| 59b032bc7f837293068a2ee0469274c6dfede801 |
|
06-Jun-2015 |
Jarkko Pöyry <jpoyry@google.com> |
Add simulated error before rounding to target precision in dfdx tests.
- Add interpolation error before rounding in GLES3 Derivate tests. If the
error is not added before rounding and reference value happens to be 1:1
representable in the target precision, both rounding directions yield the
same value and error bounds will be lost.
Bug: 21666909
Change-Id: Ie948ea32075439c962d8ca52362069ca030a172e
/external/deqp/modules/gles3/functional/es3fShaderDerivateTests.cpp
|
| d9551dc8ecff19b0bfe50e0a463c49e99997b4a2 |
|
12-Mar-2015 |
Jarkko Pöyry <jpoyry@google.com> |
Fix assertion failure in dfdx and dfdy lost bit calculations.
- Fix lost-bits calculation returning negative values when operand
component range intercepted zero (i.e. operands had different signs).
This caused verification to expect more significant bits than the
native (lowp/mediump/highp) float format could store. With highp
values, error calculation tried to create a error compontent of a
floating point value with 24 mantissa bits which cannot be stored
to a native float32.
- Use better terminology, s/nominator/numerator/.
Bug: 19694789
Change-Id: Ifa88fb1c8cb333bab22a550d01f5e310d2d2673d
/external/deqp/modules/gles3/functional/es3fShaderDerivateTests.cpp
|
| 7d6ad9e003b52cfc23626b216b7580babf423c8f |
|
03-Feb-2015 |
Pyry Haulos <phaulos@google.com> |
Fix couple of implicit conversion warnings reported by MSVC
Change-Id: If6dd153b8927e4f8e0cc5629d4dd051a9a96dcdf
/external/deqp/modules/gles3/functional/es3fShaderDerivateTests.cpp
|
| bd4d098088a5b205c7e7d8dc0bec140d2fa55446 |
|
04-Dec-2014 |
Jarkko Pöyry <jpoyry@google.com> |
Relax shader derivative test verification.
- Disable GL_DITHER. Dithering only adds noise to result images.
- Add to dfdx and dfdy cases a flush-to-zero aware verification phase after
preliminary verification to check if seemingly invalid derivatives could
be caused by flushing or other relaxations allowed by the spec. This phase
is NOT added to the fwidth cases since they operate with significantly higher
thresholds which are likely to hide flushing anyway.
Bug: 18539021
Change-Id: Ic17909d30de25c8bae9d9996ce8df15c9ad5bc5e
/external/deqp/modules/gles3/functional/es3fShaderDerivateTests.cpp
|
| 3c827367444ee418f129b2c238299f49d3264554 |
|
02-Sep-2014 |
Jarkko Poyry <jpoyry@google.com> |
Import dEQP.
Import drawElements Quality Program from an internal repository.
Bug: 17388917
Change-Id: Ic109fe4a57e31b2a816113d90fbdf51a43e7abeb
/external/deqp/modules/gles3/functional/es3fShaderDerivateTests.cpp
|