intel cpu performance and pricing is enabled by rigged benchmarks that lack disclaimers about optimizations for intel's uarch through software memory footprint, compiler settings/build(using ICC), and biased "reviewers" such as toms & anandtech;
amd offers ZERO dev-support and does not pay for play when it comes to software optimization for bentmarks.
in contrast to int-el: http://youtu.be/-M1eX5Vvrg8
see more intel-sponsored software optimization:
http://newsroom.intel.com/servlet/JiveServlet/previewBody/1043-102-1-1033/10-03-Intel_Core_i7-980X_EnabledSoftware.pdf
amd wont pay for software optimization, as intel does:
http://www.sdtimes.com/AMD_HELPS_OPENCL_GAIN_GROUND_IN_HPC_SPACE/By_Alex_Handy/About_AMD_and_HPC_and_NVIDIA_and_OPENCL/35956
"AMD won't be offering paid-for consulting services to help optimize applications, as Intel does."
amd tried working w/ bentmark devs, but quit BapCo due to unfair intel bias in points-scoring:
http://semiaccurate.com/2011/06/20/nvidia-amd-and-via-quit-bapco-over-sysmark-2012/
see also nigel dessau blog: http://blogs.amd.com/nigel-dessau/2011/06/21/1006/
bapco criminals/deceivers: http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=137432&p=185369#p185307
intel compiler uses suboptimal code paths resulting in poor performance on non-intel cpu:
http://www.agner.org/optimize/blog/read.php?i=49#49
even to this day, after antitrust settlement, intel compiler still will use suboptimal code-path if cpuid!=genuineintel
http://www.agner.org/optimize/blog/read.php?i=49#127
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138574&p=204319#p204281
see montarayjack's comment for ICC12: http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&#p210722
compiler smackdown, ICC regresses in performance from ver10 to ver11, gets beat by LLVM.
http://multimedia.cx/eggs/compiler-smackdown-2010-1-64-bit/
Cinebench 11.5 has cpuid check, likely uses sub-optimal code path for amd:
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&#p210650
Cinebench 11.5 is optimized for intel uarch:
http://i664.photobucket.com/albums/vv4/wuttzi/Snapbucket/CB115.png
crysis2 excessive tessellation to favor nV: http://techreport.com/articles.x/21404/1
software optimization goes beyond using a biased compiler, it also involves memory management, cache strategy, algorithms, data structures, etc...
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&#p210572
even athlon is better than core i7-nehalem in integer toom multiplication:
http://en.wikipedia.org/wiki/Toom–Cook_multiplication
http://gmplib.org/devel/
"FFT takes over for larger operands for machines with faster hardware multiplication. AMD's processors have great hardware multiplication, Intel's Nehalem's multiplication has twice Athlon's latency, and the venerable Pentium 4 has really poor hardware multiplication."
what is a fair benchmark? when can a comparison of CPU performance be considered fair?
"My personal opinion is that the best results obtainable for a given application on a given architecture is the best comparison between them."
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&p=211796#p211792
IPC is not everything, Amdahl's law favors CPU with just the right IPC(not too much, not too little) but with higher clocks;
http://abinstein.blogspot.com/2010/09/ipc-myths.html
changing cpuid to GenuineIntel nets 47.4% increase in performance:
http://www.osnews.com/story/22683/Intel_Forced_to_Remove_quot_Cripple_AMD_quot_Function_from_Compiler_
PCMark/Futuremark rigged bentmark to favor intel:
http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=135382#p139712
http://arstechnica.com/hardware/reviews/2008/07/atom-nano-review.ars/6
intel cheating at 3DMark vantage via driver: http://techreport.com/articles.x/17732/2
intel hypethreading chokes in heavy workload environments: http://youtu.be/mdPi4GPEI74
intel sandy bridge image quality is as good as 2005, in 2011: http://techreport.com/articles.x/21099/11
intel 24p bug, duplicated frame every 40s: http://www.avsforum.com/avs-vb/showthread.php?p=19851650#post19851650
intel sandy bridge chipset bug results in lost data: http://techreport.com/discussions.x/20326
intel cpu uarch has worse decoder design than amd: "The decoding of instruction lengths has been a problem in Intel processors for many years. They tried to fix the problem with the trace cache in the Pentium 4, which turned out to be a dead end street, and now the apparently more successful micro-op cache in the Sandy Bridge. AMD have solved the problem of detecting instruction lengths in their processors by marking instruction boundaries in the code cache. Intel did the same in the Pentium MMX back in 1996, and it is a mystery to me why they are not using this solution today. There would hardly be a need for the micro-op cache if they had instruction boundaries marked in the code cache."
http://www.agner.org/optimize/blog/read.php?i=142
amd llano has better power management/better battery life than sandy bridge:
http://www.tomshardware.com/reviews/a8-3500m-llano-apu,2959-22.html
amd llano has better thermal dissipation than SB:
The key take away is that Llano’s power management is significantly more advanced than previously indicated and includes a few novel features, such as taking advantage of indirect thermal dissipation.
http://realworldtech.com/page.cfm?ArticleID=RWT083111130632
"One novel trick is that Llano’s algorithms recognize that idle blocks in the system will act as indirect thermal conduits for active components. Heat will flow from hotter (active) to cooler (idle) regions in the chip and the idle region will still dissipate heat into the heat sink, effectively creating a greater surface area for cooling. The credit-based power management and especially the indirect thermal dissipation are substantially more advanced than we had described in our earlier article on Llano and it is important to acknowledge this oversight."
intel 32nm still cold bugged, amd no cold bugs:
http://www.overclocking-tv.com/content/news/11863/amd-continues-with-cold-bug-free-cpu-extreme-bulldozer/
intel antitrust lawsuit, paid off DELL not to use amd chips
http://www.nytimes.com/2009/11/05/technology/companies/05chip.html
intel continues to coerce OEM's even after settlement not to adopt AMD:
http://www.digitimes.com/news/a20100708VL204.html
intel fakes CES2012 DX11 demo: http://semiaccurate.com/2012/01/09/intel-fakes-ivy-bridge-graphics-on-stage-at-ces/
amd fx gamexperience, better gaming polled on amd systems: http://www.legitreviews.com/article/1838/1/
amd trinity video- (17w tdp for same performance as 35w llano): http://youtu.be/lsmTDb-Mlws
AMD Trinity blind test, AMD preferred ~80% of the time over intel; http://www.hardwarezone.com.sg/feature-amd-confidently-shares-preview-upcoming-trinity-apu-experience
youtube uses opensource ffmpeg in back-end;
http://multimedia.cx/eggs/googles-youtube-uses-ffmpeg/
amd offers ZERO dev-support and does not pay for play when it comes to software optimization for bentmarks.
in contrast to int-el: http://youtu.be/-M1eX5Vvrg8
see more intel-sponsored software optimization:
http://newsroom.intel.com/servlet/JiveServlet/previewBody/1043-102-1-1033/10-03-Intel_Core_i7-980X_EnabledSoftware.pdf
amd wont pay for software optimization, as intel does:
http://www.sdtimes.com/AMD_HELPS_OPENCL_GAIN_GROUND_IN_HPC_SPACE/By_Alex_Handy/About_AMD_and_HPC_and_NVIDIA_and_OPENCL/35956
"AMD won't be offering paid-for consulting services to help optimize applications, as Intel does."
amd tried working w/ bentmark devs, but quit BapCo due to unfair intel bias in points-scoring:
http://semiaccurate.com/2011/06/20/nvidia-amd-and-via-quit-bapco-over-sysmark-2012/
see also nigel dessau blog: http://blogs.amd.com/nigel-dessau/2011/06/21/1006/
bapco criminals/deceivers: http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=137432&p=185369#p185307
intel compiler uses suboptimal code paths resulting in poor performance on non-intel cpu:
http://www.agner.org/optimize/blog/read.php?i=49#49
even to this day, after antitrust settlement, intel compiler still will use suboptimal code-path if cpuid!=genuineintel
http://www.agner.org/optimize/blog/read.php?i=49#127
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138574&p=204319#p204281
see montarayjack's comment for ICC12: http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&#p210722
compiler smackdown, ICC regresses in performance from ver10 to ver11, gets beat by LLVM.
http://multimedia.cx/eggs/compiler-smackdown-2010-1-64-bit/
Cinebench 11.5 has cpuid check, likely uses sub-optimal code path for amd:
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&#p210650
Cinebench 11.5 is optimized for intel uarch:
http://i664.photobucket.com/albums/vv4/wuttzi/Snapbucket/CB115.png
crysis2 excessive tessellation to favor nV: http://techreport.com/articles.x/21404/1
software optimization goes beyond using a biased compiler, it also involves memory management, cache strategy, algorithms, data structures, etc...
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&#p210572
even athlon is better than core i7-nehalem in integer toom multiplication:
http://en.wikipedia.org/wiki/Toom–Cook_multiplication
http://gmplib.org/devel/
"FFT takes over for larger operands for machines with faster hardware multiplication. AMD's processors have great hardware multiplication, Intel's Nehalem's multiplication has twice Athlon's latency, and the venerable Pentium 4 has really poor hardware multiplication."
what is a fair benchmark? when can a comparison of CPU performance be considered fair?
"My personal opinion is that the best results obtainable for a given application on a given architecture is the best comparison between them."
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138786&p=211796#p211792
IPC is not everything, Amdahl's law favors CPU with just the right IPC(not too much, not too little) but with higher clocks;
http://abinstein.blogspot.com/2010/09/ipc-myths.html
changing cpuid to GenuineIntel nets 47.4% increase in performance:
http://www.osnews.com/story/22683/Intel_Forced_to_Remove_quot_Cripple_AMD_quot_Function_from_Compiler_
PCMark/Futuremark rigged bentmark to favor intel:
http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=135382#p139712
http://arstechnica.com/hardware/reviews/2008/07/atom-nano-review.ars/6
intel cheating at 3DMark vantage via driver: http://techreport.com/articles.x/17732/2
intel hypethreading chokes in heavy workload environments: http://youtu.be/mdPi4GPEI74
intel sandy bridge image quality is as good as 2005, in 2011: http://techreport.com/articles.x/21099/11
intel 24p bug, duplicated frame every 40s: http://www.avsforum.com/avs-vb/showthread.php?p=19851650#post19851650
intel sandy bridge chipset bug results in lost data: http://techreport.com/discussions.x/20326
intel cpu uarch has worse decoder design than amd: "The decoding of instruction lengths has been a problem in Intel processors for many years. They tried to fix the problem with the trace cache in the Pentium 4, which turned out to be a dead end street, and now the apparently more successful micro-op cache in the Sandy Bridge. AMD have solved the problem of detecting instruction lengths in their processors by marking instruction boundaries in the code cache. Intel did the same in the Pentium MMX back in 1996, and it is a mystery to me why they are not using this solution today. There would hardly be a need for the micro-op cache if they had instruction boundaries marked in the code cache."
http://www.agner.org/optimize/blog/read.php?i=142
amd llano has better power management/better battery life than sandy bridge:
http://www.tomshardware.com/reviews/a8-3500m-llano-apu,2959-22.html
amd llano has better thermal dissipation than SB:
The key take away is that Llano’s power management is significantly more advanced than previously indicated and includes a few novel features, such as taking advantage of indirect thermal dissipation.
http://realworldtech.com/page.cfm?ArticleID=RWT083111130632
"One novel trick is that Llano’s algorithms recognize that idle blocks in the system will act as indirect thermal conduits for active components. Heat will flow from hotter (active) to cooler (idle) regions in the chip and the idle region will still dissipate heat into the heat sink, effectively creating a greater surface area for cooling. The credit-based power management and especially the indirect thermal dissipation are substantially more advanced than we had described in our earlier article on Llano and it is important to acknowledge this oversight."
intel 32nm still cold bugged, amd no cold bugs:
http://www.overclocking-tv.com/content/news/11863/amd-continues-with-cold-bug-free-cpu-extreme-bulldozer/
intel antitrust lawsuit, paid off DELL not to use amd chips
http://www.nytimes.com/2009/11/05/technology/companies/05chip.html
intel continues to coerce OEM's even after settlement not to adopt AMD:
http://www.digitimes.com/news/a20100708VL204.html
intel fakes CES2012 DX11 demo: http://semiaccurate.com/2012/01/09/intel-fakes-ivy-bridge-graphics-on-stage-at-ces/
amd fx gamexperience, better gaming polled on amd systems: http://www.legitreviews.com/article/1838/1/
amd trinity video- (17w tdp for same performance as 35w llano): http://youtu.be/lsmTDb-Mlws
AMD Trinity blind test, AMD preferred ~80% of the time over intel; http://www.hardwarezone.com.sg/feature-amd-confidently-shares-preview-upcoming-trinity-apu-experience
youtube uses opensource ffmpeg in back-end;
http://multimedia.cx/eggs/googles-youtube-uses-ffmpeg/
- rev.5.2012