Note 1: Integer Math big increase is from the working DIV/IDIV unit
Note 2: A10-5700 starts @ 3.4 GHz and ends @ 4.0 GHz/FX-4100 starts @ 3.6 GHz and ends @ 3.8 GHz
 

very interesting

Decent for the same lithography, almost on par with Ivy versus Sandy in terms of the ratio between them.
 
Cue in the typical Intel fan that asks "what about games?" when most of us play GPU-intensive titles where the CPU type doesn't really matter; where we'll see a result like this more often than not at high resolutions and use the fact that there won't be much difference between FX4100 in order to claim Piledriver failed...

lehpron
Decent for the same lithography, almost on par with Ivy versus Sandy in terms of the ratio between them.

Everything is pretty much like Ivy vs Sandy.
L2 went from 20 cycles to 16 cycles/(L3 I'm not sure about)
Some SSE/AVX instructions had a 2x/100% increase
Mov/+misc instructions had an 1.5x(for 64-bit which makes those instructions Phenom II level) 2x(32-bit which makes those instructions beyond Phenom II level) increase
A10-5700 65W TDP/A10-5800K 100W TDP/FX-4100 95W TDP
 
The FX-4320 is going to have a high clock rate if the A10-5800K has four 3.8GHz CPU cores and 384 SPs running at .8 GHz for 100W TDP.

sounds nice i cant wait to see the new desktop cpu's

Are these new things AM3+?

Brad_Hawthorne
Are these new things AM3+?

FX Vishera will support drop in support for 990FX/990X/970 chipsets as well as the 1090FX/1070 chipsets which are basically just the 990FX/990X/970 chipsets with AMD USB 3.0 controllers.  FX Zambezi is supported in 1090FX/1070 as well.  Then next year the FX platform joins the A series platform.

I'm having flashbacks of Phenom -> Phenom II at the moment. I really wish AMD would get it right the first time. Good to see this chip series though.

Brad_Hawthorne
I'm having flashbacks of Phenom -> Phenom II at the moment. I really wish AMD would get it right the first time. Good to see this chip series though.

It's not really a Phenom I -> Phenom II
 
65-nm was bad for AMD while 45-nm was better.  32-nm isn't as bad as 65-nm but it isn't as good as 45-nm.
 
Bulldozer had a lot of things "disabled."  Bulldozer has 4 AGLUs(per core in a module), 2 AGLUs are called EX0/1 and execute all instructions, 2 AGLUs are called AGLU0/1 and execute CALL/LEA(Mov/misc with Piledriver).  Bulldozer has a HW Divider(was disabled in Bulldozer but is enabled and working in Piledriver).  The trace cache and L3 didn't really do anything in Bulldozer but the trace cache does something in Piledriver.  I won't know L3 till Vishera.  There is no word on if the L1i cache failing has been fixed.

Sign me up for a FX 4320. 

This is more of a chipset thing.  A85 chipsets support Nvidia SLI & Virtu MVP from lucidlogix...
A70
A55
Didn't support it.

Not sure if you guys have seen this: -

 
Toms Hardware has done an initial preview benchmark of the upcoming Trinity - A6, A8, and A10.
 
Most interesting: -

 
15% better clock-for-clock performance for Trinity against Bulldozer.
The rest of the benchmarks either slightly exceed the Stars platform (llano A8-3850) or at least match its performance.
 
 

Grr, looks like I can't post links.
 
Nonetheless, check out TomsHardware for the Trinity benchmark (preview).
 

http://www.tomshardware.c...k-a6-5400k,3224-2.html
 
You mean this?

hey seronx, do you happen to know if AMD has similar overclock restrictions as Intel like a maximum multipliers and such?  Pertaining to APUs.

lehpron
hey seronx, do you happen to know if AMD has similar overclock restrictions as Intel like a maximum multipliers and such?  Pertaining to APUs.

Depends on the bios and motherboard.
Value Motherboards will be locked to 47x-50x.
Enthusiast Motherboards the sky is the limit.

Again 
http://www.obr-hardware.com/2012/09/i-was-right-fx-8350-or-c0-is-piledriver.html