Web browsing

No need to comment these results today. These processors offer more performance than browsers may need. It would've been more interesting if we tested Atom products. But then again changing browsers is easier than changing processors anyway. Or at least free.

Audio encoding

In this group of tests much depends on how many encoding threads you run, six or eight. The latter obviously drives Phenom II X6 to the last place. Besides, while being performed fast, these tasks put much load on the processor. This prevents the Core i7-800 series from using the advanced boost mode to the max. Moreover, these processors are constrained by stricter TDP limits. So from a researcher's point of view, the winner is Core i7-950. As for the point of view of a user that just needs to encode a few albums and copy them onto a mobile phone, all today's processors are fast enough. Selecting songs may take more time than encoding itself.

Video encoding

Phenom II X6 looks good again. It outperforms all direct rivals and only loses to the much more expensive Core i7-880. If you look at the detailed results, you will see that the better is application's multi-core optimization the more noticeable is the victory. We believe that's because video encoders not just run a lot of computing threads, they run a lot of similar threads. And that's very inconvenient for Hyper-Threading, because two similar threads may compete for shared blocks. The result is still better than that obtained without SMT support, but worse than it could've been without such competition. And Phenom II just runs six threads which do not interfere with each other.

Games

Games do not need six cores. And even if a title can use that many, virtual threads will do fine. Why, even four cores are often too many, which allows Turbo Boost to help. But as you can see the final difference is negligible anyway. A single-GPU graphics card will perform fine with a processor cheaper than what we have today. If you look at the detailed results, you will see that even the reference Athlon II X4 620 for less than $100 may offer acceptable gameplay.

Conclusions

First, let's say a few words about the competing AMD Phenom II X6 1090T. As you can see, the hopes that a true hexacore solution is bound to outperform Intel's quad-core CPUs are not justified. Yes, it can sometimes perform on a par, it even wins in one test. But that's it. AMD managed to catch up with the junior Core i7 solutions, but then Intel just cut prices for the next higher-end models. Increasing Phenom II X6 clock rate considerably (by more than 100-200 MHz) can be a decent reply from AMD, but we'll hardly see this in the nearest future. The reasons are obvious: any product, be it a car or a processor, has limited upgradability. Each new improvement is harder to achieve. And Phenom II is essentially a rather old product that is nearing its limit. Efforts to implement technologies like Turbo CORE cannot radically improve situation. Intel has always been a leader in the "core race", and at lower prices at that, because they were the first to move to finer process technologies (hence better clock rates). In other words, it's high time for a radical changes, and AMD understands that well. Anyway, there are positive sides as well. Last Christmas AMD had nothing to offer in the over $200 segment. Now it has "mainstream hexacore" solutions, which will do well if you're upgrading and whatnot.

Now let's return to the processors we reviewed today. Perhaps, the most interesting result is that the Core i7-870 was slightly behind the Core i7-950 last year, but today both finished on a par. We believe that's because Turbo Boost quality depends much on motherboard, primarily its BIOS. Motherboard vendors released numerous firmware upgrades over the year, some of which were only called to "Improve Intel Turbo Boost compatibility." So it seems they've finally managed to improve support for this technology to the initially planned state. Well, it took them some time. And the boost mode is a standard feature of all processor architectures from Nehalem on, mind you. By the way, this explains why Turbo CORE hasn't had the desired effect yet. It's a later addition for the AM3 platform, implemented into only three CPUs out of dozens at that. There are LGA1366/LGA1156 processors without Turbo Boost as well, but only a few, and they were rolled out later than the full-fledged models. Partners of AMD have no special need to put much effort into Turbo CORE. This is another reason to overhaul the platform.

The Core i7-870/875K should've demonstrated equal performance, and they did it. It's clear that there's no sense in buying the latter just to use it in the typical mode, because it's more expensive. But if you want to fine-tune it, then the price is very well justified. For example, a Core i7-875K can be easily transformed into almost a complete analog of Core i7-880 or, vice versa, Core i7-860S. And we're not just talking about mere clock rate adjustment, the boost mode can be altered as well. TDP may be the only exception: unlike extreme models for LGA1366, adjustment of power consumption thresholds is a rare thing in LGA1156 motherboards. In other words, you can transform a Core i7-875K into a solution with a completely unique clock rate for your personal needs — just add another $25. But will it interest overclockers? We doubt it, it's still a Lynnfield with the corresponding overclocking limit. Besides, overclocking varies greatly with specific CPU samples, so a Core i7-875K may overclock worse than any Core i7-870 under the same conditions.

The rest is simple and clear. As we have repeatedly mentioned, Hyper-Threading and Turbo Boost complement each other fine. The "updated" Turbo Boost of the Core i7-800 series is especially nice in that aspect, adjusting clock rate within the 600 MHz range depending on the number of active cores. So this couple helps higher-end Intel processors dominate the charts. That's not a problem, really, given there are no true competitors. All Core i7 products have so much potential computing power that the problem is how to use it all properly, not how to overcome the lack of resources. And it probably should be so. Because if you're willing to pay about $300 for a CPU alone (the price of a decent entry-level notebook), you will expect it to offer enough computing power for every reasonable task. But that's also a reason why reviewing high-end processors is somewhat boring. Testing lower-end solutions is far more interesting — you have to sacrifice certain features and it's not always easy to understand how critical those are. But that's a matter of reviews to come.

We express gratitude to Gigabyte for providing equipment for the testbeds.

Write a comment below. No registration needed!