Closed Bug 599351 Opened 15 years ago Closed 13 years ago

Rotation in Chrome Experiment Colorscube is not as smooth in Minefield compared to Chromium (pre3d)

Categories

(Core :: Graphics: Canvas2D, defect)

x86_64
Windows Vista
defect
Not set
normal

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: klasconia, Unassigned)

References

()

Details

(Whiteboard: [chromeexperiments])

User-Agent: Mozilla/5.0 (Windows NT 6.0; Win64; x64; rv:2.0b7pre) Gecko/20100916 Firefox/4.0b7pre Build Identifier: Mozilla/5.0 (Windows NT 6.0; Win64; x64; rv:2.0b7pre) Gecko/20100916 Firefox/4.0b7pre This Chrome experiment uses pre3d : http://deanm.github.com/pre3d/ and canvas to generate a cube that can be rotated in the browser. When testing it in Minefield the rotation seemed to be choppy and "laggy". When the same experiment was tested in Chromium the rotation was incredibly smooth. Enabling/disabling hardware acceleration did not appear to help. Reproducible: Always Steps to Reproduce: 1. Navigate to the experiment 2. Start the experiment 3. Rotate the cube by clicking and moving the mouse in different directions Actual Results: When the cube is rotated the animation appears to be very choppy. Expected Results: Rotation should be smoother just as Chrome is.
Whiteboard: [chromeexperiments]
CC'ing the Direct2D guru...
So over here, on Mac, we spend about 38% of the time doing Fill() on the canvas and another 40% or so running JS (more on that below). 5% is vm_fault, 4% is setting style on canvas, 3% is LineTo/MoveTo/BeginPath/FillRect on the canvas. 5% is painting to screen. Then again, on Mac this seems pretty smooth. The JS engine time (all percentages are now of time spent in the js engine itself) is about: 42% in methodjit-generated code 15% stubs::NewInitObject, mostly GC 6% stubs::CallName (lots of NameOp, js_NativeGet, getting call vars, etc) 6% InitPropOrMethod (mostly self time) 6% array_join 6% array_sort (about half calling back out to page code for the sort compare) 5% stubs::SetName<0> (2/3 is allocSlot, the rest is self time and js_NativeSet) 4% stubs::NewArray (about a quarter of this is GC; the rest is just "stuff", e.g. reallocs). The rest is little stuff.
Status: UNCONFIRMED → NEW
Ever confirmed: true
For me, on Windows 7 x64 with a T4300 processor, it is smooth with or without HW acceleration. Kevin, What is your processor ? Provide also the graphic info of the "about:support" page.
Scoobidiver: My processor is an Intel Core 2 Duo T6500 @ 2.10Ghz. The info from about:support is: Adapter Description: ATI Mobility Radeon HD 4530 Vendor ID: 1002 Device ID: 9553 Adapter RAM: Unknown Adapter Drivers: atidxx32 atidxx64 atiumdag atiumdva atiumd64 atiumd6a atitmm64 Driver Version: 8.563.1.3000 Driver Date: 2-9-2009 Direct2D Enabled: false DirectWrite Enabled: false GPU Accelerated Windows: 1/1 Direct3D 9
Kevin, It seems that D2D/DW is disabled for you with no reason for that. Here is a list of actions to do by order of preference: 1. Try with a new profile in the latest nightlies, 2. Download the graphic driver 8.762 from ATI,
This now seems to be pretty good, even with hardware acceleration disabled; with hardware acceleration enabled, it's *great*.
Status: NEW → RESOLVED
Closed: 13 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.