Simulation performance #961

ydnatag · 2019-06-25T21:20:40Z

Hi,
At work, i'm porting all the repositories to mainline cocotb with python3 and i'm seeing that the performance using python3 is ~4 times worse than python2 For executing the tests, i just changed the PYTHON_BIN env,

I can't share a testcase now, i will try to recreate this performance issue with a sharable test.

My environment:

debian stretch
python2.7
python3.5
current cocotb master

Thanks

The text was updated successfully, but these errors were encountered:

ydnatag · 2019-06-25T21:36:04Z

Some results:

Python2

*************************************************************************************************************
** TEST                                                 PASS/FAIL  SIM TIME(NS)  REAL TIME(S)  RATIO(NS/S) **
*************************************************************************************************************
** test.xxxxxxxxxxxxxxxxxxxx                              PASS        60983.81          1.20     50900.73  **
** test.xxxxxxxxxxxxxxxxxx                                PASS        79006.15          4.00     19732.29  **
** test.xxxxxxxxxxxxxx                                    PASS          566.95          0.02     25764.83  **
** test.xxxxxxxxxxxxxxxxxxxxxxx                           PASS        64058.68          1.28     49926.37  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxx                         PASS       160903.77          3.10     51852.69  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       220670.00         44.15      4998.32  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       270395.00         54.00      5007.50  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx    PASS       394325.00         86.88      4538.70  **
** test.xxxxxxxxxx                                        PASS       160353.47          2.99     53638.34  **
** test.xxxxxxxxxxxxxx                                    PASS         5362.68          0.16     33913.34  **
*************************************************************************************************************

*************************************************************************************
**                                 ERRORS : 0                                      **
*************************************************************************************
**                               SIM TIME : 1485586.65 NS                          **
**                              REAL TIME : 199.12 S                               **
**                        SIM / REAL TIME : 7460.66 NS/S                           **
*************************************************************************************

Python3

*************************************************************************************************************
** TEST                                                 PASS/FAIL  SIM TIME(NS)  REAL TIME(S)  RATIO(NS/S) **
*************************************************************************************************************
** test.xxxxxxxxxxxxxxxxxxxx                              PASS        61183.91          3.54     17286.30  **
** test.xxxxxxxxxxxxxxxxxx                                PASS        79206.25          9.50      8340.80  **
** test.xxxxxxxxxxxxxx                                    PASS          767.05          0.06     13165.36  **
** test.xxxxxxxxxxxxxxxxxxxxxxx                           PASS        64258.78          3.16     20356.97  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxx                         PASS       161103.87          7.86     20496.36  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       221070.00        113.51      1947.57  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       270795.00        142.54      1899.82  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx    PASS       568430.00        343.62      1654.23  **
** test.xxxxxxxxxx                                        PASS       160553.57          8.03     20004.19  **
** test.xxxxxxxxxxxxxx                                    PASS         5562.78          0.38     14753.74  **
*************************************************************************************************************

*************************************************************************************
**                                 ERRORS : 0                                      **
*************************************************************************************
**                               SIM TIME : 1940301.22 NS                          **
**                              REAL TIME : 849.55 S                               **
**                        SIM / REAL TIME : 2283.91 NS/S                           **
*************************************************************************************

eric-wieser · 2019-06-25T21:45:11Z

Which version of python 2 and 3?

ydnatag · 2019-06-26T14:14:40Z

Python versions:

python 3.5.3
python 2.7.13

imphil · 2019-06-26T14:19:03Z

Could you possibly test with a newer Python 3 version, Python 3.7 being preferred? (Only to narrow down the problem a bit, we are still aiming for good performance in Python 3.5.)

cmarqu · 2019-06-26T15:45:54Z

There is also an environment variable COCOTB_ENABLE_PROFILING:
https://cocotb.readthedocs.io/en/latest/building.html?highlight=profile#additional-environment-variables

ydnatag · 2019-06-26T16:39:35Z

I supposed that i was using the same COCOTB but in one terminal i was inside a docker with an older cocotb. So, I rerun all the test with current and old commit and i got the following results:

Current master

Python 3.7.3

*************************************************************************************************************
** TEST                                                 PASS/FAIL  SIM TIME(NS)  REAL TIME(S)  RATIO(NS/S) **
*************************************************************************************************************
** test.xxxxxxxxxxxxxxxxxxxx                              PASS        61183.91          2.48     24695.05  **
** test.xxxxxxxxxxxxxxxxxx                                PASS        79206.25          6.52     12139.58  **
** test.xxxxxxxxxxxxxx                                    PASS          767.05          0.04     17737.50  **
** test.xxxxxxxxxxxxxxxxxxxxxxx                           PASS        64258.78          2.13     30143.39  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxx                         PASS       161103.87          5.47     29432.25  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       252351.29         92.17      2737.78  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       302076.66        109.01      2771.22  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx    PASS       947690.01        421.32      2249.33  **
** test.xxxxxxxxxx                                        PASS       160553.57          5.66     28375.40  **
** test.xxxxxxxxxxxxxx                                    PASS         5562.78          0.26     21150.07  **
*************************************************************************************************************
                                                                                                             
*************************************************************************************
**                                 ERRORS : 0                                      **
*************************************************************************************
**                               SIM TIME : 2034754.18 NS                          **
**                              REAL TIME : 645.18 S                               **
**                        SIM / REAL TIME : 3153.80 NS/S                           **
*************************************************************************************

Python 2.7.3

*************************************************************************************************************
** TEST                                                 PASS/FAIL  SIM TIME(NS)  REAL TIME(S)  RATIO(NS/S) **
*************************************************************************************************************
** test.xxxxxxxxxxxxxxxxxxxx                              PASS        61183.91          3.54     17282.70  **
** test.xxxxxxxxxxxxxxxxxx                                PASS        79206.25          8.68      9128.62  **
** test.xxxxxxxxxxxxxx                                    PASS          767.05          0.05     14443.75  **
** test.xxxxxxxxxxxxxxxxxxxxxxx                           PASS        64258.78          3.12     20622.67  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxx                         PASS       161103.87          7.78     20709.44  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       252351.29        117.08      2155.40  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       302076.66        147.67      2045.64  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx    PASS       600020.65        329.24      1822.44  **
** test.xxxxxxxxxx                                        PASS       160553.57          7.75     20720.70  **
** test.xxxxxxxxxxxxxx                                    PASS         5562.78          0.36     15554.01  **
*************************************************************************************************************
                                                                                                             
*************************************************************************************
**                                 ERRORS : 0                                      **
*************************************************************************************
**                               SIM TIME : 1687084.81 NS                          **
**                              REAL TIME : 625.35 S                               **
**                        SIM / REAL TIME : 2697.80 NS/S                           **
*************************************************************************************

Old Cocotb commit: `2744c4f`

Python 2.7

*************************************************************************************************************
** TEST                                                 PASS/FAIL  SIM TIME(NS)  REAL TIME(S)  RATIO(NS/S) **
*************************************************************************************************************
** test.xxxxxxxxxxxxxxxxxxxxxxxxx                          N/A            0.00          0.00         0.00  **
** test.xxxxxxxxxxxxxxxxxxxx                              PASS        61183.91          1.94     31557.20  **
** test.xxxxxxxxxxxxxxxxxx                                PASS        79206.25          5.59     14173.21  **
** test.xxxxxxxxxxxxxx                                    PASS          767.05          0.04     18289.67  **
** test.xxxxxxxxxxxxxxxxxxxxxxx                           PASS        64258.78          1.93     33259.93  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxx                         PASS       161103.87          4.79     33622.94  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       252351.29         58.97      4279.14  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       302076.66         74.35      4062.96  **
** test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx    PASS       600020.65        171.52      3498.33  **
** test.xxxxxxxxxx                                        PASS       160553.57          4.81     33392.15  **
** test.xxxxxxxxxxxxxx                                    PASS         5562.78          0.25     22497.06  **
*************************************************************************************************************
           
*************************************************************************************
**                                 ERRORS : 0                                      **
*************************************************************************************
**                               SIM TIME : 1687084.81 NS                          **
**                              REAL TIME : 324.24 S                               **
**                        SIM / REAL TIME : 5203.12 NS/S                           **
*************************************************************************************

Python 3.5.3:

* TEST                                                 PASS/FAIL  SIM TIME(NS)  REAL TIME(S)  RATIO(NS/S) **
************************************************************************************************************
* test.xxxxxxxxxxxxxxxxxxxxxxxxx                          N/A            0.00          0.00         0.00  **
* test.xxxxxxxxxxxxxxxxxxxx                              PASS        61183.91          1.95     31354.22  **
* test.xxxxxxxxxxxxxxxxxx                                PASS        79206.25          5.48     14465.62  **
* test.xxxxxxxxxxxxxx                                    PASS          767.05          0.04     17940.26  **
* test.xxxxxxxxxxxxxxxxxxxxxxx                           PASS        64258.78          1.96     32822.68  **
* test.xxxxxxxxxxxxxxxxxxxxxxxxx                         PASS       161103.87          4.88     33025.67  **
* test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       252351.29         60.02      4204.34  **
* test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx              PASS       302076.66         74.06      4078.62  **
* test.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx    PASS       773851.17        241.63      3202.65  **
* test.xxxxxxxxxx                                        PASS       160553.57          5.01     32063.15  **
* test.xxxxxxxxxxxxxx                                    PASS         5562.78          0.25     21906.17  **
************************************************************************************************************
                                                          
************************************************************************************
*                                 ERRORS : 0                                      **
************************************************************************************
*                               SIM TIME : 1860915.33 NS                          **
*                              REAL TIME : 395.37 S                               **
*                        SIM / REAL TIME : 4706.75 NS/S                           **
************************************************************************************

Summary

The performance doesn't depends on the python version. Cocotb suffered a loss of performance in the last year. Maibe it is expected, i don't know. I will do a the profiling and share the results.

imphil · 2019-06-26T17:06:07Z

The simulated time differs between your test runs, so things are not fully deterministic.

My observations

Looking at the "SIM / REAL TIME" metric:
In the newer commit, Python 3.7 is faster than Python 2.7.
In the older commit, Python 3.5 is slower than Python 2.7.
So this could suggest: "Python 3.5 is slower than Python 3.7, with Python 3.7 being faster than Python 2.7."
Independent of the Python version, cocotb is slower with today's master than it was with 2744c4f (December 2018).

Can you confirm the latter two observations with a more deterministic test case? The first item would then be a documentation issue (suggest to people to use a modern Python version), while the second observation would require more investigation. It would be very helpful to have a deterministic test case and bisect the cocotb history between 2744c4f and master to figure out in which commit (if it is a single one) the performance regression was introduced.

imphil · 2019-06-26T17:06:38Z

And no, the performance regression is not expected.

ydnatag · 2019-06-26T18:00:32Z

Just one testcase has randomization (the testcase with the longest name), the others are deterministic. I can confirm both observations. These results are similar in all our tests (we have around 50 cores that are currently tested with cocotb).

Currently i'm looking for a shorter test case without confidential information for sharing the profiling.

eric-wieser · 2019-06-26T18:04:28Z

As a hunch, #870 might be a good starting point to test before and after

ydnatag · 2019-06-26T18:24:25Z

Using python3.7

Current master

commit: 714e129

SIM TIME : 270162.40 NS 
REAL TIME : 16.99 S      
SIM / REAL TIME : 15900.06 NS/S

After `Fix scheduler issues caused by compound triggers`

commit: a354327

SIM TIME : 270162.40 NS 
REAL TIME : 14.08 S      
SIM / REAL TIME : 19192.90 NS/S

Before

commit: 5ac9bfb

SIM TIME : 270162.40 NS 
REAL TIME : 13.78 S      
SIM / REAL TIME : 19599.73 NS/S

My freezed version:

commit: 2744c4f

SIM TIME : 270162.40 NS 
REAL TIME : 8.52 S       
SIM / REAL TIME : 31705.06 NS/S

Observation

#870 doesn't affect the performance considerably.

ydnatag · 2019-06-26T18:46:23Z

I'm searching the guilty commits with brute force: An script which executes and grep the performance for each commit.

imphil · 2019-06-26T20:08:48Z

For the difference between the Python versions I've opened #965 to track the further investigation and implementation, leaving this bug to actually figure out the performance regression within cocotb.

ydnatag · 2019-06-26T20:15:40Z

Here the results:

CSV file with the results:
performance.txt

The outlier near 2019-01-04 was my fault. I slept the process.

As is shown in the image, the performance has a negative derivative along the time and 2/3 commits with higher performance loss:

f692e8d 2019-01-12
a354327 2019-04-21
07d28be 2019-04-28

Of course, those results are relative to the test but i think they are a good estimation.

eric-wieser · 2019-06-26T21:03:05Z

What's the y axis on that graph?

ydnatag · 2019-06-27T00:32:55Z

The Y axis is SIM / REAL TIME measument. I call it "performance".

For having a better performane estimator, It should be normalized to the max SIM / REAL TIME. But for this investigation i think it is enough.

eric-wieser · 2019-06-27T02:25:01Z

It would be good to confirm that the sim time is the same across all commits, in case the cause of the increased runtime is something like "we fixed a bug where the test ended silently too soon"

ydnatag · 2019-06-27T02:30:06Z

I checked that simtime is the same. . If you want i can share the scripts to recreate these results. The scripts are so ugly but useful.

eric-wieser · 2019-06-27T06:57:09Z

So, I've found one obvious performance cost already - a logger is created for every single coroutine, which for a very simple test-case of:

import cocotb
from cocotb.triggers import Timer, RisingEdge, First
from cocotb.clock import Clock

@cocotb.test()
def test1(dut):
	clk_gen = cocotb.fork(Clock(dut.clk, 100).start())
	for i in range(10000):
		yield [RisingEdge(dut.clk), Timer(200)]

Seems to spend 12% of it's time calling logging.getLogger

Some of the scheduler refactors replaced some special-casing with more coroutines. What I didn't realize is that the each mini coroutine creates and destroys a full logger instance, which is not cheap at all.

eric-wieser · 2019-06-27T07:51:52Z

Patch for that at #967. @andresdemski, please let me know how much that helps your more realistic suite.

ydnatag · 2019-06-27T12:20:55Z

I reprocess all the commits but this time i recorded simtime.
performance.txt

As is shown in the picture, simtime is constant

ydnatag · 2019-06-27T12:33:52Z

I tried #967 . The performance is improved a 10% but still far from the best performance

Improve performance by avoiding calling logging.getLogger unless something is actually being logged #967 performance: 19937.99 NS/S
Current master: 17860.06 NS/S
Best performance: ~35000 NS/S

ydnatag · 2019-06-27T13:38:05Z

Here the profiling:

Old cocotb version

Current master

eric-wieser · 2019-06-27T15:42:53Z

What's the tool you used to produce those diagrams?

ydnatag · 2019-06-27T15:51:11Z

The same command that it is in endian_swapper example

$ make COCOTB_ENABLE_PROFILING=true
$ python3 -m gprof2dot -f pstats test_profile.pstat | dot -Tsvg -o profile.svg

eric-wieser · 2019-06-27T15:55:18Z

Any chance you could upload the svg version?

From the investigation at cocotbgh-961, this is shown to be 3% of the runtime on master. The version of python does not change at runtime, so we can replace this function with a simple constant. Doing so also brings our name in line with six's.

imphil · 2019-06-28T14:51:50Z

Let's please keep the discussion in this issue focused on understanding and fixing the performance issues.

ydnatag · 2019-07-03T00:14:05Z

I made a simple tool which allow us to do performance analysis in an easy way: https://github.com/andresdemski/cocotb-performance

Terminal output:

$ ./run_in_docker.sh python3 compare.py                                                                                                                                                                     
sha256:7a4d132d15ca92f95b7f75b637c2c95b277bf7a3609918815f276bbe9fb8c5e4

 url https://github.com/potentialventures/cocotb/
 commit: master
 path: /tmp/cocotb-master
 results: 
    simtime: 142610.10 ns
    realtime: 18.65s
    performance: 7648.49 ns/s


 url https://github.com/eric-wieser/cocotb/
 commit: avoid-runtime-coroutine-creation
 path: /tmp/cocotb-eric-wieser
 results: 
    simtime: 142610.10 ns
    realtime: 18.20s
    performance: 7833.63 ns/s


 url https://github.com/potentialventures/cocotb/
 commit: 2744c4fec607bf598d677484e6c0290183360f7a
 path: /tmp/cocotb-master
 results: 
    simtime: 142610.10 ns
    realtime: 15.53s
    performance: 9182.79 ns/s

imphil · 2019-07-03T16:56:13Z

@andresdemski now that all outstanding performance PRs have been merged, how close to 1.1 performance are we with current master? I'd like to use this information to make a judgement call if we can actually release master right now, or if we need to do further performance work. (A small performance regression is fine, a huge regression would be a bad thing.)

eric-wieser · 2019-07-03T17:19:35Z

If #764 (f692e8d) is really significant, it would be fairly straightforward to revert - most of the value in it was to provide the simplicity needed to refactor. The refactor has now happened, and the three profiler.disable()s that touched became one. I suppose the cost of nullcontext() might be further from 0 than I hoped.

ydnatag · 2019-07-03T18:14:42Z

our current situation.

eric-wieser · 2019-07-03T18:25:31Z

Can you replot that graph with an origin of 0?

ydnatag · 2019-07-03T18:28:57Z

performance_2019_07_03.txt

imphil · 2019-07-03T19:45:46Z

So we're roughly 30 % slower than we were in the 1.1 release.

eric-wieser · 2019-07-03T20:23:44Z

Phrased another way, (that) simulation will take 1 / (1 - 0.3) = 1.4 times the time it did in 1.1.

At the point this issue was opened, the simulation took 2.0 times longer

eric-wieser · 2019-07-04T05:16:13Z

I investigated some more - the graph is misleading, as it doesn't handle branches well. Here's a better plot, that draws the branches separately:

The labels are linked to the commits on github.

Edit: updated to use the new data

imphil · 2019-07-05T08:45:21Z

Thanks @andresdemski for all the analysis runs, and @eric-wieser for the great work on performance patches! Are there any more patches upcoming in the next couple of days which target performance? (Otherwise I'm going to release 1.2 with what we have now, and we can build on top of that without time pressure.)

eric-wieser · 2019-07-11T06:09:04Z

@imphil: Came up with another one at #1012.

Thinking back, I think #727 (and perhaps some of #723) might have been a performance mistake, and the cost of the weak dictionaries may not justify the minor gain of avoiding reference cycles. I don't really want to look at reverting those just yet.

In the long term though, there's a lot of efficiency to be gained by eliminating internal uses of Join triggers where using await / yield from directly is possible.

cmarqu · 2019-07-15T22:54:16Z

With these real-life testcases and current performance numbers, it may be worth re-investigating gcc options like in #178 (this would be set in
https://github.com/potentialventures/cocotb/blob/master/cocotb/share/lib/simulator/Makefile#L33 I suppose).

ydnatag · 2019-11-27T03:16:10Z

Update 11/26/19

No considerable changes since July.

Profile

profile_11_26_19.zip

github-actions · 2020-11-17T02:15:48Z

Has your question been resolved? If so please close this issue. If it has not been resolved, you may need to provide more information. If no more activity on this issue occurs in 7 days, it will be closed.

themperek · 2020-12-06T12:29:30Z

One can also use pprofile to generate reports.

Example for matrix_multiplier commit: f8c0e40

pprofile_matrix_multiplier_f8c0e40.log

themperek · 2020-12-08T08:59:17Z

Maybe something more like a real case using https://github.com/alexforencich/verilog-axi/tree/master/tb/axil_ram
pprofile_axi_ram_f8c0e40.log

imphil added the type:question a support question label Jun 25, 2019

imphil mentioned this issue Jun 26, 2019

Recommend using latest Python 3 for performance reasons #965

Closed

imphil mentioned this issue Jun 27, 2019

Improve performance by avoiding calling logging.getLogger unless something is actually being logged #967

Merged

eric-wieser mentioned this issue Jun 29, 2019

Extract closure functions to a private method, so that cocotb.coroutine.__new__ is not invoked each time #980

Merged

eric-wieser mentioned this issue Jul 4, 2019

Dictionary-related optimizations #996

Merged

ydnatag closed this as completed Jul 4, 2019

ydnatag reopened this Jul 4, 2019

eric-wieser mentioned this issue Jul 11, 2019

Use slots in triggers for speed and memory savings #1012

Merged

cmarqu mentioned this issue Jul 23, 2019

Performance regression test #126

Closed

themperek mentioned this issue Oct 16, 2019

Possible Optimisations - cache get_sim_time for performance #171

Closed

cmarqu changed the title ~~Simulation performace~~ Simulation performance May 1, 2020

cmarqu added the category:performance performance topics label May 1, 2020

github-actions bot added the status:stale label Nov 17, 2020

ktbarrett removed status:stale type:question a support question labels Nov 17, 2020

marlonjames mentioned this issue Dec 15, 2020

Scheduler._task_context performance #2291

Closed

Simulation performance #961

Simulation performance #961

Comments

ydnatag commented Jun 25, 2019 • edited

ydnatag commented Jun 25, 2019

Python2

Python3

eric-wieser commented Jun 25, 2019

ydnatag commented Jun 26, 2019

imphil commented Jun 26, 2019

cmarqu commented Jun 26, 2019

ydnatag commented Jun 26, 2019

Current master

Python 3.7.3

Python 2.7.3

Old Cocotb commit: 2744c4f

Python 2.7

Python 3.5.3:

Summary

imphil commented Jun 26, 2019

imphil commented Jun 26, 2019

ydnatag commented Jun 26, 2019

eric-wieser commented Jun 26, 2019

ydnatag commented Jun 26, 2019

Current master

After Fix scheduler issues caused by compound triggers

Before

My freezed version:

Observation

ydnatag commented Jun 26, 2019 • edited

imphil commented Jun 26, 2019

ydnatag commented Jun 26, 2019 • edited

eric-wieser commented Jun 26, 2019

ydnatag commented Jun 27, 2019

eric-wieser commented Jun 27, 2019 • edited

ydnatag commented Jun 27, 2019

eric-wieser commented Jun 27, 2019 • edited

eric-wieser commented Jun 27, 2019

ydnatag commented Jun 27, 2019 • edited

ydnatag commented Jun 27, 2019

ydnatag commented Jun 27, 2019

Old cocotb version

Current master

eric-wieser commented Jun 27, 2019

ydnatag commented Jun 27, 2019 • edited

eric-wieser commented Jun 27, 2019

imphil commented Jun 28, 2019

ydnatag commented Jul 3, 2019

imphil commented Jul 3, 2019

eric-wieser commented Jul 3, 2019

ydnatag commented Jul 3, 2019

eric-wieser commented Jul 3, 2019

ydnatag commented Jul 3, 2019 • edited

imphil commented Jul 3, 2019

eric-wieser commented Jul 3, 2019 • edited

eric-wieser commented Jul 4, 2019 • edited

imphil commented Jul 5, 2019

eric-wieser commented Jul 11, 2019

cmarqu commented Jul 15, 2019

ydnatag commented Nov 27, 2019

Update 11/26/19

Profile

github-actions bot commented Nov 17, 2020

themperek commented Dec 6, 2020 • edited

themperek commented Dec 8, 2020

ydnatag commented Jun 25, 2019 •

edited

Old Cocotb commit: `2744c4f`

After `Fix scheduler issues caused by compound triggers`

ydnatag commented Jun 26, 2019 •

edited

ydnatag commented Jun 26, 2019 •

edited

eric-wieser commented Jun 27, 2019 •

edited

eric-wieser commented Jun 27, 2019 •

edited

ydnatag commented Jun 27, 2019 •

edited

ydnatag commented Jun 27, 2019 •

edited

ydnatag commented Jul 3, 2019 •

edited

eric-wieser commented Jul 3, 2019 •

edited

eric-wieser commented Jul 4, 2019 •

edited

themperek commented Dec 6, 2020 •

edited