Core -I7-820QM review

The 820QM will be faster than Q9000 even in highly threaded applications. Its not going to be by a lot but faster none the less.
However, the 820QM will be much faster than a Q9000 at times when running only 1 or 2 cores, which is more often than not.

As an example of what I'm talking about, compare Cinebench R10 single core vs multi-core. http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3647&p=5
As you can see in single core benchmark the 920QM beats the QX9300 easily.
However the QX9300 catches up in the multicore test because the 920QM is running at a lower frequency, regardless, The 920QM is still faster.

 

Let's assume for the sake of argument that the Ci7 2.0GHz = C2QX 2.53GHz. Numerically speaking this means Ci7 at 1.73GHz is approx equal to a C2Q w/ 2.2 GHz. Considering the Q9000 is less than that and the Ci7 2.0GHz is slightly better than the C2QX 2.53GHz, I think it's fair to say that the 820QM would even be competitive against the Q9100 (2.26GHz). The numbers are just for demonstration and won't accurately reflect performance, so don't take this as the rule or anything, but I think it proves the point effectively.

 

Yeah so basically these i7s are fusions of the current quads and duos..

 

^ no, way off. Entirely new architecture.

 

I think he meant in reference to the fact that Turbo can OC the quad core turning into a faster dual core when the other cores are unused.

 

Yeah it's kind of having inter-switchable QX9300 and T9900.

 

so the conclusion is , the mobile i7s are not outperforming the quads like the desktop version eh?

 

yeah you could say that, it can also switch off cores and downclock to 1.2ghz to save power, so it's a lot more efficient than the current quads.

 

It would kind of dumb of Intel to release new quad cores that don't beat the current generation don't you think... the Clarksfield i7s are better than current C2Q but not by that much, hence if you have a current notebook w/ a quad core in it, then you won't gain much by going to an i7.

 

Don't forget these have hyperthreading. Quads don't

 

Those benchmarks already take account that i7 has hyper-threading.

 

Clevo preview:

<width='480' height='295'><param name="movie" value="http://www.youtube-nocookie.com/v/nQEPvWk91fg&hl=en&fs=1&rel=0"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube-nocookie.com/v/nQEPvWk91fg&hl=en&fs=1&rel=0" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width='480' height='295'></embed></object>

 

I have to say I completely agree. Currently I have an NP8662 with the 2.0 Ghz Q9000 and the GTX 260M and you can see from the benchmarks in my sig that it performs really well in gaming, even alongside the higher clocked 2.85 ghz Core2Duo. Regardless of how much better the recently released mobile i7 CPU's are, until we deal with the real bottleneck in gaming, i.e the GPU, I don't see it as a worthy upgrade at all.

I am glad to see this release and am excited to see what is next but I am waiting for GPU improvements since that is what will give us serious performance boosts in games. Nice to see the GTX 280M officially released in a 15'' chassis as well.

 

Isn't the bigger hurdle for laptop owners is that they usually need to replace their whole laptop in cases like these? I guess you can sell the original laptop but even then it'll be a significant cost. If I guess the buyer is looking for a high end laptop it'll end up the best choice.

 

they are using the same socket??

 

No, they are incompatible.

 

At the same clocks, i7 is faster marginally, as it is with new generations, but the benefit of the Turbo Boost is the real kicker. I know my i7 clocks up to 3.4 Ghz sometimes and it doesn't get too much hotter.

We have 3Ghz Core 2's at school and they are no where near as responsive or as quick as my i7, which could be due to other factors, but I had the same experience on my laptop when I had the X7800 @ 3.0Ghz. There is always a core available to do something, which is the current advantage over dual cores. I still dont think the QX9300 would be faster, even when over clocked, and the fact that they are that expensive is another strike against it, I see no reason to spend $500+ on the QX9300 when this mid-range offering can match it.

 

Do what use the same socket?

 

A few things to note here:

1) I was thoroughly impressed with the power consumption levels based off the i7 tests. A user running battery would be in a phenomenally better position for portable performance than a user with a C2Q / C2QX. This however is majorly wrong.

Scientific experiments must have a control unit (the C2Q base) and then change ONE factor to see the implementation that has on the hypothesis (in this case that the i7 is more efficient).

One can see, at partial load the i7 shows it is very capable of cutting power consumption and thus increase battery life significantly.

What didn't make sense to me was how a Q9000 w/ a 45 Watt max TDP would pull more power than an i7 w/ the same TDP that will use TurboBoost to maximize performance while remaining in that window. Following that logic, I began to research the different models tested.

a) A 16" i7 based machine, running 2 x DDR3 sodimm's, and a NVidia GT240m
b) A 17" Asus G71Gx w/ Q9000 processor, running 3 x 2GB DDR3

The G71Gx is only available with one graphics card: a GTX 260M, which draws 75 watts ( http://www.notebookcheck.net/NVIDIA-GeForce-GTX-260M.14559.0.html)

The GT240 has only a 23 watt power draw ( http://www.notebookcheck.net/NVIDIA-GeForce-GT-240M.17654.0.html), which leads to a 52 watt difference.

In addition, the extra stick of RAM in the Asus, 2nd hard drive in the Asus vs 1 in the i7 test machine, larger Asus screen (which is also CCFL backlit) and DDR2 vs DDR3 power consumption are what make up the difference in power draw, not the processor itself.

2) The multithreaded bechmarks will not replicate real-world performance unless you are using the machine only for ray-tracing or transcoding, etc that you would be running parallel threads 100% of the time. The hyper-threading will give much more real-world performance with multiple, smaller execution threads (many programs) running simultaneously.

3) With the current speed of processors, it's not as big an issue as it will be in the future, but FSB majorly restricts a system's performance. By having the memory controller onboard rather than having to relay through a northbridge, the i7 can get significantly improved memory speeds. As the clock speeds go up and RAM speeds increase, the i7 will really take off. It may take a little bit of time for the i7 to catch up to the clock speeds of a Core2, but it has much more potential than the Core2 line.

 

I would think the onboard memory controller is responsible for the responsiveness of the i7 more than anything else.

 

You forgot that on i7 CPUs it includes entire Northbridge on the CPU and counts it as part of the TDP. So a i7 with 45W TDP is effectively like a Core 2 with 35W TDP. There's one there, another is that the core itself has better power management. It won't be drastically lower, but it'll still be noticeably lower(I'm guessing 15-25% on load).

Anandtech has a more fair benchmark( PCPro is kinda invalid by the fact that the systems were still pre-release at that stage):

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3647&p=8

Notice how even a 55W i7 920XM has lower power consumption than 45W QX9300.