09-16-2016 06:12 PM
I know that technically the M3000M is better than the M2000M, and the M3000M is compared to a GTX 965M while the M2000M is compared to a GTX 960M.
Note that nvidia (http://www.nvidia.com/object/quadro-for-mobile-workstations.html) classifies the M3000M as the lowest GPU is the 17.3" platform, while the M2000M is the highest in the 15.6" platform.
Obviously, if you had a laptop with the M3000M, and an identical laptop (identical in every respect except GPU) with the M2000M, the M3000M would outpace the M2000M. However, they aren't in identical laptops. Which leads to my question(s).
With CPU, RAM, resolution, and storage drives being equal, how does the M3000M's practical performance in a 17" laptop match up to the M2000M's practical performance in a 15" laptop?
Does the extra screen size use most-to-all of the extra video processing power?
Does the M3000M lose most of its inherent advantage [over the M2000M] because it is powering a bigger display?
Does the M2000M in a 15" laptop outperform the M3000M in a 17" beast?
Would the M2000M @1080p outperform the M3000M at 4K?
For many purposes I imagine the difference between the two chips would be a wash, yet is this the case on the top end of their capabilities (in their respective machines)?
And maybe this isn't the place in this forum to ask this, or even the forum to ask it in, but I was wondering, and thought I'd see what you people say. Thanks.
Solved! Go to Solution.
09-16-2016 08:06 PM - edited 09-16-2016 08:16 PM
You're looking at the wrong specs. The laptop screen size is basically irrelevant.
There is absolutely no difference in processing performance required to drive a 15" vs a 17" screen; what matters is how many pixels need to be serviced (e.g. data throughput needed) and how much processing is needed to support intended applications with adequate levels of performance. 4K screens obviously have a higher bandwidth requirement than 1080P. All else being equal, different sized panels themselves may consume more electrical power, depending on technology used -- but that has nothing to do with display adapter chips.
Meaningful characteristics of display adapters include: memory, GPU architecture (quantity of processing cores, pipelines, clock speed, bus bandwidth, etc.). Higher numbered Quadro Series chips have more of these elements and consequently, can perform typical mathematical processing tasks at higher speeds and deliver more rapidly changing display information faster, to whatever panel which they are attached.
The screen size differentiation in this table has nothing to do with the actual screens used, only suggesting correlations with ability of laptop designers to deliver adequate power supply current and cooling capacity to the display controller chips. It is presumed that larger screens will be part of larger laptops and therefore can be equipped with with more potent cooling capabilities -- hence capable of supporting chips with higher TDP requirement (total thermal load). P70 machines, by way of example, have dual fans and an elaborate connecting heat pipe along with control logic to differentially cool either the GPU or CPU, as needed -- it takes a large chassis to accomodate all the plumbing. Consequently, the higher-end chips graphics can be supported whereas a smaller box might not suffice.
There is no reason otherwise to group these chips by screen size. The distinction has little practical significance to end-users, as long as good engineering design principles have been followed by a reliable manufacturer. Buy the capabilities you need in packaging that makes sense for your particular usage.
09-16-2016 11:10 PM - edited 09-16-2016 11:18 PM
I suspect that one of the reasons that M3000M is spec'd for bigger laptops is that it has higher power and cooling requirements.
As stated by practical, it is the screen resolution, not size in inches that can affect performance.
That said, the M3000M has 60% more cores, double the memory bus width and 36% more power consumption versus the M2000M. Does this mean the M3000M is automatically 60% faster? Not necessarily, as not all tasks scale in a linear manner with the numben of cores.
09-17-2016 02:53 PM