[cogl/wip/cogl-1.14: 65/177] cogl: Enable PBO path for all mesa versions when using intel
- From: Robert Bragg <rbragg src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [cogl/wip/cogl-1.14: 65/177] cogl: Enable PBO path for all mesa versions when using intel
- Date: Mon, 21 Jan 2013 15:50:24 +0000 (UTC)
commit 3a0e57fd9344f061b60ba4ba1168377c5b92611d
Author: Adel Gadllah <adel gadllah gmail com>
Date: Thu Oct 11 17:59:55 2012 +0200
cogl: Enable PBO path for all mesa versions when using intel
There seem to be other cases where the slow path is hit even with mesa 9.0.
https://bugzilla.gnome.org/show_bug.cgi?id=685915
(cherry picked from commit 6b14fa222bafb7e67c4845b26439322b34ac26a2)
cogl/cogl-gpu-info.c | 8 +++-----
1 files changed, 3 insertions(+), 5 deletions(-)
---
diff --git a/cogl/cogl-gpu-info.c b/cogl/cogl-gpu-info.c
index 5bb0e65..0d23c97 100644
--- a/cogl/cogl-gpu-info.c
+++ b/cogl/cogl-gpu-info.c
@@ -540,14 +540,12 @@ probed:
/* Determine the driver bugs */
- /* In Mesa < 8.0.2 the glReadPixels implementation is really slow
- because it converts each pixel to a floating point representation
- and back even if the data could just be memcpy'd. The Intel
+ /* In Mesa the glReadPixels implementation is really slow
+ when using the Intel driver. The Intel
driver has a fast blit path when reading into a PBO. Reading into
a temporary PBO and then memcpying back out to the application's
memory is faster than a regular glReadPixels in this case */
if (gpu->vendor == COGL_GPU_INFO_VENDOR_INTEL &&
- gpu->driver_package == COGL_GPU_INFO_DRIVER_PACKAGE_MESA &&
- gpu->driver_package_version < COGL_VERSION_ENCODE (8, 0, 2))
+ gpu->driver_package == COGL_GPU_INFO_DRIVER_PACKAGE_MESA)
gpu->driver_bugs |= COGL_GPU_INFO_DRIVER_BUG_MESA_46631_SLOW_READ_PIXELS;
}
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]