Add column mux ratio for ratio = 1, 2, 4 by tnguy19 · Pull Request #10016 · The-OpenROAD-Project/OpenROAD

tnguy19 · 2026-03-31T23:16:42Z

Summary

Add -column_mux_ratio parameter to generate_ram and generate_ram_netlist with default value = 1.

The column_mux_ratio value indicate the number of words that will share the same physical row, and currently this implementation will work correctly for values = 1, 2, and 4*.

Muxes are implemented using AOI22s for ratio =2 (one AOI22+inverter) and ratio =4 (2 stage-1 AOI22s feeding a stage-2 AOI22).

Some changes to the code to accommodate this feature additions are also added.

*Mux_ratio=4 currently have correct functionality but faces congestion issue during cell placement

Type of Change

New feature

Impact

generate_ram and generate_ram_netlist can be use with option -column_mux_ratio with default value = 1. Scripts that do not have column_mux_ratio are not affected
Change from using num_words to num_rows to indicate the number of physical row that will be created. For column_mux_ratio >1, num_rows = num_words / column_mux_ratio physical rows are generated. Column mux are placed in the buffer row
Change naming instances with format storage_{row}_{bit} to storage_{row}_{word}_{bit} to reflect the placement of multiple words in one row when column_mux_ratio > 1

Verification

I have verified that the local build succeeds (./etc/Build.sh).
I have run the relevant tests and they pass.
My code follows the repository's formatting guidelines.
I have signed my commits (DCO).

Related Issues

N/A

@rovinski

NOTE: PR is reopened in #10018

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

…OI22 Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

gemini-code-assist

Code Review

This pull request introduces support for column muxing (ratios 1, 2, and 4) in the RAM generator, involving updates to layout grid calculations and the addition of AOI22-based multiplexing logic. Feedback identifies a potential out-of-bounds access when the number of rows is less than 2 and suggests moving local lambdas to a namespace-level scope. Additionally, the reviewer recommended improving the robustness of the port extraction logic and replacing hardcoded magic numbers with named constants for better maintainability.

gemini-code-assist · 2026-03-31T23:19:26Z

+
+  int num_inputs = std::ceil(std::log2(num_words));
+  // compute information to support col/mux ratio feature
+  int num_rows = num_words / column_mux_ratio;


The current implementation does not support cases where num_rows (calculated as num_words / column_mux_ratio) is less than 2. For example, if num_words=4 and column_mux_ratio=4, num_rows will be 1, which leads to an out-of-bounds access in addr_inputs_ at line 1136 (addr_inputs_[num_word_bits]) because num_word_bits will equal num_inputs. Please add a check to ensure num_rows >= 2.

Suggested change

int num_rows = num_words / column_mux_ratio;

int num_rows = num_words / column_mux_ratio;

if (num_rows < 2) {

logger_->error(RAM,

35,

"The ram generator currently requires at least 2 rows "

"(num_words / column_mux_ratio >= 2).");

}

References

Calls to logger_->error are expected to terminate the program, so additional error handling like returning an error code or throwing an exception is not necessary.

Fine to include this suggestion (but can remove the word "currently" - there is no reason to support a single-row RAM)

gemini-code-assist · 2026-03-31T23:19:26Z

+    auto isNotPort = [](const sta::FuncExpr* e) {
+      return e && e->op() == sta::FuncExpr::Op::not_ && e->left()
+             && e->left()->op() == sta::FuncExpr::Op::port;
+    };
+    auto isAndOfNots = [&isNotPort](const sta::FuncExpr* e) {
+      return e && e->op() == sta::FuncExpr::Op::and_ && isNotPort(e->left())
+             && isNotPort(e->right());
+    };


According to the general rules, helper logic should be defined as a free function in a namespace rather than as a local lambda within a function. These lambdas (isNotPort and isAndOfNots) should be moved to a namespace-level scope.

References

Helper logic should be defined as a free function in a namespace rather than as a local lambda within a function.

gemini-code-assist · 2026-03-31T23:19:27Z

+        if (f->op() == sta::FuncExpr::Op::not_) {
+          // Compact form: not_(or_(and_(A1,A2), and_(B1,B2)))
+          auto or_expr = f->left();
+          auto and_a = or_expr->left();
+          auto and_b = or_expr->right();
+          aoi22_in_a1_ = and_a->left()->port()->name();
+          aoi22_in_a2_ = and_a->right()->port()->name();
+          aoi22_in_b1_ = and_b->left()->port()->name();
+          aoi22_in_b2_ = and_b->right()->port()->name();
+        } else {
+          // SOP expanded form used by sky130hd:
+          // or_(or_(or_(and_(not_(A1),not_(B1)), and_(not_(A1),not_(B2))),
+          //     and_(not_(A2),not_(B1))),
+          //     and_(not_(A2),not_(B2)))
+          // Port names are inside not_() wrappers, requiring an extra
+          // ->left() to reach the actual port compared to the compact form.
+          // e.g. and_(not_(A1), not_(B1))->left()->left()->port() gives A1
+          // NOTE: this branch is PDK-specific to sky130hd's liberty
+          // representation. If adding support for other PDKs, verify their
+          // AOI22 function expression form and extend this extraction
+          // accordingly.
+          auto t1 = f->left()->left()->left();
+          auto t2 = f->left()->left()->right();
+          auto t3 = f->left()->right();
+          aoi22_in_a1_ = t1->left()->left()->port()->name();
+          aoi22_in_a2_ = t3->left()->left()->port()->name();
+          aoi22_in_b1_ = t1->right()->left()->port()->name();
+          aoi22_in_b2_ = t2->right()->left()->port()->name();
+        }


The logic for extracting port names from the AOI22 function expression is brittle as it relies on a specific nesting structure of the sta::FuncExpr tree (especially for the SOP expanded form). If a library represents the AOI22 function differently, this extraction will fail or produce incorrect results. Consider implementing a more robust matching algorithm or providing a way to specify these port names via configuration if the automatic detection fails.

gemini-code-assist · 2026-03-31T23:19:27Z

+              "inv",
+              inv_cell_,
+              {{"A", aoi_out}, {"Y", q_outputs_[0][global_bit]->getNet()}});
+          ram_grid_.addCell(std::move(mux_cell), base_col + 1);


The column offsets used for cell placement (e.g., base_col + 1, base_col + 2, base_col + 3) are hardcoded magic numbers. Following the general rules, these should be defined as named constants to improve code clarity and maintainability.

References

Define tunable parameters as named constants instead of using hardcoded magic numbers.

github-actions

clang-tidy made some suggestions

github-actions · 2026-03-31T23:21:46Z

+             && isNotPort(e->right());
+    };
+    aoi22_cell_ = findMaster(
+        [&isNotPort, &isAndOfNots](sta::LibertyPort* port) {


warning: lambda capture 'isNotPort' is not used [clang-diagnostic-unused-lambda-capture]

Suggested change

[&isNotPort, &isAndOfNots](sta::LibertyPort* port) {

[ &isAndOfNots](sta::LibertyPort* port) {

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

1. Add -column_mux_ratio flag to generate_ram and generate_ram_netlist (default=1) 2. Mux created from inline AOI22 mux placement for ratio=2 (single AOI22+INV per bit) and ratio=4 (2-stage AOI22 tree) 3. Fix addr BTerm naming: addr[i] -> addr_rw[i] 4. Fix makeSlice prefix to include word_idx: storage_{row}_{word}_{bit} 5. Use num_rows for column generation instead of num_words 6. Update make_8x8.ok with new naming convention for cells 7. Add make_8x8_mux2.ok for test with col_mux_ratio=2 Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

tnguy19 added 7 commits March 17, 2026 11:57

add column mux support with flag -mux_col_ratio parameter

de1551a

add check to only search for aoi master when mux_col_ratio > 1

ec7e9ee

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

Fix AOI22 FuncExpr matcher to handle SOP expanded form

315d88a

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

ram: add PDN strap pitch vs die size error check to prevent segfault

cb55753

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

Merge upstream master - accept refactored RAM structure

07f2492

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

add code to make mux for col/mux featureand findMaster function for A…

cfefab1

…OI22 Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

Merge upstream master into add-column-mux

4a8aa15

tnguy19 force-pushed the add-column-mux branch from dd740f6 to 32ea550 Compare March 31, 2026 23:19

gemini-code-assist bot reviewed Mar 31, 2026

View reviewed changes

github-actions bot reviewed Mar 31, 2026

View reviewed changes

tnguy19 force-pushed the add-column-mux branch from 32ea550 to fdcebf9 Compare March 31, 2026 23:25

tnguy19 added 2 commits March 31, 2026 19:28

add column_mux_ratio support

1268bd1

Signed-off-by: Thinh Nguyen <nguyenthinh19011@gmail.com>

tnguy19 force-pushed the add-column-mux branch from fdcebf9 to 5dd5278 Compare March 31, 2026 23:28

tnguy19 closed this Mar 31, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add column mux ratio for ratio = 1, 2, 4#10016

Add column mux ratio for ratio = 1, 2, 4#10016
tnguy19 wants to merge 9 commits intoThe-OpenROAD-Project:masterfrom
tnguy19:add-column-mux

tnguy19 commented Mar 31, 2026 •

edited

Loading

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Mar 31, 2026

Uh oh!

rovinski Mar 31, 2026

Uh oh!

gemini-code-assist bot Mar 31, 2026

Uh oh!

gemini-code-assist bot Mar 31, 2026

Uh oh!

gemini-code-assist bot Mar 31, 2026

Uh oh!

github-actions bot left a comment

Uh oh!

github-actions bot Mar 31, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-  int num_rows = num_words / column_mux_ratio;
+  int num_rows = num_words / column_mux_ratio;
+  if (num_rows < 2) {
+    logger_->error(RAM,
+,
+                   "The ram generator currently requires at least 2 rows "
+                   "(num_words / column_mux_ratio >= 2).");
+  }

	[&isNotPort, &isAndOfNots](sta::LibertyPort* port) {
	[ &isAndOfNots](sta::LibertyPort* port) {

Conversation

tnguy19 commented Mar 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Type of Change

Impact

Verification

Related Issues

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

rovinski Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Uh oh!

github-actions bot Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

tnguy19 commented Mar 31, 2026 •

edited

Loading