YABM – yet another benchmark
Posted by Jon Peddie on September 15th 2008 | Discuss
Categories:
Tags:
We bought a nifty performance test program from PassMark. (See the mention of it in this in the review of the Lenovo ThinkPad.) It’s a full suite of tests, and totally affordable ($24.) Included in the suite is a 2D and a 3D test set. They run on DirectX 9.
FIGURE 1:PassMark 2D and 3D test results
(Source: Jon Peddie Research)
FIGURE 2:3DMark06 multi-res benchmarks.
(Source: Jon Peddie Research)
Since we had a couple of laptops on the bench for some other testing, we decided to run the graphics benchmarks on them too. Why not? The results were amazing. So much so we decided to run Futuremark benchmarks as well to see if they corresponded.
As Figure 1 indicates there’s no correlation between 2D and 3D performance using the PassMark tests, with the Think- Pad doing the best in 2D and not so in 3D. Also, the ThinkPad shows up as inferior to the Mini Note in 3D, which just doesn’t seem logical.
We decided to use 3DMark06 to test the laptops for DirectX9 performance. The results in Figure 2 seem more in line with what we would expect. To be fair, the FireGL in the ThinkPad is not tuned for DirectX. We have experienced this with other FireGL AIBs in desktop machines as well.
It should be also be noted that T60P is no longer available as a product offered by Lenovo so getting too worked up about its graphics performance is moot. Also, the chip on the graphic card is R5xxx series and ATI is now on to R6xxx (for workstation mobility)
We are going to look at a Lenovo W500, which also uses a FireGL Mobility card but with the R6XXX chip which shows considerably better performance.
One other point is that the T60P has the annoying habit of turning down graphics performance unless specifically set up on the power management panel not to do so. This happens even under max performance unless otherwise setup.
What do we think?
PassMark doesn’t seem to be a reliable 3D benchmark. However, it may be useful for 2D. We’ll try it again on our next series of tests to see if it gives any insights on a chip or AIB or system’s graphics behavior. It would be nice to have more than just one synthetic benchmark.—JP
