment. Passmark remains the best single unit benchmark. If you had to give one number to most accurately describe a chip's performance, yes it would be its passmark rating. It is very accurate for a single number. Take 2 different CPUs that score 4000 and you will find that they perform very similarly overall. Same goes with two chips that score 1000. Notice for example that an AMD E2-2000 and a Pentium D 3.73GHz both score the same.

Only for applications that stress all cores to the max, most consumer software does not do this and focuses on only a few threads.

ex) 8150 is ahead of the i5-3570k but for games the i5 is much better.

 

[SPBHM]

passmark does not represent well CPU performance, for most uses.

the source of their results is... questionable, their testing method or whatever, there are quite a few unrealistic results with no explanation.

 

[ShintaiDK]

lol @ the useless site comment. Passmark remains the best single unit benchmark. If you had to give one number to most accurately describe a chip's performance, yes it would be its passmark rating. It is very accurate for a single number. Take 2 different CPUs that score 4000 and you will find that they perform very similarly overall. Same goes with two chips that score 1000. Notice for example that an AMD E2-2000 and a Pentium D 3.73GHz both score the same.

The problem is the random sample size and where the samples come from.

Hence you get results that you even explained yourself is an old post:

AMD Athlon II X2 4400e 1516
AMD Athlon 64 X2 Dual Core 5600+ 1515
Intel Celeron B815 @ 1.60GHz 1504
Intel Celeron B840 @ 1.90GHz 1504
Intel Pentium E5800 @ 3.20GHz 1469

 

[Turbonium]

The AT site is obviously better and more accurate, but the thing is that it doesn't have things like a 486 in the lineup. I literally want to be able to compare any CPU to any CPU, and get some idea of the power difference in a given application.

Purely for nerdy reasons, but still.

 

[R0H1T]

The AT site is obviously better and more accurate, but the thing is that it doesn't have things like a 486 in the lineup. I literally want to be able to compare any CPU to any CPU, and get some idea of the power difference in a given application.

Purely for nerdy reasons, but still.

I guess for raw numbers you can try cpu-world like e.g.
http://www.cpu-world.com/Compare/320/AMD_A10-Series_A10-5700_vs_AMD_A10-Series_A10-6700.html

 

[Maximilian]

Its interesting to look at but thats all, just because CPU X has 10x higher numbers than CPU Y does not make it 10x faster across the board.

 

[gevorg]

Passmark is fine but its very synthetic. I find Anand's and Xbitlabs results to be the most realistic in performance analysis.