Difference between revisions of "Instruction Set/fmafe"

From Mill Computing Wiki
Jump to: navigation, search
 
Line 1:Line 1:
 
{{DISPLAYTITLE:fmafe}}
 
{{DISPLAYTITLE:fmafe}}
 
<div style="font-size:80%;line-height:90%;margin-bottom:2em">[[Speculation|speculable]]&nbsp;&nbsp;[[Encoding|exu stream]]&nbsp;&nbsp;[[Decode|exu block]]&nbsp;&nbsp;[[Phasing|compute phase]]&nbsp;&nbsp; operation&nbsp;&nbsp; [[Domains|in the binary floating point value domain]]&nbsp;&nbsp; [[Rounding|and rounds to nearest, ties toward even adjacent value]]<br />
 
<div style="font-size:80%;line-height:90%;margin-bottom:2em">[[Speculation|speculable]]&nbsp;&nbsp;[[Encoding|exu stream]]&nbsp;&nbsp;[[Decode|exu block]]&nbsp;&nbsp;[[Phasing|compute phase]]&nbsp;&nbsp; operation&nbsp;&nbsp; [[Domains|in the binary floating point value domain]]&nbsp;&nbsp; [[Rounding|and rounds to nearest, ties toward even adjacent value]]<br />
'''native on:''' [[Cores/Silver|Silver]] [[Cores/Gold|Gold]] <br />
+
'''native on:''' [[Cores/Silver|Silver]] <br />
 
</div>
 
</div>
  
Line 14:Line 14:
 
</div>
 
</div>
 
Returns x*y+z on the belt.<br />
 
Returns x*y+z on the belt.<br />
 
'''encoding:'''
 
<code style="font-size:100%"><b style="color:#050">fmafe</b>(<span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops window">x</span>)</code>
 
,
 
<code style="font-size:100%"><b style="color:#050">exuArgs</b>(<span style="color:#009">[[Domains#op|op]]</span> <span title="belt operand from ops window">arg0</span>, <span style="color:#009">[[Domains#op|op]]</span> <span title="belt operand from ops window">arg1</span>)</code>
 
<br />
 
  
 
{| class="mw-collapsible mw-collapsed wikitable" data-expandtext="&#9656;" data-collapsetext="&#9662;"
 
{| class="mw-collapsible mw-collapsed wikitable" data-expandtext="&#9656;" data-collapsetext="&#9662;"
 
! [[Cores|Core]] || [[Slot|In Slots]]|| [[Latency|Latencies]]
 
! [[Cores|Core]] || [[Slot|In Slots]]|| [[Latency|Latencies]]
 
|-
 
|-
| [[Cores/Silver/Encoding#fmafe|Silver]] || E0 E1 || w,w:w=6 wv,wv:wv=6 d,d:d=7 dv,dv:dv=7 q,q:q=8 qv,qv:qv=8
+
| [[Cores/Silver/Encoding#fmafe|Silver]] || E0 E1 ||  
|-
+
| [[Cores/Gold/Encoding#fmafe|Gold]] || E0 E1 E2 E3 || w,w:w=6 wv,wv:wv=6 d,d:d=7 dv,dv:dv=7 q,q:q=8 qv,qv:qv=8
+
|}
+
 
+
----
+
<code style="font-size:130%"><b style="color:#050">fmafe</b>(<span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops window">x</span>, <span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops window">y</span>, <span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops
+
            window">z</span>, <span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops window">w</span>) &#8594; [[Domains#f|f]] r<sub>0</sub>, [[Domains#f|f]] r<sub>1</sub></code>
+
<div style="font-size:80%;line-height:90%;margin-bottom:2em">'''operands:''' [[Operands#likeFmasf|like Fmasf [ff:f]]]
+
</div>
+
This is a fused multiply-add-subtract. An excellent way to make full use of all [[Functional Unit]]s in the 2 [[Slot]]s.<br />
+
r<sub>0</sub> is x*y+z*w<br />
+
r<sub>1</sub> is x*y-z*w<br /><br />
+
 
+
'''encoding:'''
+
<code style="font-size:100%"><b style="color:#050">fmafe</b>(<span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops window">x</span>, <span style="color:#009">[[Domains#f|f]]</span> <span title="belt operand from ops window">y</span>)</code>
+
,
+
<code style="font-size:100%"><b style="color:#050">exuArgs</b>(<span style="color:#009">[[Domains#op|op]]</span> <span title="belt operand from ops window">arg0</span>, <span style="color:#009">[[Domains#op|op]]</span> <span title="belt operand from ops window">arg1</span>)</code>
+
<br />
+
 
+
{| class="mw-collapsible mw-collapsed wikitable" data-expandtext="&#9656;" data-collapsetext="&#9662;"
+
! [[Cores|Core]] || [[Slot|In Slots]]|| [[Latency|Latencies]]
+
|-
+
| [[Cores/Silver/Encoding#fmafe|Silver]] || E0 || w,w:w,w=6,6 wv,wv:wv,wv=6,6 d,d:d,d=7,7 dv,dv:dv,dv=7,7 q,q:q,q=8,8 qv,qv:qv,qv=8,8
+
|-
+
| [[Cores/Gold/Encoding#fmafe|Gold]] || E0 E2 || w,w:w,w=6,6 wv,wv:wv,wv=6,6 d,d:d,d=7,7 dv,dv:dv,dv=7,7 q,q:q,q=8,8 qv,qv:qv,qv=8,8
+
 
|}
 
|}
  
  
 
[[Instruction_Set|Instruction Set, alphabetical]], [[Instruction Set by Category]], [http://millcomputing.com/instructions.html?collapse=7#ops Instruction Set, sortable, filterable]
 
[[Instruction_Set|Instruction Set, alphabetical]], [[Instruction Set by Category]], [http://millcomputing.com/instructions.html?collapse=7#ops Instruction Set, sortable, filterable]

Latest revision as of 14:12, 23 February 2021

speculable  exu stream  exu block  compute phase   operation   in the binary floating point value domain   and rounds to nearest, ties toward even adjacent value

native on: Silver

Binary floating point fused multiply-add. As usual for those, it yields a higher precision than doing it separately, and is faster too. Rounds towards even.


fmafe(f x, f y, f z) → f r0

operands: like Addf [ff:f]

Returns x*y+z on the belt.

Core In Slots Latencies
Silver E0 E1


Instruction Set, alphabetical, Instruction Set by Category, Instruction Set, sortable, filterable