Intel HD i7-2760QM video benchmarks

[dasaki]

Hi,

I've noticed a high jump in the video performance of the i7-2760qm compared to all other mobile Sandy Bridge processors, according to: http://www.videocardbenchmark.net/gp...l+HD+i7-2760QM

Is there any special reason for this? I though all of the integrated HD 3000 were the same.

Thanks.

[David (PassMark)]

Video performance is also CPU dependent.

[dasaki]

Video performance is also CPU dependent.

That is precisely why I wondered, the 2760qm video results are higher than the 2920xm processor that is ranked better in cpu performance. Also the cpu increase over others is not a 60%, like the gpu increase:

GPUs


CPUs

[David (PassMark)]

What I think is happening is that we are seeing some of the effects of the NVIDIA Optimus technology. The 2920XM GPU is less commonly used in high end laptops and isn't always paired up with a faster AMD or nVidia GPU. So the results in the chart reflect the fact that the benchmark is run on the 2920XM GPU.

For the 2760QM CPU/GPU, it is commonly used in high end laptops. In particular all the top results are from the Alienware M17x machine. This laptop comes with NVIDIA Optimus, and a 2nd video chip. With Optimus the GPU used can be switched automatically by the drivers (sometimes doesn't work) or switched manually by the user. So PerformanceTest detects the first GPU in the system, the Intel HD 2760QM, and reports that in the charts. But then the actual benchmark test might or might not be run on the Intel chip. Optimus is pretty new and we don't know at the moment if a 3D application has any way to knowing which GPU has been allocated to run a particular task. Also at the moment PerformanceTest will only report on the first video card found. This hasn't been a problem in the past because SLI always used matched pairs of cards, and integrated GPUs were traditionally disabled when a discrete cards were installed.

[dasaki]

Does that mean the numbers shown in the videocard passmarkare are not a reliable reference for the performance of this specific CPU+GPU pair?
I was about to spend almost twice as much money in a 2760qm (compared to a 2630qm), to double the graphic performance, relying in your numbers.

[David (PassMark)]

Yes and no.

For the machines with Intel HD and Optimus and a fast 2nd GPU, the numbers are not entirely accurate. We will be looking to try and improve on this for the next major release.

On the other hand these machines will probably outperform machines without Optimus, on average, but not because the integrated GPU is faster. So in this sense the chart reflects the performance from these type of machines. But we agree is confusing and it would be better if it wasn't.

[dasaki]

Yes and no.

For the machines with Intel HD and Optimus and a fast 2nd GPU, the numbers are not entirely accurate. We will be looking to try and improve on this for the next major release.

On the other hand these machines will probably outperform machines without Optimus, on average, but not because the integrated GPU is faster. So in this sense the chart reflects the performance from these type of machines. But we agree is confusing and it would be better if it wasn't.

I understand. Thanks for the reply and the clarification. Maybe in the meanwhile until this issue is improved, you could add an end note with an explanation about it on your graphs, just to avoid confusions.

Thanks!