Athlon II (only X2 at that time) was designed to compete with processors up to Intel Core 2 Duo. But these processors managed to compete with Pentium and E7000 models. Moreover, top E7000 models and almost all E8000 processors could be assaulted only with "heavy artillery" (Phenom II), based on the expensive (for that market segment) core with large L3 Cache. However, the new cheap multi-core die has rapidly raised Athlons to new heights. Athlon II X3 processors completely "cover" the C2D E7000 in performance (even with a margin). And Athlon II X4 does the same with Core 2 Duo E8000 and even invades the area of sterling quads, having outperformed the Q8200 and barely missing the Q9300. This comparison holds true only when we don't touch prices: if we do, Athlon II X4 will compete with the E7000, not with the E8000, and its lower model can attract users, who have enough money only for Pentium. Well, we can only congratulate AMD, which is going to play by its own rules in the strategic segment (declared two years ago). As the overhauled Pentium for LGA1156 with the new Core i3 and i5 appear only in January 8. Christmas sales will be dominated by AMD. Alas, that's the modern market economy -- victory in the top segment does not at all mean prosperity in other market segments: owners of slim purses should also have their share of adequate offers.
But let's leave business analytics to business analysts and return to technical matters. Our today's subject is processors from Intel, not AMD products. Let's address them from bottom to top.
Pentium fares well against its dual-core counterpart from AMD: the E6300 performs just as the Athlon II X2 250, the new E6500 is naturally faster, and the coming E6600 will be even faster. But there are two problems -- for one, the X2 250 is going to be replaced with a faster X2 255. Besides, we've heard rumors of the X2 260 3.2GHz (why not, if Phenom II processors have mastered 3.4GHz long ago). It's not a disaster, just a nuisance -- the main problem is in Athlon II X3, which is squeezed into the same segment. As we already know, dual-core processors can compete with triple-core processors only in applications with one or two threads. As their number grows, top Athlon II X3 processors can defeat even low quad-core processors, to say nothing of top dual-core models. So Pentium has a limited field of application -- a good and inexpensive processor for all applications, which do not support multithreading (or support it poorly). But it's a very good processor here! Let's hope that the coming Pentium G6950 will be no worse.
What concerns Core 2 Duo E7000, there are no reasons to look in its direction. This used to be a very attractive series, providing a balance between the expensive E8000 and the outdated 65-nm Core 2 Duo. Pentiums were also based on the old core at that time, and they were outscored by the E7000 by all parameters: lower clock rates of their cores and FSB, three times as small cache, etc. Having upgraded to 45-nm, Pentium got a 2MB L2 Cache. Then these processors mastered FSB 1066, and they are still growing their clock rates. What is left for Core 2 Duo E7000? Its cache is bigger by 1MB. And it's faster by 133MHz so far. Clock rates being equal, this extra MB yields too little, as you can see it in test results. But it makes this processor more expensive. What do manufacturers usually do in this case? That's right. However, we shouldn't worry about the E7000 in the nearest future -- they won't leave the market, as long as Core 2 Quad Q9000 are present. But they are getting less and less useful: with the same clock rates, they are noticeably more expensive than Pentium, but they are inconspicuously faster.
It's very difficult to write about the Core 2 Duo E8000 and especially about the top representative of this series - E8600. Pentium is a good inexpensive processor for low-thread applications, but the E8000 is the best choice for the latter. The E8600 will leave the market unconquered in this respect. Moreover, a very high clock rate and large cache (full speed at that) allow this processor to demonstrate an acceptable average result even with our mix of various applications. Even though we use several applications with perfect optimizations for multithreading, the rest of the applications make sure that the E8600 scores no worse than some quad-core processors. However, there is a reverse side as well -- such technical characteristics come at a great cost. The number of applications that can use only two cores is waning. To be more exact, there are a great many of them, but few of them really interest users. And it's no use to accelerate games with old engines: Core 2 Duo E8600 manages to outperform the Pentium E5300 by one and a half in Unreal Tournament 3. So what? Even the latter provides about 90 frames per second. And it costs less than $100, while the E8600 is somewhere beyond the line of good and bad -- you can buy a Core i7-860 which can run any applications instead. On the other hand, it provides impressive performance with one or two threads. Quintessence of the Core 2 architecture and an excellent illustration of the blind alley, where it has gone after so many glorious years.
Core 2 Quad Q9505 produces a strange impression. On the one hand, it's a very useful processor for the remaining users of the LGA775 platform -- it's noticeably cheaper than the Q9550 even after all price drops, and performance differences are insignificant, as we can see. On the other hand, we don't understand why we had to wait for this model for so long. Old times, when the Q9550 was a top product for $500, the only faster product being the extreme QX9650, is quite another matter -- each percent of gain was important in these conditions. But much time has passed since that time -- the QX9650 has split into a faster QX9770 and a (relatively) cheap Q9650, having pushed the Q9550 to the third place (formally even to the fourth). Then the LGA1366 platform appeared to attract those users, who need top performance. The Q9505 appeared only six months later. Moreover, if the initial plans about the LGA1156 hadn't been changed, this processor would have appeared together with Core i5-750. Why do we need this processor at all? It would have been different a year ago. As for now, this model may come in handy only to those users, who already invested into LGA775, now having a good motherboard and enough memory. They may want to raise performance at little cost -- without changing the platform. After all, this is a much smaller market than the segment of ready systems.
The memory issue is also painful, by the way -- LGA775 processors don't need DDR3 memory, it even slows them down. It was fine a year ago -- you didn't have to buy expensive memory. Capacity to work well with cheap memory is a competitive advantage. But the situation has changed now, as the new memory type costs practically just as much. Moreover, to buy DDR2 memory now means to stay with it: it's impossible to use it with LGA1366/LGA1156. Besides, AMD focuses on AM3 instead of AM2+ (especially as all new processors from this company gain from using DDR3 memory). So it's not a surprise that DDR3 sales in most countries have already exceeded those of DDR2. But there is nothing good about it for our contenders either.
That's the current state of affairs for the LGA775 platform. It has changed several times together with its processors. By the way, this is why it's so difficult to speak about this platform as a single whole from the historical point of view: the first processors for this platform simply cannot work with modern motherboards, to say nothing of modern processors with old motherboards. However, if we don't take it into account, we've come a long way with this design. From DDR to DDR3. From single-core processors with the NetBurst architecture to dual- and quad-core processors Core 2. This very platform has to do with the beginnings and bloom of PCIe. This very platform demonstrated top performance for two years. That is it has lots of old achievements. But it's impossible to ride into the future on the achievements of the past. The LGA775 platform has lost its relevance. It still holds positions in the value segment, but only until Intel has saturated the LGA1156 product range with $100-$150 models. Top quad-core processors are necessary only for upgrades -- when you buy a new computer, the other platforms can offer much more benefits for each dollar spent. On the other hand, this platform is not too old to be immediately replaced with something newer. As we can see, its overall performance level is quite good, so if you have already invested into this platform, you can still rip dividends from it.
We express our gratitude to Kingston Technology for the contribution to our testbeds.
Write a comment below. No registration needed!