Tom2D: Line Output
This test exercises line drawing functions in GDI. In Part 1 of this story, these tests produced the most dramatic and obvious differences among graphics cards. Please read the summary at the end of this test, because the problems described in our first exploration were also reproduced on more than 40 systems in our follow-up round, and in results posted later by our readers.
Test Content:
A total of about 150,000 lines of random length and color are drawn in random positions on-screen as this test runs. Because the seed value for the random number generator remains constant, all test runs should produce the same outputs as well. All objects will appear inside the drawing surface, so no clipping is needed.
GDI Functions Called
For the direct line drawing benchmark:
CreatePenMoveToExLineToSelectObjectDeleteObject
For drawing lines from the DIB buffer:
CreateCompatibleDCCreateDIBSectionCreateSolidBrushRectangleBitBlt
Summary:
Here's where we're able to re-confirm the results seen in Part 1, where ATI's Radeon HD 5000-series cards don't perform well.
Interestingly the measurements for output with Aero turned off increased when we opened a second window and maximized it (something like Mozilla Firefox, for example) before running this test, and then switched back to the test window. After that, a measurably higher value is produced on test machines in a completely reproducible fashion. A second test run after a brief period of time no longer shows this effect, however, and performance losses once again appear across the board.
Again, this is indicative of a driver issue, and we can only hope that the test driver we're working with currently addresses the issue. The rest of the test results speak for themselves. Once again, the Atom with a GMA 950 (and Aero enabled) generates surprisingly good results, and the low performance from the GeForce GTX 285 is equally noteworthy for direct output with Aero disabled. Anything that goes over 20,000 lines drawn probably won’t impact real-world usage scenarios.