-
Notifications
You must be signed in to change notification settings - Fork 60
This issue was moved to a discussion.
You can continue the conversation there. Go to discussion →
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ASAP7 tutorial: make buildfile CONFIG=TinyRocketConfig #643
Comments
This is actually because barstools' MacroCompiler is too verbose and expects a certain order of 1-port and 2-port RAMs: https://github.com/ucb-bar/barstools/blob/master/src/main/scala/barstools/macros/MacroCompiler.scala#L640. As long as you have the generated As for the Make error, it looks like for some reason, |
|
The ASAP7 SRAM compiler doesn't use the Yes, you are correct about the |
Hi, Harrison: my linux server version is Ubuntu18.04 and using VCS2017 to simulate default config. Due to the version of VCS is not so suitable for ubuntu18, I revise vcs.mk according to the workaround here. [https://github.com/chipsalliance/rocket-chip/issues/1377] VCS_CC_OPTS = I can use VCS to simulate other project wrote by myself. So VCS works well outsides hammer. /usr/local/bin/ld: /home/liqiang/eda/app/synopsys/vcs/N-2017.12-1/linux64/lib/libvcsnew.so: undefined reference to /usr/local/bin/ld: /home/liqiang/eda/app/synopsys/vcs/N-2017.12-1/linux64/lib/libvcsnew.so: undefined reference to What do I need to do to fix this problem? maybe I miss some settings? @harrisonliew full-log is here: [ Best, |
It looks to me like you are trying to use the workaround to use gcc-4.8 but your dynamic linker/loader is still the default on in |
Hi,harrison: First, I validate the VCS and system environment by running make and simulation in g++ -w -pipe -fPIC -I/export/Apps/Synopsys/vcs/O-2018.09-SP2-2/include -std=c++11 -I/export1/Workspace/liqiang/socgen/hwgen/chipyard/riscv-tools-install/include -I/export1/Workspace/liqiang/socgen/hwgen/chipyard/tools/DRAMSim2 -I/export1/Workspace/liqiang/socgen/hwgen/chipyard/vlsi/generated-src/chipyard.TestHarness.MediumBoomConfig -std=c++11 -I/export1/Workspace/liqiang/socgen/hwgen/chipyard/riscv-tools-install/include -I/export1/Workspace/liqiang/socgen/hwgen/chipyard/tools/DRAMSim2 -I/export1/Workspace/liqiang/socgen/hwgen/chipyard/vlsi/generated-src/chipyard.TestHarness.MediumBoomConfig -O3 -I/export/Apps/Synopsys/vcs/O-2018.09-SP2-2/include -c /export1/Workspace/liqiang/socgen/hwgen/chipyard/vlsi/generated-src/chipyard.TestHarness.MediumBoomConfig/uart.cc The full log is above, How Can I fix this Multiple definition warning here? |
Thanks for uploading the full log. Can you also upload the full log from compiling the simulator in |
I finally fix this issue. I found out hammer-synopsys_plugin/sim/vcs plugin file: I was checking all options resulting in the error , so the log has been sorted out. mediumboom-VCScompileOptions.log line 15-18 , line 23 are duplicated, it come from I dont know whether only me has this problem or not |
and according to #642 (comment) Error-[FGP_AFFINITY_FAILED] cpu_affinity/auto_affinity failed then I try with simulation option |
So the update to the Synopsys plugin was enabling VCS Fine-Grained Paralellism. This is supposed to work in VCS 2018, though based on the log it looks like it doesn't support hyper-threading (not in the documentation) and all of your cores are busy with other jobs. From the documentation, you can try adding the Also, about the duplicated options: did this go away when you updated the Synopsys plugin? We have not seen this before. If it still is a problem, can you attach the generated |
|
The duplicated files actually comes from the Makefile: https://github.com/ucb-bar/chipyard/blob/main/vlsi/Makefile#L229-L232 and https://github.com/ucb-bar/chipyard/blob/main/vlsi/Makefile#L129-L131 and the `$(VCS_NONCC_OPTS). This isn't a problem for VCS - it will ignore previously defined modules. Judging from your logs, this icache-caused fatal error seems similar to #642. Can you pull up some waveforms to see if there's an issue with an incorrect icache tag read/write? I'm thinking there's some bug in the VCS FGP or incompatibility with our generated binaries that is causing simulations to fail. If this is confirmed, then I will revert the Chipyard + Synopsys plugin PRs that enabled VCS FGP until we find a fix. |
I have these test cases below,why dhrystone can not pass in the original plugin? but succeed in sim/vcs mb-sim-dhry.log under VLSI folder : make sim-rtl-debug CONFIG=MediumBoomConfig BINARY=*
original *: before updated sim plugin,ie, no fgp updated*: with fgp |
Right, fgp is not working, somehow, and we're still debugging in #642. Can you use the Synopsys plugin before the FGP update: (commit hash 81720fdd5b7fd747f34606f09b8a7d0da438a3ef) for now? |
Yes, for now this fgp feature is not necessary for debuging. But later, this feature would be exciting. However, in the before-updated case, why the dhrystone benchmark can not pass the test? and another thing I wanna ask for help. because I need to make customed-config many times, how can I shorten the time of make stage, Maybe I need to revise |
Can you zip up and email me the various files of the dhrystone failure case for MediumBoom? Specifically, the waveform and the generated verilog for the design. |
sure ! I have send the tarball to you. |
fixed ! close here |
This issue was moved to a discussion.
You can continue the conversation there. Go to discussion →
when I run command: make buildfile CONFIG=TinyRocketConfig. Errors below came out and terminate in the end
[error] INFO: unable to compile data_arrays_0_0_ext using SRAM2RW64x4 port count must match
[error] INFO: unable to compile data_arrays_0_0_ext using SRAM2RW64x8 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW128x16 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW128x32 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW128x4 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW128x8 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW16x16 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW16x32 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW16x4 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW16x8 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x16 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x22 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x32 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x33 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x39 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x4 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW32x8 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW64x16 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW64x24 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW64x32 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW64x4 port count must match
[error] INFO: unable to compile l2_tlb_ram_ext using SRAM2RW64x8 port count must match
[success] Total time: 6 s, completed Mar 28, 2022, 9:47:37 PM
make: *** No rule to make target '/home/lx/hwgen0327/chipyard/vlsi/generated-src/chipyard.TestHarness.TinyRocketConfig/EICG_wrapper.v', needed by '/home/lx/hwgen0327/chipyard/vlsi/build/chipyard.TestHarness.TinyRocketConfig-ChipTop/inputs.yml'. Stop.
the generated-src fold has contain top.mems.v. but didn't output build folder. What should I do next?@harrisonliew
Thank you,
Best regards,
Leon.
The text was updated successfully, but these errors were encountered: