Add more NV12 and YUYV tests
Added clip tests for NV12 and YUYV.
Added reference vectors for NV12 and YUYV tests
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Added clip tests for NV12 and YUYV.
Added reference vectors for NV12 and YUYV tests
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[Linux] Build update
Adding more paths as needed tby Yocto recipes
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Adding more paths as needed tby Yocto recipes
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[Linux] Removed auto detection on serch path
Yocto builds failed because it was not picking the right serach paths
using pkg-config.
This patch revert ths usage of pkg-config binary.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Yocto builds failed because it was not picking the right serach paths
using pkg-config.
This patch revert ths usage of pkg-config binary.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[Linux] Add DMA buf functionality
This patch adds DMA buf functionality to the existing bvtest
application.
Surfaces are now allocated using OMAPDRM which is an exporter
of DMA buf. The changes have been added under ENABLE_OMAPDRM
compile flag that is enabled in makefile.linux
usage: make -C src/bvtest -f makefile.linux
or
cd src/bvtest
./autogen.sh --host=arm-linux-gnueabihf --prefix=<install-path>
make
make install
A binary 'bvtest' will be created in src/bvtest directory which
can be used on target.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[Linux] Use DMA buf for buffer allocation
This patch adds dma buf support in application.
The buffers will be allocated from OMAPDRM and dma bufs are exported.
This patch should go along with kernel patch
http://review.omapzoom.org/#/c/34441/
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
This patch adds DMA buf functionality to the existing bvtest
application.
Surfaces are now allocated using OMAPDRM which is an exporter
of DMA buf. The changes have been added under ENABLE_OMAPDRM
compile flag that is enabled in makefile.linux
usage: make -C src/bvtest -f makefile.linux
or
cd src/bvtest
./autogen.sh --host=arm-linux-gnueabihf --prefix=<install-path>
make
make install
A binary 'bvtest' will be created in src/bvtest directory which
can be used on target.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[Linux] Use DMA buf for buffer allocation
This patch adds dma buf support in application.
The buffers will be allocated from OMAPDRM and dma bufs are exported.
This patch should go along with kernel patch
http://review.omapzoom.org/#/c/34441/
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[test] Add NV12 and YUYV inputs to repo
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Build setup
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[TEST] Reference vectors for global alpha updated
The existing reference test vectors for global aplha mismatched
with the reference simulator outputs and hence were considered
to be stale. Hence updating the global alpha reference vectors
with the current outputs to be considered as reference for
new set of changes.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
The existing reference test vectors for global aplha mismatched
with the reference simulator outputs and hence were considered
to be stale. Hence updating the global alpha reference vectors
with the current outputs to be considered as reference for
new set of changes.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
[PERF] Performance Measurements changes
In order to get good estimate of the performance of
hw2d, the IP needs to be up and running for a while.
If the measurements are done as soon as a first
iteration of the test case is run, the measured values
would be low. However, if the same test case is run for
about 500 times, the measured performance figures are comparable
with the data sheet values.
This is suspected to be due to high ramp up, power on and
initialization sequence during GC320 boot up.
Suitable changes are made in scripts and application
to run the test case for sufficient number of times before
estimating performance.
A new configuration is added to bvtest in order to facilitate
running multiple iterations.
eg: bvtest -loops 500 <other test parameters>
would run the test for 500 times.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
In order to get good estimate of the performance of
hw2d, the IP needs to be up and running for a while.
If the measurements are done as soon as a first
iteration of the test case is run, the measured values
would be low. However, if the same test case is run for
about 500 times, the measured performance figures are comparable
with the data sheet values.
This is suspected to be due to high ramp up, power on and
initialization sequence during GC320 boot up.
Suitable changes are made in scripts and application
to run the test case for sufficient number of times before
estimating performance.
A new configuration is added to bvtest in order to facilitate
running multiple iterations.
eg: bvtest -loops 500 <other test parameters>
would run the test for 500 times.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Disabled surface dump by default
This patch Defaults to disabled prints to improve efficiency
White space errors are removed.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
This patch Defaults to disabled prints to improve efficiency
White space errors are removed.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Update makefile and README
Patch adapts bvtest linux build and
Updates README with build procedure for
linux and android.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Patch adapts bvtest linux build and
Updates README with build procedure for
linux and android.
Signed-off-by: Pradeep Venkatasubbarao <pradeepv@ti.com>
Initial commit: Added bvtest to empty repository
Signed-off-by: Sundar Raman <sunds@ti.com>
Signed-off-by: Sundar Raman <sunds@ti.com>