iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Intel Core 2 Duo Processors Revisited

And retested with the new test method.

September 23, 2009



<< Previous page

     Next page >>

Java



Results here are canonically correct, because triple-core processors do not try to compete with quad-core models. However, it does not make the life of dual-core processors any easier. And if we also take into account lower requirements of the virtual Java machine to cache size, their top representatives are screwed.

Audio encoding



Another striking blow, but not the last one. We have seen many times that processors from AMD perform worse here than their counterparts from Intel. However, the third core allows them to compete practically on a par in the Mid-End segment. It's a pity, of course, that we failed to get the Core 2 Duo E8600 to shift this situation a little towards usual. However, the best this expensive processor could have done is to outperform the Х3 720 a little, but not get closer to the Phenom level at the cost of Core 2 Quad Q8200.

But we can also take a look at this situation from a different angle. Athlon II X2 250 turned out to be the slowest. It demonstrates the worst result with OGG Vorbis. It equals just 32, which means that a one-hour album will be encoded by this processor... for less than two minutes. That is in terms of absolute results it's difficult to come up with a situation, where the speed of audio encoding will really matter. Copying an audio disc had taken thirty minutes ten years ago, and then it had taken another several hours to encode audio files into MР3. But now the slowest operation will practically always be getting the source audio, not encoding. For example, it can be done simultaneously with getting the source audio or transferring audio files into your player.

Video encoding



Results in these tests go a little beyond the idea of common users that video encoding takes a multi-core processor. It happens so because two codecs out of five (at least the versions we use here) are relatively indifferent to more than two cores. One of them is even content to use only one core, the other two codecs do not utilize the third and the fourth cores in the same manner. When upgraded from the C2D E7600 to the C2Q Q8200, mainconcept gains only 20% (that is the doubled number of cores is noticeably compensated by the difference in clock rates), but x264 gets proper performance gains -- over 1.5 times in the same conditions! If the other codecs had followed suit, we would have got the same situation as in the previous group. But the baggage of years complicates things. On the other hand, one of the codecs is so different (another two demonstrate practically an even exchange of cores for frequency) that it becomes clear that the best dual-core processors can do it get close to the slowest quad-core processors, outperforming them is out of the question. Such results are mostly conditioned by the heaviest tasks, which should be accelerated at any costs.

3D games



Up to recently games have been considered the field, where high-clocked dual-core processors with large cache and fast FSB (the E8000 family meet all these requirements) can easily defeat cut-down quad-core processors. But it's not quite true. As in case of video encoding, Core 2 Duo or Athlon II X2 generally perform well, but as we take a look at detailed results, our optimism fades away. Just because it's much harder to compare frame rates in games than render times in 3D modeling packages (using plain arithmetic rules). Games are interactive applications, so they always have a certain lower limit of comfort that cannot be crossed. What concerns encoding or rendering, lower results are just lower results. For example, if you encode movies at night, not many and not very often -- it does not make a difference whether the job will be done for three or for five hours: you will see results only in the morning, and it will be impossible to add another job to the batch, as there is no other job. Games are different, when the low limit is crossed, it means that you practically cannot play this game with these settings on a given computer. For example, you shouldn't try to play GTA IV with our test settings with Pentium or Athlon II. The average FPS of about 30 or lower with a corresponding minimum FPS is not what you'd like to have. The same situation is with FarCry2, but it's less catastrophic. Even if we upgrade to Core 2 Duo E7600, we still cannot go beyond 35 FPS in these games. Compare the following: Core 2 Quad Q8200 -- ~49 and ~39 FPS, Phenom II X3 720 -- 52 and 39 FPS correspondingly. Only Core 2 Duo E8200 demonstrates good results, especially considering that it's a low (discontinued) processor from Series E8000, more expensive models will be even faster. But don't forget that they will be more expensive. So it's a rhetoric question what processor to choose for modern games in case of a limited budget. Older games will run fine even with a Pentium or Celeron.

Conclusions



We've deliberately published no comments on results of our old processors -- everything is clear. Indeed, Core 2 Duo E6600 used to be a cherished dream of many users, but now it can compete only with Pentium. However, it's still competitive, even though it had been designed three years ago. So it apparently makes no sense to upgrade it to a modern dual-core model. If you want to improve performance of your computer (that is you really need it), you shouldn't be too economic.

Especially as modern dual-core processors are not always justified, even if you build a computer from scratch (that is when you don't have a computer, or it's too old -- for example, based on Pentium 4 or similar). It often makes no sense to install a quad-core processor, but it's not the worst choice, if it has a similar price. At least you won't regret it, when you run GTA IV or some other new game. Such applications need many cores, because they are products of bad optimizations, not because programmers use multithreading well, but it does not really matter. The question "why so slow" is not interesting to many users -- most of them just want to solve their tasks, without charging memory with details (besides, you won't get your money back, even if you know what component to blame).

This is all true, if we speak of purchases. From the point of view of pure theory we simply run into the fact that optimizations for several processor cores are still imperfect. That's why performance gains from the number of cores increased to three or four are not always noticeable. Or such performance gains can be reached by increasing the clock rate, and processors with fewer cores do it better. From this point of view, Core 2 Duo E8000 processors could have been the best choice for a usual home computer. If it weren't for their very high prices. Full-speed 6M cache is great for performance, but it's very expensive. So expensive that two dice with 3M each can be cheaper. And they will be faster even with a lower clock rate. So our main recommendation used to sound like this: "Buy a quad-core processor, if you know what you need it for, and buy a dual-core processor in all other cases." But now everything has changed. "Buy a dual-core processor, when you are sure it will suit your programs, or buy a quad-core processor in all other cases." You can also buy a triple-core processor: as we can see, Phenom II X3 720 performs well in our conditions -- it's not as cut-down in clock rate and cache size as Core 2 Quad Q8200, so sometimes it can outperform the latter even in multithreaded applications.

All these "torments of choice" hold true only for one popular segment: 130-200 dollars. Everything is quite simple in the higher segment: it's the reign of medium and top quad-core processors. Core 2 Duo E8500/E8600 used to invade that segment not long ago. However, they have nothing to compete with Core i5 750. So this family won't live long, and the same concerns Core 2 Quad Q9x50. What concerns the price range below $130, there are no triple-core processors there so far (maybe only old models like Phenom X3) -- the entire budget sector is occupied by dual-core models with a rare exception of old single-core models. However, people have to answer other questions in that segment -- it's not about what processor will be faster, but how much money can be saved without much trouble. If you choose Pentium, you'll get good results -- similar to those obtained a couple years ago with Mid- and even High-End processors (no Extreme models, of course). We'll see what we can expect from the overhauled Celeron a tad later. Fortunately, there are still several blind spots in the family of processors for LGA775 for us to explore.


We express our gratitude to Kingston Technology Russia for a contribution to our testbeds.

Write a comment below. No registration needed!


<< Previous page

Article navigation:

Page 1: Introduction, testbeds

Page 2: Tests, part 1

Page 3: Tests, part 2, conclusions



blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook


Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.