Logic Array Blocks and Adaptive Logic Modules in Cyclone ... - Altera
Logic Array Blocks and Adaptive Logic Modules in Cyclone ... - Altera
Logic Array Blocks and Adaptive Logic Modules in Cyclone ... - Altera
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
1-10<br />
Figure 1-9: ALM <strong>in</strong> Shared Arithmetic Mode for <strong>Cyclone</strong> V Devices<br />
Shared Arithmetic Cha<strong>in</strong><br />
datae0<br />
datac<br />
datab<br />
dataa<br />
datad<br />
datae1<br />
4-Input<br />
LUT<br />
4-Input<br />
LUT<br />
4-Input<br />
LUT<br />
4-Input<br />
LUT<br />
shared_arith_<strong>in</strong><br />
shared_arith_out<br />
carry_<strong>in</strong><br />
carry_out<br />
labclk<br />
reg0<br />
reg1<br />
reg2<br />
reg3<br />
To General or<br />
Local Rout<strong>in</strong>g<br />
The shared arithmetic cha<strong>in</strong> available <strong>in</strong> enhanced arithmetic mode allows the ALM to implement a 3-<strong>in</strong>put<br />
adder. This significantly reduces the resources necessary to implement large adder trees or correlator<br />
functions.<br />
The shared arithmetic cha<strong>in</strong> can beg<strong>in</strong> <strong>in</strong> either the first or sixth ALM <strong>in</strong> a LAB.<br />
Similar to carry cha<strong>in</strong>s, the top <strong>and</strong> bottom half of the shared arithmetic cha<strong>in</strong>s <strong>in</strong> alternate LAB columns<br />
can be bypassed. This capability allows the shared arithmetic cha<strong>in</strong> to cascade through half of the ALMs <strong>in</strong><br />
an LAB while leav<strong>in</strong>g the other half available for narrower fan-<strong>in</strong> functionality. In every LAB, the column<br />
is top-half bypassable; while <strong>in</strong> MLAB, columns are bottom-half bypassable.<br />
The Quartus II Compiler creates shared arithmetic cha<strong>in</strong>s longer than 20 ALMs (10 ALMs <strong>in</strong> arithmetic or<br />
shared arithmetic mode) by l<strong>in</strong>k<strong>in</strong>g LABs together automatically. To enhance fitt<strong>in</strong>g, a long shared arithmetic<br />
cha<strong>in</strong> runs vertically, allow<strong>in</strong>g fast horizontal connections to the TriMatrix memory <strong>and</strong> DSP blocks. A<br />
shared arithmetic cha<strong>in</strong> can cont<strong>in</strong>ue as far as a full column.<br />
Document Revision History<br />
Date<br />
May 2013<br />
<strong>Altera</strong> Corporation<br />
Document Revision History<br />
Version<br />
2013.05.06<br />
Changes<br />
• Added l<strong>in</strong>k to the known document issues <strong>in</strong> the Knowledge Base.<br />
• Removed register cha<strong>in</strong> outputs <strong>in</strong>formation <strong>in</strong> ALM output section.<br />
• Removed reg_cha<strong>in</strong>_<strong>in</strong> <strong>and</strong> reg_cha<strong>in</strong>_out ports <strong>in</strong> ALM<br />
high-level block diagram <strong>and</strong> ALM connection details diagram.<br />
<strong>Logic</strong> <strong>Array</strong> <strong>Blocks</strong> <strong>and</strong> <strong>Adaptive</strong> <strong>Logic</strong> <strong>Modules</strong> <strong>in</strong> <strong>Cyclone</strong> V Devices<br />
CV-52001<br />
2013.05.06<br />
Feedback