|Intro, Specifications, and Related Info|
What goes faster than a Core 2 Quad processor at 3GHz? Unfortunately, or fortunately depending on your perspective, nothing from AMD at the moment. Thankfully Intel has a few more "MHz in the can", or so to speak. Actually, it appears they may have a lot more, as you'll see in the pages that follow. Let's not get too far ahead of ourselves though. First, Intel had to lay the platform foundation to support higher performance. After all, a faster engine is no good to you, sitting idle.
The introduction of the Intel P35 "Bearlake" chipset and soon-to-be-released X38 chipset back in May, allowed a platform goose of the Front Side Bus speed to a snappy 1,333MHz and ushered in support for DDR3 memory synchronously, pun intended, at the same speed in concert with the FSB. It also afforded Intel the opportunity to characterize their Core 2 Quad architecture at the same quad-pumped FSB speed--perhaps a lot like many enthusiast overclockers have already been doing for a while now? You can bet on it.
So the answer to our initial question is fairly straight-forward and Intel is answering that definitively today with the launch of the new Core 2 Extreme QX6850 quad-core processor. What goes faster than a Core 2 Quad QX6800? Like a Ferrari or Lamborghini it's going to cost you a few Lira but read on paisano and we'll wind her out and see how she handles.
We've published many articles relating to Intel's Core microarchitecture, Core 2 Duo and Extreme family of processors here at HotHardware.com. For more detail or a refresh on the technologies employed by the Core microarchitecture and Intel's platform as a whole, we suggest taking a look at the following related articles. These articles contain detailed explanations of some of the features common to Intel's legacy products, compatible chipsets, and the new Core 2 Duo and Core 2 Extreme processors:
We cover some specifics regarding Intel's 65nm manufacturing process in our 955XE / i975X evaluation and outline Intel's AMT (Active Management Technology) and IVT (Intel Virtualization Technology), among other things inherent to the Core microarchitecture, in our Core 2 Duo E6700 & Extreme X6800 Evaluation . Our Intel P35 Bearlake and DDR3 launch article will give you the background you need for the new higher front side bus speed and companion memory technology. The other articles listed above will also give you some background as to how the Core 2 has matured, leading up to today.
Beyond the increase in FSB speed, nothing has changed for the Core 2 Extreme QX6850, other than the fact that the chip is based on a new stepping of Intel's quad-core Conroe architecture. Even the TDP rating (Thermal Design Power) for the chip hasn't changed. This new chip will require a motherboard and chipset capable of supporting the new 1,333MHz FSB. Motherboards based on these chipsets, such as Intel Bearlake (P3x, G3x) or NVIDIA's nForce 6, are on the market today, with a healthy offering of new P35-based boards from the likes of Asus, Abit, Gigabyte, MSI and others.
|Vital Signs and Overclocking|
Like all Core 2 Quad CPUs, the QX6850 runs a pair of Conroe dual-core processors under its heat spreader - the only difference is that its stock FSB speed is clocked at 1,333MHz. This processor runs a 9X multiplier and a 333MHz system bus speed that is then "quad-pumped" to its rated FSB of 1,333MHz.
CPU-Z shows a "B" stepping and a die revision of G0. Each dual-core die is equipped with 4MB of cache for a total of 8MB. The QX6850, like the QX6800, is also built on Intel's 65nm fab process. With four cores operating at 3GHz, this cutting-edge die geomtry is required for mitigating power consumption and heat, among other things.
In the above screen shots you can see we've tested the QX6850 under both idle and full-load conditions. However, we disabled Intel's Advanced Speedstep technology in the BIOS of the motherboard, to show you temperatures at idle but also at full clock speed. At 41°C and at a full 3GHz, the QX6850 is cool and comfortable. At full load, temperatures scale up to 56°C but that's pretty impressive actually, since there are four CPU cores under the hood running at 3GHz. To think Intels old Prescott Pentium 4 single-core CPUs reach 60°C under load... How far we have come, Intel.
Finally, with Intel's Speedstep technology enabled, at idle the core drops its voltage to 1.1V and clock speed to 2GHz. This allows the core even lower temperatures, reported here at 21°C.
And of course, since Intel has come this far, we just had to see if we could take things up a notch or two more. Our overclocking efforts are pictured below. Needless to say, we were rather impressed.
It took a small voltage bump to 1.45V in the BIOS to hit 3.73GHz. To achieve this, we raised the FSB to 466MHz and dropped the multiplier to X8. Once we hit our highest, stable clock speed, we were surprised how stable things were even with a relatively low cost standard HSF cooler. We used an Artic Cooling Freezer 7 to achieve these results. Temperatures under full load were getting a little toasty though, at 69°C. With more exotic cooling, or water, the QX6850 most likely would top out even faster. That said, even with a stock retail heatsink and fan assembly, 3.4 - 3.6GHz is most likely a walk in the park for many QX6850 chips. Is Intel holding back? Probably, because they can.
|Test Systems and SiSoft SANDRA|
How we configured our test systems: When configuring our test systems for this article, we first entered their respective system BIOS and set each board to its "Optimized" or "High performance Defaults". We then saved the settings, re-entered the BIOS and set memory timings for DDR2-800 with 4,4,4,12 timings. The hard drives were then formatted, and Windows XP Professional (SP2) was installed. When the Windows installation was complete, we installed the drivers necessary for our components, and removed Windows Messenger from the system. Auto-Updating and System Restore were then disabled and we set up a 1024MB permanent page file on the same partition as the Windows installation. Lastly, we set Windows XP's Visual Effects to "best performance," installed all of our benchmarking software, defragged the hard drives, and ran all of the tests .
We began our testing with SiSoftware's SANDRA XI, the System ANalyzer, Diagnostic and Reporting Assistant. We ran six of the built-in subsystem tests that partially comprise the SANDRA XI suite with the Core 2 Extreme QX6850 ( CPU Arithmetic, Multimedia, Multi-Core Efficiency, Memory, Cache, and Memory Latency) . All of the scores reported below were taken with the processor running at its default clock speed of 3.0GHz.
Not surprisingly, the new Core 2 Extreme QX6850 smoked all reference scores from processors listed in SANDRA's database. The version of SANDRA XI we were running had a QX6700 as its fastest desktop chip, for reference, though a QX6800 would undoubtedly report higher bandwidth. Regardless, you can see that AMD's fastest single chip processor configuration can't even hold a candle to the QX6850 in either the processor Arithmetic or Multi-Media tests. An interesting reference point would perhaps be that the Core 2 Duo Extreme X6800 offers basically half the performance of the QX6850 quad-core in these tests, which stands to reason obviously, with half the number of CPU cores available.
In terms of memory bandwidth, SANDRA reports a nice boost from the chip's 1,333MHz FSB. The CPU and DDR2-800 memory combination drives about 6.4GB/sec of bandwidth.
|PCMark05: CPU and Memory|
For our next round of synthetic benchmarks, we ran the CPU and Memory performance modules built into Futuremark's PCMark05 suite. The following tests are synthetic benchmarks designed to show relative performance metrics but may or may not equate to "real-world" performance.
The CPU performance module test shows the Core 2 Extreme QX6850 ahead of its 1066MHz FSB-based quad-core brethren, the QX6800 on the 975x chipset, by about 7% - that lead is much smaller when testing the QX6800 on the P35 chipset, however. Even though the two processors have nearly the same core frequency (2.93GHz for the QX6800 and an even 3GHz for the QX6850), the additional front side bus bandwidth of the QX6850 in conjunction with the newer chipset give it a more significant advantage in this test.
"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing." - Courtesy FutureMark Corp.
The inherent nature of Intel's current quad-core architecture requires all cores on the chip to share a single system bus. As a result, we actually see memory bandwidth on in the PCMark Memory performance module scale in favor slightly of the dual architectures, when you compare scores between the QX6800 and X6800, for example. However, the new QX6850's extra FSB bandwidth at 1,333MHz compensates for this nicely. As a result, the QX6850 is able to boast a 4% advantage over a similarly clocked X6800 dual-core processor and a roughly a 8-10% advantage over a QX6800 quad-core CPU.
|Office XP and Photoshop|
PC World Magazine's Worldbench 5.0 is a Business and Professional application benchmark. The tests consist of a number of performance modules that each utilize one, or a group of popular desktop applications to gauge performance.
Below we have the results from WB 5.0's Office XP SP2 and Photoshop 7 performance modules, recorded in seconds. Lower times indicate better performance here, so the shorter the bar the better.
Though this is the fastest score we've ever recorded in the Office XP test module for World Bench, the QX6850 is afforded less than a 2% gain over the QX6800 with a 6 second overall advantage. To be honest, Office applications really don't stress any of the processors we tested here.
The WB5 Photoshop 7 test shows a similar scale but the scores ended up even tighter. In short, when it comes to Photoshop 7, a 3GHz dual-core Core 2 chip is nearly as fast as a 3GHz Core 2 Quad. And the 1,333MHz FSB of the new QX6850 didn't offer any tangible performance increase. In the future, we hope to migrate to World Bench 6, though it is still in the beta stage. Newer versions of Photoshop with better multi-threading capability are incorporated into that release, so we hope to be able to exercise these new high-end processors more then, once the benchmark suite is finalized.
|LAME MT and Sony Vegas|
In our custom LAME MT MP3 encoding test, we convert a large WAV file to the MP3 format, which is a very popular scenario that many end users work with on a day-to-day basis to provide portability and storage of their digital audio content. LAME is an open-source mid to high bit-rate and VBR (variable bit rate) MP3 audio encoder that is used widely around the world in lots of third party applications.
In this test, we created our own 223MB WAV file (a hallucinogenically-induced Grateful Dead jam) and converted it to the MP3 format using the multi-thread capable LAME MT application in single and multi-thread modes. Processing times are recorded below. Once again, shorter times equate to better performance.
Though this version of the LAME encoder is multi-threaded, the application only supports two threads. As a result, all dual and quad-core processors at the same core frequency perform similarly. Again we do see a slight advantage to the dual-core X6800 however, where the system bus is only shared by two cores. Even the QX6850's 1,333MHz FSB speed doesn't make up for this in LAME testing.
Sony's Vegas DV editing software is heavily multi-threaded as it processes and mixes both audio and video streams. This is a new breed of digital video editing software that takes full advantage of current dual and multi-core processor architectures.
With Sony Vegas, the quad-core CPUs in our test reined supreme, offering almost 2X the performance of their dual-core counterparts. In fact, the Core 2 Extreme QX6850 is more than twice as fast as the Core 2 Duo E6750 in Sony Vegas workload processing. What's much more interesting though is that the QX6850 pulled out the win over the 1066MHz FSB / 975X driven QX6800 by a 13% margin of victory. This performance variance is sizeable to be sure and is the sort of gain that would be easily perceptible to the end user within this application. That margin of victory decreases to 15 seconds with the QX6800 running on the P35 chipset, however, which equates to about an 8% lead.
Next we ran Kribibench v1.1, a 3D rendering benchmark produced by Adept Development. Kribibench is an SSE aware software renderer in which a 3D model is rendered and animated by the host CPU rather than the GPU, then the average frame rate is reported.
We used two of the included models with this benchmark: a "Sponge Explode" model consisting of over 19.2 million polygons and the test suite's "Ultra" model that is comprised of over 16 billion polys.
In both the medium complexity polygon count and high complexity/high polygon count tests, the QX6850 showed only a minor gain versus the QX6800. Clearly the quad-core CPUs took a huge lead in this test at nearly 2X the fastest dual-core score we recorded. However, the QX6850 offered a less than 1% performance advantage in this test versus the QX6800. The scores here are indicative of the 70MHz clock speed advantage the 3GHz QX6850 has over the 2.93GHz QX6800. Front Side Bus speed doesn't even enter into the equation.
|Cinebench R9.5 and 3DMark06|
Cinebench 9.5 is an OpenGL 3D rendering performance test based on Cinema 4D. Cinema 4D from Maxon is a 3D rendering and animation tool suite used by 3D animation houses and producers like Sony Animation and many others. It's very demanding of system processor resources and is an excellent gauge of pure computational throughput.
This is a multi-threaded, multi-processor aware benchmark that renders a single 3D scene and tracks the length of the entire process. The time it took each test system to render the entire scene is represented in the graph below, listed in seconds.
In our multi-threaded test, the quad-core processor enabled systems both took the lead by a substantial margin. The QX6850 and QX6800 were both over 40% faster than the nearest dual-core driven score. Obviously the QX6850's 267MHz faster FSB speed afforded it no advantage whatsoever over the QX6800. In our single-threaded tests however, multi-core bus contention caused the QX6800 to clock in a second slower than the dual core X6800. The QX6850's higher FSB bandwidth did actually offer a bit more performance in this situation however and the new processor was able to edge out the Core 2 Extreme X6800 dual-core in this test.
3DMark06's built-in CPU test is a multi-threaded DirectX gaming metric that's useful for comparing relative performance between similarly equipped systems. This test consists of two different 3D scenes that are processed with a software renderer that is dependent on the host CPU's performance. Calculations that are normally reserved for your 3D accelerator are instead sent to the CPU for processing and rendering. The frame-rate generated in each test is used to determine the final score.
Nearly twice as fast as the Core 2 Extreme X6800 dual-core CPU and about 13% faster than the Core 2 Extreme QX6800 quad-core / 975x combo, the new QX6850 once again puts up the fastest score we've recorded to date in 3DMark06's CPU test.
|Quake 4 and F.E.A.R.|
For our last set of game tests, we moved on to some in-game benchmarking with Quake 4 and F.E.A.R. When testing processors and motherboards with Q4 or F.E.A.R, we drop the screen resolution and reduce all of the in-game graphical options to their minimum values, to isolate CPU and memory performance as much as possible. However, the in-game effects and the level of detail for processing workloads such as physics calculations and particle systems, are left at their maximum values, since these actually do place some load on the CPU rather than GPU.
The Core 2 Extreme QX6850 just sneaks past the Core 2 Extreme X6800, once again proving itself to be the fastest desktop processor on the planet currently for gaming requirements. In Quake 4 we see the QX6800 quad-core / 975x combo actually drops in behind the X6800 dual-core chip, due once again in part to bus bandwidth limitations and the fact that Quake 4 doesn't efficiently use all processing resources, though the engine does somewhat make use of multi-threading. The new Core 2 Extreme 6850 does offer a bit more performance however and it shows itself to be about 3%-5% faster than the X6800 in Q4.
In our F.E.A.R. tests on the other hand, the new QX6850 bests the QX6800 quad-core for a first place finish with a 7.5% gain. The Core 2 Extreme QX6850 was also 34% faster than the Core 2 Extreme X6800 dual-core. All told, the new QX6850 from Intel serves up the best gaming performance we've recorded to date.
Before we bing this analysis to a close we wanted to give you an idea of how much power each of the system configurations we tested used, while idling and while under a workload.
Please keep in mind that we are looking at total system power consumption here at the electrical outlet, not just the power being drawn by the processors alone. In this test, we're showing you a ramp-up of power from idle on the desktop to 100% processor load. We tested with a combination of Cinebench 9.5 and SANDRA XI running on the CPU.
We also should remind you that our QX6850 and E6750 power consumption readings were taken on an Asus P5K3 motherboard, versus the scores taken on the 975X-based board for the other Intel processors in the test. Specifically, the P35-based Asus P5K3 motherboard we used to test the QX6850, is loaded with extra peripherals, bells and whistles; so natually total system power consumption is going to be higher for both of these data points.
A more interesting variance we'd like to call your attention to rather, is the difference between idle system power consumption and full-load system power consumption. Specifically, we see the Core 2 Extreme QX6850 ramp up an increase in power consumption on the order of 102 Watts under full load. Conversely, the Core 2 Extreme QX6800 ramps up 108 Watts over its idle power consumption. In short, the new stepping and manufacturing process optimizations that Intel has brought to the Core 2 quad-core architecture over the last few months, has brought forth tangible power consumption reduction as well, even with a higher FSB speed driving the chip.
|Our Summary and Conclusion|
With today's launch Intel is also introducing a few other variants of their Core 2 desktop and workstation line-up with 1,333MHz system bus speeds. They're also ratcheting down the price of the Core 2 Extreme QX6700, now branding it as a standard or "mainstream" offering dubbed the Core 2 Quad Q6700, for a significantly more palatable $530 price tag. As always however, the new flagship processor, the QX6850, will retail for $999. As usual, only those with money to burn need apply. Here's the line-up of new Intel Core 2 processors:
The introduction of the Core 2 Extreme QX6850 and these other 1,333MHz FSB driven chips is as much about a platform migration as anything else. Without question, as more and more applications make heavier use of multi-threading techniques, Intel's shared bus architecture will have to be pushed higher and higher to afford more off-chip system and memory bandwidth for these multi-core CPUs. There will come a day where Intel will finally go the way of a serial interface for their X86 processor offering but today another 267MHz or so of bus bandwidth will keep their processor line in the forefront while AMD is still running to play catch-up.