i've taken the liberty of reading through every piledriver review i could find and compiling just the encoding benchmarks in one post.
one thing that really stands out is how different website can take the exact same hardware and slant the article to either make piledriver look like a piece of garbage or like amd has finally gotten it right.
http://www.maximumpc.com/article/features/vishera_review?page=0,3
http://www.legitreviews.com/article/2055/10/
http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html
http://www.anandtech.com/show/6396/the-vishera-review-amd-fx8350-fx8320-fx6300-and-fx4300-tested/3
http://www.xbitlabs.com/articles/cpu/display/amd-fx-8350_7.html#sect0
http://www.pcper.com/reviews/Processors/AMD-FX-8350-and-FX-6300-Processor-Review-Vishe...ts-WME-64-and-
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/57446-fx-8350-cpu-review...rrives-11.html
http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/11
and finally the one that comes off like it's meant to show amd's processors in as poor a light as possible:
http://hardocp.com/article/2012/10/22/amd_fx8350_piledriver_processor_ipc_overclocking/4
this last one does raise a few interesting points; they basically took an 8150, 8350, 2600k and a 3770k and clocked them all at 4ghz and ran their benchmarks. not surprising, the intel cpu's slaughtered amd's offerings.
on the one hand they do have a point, when clocked at the same speeds, amd's processors are way slower. on the other hand, a 2600k has a base clock of 3.4ghz, so you have to overclock it by 600mhz to get to 4ghz, the 8350 is already clocked at 4ghz so if you're going to overclock the intel cpu you should also overclock the amd cpu. furthermore, the 4ghz oc is just about the limit for most 2600k's on air, while the 8350 has more headroom for overclocking on air.
but all that aside, it doesn't really address one thing: if you have an amd based system that has a chipset that will work with piledriver, then vishera based processor are the way to go. the next intel cpu, the haswell, is slated to bring 256 bit integer simd via avx extensions to desktop cpu's. considering video encoding and processing apps make heavy use of integer simd, avx2 promises to nearly double performance in applications, such as x264, that use lot's of simd instructions. as such, it's kind of silly to invest lots of money in a high end cpu + motherboard if you can just upgrade your existing processor and be good until next summer.
+ Reply to Thread
Results 1 to 6 of 6
-
-
AMD is so far behind Intel in IPC per core that they are doing the only thing they can do: piling on more cores per die and selling them at low prices in a desperate attempt to stay alive. That's an expensive way of competing because silicon real estate is expensive, especially when you are a full process technology behind. Unfortunately, Intel has made a mistake in pricing so AMD is now losing money on each CPU they sell. (Yes, Intel sets AMD's prices by setting the prices of "equivalent" performing Intel CPUs. AMD can't charge more or they will sell no CPUs. They can't charge less or they'll lose more money.) With no competition on the x86 performance front Intel is concentrating on lower power consumption in an attempt to make x86 competitive with ARM, for the tablet and smart phone market.
-
there is a bit of hope for amd, rumor has it that they are planning on releasing a 5 module 10 core desktop cpu, such a processor should be able to beat any intel quad core cpu and thus it should stimulate sales of amd processors. i know i would buy a sub $300 5m/10c piledriver based cpu.
of course, maybe amd should rethink it's strategy and go back to a traditional dual core cpu, but crank up the clock rates to over 5ghz and possibly introduce 512 bit simd integer and fp instruction sets, now that would be a tasty proposition. -
Of course, Intel could pile on cores if they wanted to too. Piledriver (315 mm^2) is twice the size of Ivy Bridge (160 mm^2). So Piledriver probably costs AMD more than twice as much to manufacture (yields decrease with die size).
Thanks for all the links though. I'll look through them tomorrow.Last edited by jagabo; 28th Oct 2012 at 22:26.
-
I think,therefore i am a hamster.
-
the fx series needs liquid cooling to oc especially the 8350 which consumes almost 300 watts @ 4.8ghz. far over the 198 watt air cooling limit(http://cdn.intechopen.com/pdfs/25491/InTech-Multi_core_cpu_air_cooling.pdf)
i'm happy for now with my 2600k hyperthreaded @ 4.0ghz with no voltage increase, air cooling, and 120 watt power consumption.--
"a lot of people are better dead" - prisoner KSC2-303
Similar Threads
-
ffmpegX benchmarks
By Case in forum ffmpegX general discussionReplies: 0Last Post: 20th Aug 2011, 11:21 -
ridiculous quick sync benchmarks
By deadrats in forum ComputerReplies: 2Last Post: 16th Feb 2011, 18:55 -
clarkdake encoding benchmarks!!!
By deadrats in forum ComputerReplies: 12Last Post: 6th Jan 2010, 17:38 -
first Core i3 benchmarks
By deadrats in forum ComputerReplies: 12Last Post: 2nd Aug 2009, 16:17 -
Core i7 benchmarks!!!!
By deadrats in forum ComputerReplies: 6Last Post: 3rd Nov 2008, 19:33