AMD vs Intel Processors In Linux

[bononos]

........
Ivy or Haswell with Hyperthreading should be the fastest, by some degree, but the dedicated int units on the AMDs should make the FX-63x0 and FX-83x0 reasonable, if your electricity costs are low.

Best I could find, with Openbenchmarking's search currently off:

http://openbenchmarking.org/result/1409038-SO-QQWE3829779
http://openbenchmarking.org/result/1409055-PL-QHQM0022984 (near the bottom)
http://openbenchmarking.org/result/1408317-SO-TESTBUILD46
http://openbenchmarking.org/result/1408319-LI-XXXLINUXC39

The edge in performance and power consumption go to Intel. A new Core i5 should be about as fast as an FX-8350, while using less power, and having better IGP support overall. HT only gives about 20-30%, it looks like (typical, but actually good for the AMD CPUs), but the performance per core of Intel's is just plain higher.

Could you explain abit on your statement here that "Phoronix's benchmarks look better for the FX because their benchmarks are worse".

 

[Phynaz]

At 5x the TDP also, with 5 8350 you would get 25s and about the same TDP for the processing parts, so the GPU scores are not that impressive considering that the CPUs are general purpose calculators.

10 times the performance with only a 5x increase in TDP implies a doubling of efficiency.

Please provide a bench of this mythical 5 CPU 8350 system.

 

[MiddleOfTheRoad]

Could you explain abit on your statement here that "Phoronix's benchmarks look better for the FX because their benchmarks are worse".

I'd love to hear the answer to that, too. Because Phoronix's suite is the most comprehensive set of benchmarks I'm aware of for testing performance under Linux.

 

[jhu]

10 times the performance with only a 5x increase in TDP implies a doubling of efficiency.

Please provide a bench of this mythical 5 CPU 8350 system.

He's just multiplying samples/s * #CPUs. But 25s is still slower than 16s. Would require 10 FX8350 CPUs to match a dual GTX 580 (2*244 W TDP). A Titan Black (250 W TDP) is even faster and more power efficient than a dual GTX 580 (not to mention 10 Fx 8350 systems {10*125 W TDP at least}).

Er... I guess this benchmark result is just for the Titan, but close enough. Also, Blender 2.71 is faster than Blender 2.66.

 

[Cerb]

Could you explain abit on your statement here that "Phoronix's benchmarks look better for the FX because their benchmarks are worse".

Where is anything remotely akin to say, PCMark, on Phoronix? They only test long-running loopy programs, with nothing to test responsiveness nor multitasking prowess. It's not Phoronix's fault, just that there's no such benchmark or set of benchmarks for Linux, like there are for Windows (and Windows hasn't even had good ones for 10 years, yet, and what we did have that long ago was what made early, and then Northwood Celerons, and PC133-starved P4s, look decent). Given the server history, there's Iometer that does well for just storage, but not much like that on the CPU side. If you want general desktop performance measured, there aren't suitable benchmarks to compare across systems with, TMK.

 

[MiddleOfTheRoad]

Where is anything remotely akin to say, PCMark, on Phoronix? They only test long-running loopy programs, with nothing to test responsiveness nor multitasking prowess. It's not Phoronix's fault, just that there's no such benchmark or set of benchmarks for Linux, like there are for Windows (and Windows hasn't even had good ones for 10 years, yet, and what we did have that long ago was what made early, and then Northwood Celerons, and PC133-starved P4s, look decent). Given the server history, there's Iometer that does well for just storage, but not much like that on the CPU side. If you want general desktop performance measured, there aren't suitable benchmarks to compare across systems with, TMK.

PCMark? Lol, yeah that's a reliable benchmark..... LOL (sarcasm off)
They are notorious and have already been caught with "rigged" benchmarks.

Please, just give up while you're ahead. Phoronix never got caught pulling this crap:

http://www.osnews.com/story/22683/I...quot_Cripple_AMD_quot_Function_from_Compiler_

My Favorite Quote:
"In fact, Fog points out that even benchmarking programs are affected by this, up to a point where benchmark results can differ greatly depending on how a processor identifies itself. Ars found out that by changing the CPUID of a VIA Nano processor to AuthenticAMD you could increase performance in PCMark 2005's memory subsystem test by 10% - changing it to GenuineIntel yields a 47.4% performance improvement! There's more on that here "

Yeah, seriously.... PCMark..... I bet I could get a 200% performance improvement with their benchmark by changing my CPUID to Optimus Prime!

 

[Cerb]

Phoronix also hasn't been caught trying to come up with anything to remotely attempt to test general productivity performance, which, today, is only partly limited by the performance per core/thread, like most of their multihtreaded benchmarks are. It's also apparently such a bad benchmark that its scores are comparable to gaming benchmarks, comparing current CPUs (because even that failure of an older version only affected a couple tests). It's not going to be super easy to make something like that, that can function as more than a pointed microbench, and like Iometer, will likely take professionals to get it going and keep it up (there are lots of more fun things to work on, if you're not being paid). No benchmark is perfect, but having one made to be comprehensive is better than not, by a long shot. It would even be better on average under Linux, because the same binaries from the repo, made with GCC, and no special dispatching, would be used for the testing, so the ICC tricks wouldn't be able to get pulled in the first place.