Merge master into 2nd-cufr branch (#96)

* Fix rapidwright_picoblazearray BRAM count * Fix another typo * Update/fix Actions; add open-source links/publications (#95) * Up Xms/Xmx to 14G (of 16G) inside GitHub Actions Signed-off-by: Eddie Hung <[email protected]> * Retry all memory-limited benchmarks again Signed-off-by: Eddie Hung <[email protected]> * Explicitly use 22.04 docker image for libtinfo5 Signed-off-by: Eddie Hung <[email protected]> * Disable mock results Signed-off-by: Eddie Hung <[email protected]> * Use JDK not JRE image Signed-off-by: Eddie Hung <[email protected]> * Add open-source and publication links Signed-off-by: Eddie Hung <[email protected]> * Remove all uses of *_MOCK_RESULT Signed-off-by: Eddie Hung <[email protected]> * Remove commented out code that's now working Signed-off-by: Eddie Hung <[email protected]> * Fix GRoute publication Signed-off-by: Eddie Hung <[email protected]> --------- Signed-off-by: Eddie Hung <[email protected]> --------- Signed-off-by: Eddie Hung <[email protected]>
Xilinx · Oct 25, 2024 · 74cbfb8 · 74cbfb8
1 parent 9968a2e
commit 74cbfb8
Show file tree

Hide file tree

Showing 7 changed files with 28 additions and 34 deletions.
diff --git a/.github/workflows/make.yml b/.github/workflows/make.yml
@@ -38,15 +38,6 @@ jobs:
           - boom_soc
           - ispd16_example2
         exclude:
-          # Insufficient memory on GitHub Actions
-          - router: cufr
-            benchmark: mlcad_d181_lefttwo3rds
-          - router: cufr
-            benchmark: koios_dla_like_large
-          - router: cufr
-            benchmark: boom_soc
-          - router: cufr
-            benchmark: ispd16_example2
           # NXRoute does not support LUT pin swapping
           - router: nxroute-poc
             lutpinswapping: true
@@ -75,9 +66,6 @@ jobs:
       - env:
           REPORT_ROUTE_STATUS_URL:      ${{ secrets.REPORT_ROUTE_STATUS_URL }}
           REPORT_ROUTE_STATUS_AUTH:     ${{ secrets.REPORT_ROUTE_STATUS_AUTH }}
-          # For certain benchmarks, wirelength_analyzer/CheckPhysNetlist requires more memory than that available in GitHub Actions
-          WIRELENGTH_ANALYZER_MOCK_RESULT: ${{ matrix.benchmark == 'koios_dla_like_large' }}
-          CHECK_PHYS_NETLIST_DIFF_MOCK_RESULT: ${{ matrix.benchmark == 'koios_dla_like_large' }}
           RWROUTE_FORCE_LUT_PINSWAPPING: ${{ (matrix.router == 'rwroute' || matrix.router == 'cufr') && matrix.lutpinswapping }}
           RWROUTE_FORCE_LUT_ROUTETHRU:   ${{ (matrix.router == 'rwroute' || matrix.router == 'cufr') && matrix.lutroutethru }}
         run: |

diff --git a/Makefile b/Makefile
@@ -53,8 +53,8 @@ else
 endif
 
 ifdef GITHUB_ACTIONS
-    # Limit Java heap size inside GitHub Actions to 6G
-    JVM_HEAP = -Xms6g -Xmx6g
+    # Limit Java heap size inside GitHub Actions to 14G
+    JVM_HEAP = -Xms14g -Xmx14g
 else
     # If not specified, limit Java heap size ~32G
     JVM_HEAP ?= -Xms32736m -Xmx32736m
@@ -119,12 +119,7 @@ fpga-interchange-schema/interchange/capnp/java.capnp:
         fi
 
 %_$(ROUTER).wirelength: %_$(ROUTER).phys | setup-wirelength_analyzer
-	if [[ "$(WIRELENGTH_ANALYZER_MOCK_RESULT)" == "true" ]]; then \
-            echo "::warning file=$<::wirelength_analyzer not run because WIRELENGTH_ANALYZER_MOCK_RESULT is set"; \
-	    echo "Wirelength: inf" > $@; \
-	else \
-	    python3 wirelength_analyzer/wa.py $< $(call log_and_or_display,$@); \
-	fi
+	python3 wirelength_analyzer/wa.py $< $(call log_and_or_display,$@); \
 
 .PHONY: score-$(ROUTER)
 score-$(ROUTER): $(foreach b,$(BENCHMARKS),$b_$(ROUTER).wirelength $b_$(ROUTER).check)

diff --git a/alpha_submission/rwroute_container.def b/alpha_submission/rwroute_container.def
@@ -1,5 +1,5 @@
 BootStrap: docker
-From: eclipse-temurin:17 # Base image with Java VM 17 on Ubuntu
+From: eclipse-temurin:17-jdk-jammy # Base image with Java VM 17 on Ubuntu 22.04
 
 %post
 	# Install remaining system dependencies

diff --git a/docs/benchmarks.md b/docs/benchmarks.md
@@ -41,8 +41,8 @@ Available from [https://github.com/Xilinx/fpga24_routing_contest/releases/latest
 
 |Source Benchmark Suite|Benchmark Name|LUTs|FFs|DSPs|BRAMs|OOC [1]|
 |----------------------|--------------|----|---|----|-----|-------|
-| [RapidWright](https://github.com/Xilinx/RapidWright)                                                                    |`picoblaze_array` (660 PicoBlaze cores)                |76k |77k  |0   |0  |Y   |
-| [Corundum](https://github.com/corundum/corundum)                                                                        |`100g` (ADM_PCIE_9V3 25G)                              |76k |104k |0   |290|N   |
+| [RapidWright](https://github.com/Xilinx/RapidWright)                                                                    |`picoblazearray` (660 PicoBlaze cores)                 |76k |77k  |0   |660|Y   |
+| [Corundum](https://github.com/corundum/corundum)                                                                        |`100g` (ADM_PCIE_9V3 100G)                             |76k |104k |0   |290|N   |
 | [Koios 2.0](https://docs.verilogtorouting.org/en/latest/vtr/benchmarks/#koios-2-0-benchmarks)                           |`clstm_like_large` (clstm_like.large)                  |89k |184k |1289|370|Y   |
 | [Titan23](https://docs.verilogtorouting.org/en/latest/vtr/benchmarks/#titan-benchmarks)                                 |`orig_gsm_x6` (Original gsm_switch replicated 6 times) |133k|160k |0   |432|Y   |
 | [CoreScore](https://github.com/olofk/corescore)                                                                         |`900` (900 SERV cores)                                 |174k|210k |0   |451|N   |

diff --git a/docs/results.md b/docs/results.md
@@ -17,6 +17,9 @@ For an overview of the reuslts, please see these [slides](fpga24-contest-slides.
 **Team members:** Dani Maarouf, Timothy Martin, Charlotte Barnes<br>
 **Advisors:** Shawki Areibi, Gary Grewal
 
+Publication:
+- [A High-Performance Routing Engine for Large-Scale FPGAs](https://doi.ieeecomputersociety.org/10.1109/FPL64840.2024.00017)
+
 | Overview | Video |
 | - | - |
 | [![GRoute-Slide](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/5b279fc6-6c58-43f1-9b51-a0aef72dcf86)](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/5b279fc6-6c58-43f1-9b51-a0aef72dcf86) | <video src="https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/edbcbd5a-f86b-47fe-ae58-a10c05b15e8a#t=0.5" controls="controls" style="max-width: 662px;"/> |
@@ -26,6 +29,12 @@ For an overview of the reuslts, please see these [slides](fpga24-contest-slides.
 **Team members:** Xinshi Zang, Wenhao Lin, Shiju Lin, Qin Luo<br>
 **Advisor:** Evangeline F.Y. Young
 
+Open-source: https://github.com/xszang/parallel-routing ([integrated upstream](https://github.com/Xilinx/fpga24_routing_contest/tree/2nd-cufr))
+
+Publications:
+- [An Open-Source Fast Parallel Routing Approach for Commercial FPGAs](https://github.com/xszang/parallel-routing/blob/main/doc/glsvlsi24-camera-ready.pdf)
+- Potter: A Parallel Overlap-Tolerant Router for UltraScale FPGAs *(to appear)*
+
 | Overview | Video |
 | - | - |
 | [![CUFR-Slide](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/9242ce96-6517-44c1-829f-f5c8f2d28339)](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/9242ce96-6517-44c1-829f-f5c8f2d28339) | <video src="https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/780a30df-b7cc-483d-9bfb-f2f354b4d5d1#t=0.5" controls="controls" style="max-width: 662px;"/> |
@@ -36,6 +45,9 @@ For an overview of the reuslts, please see these [slides](fpga24-contest-slides.
 **Advisor:** Guojie Luo<sup>*</sup><br>
 *<sup>\*</sup>Peking University, <sup>+</sup>DeePoly Technology Inc.*
 
+Publication:
+- [AceRoute: Adaptive Compute-Efficient FPGA Routing with Pluggable Intra-Connection Bidirectional Exploration](https://xmwei.com/assets/pdf/wei2024aceroute.pdf)
+
 | Overview | Video |
 | - | - |
 | [![AceRoute-Slide](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/9d7dc7b6-e31d-44df-8e30-90a3f1f19daa)](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/9d7dc7b6-e31d-44df-8e30-90a3f1f19daa) | <video src="https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/2f1e36da-80cd-4859-8ce8-2bfdaeb3075a#t=0.5" controls="controls" style="max-width: 662px;"/> |
@@ -45,6 +57,8 @@ For an overview of the reuslts, please see these [slides](fpga24-contest-slides.
 **Team members:** Jiarui Wang, Xun Jiang, Chunyuan Zhao<br>
 **Advisor:** Yibo Lin
 
+Open-source: https://github.com/PKU-IDEA/OpenPARF/tree/master/fpga24contest ([integrated upstream](https://github.com/Xilinx/fpga24_routing_contest/tree/4th-cuckoo))
+
 | Overview | Video |
 | - | - |
 | [![TeamCuckoo-Slide](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/6483ab18-be08-4be3-ad46-8b69a5d13a55)](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/6483ab18-be08-4be3-ad46-8b69a5d13a55) | <video src="https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/af84a46e-d73a-4b87-be18-9652621f6b5c" controls="controls" style="max-width: 662px;"/> |
@@ -54,6 +68,8 @@ For an overview of the reuslts, please see these [slides](fpga24-contest-slides.
 **Team members:** Wenbin Teng, Qianyu Cheng, Zhendong Zheng, Binze Jiang, Yixuan Zhu, Zihan Wang<br>
 **Advisors:** Chao Wang, Teng Wang
 
+Open-source: https://github.com/Reconfigurable-Computing/RapidWright ([integrated upstream](https://github.com/Xilinx/fpga24_routing_contest/tree/5th-hao3))
+
 | Overview | Video |
 | - | - |
 | [![Hao3-Slide](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/14b148e3-55a4-48e8-a0ee-88f19498b253)](https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/14b148e3-55a4-48e8-a0ee-88f19498b253) | <video src="https://github.com/Xilinx/fpga24_routing_contest/assets/90657806/00418fb1-2bdc-4a8b-93d0-125f8726ec00" controls="controls" style="max-width: 662px;"/> |
diff --git a/final_submission/rwroute_container.def b/final_submission/rwroute_container.def
@@ -1,5 +1,5 @@
 BootStrap: docker
-From: eclipse-temurin:17 # Base image with Java VM 17 on Ubuntu
+From: eclipse-temurin:17-jdk-jammy # Base image with Java VM 17 on Ubuntu 22.04
 
 %files
 	## Example copy of /dir1 into /opt inside container

diff --git a/src/com/xilinx/fpga24_routing_contest/CheckPhysNetlist.java b/src/com/xilinx/fpga24_routing_contest/CheckPhysNetlist.java
@@ -57,17 +57,12 @@ public static void main(String[] args) throws IOException, InterruptedException
 
         // Read the routed and unrouted Physical Netlists
         Design routedDesign = PhysNetlistReader.readPhysNetlist(args[1]);
-        int numDiffs = 0;
-        if ("true".equals(System.getenv("CHECK_PHYS_NETLIST_DIFF_MOCK_RESULT"))) {
-            System.out.println("::warning file=" + args[1] + "::CheckPhysNetlist's DesignComparator not run because CHECK_PHYS_NETLIST_DIFF_MOCK_RESULT is set");
-        } else {
-            Design unroutedDesign = PhysNetlistReader.readPhysNetlist(args[2]);
+        Design unroutedDesign = PhysNetlistReader.readPhysNetlist(args[2]);
 
-            DesignComparator dc = new DesignComparator();
-            // Only compare PIPs on static and clock nets
-            dc.setComparePIPs((net) -> net.isStaticNet() || net.isClockNet());
-            numDiffs = dc.compareDesigns(unroutedDesign, routedDesign);
-        }
+        DesignComparator dc = new DesignComparator();
+        // Only compare PIPs on static and clock nets
+        dc.setComparePIPs((net) -> net.isStaticNet() || net.isClockNet());
+        int numDiffs = dc.compareDesigns(unroutedDesign, routedDesign);
         if (numDiffs == 0) {
             System.out.println("INFO: No differences found between routed and unrouted netlists");
         } else {