DB
David Banks
Thu, Jan 17, 2019 5:00 PM
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
MemoryError
- Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
David Banks
dbanks@geontech.com
Geon Technologies, LLC
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
*MemoryError*
* Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
--
David Banks
dbanks@geontech.com
Geon Technologies, LLC
MA
Marasco, Aaron
Fri, Jan 18, 2019 11:59 AM
David,
Thanks for the feedback. Looking it over, I can see some places where it can be improved because it seems to slurp up the whole results file without iterating over it with something like numpy.memmap. At this time, we don't have the resources to dig into it, but I've assigned it bug #AV-5098.
Thanks,
Aaron
From: discuss discuss-bounces@lists.opencpi.org on behalf of David Banks dbanks@geontech.com
Sent: Thursday, January 17, 2019 12:00:47 PM
To: discuss@lists.opencpi.org
Subject: [Discuss OpenCPI] data_src.test verification using up 24GB RAM and failing
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
MemoryError
- Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
David Banks
dbanks@geontech.com
Geon Technologies, LLC
David,
Thanks for the feedback. Looking it over, I can see some places where it can be improved because it seems to slurp up the whole results file without iterating over it with something like numpy.memmap. At this time, we don't have the resources to dig into it, but I've assigned it bug #AV-5098.
Thanks,
Aaron
________________________________
From: discuss <discuss-bounces@lists.opencpi.org> on behalf of David Banks <dbanks@geontech.com>
Sent: Thursday, January 17, 2019 12:00:47 PM
To: discuss@lists.opencpi.org
Subject: [Discuss OpenCPI] data_src.test verification using up 24GB RAM and failing
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
*MemoryError*
* Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
--
David Banks
dbanks@geontech.com
Geon Technologies, LLC
DB
David Banks
Fri, Jan 18, 2019 1:47 PM
No problem at all. I just want to make sure someone is aware of this
bug/limitation of that unit test. The only other problem I have experienced
with OpenCPI Component Unit Tests is that zero_pad.test case00.01 will hang
indefinitely during execution on hardware. Has anyone seen this behavior
before?
Thanks,
David
On Fri, Jan 18, 2019 at 6:59 AM Marasco, Aaron aaron.marasco@bia-boeing.com
wrote:
David,
Thanks for the feedback. Looking it over, I can see some places where it
can be improved because it seems to slurp up the whole results file without
iterating over it with something like numpy.memmap. At this time, we don't
have the resources to dig into it, but I've assigned it bug #AV-5098.
Thanks,
Aaron
From: discuss discuss-bounces@lists.opencpi.org on behalf of David
Banks dbanks@geontech.com
Sent: Thursday, January 17, 2019 12:00:47 PM
To: discuss@lists.opencpi.org
Subject: [Discuss OpenCPI] data_src.test verification using up 24GB RAM
and failing
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
MemoryError
- Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
David Banks
dbanks@geontech.com
Geon Technologies, LLC
No problem at all. I just want to make sure someone is aware of this
bug/limitation of that unit test. The only other problem I have experienced
with OpenCPI Component Unit Tests is that zero_pad.test case00.01 will hang
indefinitely during execution on hardware. Has anyone seen this behavior
before?
Thanks,
David
On Fri, Jan 18, 2019 at 6:59 AM Marasco, Aaron <aaron.marasco@bia-boeing.com>
wrote:
> David,
>
>
> Thanks for the feedback. Looking it over, I can see some places where it
> can be improved because it seems to slurp up the whole results file without
> iterating over it with something like numpy.memmap. At this time, we don't
> have the resources to dig into it, but I've assigned it bug #AV-5098.
>
>
> Thanks,
>
>
> Aaron
>
>
> ------------------------------
> *From:* discuss <discuss-bounces@lists.opencpi.org> on behalf of David
> Banks <dbanks@geontech.com>
> *Sent:* Thursday, January 17, 2019 12:00:47 PM
> *To:* discuss@lists.opencpi.org
> *Subject:* [Discuss OpenCPI] data_src.test verification using up 24GB RAM
> and failing
>
> On my system with 24GB of RAM, data_src.test (from release_1.4)
> verification fails for case00.02 with the following python error:
>
> Executing case case00.02 using worker data_src.hdl...
> Execution succeeded, time was 00:01:07 (67 seconds).
> Verifying case case00.02 for worker data_src.hdl using script on output
> file: case00.02.data_src.hdl.out.out
> PASS: output file length >= 1 samples for 'num_samples' property value
> = -1
> Traceback (most recent call last):
> File "../../verify.py", line 187, in <module>
> main()
> File "../../verify.py", line 184, in main
> test_odata_file_binary_contents_are_as_expected()
> File "../../verify.py", line 115, in
> test_odata_file_binary_contents_are_as_expected
> data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
> *MemoryError*
> * Verification for port out: FAILED*
> make: *** [run] Error 1
>
> The first two cases succeed and pass verification. I monitored my system
> resources while it was running case00.02 and saw that the python3 process
> used up ~100% of my system's 24GB of memory.
> --
> David Banks
> dbanks@geontech.com
> Geon Technologies, LLC
>
BL
Brandon Luquette
Fri, Jan 18, 2019 2:09 PM
Hi David,
Multiple issues with this worker were identified after the 1.4 release. While they didn't not manifest themselves in the failures you are describing, it could be worth trying the fixes. The fixes can be found in the develop branch.
Thanks,
Brandon
From: discuss discuss-bounces@lists.opencpi.org on behalf of David Banks dbanks@geontech.com
Sent: Friday, January 18, 2019 8:47:20 AM
To: discuss@lists.opencpi.org
Subject: Re: [Discuss OpenCPI] data_src.test verification using up 24GB RAM and failing
No problem at all. I just want to make sure someone is aware of this
bug/limitation of that unit test. The only other problem I have experienced
with OpenCPI Component Unit Tests is that zero_pad.test case00.01 will hang
indefinitely during execution on hardware. Has anyone seen this behavior
before?
Thanks,
David
On Fri, Jan 18, 2019 at 6:59 AM Marasco, Aaron aaron.marasco@bia-boeing.com
wrote:
David,
Thanks for the feedback. Looking it over, I can see some places where it
can be improved because it seems to slurp up the whole results file without
iterating over it with something like numpy.memmap. At this time, we don't
have the resources to dig into it, but I've assigned it bug #AV-5098.
Thanks,
Aaron
From: discuss discuss-bounces@lists.opencpi.org on behalf of David
Banks dbanks@geontech.com
Sent: Thursday, January 17, 2019 12:00:47 PM
To: discuss@lists.opencpi.org
Subject: [Discuss OpenCPI] data_src.test verification using up 24GB RAM
and failing
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
MemoryError
- Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
David Banks
dbanks@geontech.com
Geon Technologies, LLC
Hi David,
Multiple issues with this worker were identified after the 1.4 release. While they didn't not manifest themselves in the failures you are describing, it could be worth trying the fixes. The fixes can be found in the develop branch.
Thanks,
Brandon
________________________________
From: discuss <discuss-bounces@lists.opencpi.org> on behalf of David Banks <dbanks@geontech.com>
Sent: Friday, January 18, 2019 8:47:20 AM
To: discuss@lists.opencpi.org
Subject: Re: [Discuss OpenCPI] data_src.test verification using up 24GB RAM and failing
No problem at all. I just want to make sure someone is aware of this
bug/limitation of that unit test. The only other problem I have experienced
with OpenCPI Component Unit Tests is that zero_pad.test case00.01 will hang
indefinitely during execution on hardware. Has anyone seen this behavior
before?
Thanks,
David
On Fri, Jan 18, 2019 at 6:59 AM Marasco, Aaron <aaron.marasco@bia-boeing.com>
wrote:
> David,
>
>
> Thanks for the feedback. Looking it over, I can see some places where it
> can be improved because it seems to slurp up the whole results file without
> iterating over it with something like numpy.memmap. At this time, we don't
> have the resources to dig into it, but I've assigned it bug #AV-5098.
>
>
> Thanks,
>
>
> Aaron
>
>
> ------------------------------
> *From:* discuss <discuss-bounces@lists.opencpi.org> on behalf of David
> Banks <dbanks@geontech.com>
> *Sent:* Thursday, January 17, 2019 12:00:47 PM
> *To:* discuss@lists.opencpi.org
> *Subject:* [Discuss OpenCPI] data_src.test verification using up 24GB RAM
> and failing
>
> On my system with 24GB of RAM, data_src.test (from release_1.4)
> verification fails for case00.02 with the following python error:
>
> Executing case case00.02 using worker data_src.hdl...
> Execution succeeded, time was 00:01:07 (67 seconds).
> Verifying case case00.02 for worker data_src.hdl using script on output
> file: case00.02.data_src.hdl.out.out
> PASS: output file length >= 1 samples for 'num_samples' property value
> = -1
> Traceback (most recent call last):
> File "../../verify.py", line 187, in <module>
> main()
> File "../../verify.py", line 184, in main
> test_odata_file_binary_contents_are_as_expected()
> File "../../verify.py", line 115, in
> test_odata_file_binary_contents_are_as_expected
> data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
> *MemoryError*
> * Verification for port out: FAILED*
> make: *** [run] Error 1
>
> The first two cases succeed and pass verification. I monitored my system
> resources while it was running case00.02 and saw that the python3 process
> used up ~100% of my system's 24GB of memory.
> --
> David Banks
> dbanks@geontech.com
> Geon Technologies, LLC
>
DB
David Banks
Fri, Jan 18, 2019 2:14 PM
Hi David,
Multiple issues with this worker were identified after the 1.4 release.
While they didn't not manifest themselves in the failures you are
describing, it could be worth trying the fixes. The fixes can be found in
the develop branch.
Thanks,
Brandon
From: discuss discuss-bounces@lists.opencpi.org on behalf of David
Banks dbanks@geontech.com
Sent: Friday, January 18, 2019 8:47:20 AM
To: discuss@lists.opencpi.org
Subject: Re: [Discuss OpenCPI] data_src.test verification using up 24GB
RAM and failing
No problem at all. I just want to make sure someone is aware of this
bug/limitation of that unit test. The only other problem I have experienced
with OpenCPI Component Unit Tests is that zero_pad.test case00.01 will hang
indefinitely during execution on hardware. Has anyone seen this behavior
before?
Thanks,
David
On Fri, Jan 18, 2019 at 6:59 AM Marasco, Aaron <
aaron.marasco@bia-boeing.com>
wrote:
David,
Thanks for the feedback. Looking it over, I can see some places where
it
can be improved because it seems to slurp up the whole results file
without
iterating over it with something like numpy.memmap. At this time, we
don't
have the resources to dig into it, but I've assigned it bug #AV-5098.
Thanks,
Aaron
From: discuss discuss-bounces@lists.opencpi.org on behalf of David
Banks dbanks@geontech.com
Sent: Thursday, January 17, 2019 12:00:47 PM
To: discuss@lists.opencpi.org
Subject: [Discuss OpenCPI] data_src.test verification using up 24GB RAM
and failing
On my system with 24GB of RAM, data_src.test (from release_1.4)
verification fails for case00.02 with the following python error:
Executing case case00.02 using worker data_src.hdl...
Execution succeeded, time was 00:01:07 (67 seconds).
Verifying case case00.02 for worker data_src.hdl using script on output
file: case00.02.data_src.hdl.out.out
PASS: output file length >= 1 samples for 'num_samples' property
value
= -1
Traceback (most recent call last):
File "../../verify.py", line 187, in <module>
main()
File "../../verify.py", line 184, in main
test_odata_file_binary_contents_are_as_expected()
File "../../verify.py", line 115, in
test_odata_file_binary_contents_are_as_expected
data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
MemoryError
- Verification for port out: FAILED*
make: *** [run] Error 1
The first two cases succeed and pass verification. I monitored my system
resources while it was running case00.02 and saw that the python3 process
used up ~100% of my system's 24GB of memory.
David Banks
dbanks@geontech.com
Geon Technologies, LLC
Thank you Brandon! I will go ahead and try the fixes in develop.
-David
On Fri, Jan 18, 2019 at 9:09 AM Brandon Luquette <
bluquette@geontech.onmicrosoft.com> wrote:
> Hi David,
>
> Multiple issues with this worker were identified after the 1.4 release.
> While they didn't not manifest themselves in the failures you are
> describing, it could be worth trying the fixes. The fixes can be found in
> the develop branch.
>
> Thanks,
>
> Brandon
> ------------------------------
> *From:* discuss <discuss-bounces@lists.opencpi.org> on behalf of David
> Banks <dbanks@geontech.com>
> *Sent:* Friday, January 18, 2019 8:47:20 AM
> *To:* discuss@lists.opencpi.org
> *Subject:* Re: [Discuss OpenCPI] data_src.test verification using up 24GB
> RAM and failing
>
> No problem at all. I just want to make sure someone is aware of this
> bug/limitation of that unit test. The only other problem I have experienced
> with OpenCPI Component Unit Tests is that zero_pad.test case00.01 will hang
> indefinitely during execution on hardware. Has anyone seen this behavior
> before?
>
> Thanks,
> David
>
>
> On Fri, Jan 18, 2019 at 6:59 AM Marasco, Aaron <
> aaron.marasco@bia-boeing.com>
> wrote:
>
> > David,
> >
> >
> > Thanks for the feedback. Looking it over, I can see some places where
> it
> > can be improved because it seems to slurp up the whole results file
> without
> > iterating over it with something like numpy.memmap. At this time, we
> don't
> > have the resources to dig into it, but I've assigned it bug #AV-5098.
> >
> >
> > Thanks,
> >
> >
> > Aaron
> >
> >
> > ------------------------------
> > *From:* discuss <discuss-bounces@lists.opencpi.org> on behalf of David
> > Banks <dbanks@geontech.com>
> > *Sent:* Thursday, January 17, 2019 12:00:47 PM
> > *To:* discuss@lists.opencpi.org
> > *Subject:* [Discuss OpenCPI] data_src.test verification using up 24GB RAM
> > and failing
> >
> > On my system with 24GB of RAM, data_src.test (from release_1.4)
> > verification fails for case00.02 with the following python error:
> >
> > Executing case case00.02 using worker data_src.hdl...
> > Execution succeeded, time was 00:01:07 (67 seconds).
> > Verifying case case00.02 for worker data_src.hdl using script on output
> > file: case00.02.data_src.hdl.out.out
> > PASS: output file length >= 1 samples for 'num_samples' property
> value
> > = -1
> > Traceback (most recent call last):
> > File "../../verify.py", line 187, in <module>
> > main()
> > File "../../verify.py", line 184, in main
> > test_odata_file_binary_contents_are_as_expected()
> > File "../../verify.py", line 115, in
> > test_odata_file_binary_contents_are_as_expected
> > data = np.remainder(np.arange(len(odata)), 2**DATA_BIT_WIDTH_p)
> > *MemoryError*
> > * Verification for port out: FAILED*
> > make: *** [run] Error 1
> >
> > The first two cases succeed and pass verification. I monitored my system
> > resources while it was running case00.02 and saw that the python3 process
> > used up ~100% of my system's 24GB of memory.
> > --
> > David Banks
> > dbanks@geontech.com
> > Geon Technologies, LLC
> >