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
[top,tlgen] fabric multi-clock support #903
Conversation
new PR from #519 |
b4c36fa
to
ad71cc9
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks OK.
I am wondering if we could try multi xbar in this PR. Only concern is how the main xbar figures out the address range to the downstream (second tier crossbar). It could be manually done at this time but later the tool can recognize.
hw/top_earlgrey/dv/xbar_main_tb.sv
Outdated
@@ -0,0 +1,334 @@ | |||
// Copyright lowRISC contributors. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This file shouldn't be checked in. @weicaiyang has been preparing the automatic TB wrapper for xbar test.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
o is that ready? Or will @weicaiyang remove the checked in stuff?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The top_earlgray xbar TB is currently at hw/ip/tlul/dv, but this one is not auto-generated
When we finish the TB automation, will check in generated files in hw/top_earlgrey/dv
o i actually wanted to do multi-xbar in a different PR. Basically the experiment to split up domains. |
ah so should i remove in this PR?
…On Wed, Nov 13, 2019 at 10:18 AM weicaiyang ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In hw/top_earlgrey/dv/xbar_main_tb.sv
<#903 (comment)>:
> @@ -0,0 +1,334 @@
+// Copyright lowRISC contributors.
The top_earlgray xbar TB is currently at hw/ip/tlul/dv, but this one is
not auto-generated
When we finish the TB automation, will check in generated files in
hw/top_earlgrey/dv
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#903?email_source=notifications&email_token=AAH2RSUESJ5C352M33YCXPLQTRABNA5CNFSM4JJ7K6W2YY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOCLOK7XY#discussion_r345919951>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAH2RSRC6XAAG7Z6CNLY4BTQTRABNANCNFSM4JJ7K6WQ>
.
|
Yes, please remove it |
ad71cc9
to
035376e
Compare
i've removed the tb file. |
could someone approve if this looks okay? |
I'm looking at this now.
…On Thu, Nov 14, 2019 at 11:48 AM tjaychen ***@***.***> wrote:
could someone approve if this looks okay?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#903?email_source=notifications&email_token=AJZQKVOJ3TOHQOUUEORJCBTQTWTQNA5CNFSM4JJ7K6W2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEDCECA#issuecomment-554050056>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AJZQKVP7PBOOGW4SNPGBT2DQTWTQNANCNFSM4JJ7K6WQ>
.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
From my understanding of what you're trying to do, this all makes sense,
so LGTMing, but I would prefer to get +1 from @eunchan as well since he's
the one whose world is changing here.
Also note this will knock us back to D1/V1 or D2/V2 for xbar (which is fine,
just we should mark that on the .prj.hjson file. @weicaiyang can do that
in a separate PR. I'll be working on the multi-entry .prj.hjson so that we can
keep the signed off one available.
Sure, will create a PR to change TLUL back to D1/V1 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
LGTM with comments. I think at this moment, it is not necessary to split the clock in the crossbar yet. But this change definitely let me engage the multi crossbars in the top sooner :) I will take a look at the multi xbar implementation after this is merged. If the feature works, it can reduce many Async FIFOs. |
Add the ability for fabric to support multiple clocks and resets. There is currently no defined relationship between clocks, so anything that is different is treated as asynchronous. This can be improved in the future. Signed-off-by: Timothy Chen <timothytim@google.com>
Also add multiple resets sync'ed to each clock. The multi-clock and resets are still connected to the same source, a separate PR will bring true clock and reset sources. Signed-off-by: Timothy Chen <timothytim@google.com>
035376e
to
e438a60
Compare
New feature was just added lowRISC#903 Current TB doesn't support multi-clock, use same clock to avoid failure Signed-off-by: Weicai Yang <weicai@google.com>
New feature was just added #903 Current TB doesn't support multi-clock, use same clock to avoid failure Signed-off-by: Weicai Yang <weicai@google.com>
Add the ability for fabric to support multiple clocks / resets
Also separate clock / reset connectivity at top.