Today we are going to review another top performer from Intel that marks a new milestone in "bus-making" - a 400 MHz CPU bus with throughput equivalent to 1600 MHz. (The previous record was 333 MHz, equivalent to 1333 MHz.) Besides, now we can have a look at the complete series of future Intel's processors based on the newer core:
Note that E8190 though seemingly identical to the E8200, according to the table above, actually differs by lacking the support of virtualization technology.
The Core 2 Extreme QX9770 processor, which we review today, is based on the Yorkfield core of the Penryn family, like the previously examined Core 2 Extreme QX9650. However unlike the latter, the QX9770 can't work (officially, at least) with any of the officially announced chipsets, including the Intel X38 Express, since it requires the 1600 MHz FSB. For this reason we had to test it on a sample of the Gigabyte GA-X48-DQ6 motherboard based on the X48 Express chipset, instead of our usual ASUS P5B Deluxe LGA775 board. Note that according to some preliminary information, the X48 will differ from the X38 only by the 1600 MHz FSB support. Like the X38, the X48 supports both DDR2 and DDR3. Our sample had only DDR2 sockets, but it bore a proud "DDRII-1066" mark besides the "DDRII-800". DDR2-1066 memory wouldn't be justified for a 1600 MHz bus though, since they'd work asynchronously. But DDR2-800, in its turn, is nearly ideal, because CPU and memory bus clocks are identical, and the throughput of DDR2-800 dual-channel controller is equal to the CPU bus throughput. Therefore we conducted our tests in this mode exactly.
Also note that the new 1600 MHz bus for the first time provides at least a theoretical opportunity for at least DDR2-800 memory to uncover the full potential in the dual-channel mode. As you might remember, the 1333 MHz bus throughput couldn't do that as described here.
Hardware and software
Attention: To test the Core 2 Extreme QX9770 CPU we used a sample of the Gigabyte GA-X48-DQ6 board based on the Intel X48 Express chipset supporting 1600 MHz FSB.
* - "2 x ..." means per core;
Essential foreword to charts
Our test method has two peculiarities of data representation: (1) all data types are reduced to one - integer relative score (performance of a given processor relative to that of Intel Core 2 Duo E4300, given its performance is 100 points), and (2) detailed results are published in this Microsoft Excel table, while the article contains only summary charts by benchmark classes. We will nevertheless focus your attention on detailed results, when needed.
3D modelling suites
The performance boost is nearly proportional to the clock rate, allowance made for measurements precision. A completely expected result for this application class. We're just happy that the scalability is so ideal.
Unfortunately, it's far from ideal in this one. For the given clock rate increase we get up to the twice as less performance boost. The QX9770 couldn't break this "tradition" as well.
Digital photo processing
The QX9770 showed the nearly 10% performance boost at the 6% clock rate increase! It's one of those rare occurrences, when a benefit of a new processor bus is obvious in a real application, not some synthetic benchmarks.
The performance of QX9770 increased proportionally to the clock rate. No surprises.
The matter we've already mentioned this article negates all the architectural and especially L2 cache advantages of Intel's modern quad-core processors.
The result is a mystery, since the performance boost of the QX9770 relative to the QX9650 exceeds the corresponding clock rate increase. In the near future we'll take a closer look at how this benchmark works with Penryn processors. And now we just have to suppose that CPU RM, previously rather indifferent to CPU bus throughput, suddenly changed its attitude towards it exactly with Penryn processors.
This test, traditionally very critical to L2 cache throughput and volume, clearly indicates that even a very large L2 doesn't make high throughput less important.
As far back as in the previous review we suggested that a throughput increase might benefit results of this test. And now we see an obvious proof of it.
An old group of tests that has nearly lost its importance due to high predictability of results. No comments.
Another rare occurrence, when a CPU bus throughput increase has a purely positive effect. It's rather strange, since usually video codecs are not that demanding to throughput. On the other hand, processor clock rate are increasing, and the higher the core clock is, the faster data must be delivered to it.
As we have already mentioned in the QX9650 review, despite the formally higher score, both QX9770 and QX9650 are too good for today games. These products score in the Low/Medium Quality modes as well, but we doubt that someone with a PC based on a QX9770/QX9650 would prefer lower quality. And what if we calculate total game score based on the highest quality settings at, say, 1280x1024? Here, have a look:
As you can see, the impressive 44% difference between the slowest and the fastest processor was reduced to the more humble 23%. Though it's still impressive, we have to admit.
The total score impresses with the value itself. But in other respects it's extremely predictable: the QX9770 outperformed the QX9650 nearly strictly according to the clock rate difference. The scalability is almost ideal, that's all we can say.
Supposed power consumption
This all looks a bit paradoxical. While showing the expected slight increase in power consumption at 100% load comparing to the QX9650, the QX9770, at the same time, consumes significantly less when idle, according to the chart. Of course, it doesn't seem true. Most likely, the reason is another motherboard we had to test the QX9770 with. Again, this section is named "Supposed power consumption". Due to the purely technical reasons we're limited to measuring consumption of board's VRM, not the processor itself.
After the announcement of the Penryn series it's been raining Intel's top-end processors. Having just seen the already perfectly fast Core 2 Extreme QX9650, we are being surprised with an even higher performing QX9770. This seems like with these two announcements Intel is trying to nail down the lid to the coffin of you-know-who. And also deliver the message like: "Noone is going to catch up with us in the segment of desktop top performers. Let the rivals fight for the Low- and Middle-End. But the desktop High-End is ours." And, you know, it's very convincing at the moment...
Testbed memory modules provided by
Corsair Memory Russia
Stanislav Garmatyuk (firstname.lastname@example.org)
November 26, 2007
Write a comment below. No registration needed!