i7 4800MQ, 4900MQ benchmarks

Hello,
I have been trying to compare the performance of the Intel Core i7 4800MQ and 4900MQ processors. It is interesting that Geekbench Browser (browser.primatelabs.com) reports a fairly wide range of scores for both of these processors. The 4800MQ has scores between 1944 and 16379, while the 4900MQ has scores between 5910 and 16333. With caching, it seems like a second run of the benchmarking tool would score a bit faster, but I am having a difficult time understanding the large variance in these scores. Do the differences in hardware (outside of the processor) and OS really account for the difference in scores? In order for the scores to be truly comparable does the hardware (outside of the processor) and operating system have to be identical? I am trying to come up with some sort of an idea of how much faster (if at all) the 4900MQ will be, compared to the 4800MQ, when really pushing the processor in scientific computing (in games or productivity apps, I would think one would be hard pressed to see a difference between the 2 CPU's). Obviously, the really big jobs would be pushed off onto a workstation, but it is nice for me to keep as much processing local (portable and on a budget, too), as possible.

 

After doing a bit of reading it appears benchmarking scores will change on a given system, run under seemingly identical circumstances(for instance, running the benchmark first thing after the system is rebooted). Obviously, other aspects can have a major impact on benchmarking, such as running off of the battery or the power supply, the temperature of the CPU, the current load on the system, the OS, etc. Ideally, the system in question would be run under real world load (for example, a series of scripted photoshop actions) for a longer period of time (perhaps 10 hours), then the CPU would be swapped (or identical system used) and the test would be repeated. A longer run time should compensate for some of the variation normally seen in benchmarking identical systems. I will try to gather the available benchmark data and put together some sort of estimation of the performance difference between the 4800MQ and 4900MQ.

 

It's challenging to isolate one component in a system. Ideally you'd have one testbed and swap parts, changing one variable at a time.

All of these could affect the numbers you're seeing:
-Improper benchmarking techniques (users running background processes at the same time, &c)
-Operating system and drivers
-Memory (RAM) - speed/latency/capacity
-Motherboard (revision, chipset)
-Presence of a graphics card
-...
The list is about endless. Too many variables.

I think the bigger question regarding these CPUs is how much the difference in cache matters (6MB for the 4800MQ, 8MB for the 4900MQ). The clockspeed differences are just about identical ...these are the official Intel specs:
4800MQ - ARK | Intel® Core
4900MQ - ARK | Intel® Core
Less than 5%. We consider 5% to be within the acceptable margin of error for our product reviews on this site.

Feel free to create a thread in this forum, asking users to post up their own benchmark numbers for these processors. You can suggest benchmarks for them to run.

P.S. don't just use any benchmark; pick one that's indicative of your needs. If you're interested in video encoding performance, then find a benchmark that includes video encoding. The best-case scenario would be to find a benchmark that runs via the specific software package you're using. There are benchmarks available for Photoshop, for example.

 

Is an i7 4700MQ good enough? Or should i have paid extra for the 4800/4900?

 

If the 4700MQ isn't good enough, then neither are the 4800MQ or 4900MQ.

 

In most of the benchmarks I have been seeing the 4800MQ and 4900MQ are within 5% of each other. Occasionally, a benchmark will show a 7-9% difference between the 4800MQ and 4900MQ. What probably muddies the waters even more is an article at Notebookcheck ( Review Intel Haswell Processors - NotebookCheck.net Reviews) that noted during benchmark testing, the 4800MQ, 4900MQ and 4930MX were all limited to a maximum of 3.4 MHz during multi-threaded tests.

In the real world, I doubt if there would be a noticeable difference between the 4700MQ, 4800MQ and 4900MQ. However, when performing a multiple sequence alignment that keeps a single core running at 100% for 4 hours, my feeling is there may be a few minutes difference in the finishing time between the 4700MQ and 4800MQ. Given a 5% margin of error, I am not confident if there would be any difference in run times between the 4800MQ and 4900MQ.

 

The stuck at 3.4 GHz part could be a Clevo pre-production model specific issue from what I gathered from the article at notebookcheck, I don't think it's an issue with the CPUs themselves. You could check the owner's lounge of Clevos, Alienwares, etc. to see if anyone has experienced that kind of behavior.