Skip to content

VCC incorrectly routed through a FAN_BOUNCE pip #1400

New issue
New issue
Closed
#1573
Closed
VCC incorrectly routed through a FAN_BOUNCE pip#1400
#1573
@acomodi

Description

@acomodi
acomodi
opened on Apr 7, 2020
Contributor

There is an issue for which the diff fasm tests fail when running the bram_test_36 test.
The failure is generated by a wrong routing to the constant VCC net of some of the pins of teh BRAM36.

The "wrong" route passes through an additional FAN_BOUNCE pip instead of connecting directly to the VCC_WIRE. As a result, fasm2bels wrongly computes the fixed route, generating a difference in the fasm files.

Even though this is not a critical error, meaning that the route is still valid, there is no straightforward way to teach fasm2bels to take into account the extra-jump to the BOUNCE pip.

One thing to notice is that, at least locally, using the binary heap for the router, helped to solve the problem, but, as discussed in #1386, the problem should not depend on the type of heap used, probably, with different seeds and multiple runs, also the binary heap would fail eventually.

Activity

litghost

litghost commented on Apr 7, 2020

@litghost
litghost
on Apr 7, 2020
Contributor

As a result, fasm2bels wrongly computes the fixed route, generating a difference in the fasm files.

The bug is in neither fasm2bels nor fasm2frames. Both tools a respecting the inputs they were given. The problem is in VPR (or other tool) and the routes it chooses. The solution is definitely not to modify fasm2bels, as the underlying bitstreeam uses the additional pip. A filter before fasm2frames or in fasm2frames is valid, or a change to VPR.

acomodi
changed the title [-]Diff tests fail for BRAM test 36[/-] [+]VCC incorrectly routed through a FAN_BOUNCE pip[/+] on Apr 16, 2020
acomodi

acomodi commented on Apr 16, 2020

@acomodi
acomodi
on Apr 16, 2020
ContributorAuthor

Seen this in another vendor tool ci run as well, this time on the murax test.

@@ -9069,7 +9069,6 @@
 INT_L_X6Y110.CLK_L1.FAN_BOUNCE5
 INT_L_X6Y110.CTRL_L0.FAN_BOUNCE1
 INT_L_X6Y110.CTRL_L1.FAN_BOUNCE1
-INT_L_X6Y110.FAN_ALT5.FAN_BOUNCE1
 INT_L_X6Y110.GFAN0.GND_WIRE
 INT_L_X6Y110.GFAN1.GND_WIRE
 INT_L_X6Y110.IMUX_L25.GFAN0
@@ -9281,7 +9280,6 @@
 INT_L_X6Y114.CTRL_L0.FAN_BOUNCE1
 INT_L_X6Y114.CTRL_L1.FAN_BOUNCE1
 INT_L_X6Y114.ER1BEG1.LOGIC_OUTS_L8
-INT_L_X6Y114.FAN_ALT5.FAN_BOUNCE1
 INT_L_X6Y114.GFAN0.GND_WIRE
 INT_L_X6Y114.GFAN1.GND_WIRE
 INT_L_X6Y114.IMUX_L10.GFAN0
@@ -9954,7 +9952,6 @@
 INT_L_X6Y135.ER1BEG2.ER1END1
 INT_L_X6Y135.ER1BEG3.ER1END2
 INT_L_X6Y135.ER1BEG_S0.SR1END3
-INT_L_X6Y135.FAN_ALT5.FAN_BOUNCE1
 INT_L_X6Y135.GFAN0.GND_WIRE
 INT_L_X6Y135.GFAN1.GND_WIRE
 INT_L_X6Y135.IMUX_L25.GFAN0
@@ -10181,7 +10178,6 @@
 INT_L_X6Y139.CLK_L1.FAN_BOUNCE5
 INT_L_X6Y139.CTRL_L0.FAN_BOUNCE1
 INT_L_X6Y139.CTRL_L1.FAN_BOUNCE1
-INT_L_X6Y139.FAN_ALT5.FAN_BOUNCE1
 INT_L_X6Y139.GFAN0.GND_WIRE
 INT_L_X6Y139.GFAN1.GND_WIRE
 INT_L_X6Y139.IMUX_L10.GFAN0
acomodi
mentioned this in 2 pull requests on Apr 20, 2020
acomodi
mentioned this on May 4, 2020
mkurc-ant
mentioned this in 2 pull requests on May 13, 2020
acomodi
mentioned this on May 14, 2020
mkurc-ant
mentioned this in 2 pull requests on May 29, 2020
tcal-x
mentioned this on Jun 15, 2020

4 remaining items

Loading
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

      Development

      Participants

      @litghost@acomodi

      Issue actions
        VCC incorrectly routed through a FAN_BOUNCE pip · Issue #1400 · f4pga/f4pga-arch-defs