Skip to content

Commit 3d9d2d1

Browse files
author
MFC Action
committed
Docs @ 68a1b77
1 parent 15dfabe commit 3d9d2d1

File tree

1 file changed

+21
-14
lines changed

1 file changed

+21
-14
lines changed

‎documentation/md_expectedPerformance.html‎

Lines changed: 21 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -140,8 +140,8 @@ <h1><a class="anchor" id="autotoc_md63"></a>
140140
Figure of merit: Grind time performance</h1>
141141
<p>The following table outlines observed performance as nanoseconds per grid point (ns/GP) per equation (eq) per right-hand side (rhs) evaluation (lower is better), also known as the grind time. We solve an example 3D, inviscid, 5-equation model problem with two advected species (8 PDEs) and 8M grid points (158-cubed uniform grid). The numerics are WENO5 finite volume reconstruction and HLLC approximate Riemann solver. This case is located in <code>examples/3D_performance_test</code>.</p>
142142
<p>Results are for MFC v4.9.3 (July 2024 release), though numbers have not changed meaningfully since then. All results are for the compiler that gave the best performance. Note:</p><ul>
143-
<li>CPU results may be performed on CPUs with more cores than reported in the table; we report results for the best performance given the full processor die by checking the performance for different core counts on that device. These are reported as (X/Y cores), where X is the used cores, and Y is the total on the die.</li>
144-
<li>GPU results on single-precision (SP) GPUs performed computation in double-precision via conversion in compiler/software; these numbers are <em>not</em> for single-precision computation. AMD MI250X GPUs have two graphics compute dies (GCDs) per MI250X device; we report results for one GCD, though one can quickly estimate full MI250X runtime by halving the single GCD grind time number.</li>
143+
<li>CPU results may be performed on CPUs with more cores than reported in the table; we report results for the best performance given the full processor die by checking the performance for different core counts on that device. CPU results are for the best performance we achieved using a single socket (or die). These are reported as (X/Y cores), where X is the used cores, and Y is the total on the die.</li>
144+
<li>GPU results are for a single GPU device. For single-precision (SP) GPUs, we performed computation in double-precision via conversion in compiler/software; these numbers are <em>not</em> for single-precision computation. AMD MI250X GPUs have two graphics compute dies (GCDs) per MI250X device; we report results for one GCD, though one can quickly estimate full MI250X runtime by halving the single GCD grind time number.</li>
145145
</ul>
146146
<table class="markdownTable">
147147
<tr class="markdownTableHead">
@@ -167,29 +167,36 @@ <h1><a class="anchor" id="autotoc_md63"></a>
167167
<tr class="markdownTableRowEven">
168168
<td class="markdownTableBodyRight">NVIDIA RTX6000 (SP GPU) </td><td class="markdownTableBodyRight">1 GPU </td><td class="markdownTableBodyRight">3.9 </td><td class="markdownTableBodyLeft">NVHPC 22.11 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
169169
<tr class="markdownTableRowOdd">
170-
<td class="markdownTableBodyRight">Apple M1 Max </td><td class="markdownTableBodyRight">8/10 cores </td><td class="markdownTableBodyRight">72 </td><td class="markdownTableBodyLeft">GNU 14.1.0 </td><td class="markdownTableBodyLeft">N/A </td></tr>
170+
<td class="markdownTableBodyRight">Apple M1 Max </td><td class="markdownTableBodyRight">8/10 cores </td><td class="markdownTableBodyRight">14.3 </td><td class="markdownTableBodyLeft">GNU 14.1.0 </td><td class="markdownTableBodyLeft">N/A </td></tr>
171171
<tr class="markdownTableRowEven">
172-
<td class="markdownTableBodyRight">AMD EPYC 9534 (Genoa) </td><td class="markdownTableBodyRight">64/64 cores </td><td class="markdownTableBodyRight">96 </td><td class="markdownTableBodyLeft">GNU 12.3.0 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
172+
<td class="markdownTableBodyRight">IBM Power9 </td><td class="markdownTableBodyRight">20/21 cores </td><td class="markdownTableBodyRight">21.2 </td><td class="markdownTableBodyLeft">GNU 9.1.0 </td><td class="markdownTableBodyLeft">OLCF Summit </td></tr>
173+
</table>
174+
<p>Processors To-do:</p>
175+
<table class="markdownTable">
176+
<tr class="markdownTableHead">
177+
<th class="markdownTableHeadRight">Hardware </th><th class="markdownTableHeadRight"></th><th class="markdownTableHeadRight">Grind Time </th><th class="markdownTableHeadLeft">Compiler </th><th class="markdownTableHeadLeft">Computer </th></tr>
173178
<tr class="markdownTableRowOdd">
174-
<td class="markdownTableBodyRight">AMD EPYC 7763 (Milan) </td><td class="markdownTableBodyRight">24/64 cores </td><td class="markdownTableBodyRight">108 </td><td class="markdownTableBodyLeft">GNU 11.4.0 </td><td class="markdownTableBodyLeft">NCSA Delta </td></tr>
179+
<td class="markdownTableBodyRight">AMD EPYC 9534 (Genoa) </td><td class="markdownTableBodyRight">64/64 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">GNU 12.3.0 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
175180
<tr class="markdownTableRowEven">
176-
<td class="markdownTableBodyRight">Intel Xeon Platinum 8462Y+ (Sapphire Rapids) </td><td class="markdownTableBodyRight">16/32 cores </td><td class="markdownTableBodyRight">110 </td><td class="markdownTableBodyLeft">GNU 12.3.0 </td><td class="markdownTableBodyLeft">GT ICE </td></tr>
181+
<td class="markdownTableBodyRight">AMD EPYC 7763 (Milan) </td><td class="markdownTableBodyRight">24/64 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">GNU 11.4.0 </td><td class="markdownTableBodyLeft">NCSA Delta </td></tr>
177182
<tr class="markdownTableRowOdd">
178-
<td class="markdownTableBodyRight">Intel Xeon Gold 6454S (Sapphire Rapids) </td><td class="markdownTableBodyRight">16/32 cores </td><td class="markdownTableBodyRight">111 </td><td class="markdownTableBodyLeft">NVHPC 24.5 </td><td class="markdownTableBodyLeft">GT Rogues Gallery </td></tr>
183+
<td class="markdownTableBodyRight">Intel Xeon Platinum 8462Y+ (Sapphire Rapids) </td><td class="markdownTableBodyRight">16/32 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">GNU 12.3.0 </td><td class="markdownTableBodyLeft">GT ICE </td></tr>
179184
<tr class="markdownTableRowEven">
180-
<td class="markdownTableBodyRight">NVIDIA Grace CPU (Arm, Neoverse V2) </td><td class="markdownTableBodyRight">18/72 cores </td><td class="markdownTableBodyRight">116 </td><td class="markdownTableBodyLeft">NVHPC 24.1 </td><td class="markdownTableBodyLeft">GT Rogues Gallery </td></tr>
185+
<td class="markdownTableBodyRight">Intel Xeon Gold 6454S (Sapphire Rapids) </td><td class="markdownTableBodyRight">16/32 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">NVHPC 24.5 </td><td class="markdownTableBodyLeft">GT Rogues Gallery </td></tr>
181186
<tr class="markdownTableRowOdd">
182-
<td class="markdownTableBodyRight">AMD EPYC 7452 (Rome) </td><td class="markdownTableBodyRight">16/32 cores </td><td class="markdownTableBodyRight">126 </td><td class="markdownTableBodyLeft">GNU 12.3.0 </td><td class="markdownTableBodyLeft">GT ICE </td></tr>
187+
<td class="markdownTableBodyRight">NVIDIA Grace CPU (Arm, Neoverse V2) </td><td class="markdownTableBodyRight">18/72 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">NVHPC 24.1 </td><td class="markdownTableBodyLeft">GT Rogues Gallery </td></tr>
183188
<tr class="markdownTableRowEven">
184-
<td class="markdownTableBodyRight">Intel Xeon Platinum 8352Y (Ice Lake) </td><td class="markdownTableBodyRight">12/32 cores </td><td class="markdownTableBodyRight">128 </td><td class="markdownTableBodyLeft">NVHPC 24.5 </td><td class="markdownTableBodyLeft">GT Rogues Gallery </td></tr>
189+
<td class="markdownTableBodyRight">AMD EPYC 7452 (Rome) </td><td class="markdownTableBodyRight">16/32 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">GNU 12.3.0 </td><td class="markdownTableBodyLeft">GT ICE </td></tr>
185190
<tr class="markdownTableRowOdd">
186-
<td class="markdownTableBodyRight">AMD EPYC 7713 (Milan) </td><td class="markdownTableBodyRight">32/64 cores </td><td class="markdownTableBodyRight">137 </td><td class="markdownTableBodyLeft">GNU 12.1.0 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
191+
<td class="markdownTableBodyRight">Intel Xeon Platinum 8352Y (Ice Lake) </td><td class="markdownTableBodyRight">12/32 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">NVHPC 24.5 </td><td class="markdownTableBodyLeft">GT Rogues Gallery </td></tr>
187192
<tr class="markdownTableRowEven">
188-
<td class="markdownTableBodyRight">Intel Xeon Gold 6226 (Cascade Lake) </td><td class="markdownTableBodyRight">12/12 cores </td><td class="markdownTableBodyRight">152 </td><td class="markdownTableBodyLeft">Intel oneAPI 2022.1 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
193+
<td class="markdownTableBodyRight">AMD EPYC 7713 (Milan) </td><td class="markdownTableBodyRight">32/64 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">GNU 12.1.0 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
189194
<tr class="markdownTableRowOdd">
190-
<td class="markdownTableBodyRight">Ampere Altra Max (Arm, Neoverse-N1) </td><td class="markdownTableBodyRight">8/80 cores </td><td class="markdownTableBodyRight">183 </td><td class="markdownTableBodyLeft">GNU 12.2.0 </td><td class="markdownTableBodyLeft">OLCF Wombat </td></tr>
195+
<td class="markdownTableBodyRight">Intel Xeon Gold 6226 (Cascade Lake) </td><td class="markdownTableBodyRight">12/12 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">Intel oneAPI 2022.1 </td><td class="markdownTableBodyLeft">GT Phoenix </td></tr>
191196
<tr class="markdownTableRowEven">
192-
<td class="markdownTableBodyRight">Intel Xeon E5-2650V4 (Broadwell) </td><td class="markdownTableBodyRight">8/12 cores </td><td class="markdownTableBodyRight">230 </td><td class="markdownTableBodyLeft">NVHPC 23.5 </td><td class="markdownTableBodyLeft">GT CSE Internal </td></tr>
197+
<td class="markdownTableBodyRight">Ampere Altra Max (Arm, Neoverse-N1) </td><td class="markdownTableBodyRight">8/80 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">GNU 12.2.0 </td><td class="markdownTableBodyLeft">OLCF Wombat </td></tr>
198+
<tr class="markdownTableRowOdd">
199+
<td class="markdownTableBodyRight">Intel Xeon E5-2650V4 (Broadwell) </td><td class="markdownTableBodyRight">8/12 cores </td><td class="markdownTableBodyRight">n/a </td><td class="markdownTableBodyLeft">NVHPC 23.5 </td><td class="markdownTableBodyLeft">GT CSE Internal </td></tr>
193200
</table>
194201
<p><b>All grind times are in nanoseconds (ns) per grid point (gp) per equation (eq) per right-hand side (rhs) evaluation, so X ns/gp/eq/rhs. Lower is better.</b></p>
195202
<h1><a class="anchor" id="autotoc_md64"></a>

0 commit comments

Comments
 (0)