AMD a8 and i7-2670qm

I'm buying a new notebook, and it's been a while.

I have some questions regarding processors. I'm curious about i7 2670qm vs AMD a8-3520m. a8 seems to be cheaper and decent in performance (and of course great in graphics).

i7 seems to be very expensive, but much better in performance.

I don't do gaming, and it therefore seems like the a8 graphics is pretty useless for me. I need a computer that's performing well when dealing with relational databases and running expensive algorithms on huge data sets. E.g. a bigO(n^2) on up to n=100,000..

i7 2670qm to amd a8-3520m is about 100 $ more. Is it worth the money? Any experiences on a8?

 

i7. That is what you want for that workload.

The A8 is good for general consumer stuff, and even some more demanding consumer stuff. But as soon as you get into heavily threaded apps and uses that you're talking about, the i7 shines.

 

Indeed, the i7 is the way to go. The CPU of the A8 is marketed as "good enough" and for things like Microsoft Word, it generally is. For anything that seriously needs the CPU, don't bother with it. Unless you intend on overclocking the A8, the i7 will be more than twice as fast and even if you do overclock, the i7 will still be more than 50% faster. $100 is a small price to pay for that, particularly since you don't need the laptop for games.

 

Thanks, guys. The reason I'm still doubting is that i7 2670qm has so bad performance/price rate.

 

For your intended usage, though, the i7 is ideal. You also have to figure in how much your time is worth. Running complex databases and simulations will take a lot longer on an A8 versus an i7. $100 is a pittance with how much more you'll get done/time you'll save.

 

depends on his software, some can utilize GPGPU which actually puts an A6 or A8 ahead of an i7. you really need to know your software. now yes if its multithreaded, CPU only, the i7 shines and on single threaded apps they are VERY similar

 

what exactly do you mean by "bad performance/price rate", if you're referring to GHz/$, than yes, it's somewhat low, however, because of it's architecture intel's offering is about 50% faster per clock than amd.

MHz quantity is indication of speed within architecture, but can't be compared directly between different cpu architectures any more.

 

GPGPU for non-video/graphics related stuff is really, really rare. I suppose it might be worth checking, but I am fairly confident that if the software in question could use GPGPU, it would be very well advertised. In general, GPGPU is a royal pain in the neck so if somebody went through the considerable trouble it takes to implement it, there's a lot to gain and they would make sure people knew it.

What are very similar? Just to be absolutely clear: single-threaded or multi-threaded, the i7 will utterly crush the A8. Is that what you meant?

 

AMD GPGPU requires OpenCL -- there's not a lot of support for OpenCL in the market as the CUDA development tools are much, much more mature. Even then what you are describing is not going to map well to parallel computing, and it is doubly worse that the AMD GPU is not going to be powerful enough to really make up for the deficiency in CPU performance for your applications. i7 is going to run circles around AMD for the loads that you are describing.

 

incorrect Greg AMD is new to OpenCL ( which is a nightmare i hear), but the old reliable OpenGL support has been around for years with Nvidia and Matrox as well as C++AMP

AMD "Llano" A8-3850: : OpenGL Performance: Turbine Demo, Cinebench 11.5, RatGPU Bench - L O S T C I R C U I T S

we have an openGL database indexor for our video clips from AVID, an A6 laptop can run it in 22:30 roughly and an i7 MBP takes a bit over 45mins ( yes im aware that GL is normally used for graphics, but im no programmer )

but your right an i7 prob is a bertter choice

 

AMD has been dabbling with Open CL for the last two years give or take, and it is a nightmare. Open GL is a different matter entirely but at the same time high performance computing applications are a different story. Video applications, fine, I can see where it might be natural to use Open GL. C++ AMP has little to no industry support given my experiences, and people are quite cold to AMD/OpenCL as well. CUDA and AVX are the primary platforms for my clients.

As for your particular i7 application, unless it is multithreaded and utilizes SSE/AVX it isn't a fair comparison but it might be accurate given the GPU has a fair amount of processing power behind it. Not to mention differences between the two platforms might influence the comparison. I'm not going to go into this further since I do not want to derail the thread anymore than it already has been.

But for what the OP needs, and what most of the scientific computing community needs, Open GL is not the appropriate tool. CUDA or Open CL is if he seriously wants to map his applications to GPGPU. That being said relational databases is going to be a task that is very GPGPU-unfriendly. Definitely digging the i7 for this.