How does mobile CPUs perform vs PC CPUs?

Hello guys, I hope this is the correct subforum for asking such kind of question: Does anyone know how does mobile CPUs fare against their PC counterparts? For example: would an i7 @ 3GHz perform equal or worse than its PC counterpart?

Or even more contemporary: are the integrated GPUs in the new sandy bridge mobile CPUs the same as their PC counterparts?

Thanks a lot.

 

Mobile CPUs perform the same as desktop CPUs with identical specs. That is, if the architecture, clock speed, amount of cache, etc. are all the same, then the mobile chip is just a low voltage, low TPD version of the desktop part. However, most mobile chips are not anywhere near desktop chips in performance. The big exception are Sandy Bridge CPUs which can boost relatively close to their desktop counterparts and are competitive with the previous generation of high-end desktop CPUs. The integrated GPUs are likewise pretty similar on both.

 

Thanks for the quick response Althernai.

Do you have any link to the comparison of mobile and PC sandy bridge comparisons?

 

You can use ark.intel.com to check out specs between similar lines (ie. SB desktop and laptop parts) and use that as a rough guide. You would have to check out certain CPU benchmarks if you want a more accurate comparison of CPUs between the same lines and across different ones.

 

I thought this was a awesome benchmark/comparison list.


What I thought was interesting was how some notebook CPU's outperformed some big hitting desktop cpu's

 

Thanks for the link, I didn't know notebookcheck how some list of desktop parts benchmarks. I know the site is not to be trusted for accuracy, but for an overall picture it is fine.

 

Um, why? I've heard and read a lot of people saying that you can't trust NBC (not the TV station lol), but I've yet to hear a legit reason why. They're extremely thorough and cover every single GPU, processor and such...And I don't find them biased at all.

Mr. Mysterious

 

Because many of the results are submitted by home users and enthusiasts who are not part of any testing team. Benchmarks also show inconsistencies due to resolutions not being specified in GPU tests, etc.

It can be great when used as a general guide, but when using results from there, take everything with a grain of salt.

 

here is an example:
AMD Radeon HD 6470M - Notebookcheck.net Tech

check the gpu benchmark of crisys, click on the medium tab

the 4670m is slower than the 6470m, and that is just wildly inaccurate, in the same result you can see the 4650m getting a 21% better performance in there. You can also see the 550v in there, and that is just a rebranded 4670m, and that nets a 26% better result than the 6470m.



this is an example that I saw today when I was discussing gpus in the forum, there are far more, like when i saw a 5850, get beat up by a gt 425m, and no it wasnt a nvidia sponsored game

 

4850's CFX mobility are not listed that's enough for me ;p

 

Like already stated above. But on top of that many benchmarks comes from setups totally different (same GPU, different CPU/RAM) or even the GPUs which have two different RAM flavors (GDDR3 vs GDDR5) are merged together. Still I regard NBC as a general overview site for mobile benchmarks.

 

Well has anyone found anything better? No? I'm going to stick with NBC then

Mr. Mysterious

 

No one said that it was bad or not to use it, only to take the results with a grain of salt.

 

use it as a general idea, and you pretty much have to see each game to derive the performance accordingly, sometimes a card gives better performance than the competition in a review, however they give the prize to the loser card. this happened in the dell vostro review

 

CPU: At equal frequency, mobile parts are negligible amount slower than desktop counterparts, maybe 5%, due to things like power management and more relaxed timings on memory, etc. Of course, they have different frequency.

GPU: For Sandy Bridge, the highest end mobile GPU on the quad core chip performs about 5-10% slower than the best desktop integrated Sandy Bridge GPU.

-Dual core versions: cut by additional 10-15%
-OEM chips are slower than retail ones
-ULV versions are 30-40% slower