View Single Post
  #22 (permalink)  
Old September 21, 2012, 08:13 PM
terrybear terrybear is offline
Banned
 
Join Date: Jan 2008
Location: midland, ontario
Posts: 519
Default

Quote:
Originally Posted by sswilson View Post
If they want to call it a core cluster then fine, that works for me, just don't call each individual unit a core unless it has all the resources it needs to act independantly. The main problem (IMO) with BD's release was that what they were calling an 8 core unit could barely hold it's own against their own previous hexacores. I also don't see how intel's naming convention isn't honest about the number of cores... it's clearly stated that hyperthreading is not a full fledged core and listed as 4 cores / 8 threads.

Now, OTOH, if (and I don't see why they shouldn't be able to demonstrate this) AMD processors can perform more simultaneous "low/medium usage" threads without a performance penalty then quite frankly there's probably a strong market for that as most folks aren't like us enthusiasts and they aren't running their gear @ 100%. Maybe we need to rethink how we benchmark, or at the very least attempt to come up with a method of testing general "snappiness" under medium loads rather than only looking at max loads.
NOW you've hit the nail on head !! .... This is the biggest problem nowadays is that there is not 1 test that can be ran that is not somehow influenced/optimized by Intel/Amd/Nvidia ...... Thats why for the most part I refer back to as skymtl has shown ...... fraps as a defaulter of things. Same as my argument with why nvidia drivers sometime preform better cause there optimized to run based on clock rates of a cpu & not the driver itself allowing the cpu indipendantly affect preformance.
Reply With Quote