discussion Intel Core i5-655K & Core i7-875K Unlocked Processors Review

[MAC]

Gaming Benchmarks: Crysis / Far Cry 2

Gaming Benchmarks: Crysis / Far Cry 2

Crysis

Crysis v1.21
Resolution: 1680x1050
Anti Aliasing: 0
Quality Settings: High
Global Settings: DX10 / 64-Bit

Test 1: Ice benchmark_CPU2 demo
Comparison: FPS (Frames per Second)

Still one of the most hardware intensive game on the market today, Crysis has been chosen for its obvious ability to be able to showcase the differences between platforms and to showcase just how far one will need to go in the quest for maximum performance. The game also features the renowned CryEngine, the power behind the incredible graphics, which is expected to be foundation of future titles.

Far Cry 2

Far Cry 2 1.02
Resolution: 1680x1050
Anti Aliasing: 0
Quality Settings: Very High
Global Settings: DX10 Enabled

Test 1: Ranch Long Demo
Comparison: FPS (Frames per Second)

Far Cry 2 is the hot new new first-person shooter from Ubisoft's Montreal studio, and the first game to utilize the new visually stunning Dunia Engine, which will undoubtedly be used by numerous future games. Using the included Benchmarking Tool, we ran the Long Ranch demo in DX10 mode at 1680x1050 with all settings set to very high.

 

[MAC]

Gaming Benchmarks: Left 4 Dead / Particle Simulation

Gaming Benchmarks: Left 4 Dead / Particle Simulation

Left 4 Dead

Left 4 Dead (Latest Update)
Resolution: 1680x1050
Filtering: 4X MSAA / Anisotropic 8X
Graphic Settings: High
Shader Detail: Very High
Test 1: HWC Custom Timedemo
Comparison: FPS (Frames per Second)

Left 4 Dead is the latest disorienting, fast-paced zombie apocalypse mega-hit from Valve. L4D uses the latest version of the Source engine with enhancements such as multi-core processor support and physics-based animation. We test here at 1680x1050 with in-game details set to their highest levels, with MSAA 4X and AA 8X. For benching, we used a pre-recorded 20 minute timedemo taken on the No Mercy campaign during The Apartments mission.

Although the Source engine scales very well with multi-core processors, we discovered that the scaling seems to stop at eight threads, hence why the Core i7-980X doesn't take the lead. We actually found that it was faster in this benchmark when we disabled two cores and ran it as a quad-core/eight-thread processor.

Valve Particle Simulation Benchmark

Valve Particle Simulation Benchmark
Default
Comparison: Particle Performance Metric

Originally intended to demonstrate new processing effects added to Half Life 2: Episode 2 and future projects, the particle benchmark condenses what can be found throughout HL2:EP2 and combines it all into one small but deadly package. This test does not symbolize the performance scale for just Episode Two exclusively, but also for many other games and applications that utilize multi-core processing and particle effects. As you will see the benchmark does not score in FPS but rather in its own "Particle Performance Metric", which is useful for direct CPU comparisons.

 

[MAC]

Gaming Benchmarks: Street Fighter 4 / World in Conflict

Gaming Benchmarks: Street Fighter 4 / World in Conflict

Street Fighter 4

Street Fighter 4 Demo
Resolution: 1680x1050
Anti-Aliasing: 0X
Graphic Settings: High
Test 1: Built-in Timedemo
Comparison: FPS (Frames per Second)

Street Fighter IV is a 2008 arcade game produced by famous developer Capcom, that has finally been released on the PC platform. This game has not been 'ported' since the Street Fighter IV arcade machines actually have PC internals, with circa 2005 components. As a result, the version of the game released on the PC is considered the definitive version. With a multi-threaded engine and an astounding hybrid 2D/3D graphics style, this game is sure to please all fans of the Street Fighter series.

Can anyone say "GPU bottleneck" ?

World in Conflict

World in Conflict v1.010
Resolution: 1680x1050
Anti-Aliasing: 4X
Anisotropic Filtering: 4X
Graphic Settings: Very High
Test 1: Built-in Benchmark
Comparison: FPS (Frames per Second)

One of the most visually stunning real-time strategy games in recent history, World in Conflict can really push systems to the brink, which is what we attempt by running the game in DirectX 10 mode at 1680x1050 with all settings maxed out. For this test we used the in-game benchmarking tool.

 

[MAC]

Overclocking Results - Air Cooling

Overclocking Results - Air Cooling

<center>

</center><p style="text-align: justify;">Our overclocking is going to be split up into three different sections, Air Cooling, Phase Change Cooling, and LN2 Cooling. This section will consist of testing 24/7 stable air clocking results. Also, since we have used different hardware from the rest of the review, here is a complete list of the hardware in play for the overclocking sections.</p><center><table border="0" bgcolor="#666666" cellpadding="5" cellspacing="1" width="735px"><tr><td colspan="4"><b><font color="#ffffff">Test Platform:</font></b></td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Motherboard:</b></td><td align="left" bgcolor="#ececec" width="75%">EVGA P55 Classified 200</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Processors:</b></td><td align="left" bgcolor="#ececec" width="75%">Intel i5 655K<br>Intel i7 875K</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Processor Cooling:</b></td><td align="left" bgcolor="#ececec" width="75%">Thermalright Ultra-120 eXtreme CU w/Scythe Ultra Kaze 120MM 2000RPM<br>Chilly1 SS Phase Change<br>KingpinCooling.com F1EE w/LN2</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Thermal Paste:</b></td><td align="left" bgcolor="#ececec" width="75%">Arctic Silver Ceramique</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Memory:</b></td><td align="left" bgcolor="#ececec" width="75%">Corsair Dominator GTX2 PC3-18000 CL8 (CMGTX2)</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Video Card:</b></td><td align="left" bgcolor="#ececec" width="75%">XFX XXX HD5870 (CCC 10.2)</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Video Card Cooling:</b></td><td align="left" bgcolor="#ececec" width="75%">Thermalright HR-03 Rev.A</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Power Supply:</b></td><td align="left" bgcolor="#ececec" width="75%">Corsair HX1000W</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Additional Fans:</b></td><td align="left" bgcolor="#ececec" width="75%">Scythe Ultra Kaze 120MM 2000RPM 87.6CFM (DFS123812L-2000)</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%"><b>Hard Drive:</b></td><td align="left" bgcolor="#ececec" width="75%">Seagate 7200.9 80GB SATAII 8MB cache</td></tr></table></center><p style="text-align: justify;">With overclocking there are always risks involved when increasing voltages. By using the voltages we have here today, we are not validating the safety of using such voltages for daily use. We have self imposed limits of 1.4125v as set in the BIOS for vCORE which results in less than 1.4v under load. This should be a safe value for extended use, but still provide ample overclocking headroom. Let's have a look at what our two processors were able to accomplish for maximum Base Clock overclocks to start.</p>Click for full size screen shot...<center>

</center><p style="text-align: justify;">Obviously our 655K had much better luck with BCLK on air. The Lynnfield result is typical of other Lynnfield results we have had with this motherboard so nothing is out of the ordinary there. The Clarkdale i5 655K processor was about average from what we have typically seen with the first stepping of Clarkdale processors. Anything over 230BCLK is a bit of a bonus in our opinions, but it isn't uncommon to see 240BCLK stable for 24/7 use. Every processor is different though and this is just a single sample, so these numbers should be considered as only a single sample out of an endless pool of results.

We will now take a look at the overall highest CPU frequency we could stabilize with each processor.</p>Click for full size screen shot...<center>

</center><p style="text-align: justify;">Our i5 655K sample seems to be able to reach a solid 4.5GHz+ without too much effort, and staying under 70C at load to boot. These little dual cores, even with hyper-threading enabled, are amazing overclockers on air. It is impossible to say whether the new K0 stepping of the Clarkdale processor is going to consistently clock to 4.5GHz on air with relatively low voltage or not based on a single sample, but it would seem that the K0 stepping won't be any worse than what we have come to expect from Clarkdale.

The Intel i7 875K on the other hand was a bit underwhelming in our air overclocking. This is a direct result of temperature managing to only eek out a 4.1GHz overclock with all four cores and Hyper-Threading enabled. Even if we were willing to go a little higher in vCORE it wouldn't have done any good as temperatures seemed to be what introduced instability during LinX testing. Just as we can't call all i5 655K processors good overclockers based on our single sample, we can't call all i7 875K processors bad based on a single sample. There will be good and bad batches of each processor so you will want to keep your eyes peeled for end user results in forums.</p>

 

[MAC]

Overclocking Results - Phase Change

Overclocking Results - Phase Change

<center>

</center><p style="text-align: justify;">The first processor we will investigate under sub-zero temperatures from a single stage Chilly 1 phase unit is the Intel i5 655K. For those unfamiliar with phase change cooling, the single stage phase unit we are using today consists of a single compressor that compresses a gas into liquid form. During this process, the temperature of the gas is cooled to sub-zero temperatures due to the compression of the particles. This sub-zero liquid is then fed through an evaporator that comes into contact with the CPU...much like a water block would. Instead of water, however, the phase unit cools the processor with a liquid at about -35C at its coldest for this unit. The liquid is then boiled off into a gas and sent back to the compressor. So at the coldest point, the processor is running at -30C, and in some of the screen shots like wPrime and Vantage, the processor is running as warm as +20C.

The setup is the same as it was for the air cooling with the small exception of using an 8600GTS for some of the 2D benchmarks as shown in the photo above. Let's have a look at some of the results from the phase change cooling on this i5 655K processor.</p>Click for full size screen shot...<center>

<br>

<br>

</center><p style="text-align: justify;">For the SuperPi and 3DMark 05 results, we ended up running the same CPU frequency for all three benchmarks. We could have run 1M a little bit higher in theory, but we would have had to use a different combination of CPU multiplier and BCLK. This actually resulted in lower results, even with slightly higher CPU clocks because the BCLK was lower, and the QPI frequency was also lower. Clarkdale processor performance at these CPU clocks hinges on one thing and one thing only...QPI frequency. Without high QPI, performance suffers drastically. As it was, benching 2D single or dual core applications was easily accomplished at a solid 5.1GHz.

Moving on to Vantage and wPrime where both cores and Hyper-Threading were enabled, we still managed to crack the 5GHz mark getting both of those benches ran at 5080MHz. So far the chip looks decent, but nothing special as we have had i5 660 and i5 670 processors bench higher on this same setup. It will all depend on how this chip handles cold to see if it is a real sub-zero benching winner.

Next up is the 4 core/8 thread Intel i7 875K. The reason I mention the extra cores and threads that the 875K has is because these will generate a lot of heat. Coupled with the fact that the i7 875K is a 45nm part, the heat load on the phase is much greater...thus will result in higher temperatures under load. Let's take a look at the same benchmark suite as we ran above for the 655K.</p>Click for full size screen shot...<center>

<br>

<br>

</center><p style="text-align: justify;">What is interesting here is that with the same cooling, the cheaper, smaller, and lighter I5 655K out-performs the i7 875K in the single threaded 2D benchmarks. This isn't really a fair comparison, but for a bencher on a budget, in a Cooling VS Cooling ratio, the Clarkdale processor is far superior in many benchmarks to the more expensive i7 lineup.

As mentioned, with the increased heat load, the disparity between single threaded benchmark clocks and multi threaded benchmarks is much greater with the i7 due to the heat output. With 2D we could bench the 875K up over 4.7GHz, but in multi threaded benches like Vantage and wPrime, we could only manage 4.4GHz due to the heat output of the processor. When we move to LN2, the multi threaded benches will get the biggest boost.</p>

 

[MAC]

Overclocking Results - Liquid Nitrogen (LN2)

Overclocking Results - Liquid Nitrogen (LN2)

<center>

</center><p style="text-align: justify;">Moving over to colder temperatures will have the biggest impact on the i5 655K Clarkdale processor. These chips scale with cold, like AMD Phenom II processors. Even without adding voltage, simply lowering CPU temperatures increases clocks. This isn't the case with Lynnfield so the i7 875K might not see much gains at all if the colder temps don't allow for higher vCORE to be used. We'll start with the fun chip for this reason, the i5 655K.</p>Click for full size screen shot...<center>

<br>

</center><p style="text-align: justify;">I won't really discuss the SuperPi results much, they are what they are. But they do indicate what I mentioned earlier about living and dying with QPI on Clarkdale for performance. You see, even with an unlocked CPU multiplier, we weren't able to maximize the CPU clocks and performance. This is illustrated in the two 3DMark 05 results I have directly above.

On the left we have a higher score, with less CPU clocks then on the right...but notice the QPI Link clock. Because of the higher BCLK and lower CPU multi, the score on the left was run at higher QPI (also memory clocks but that is less significant). Even with 115MHz higher CPU clocks, the higher QPI result actually out performs the higher CPU frequency. This is also true for 32M so while we could have run higher CPU clocks in SuperPi...the high QPI clocks perform better, so that is what we have posted. Obviously an unlocked CPU multi helps find the magic combination, but with a locked multi i5 660 processor, we have managed to maximize CPU/QPI/RAM clocks for the ultimate in performance...so an un-locked multi isn't the end all say all.</p>Click for full size screen shot...<center>

</center><p style="text-align: justify;">Moving to LN2 didn't help Vantage much as the CPU score was really only affected as the CPU isn't the bottleneck for a single HD5870, even for an i5 655K. wPrime on the other hand got a very nice boost from the massive boost in clocks. wPrime was actually ran at -180C with the pot full of LN2. Under load, this chip showed no cold bug, even with high QPI. The SuperPi results were ran around -160C before the cold bug would show up.

As impressive as these little chips are, this 655K wasn't particularly impressive for us. We have seen much higher clocking Clarkdale processors, but like always...this is just a single sample. Retail K0 stepping Clarkdale’s could be interesting little monsters in the right hands. Now on to our disappointing 875K.</p>Click for full size screen shot...<center>

</center><p style="text-align: justify;">That is all she wrote folks. As we almost suspected, we couldn't get higher clocks out of this 875K dropping temps from -30C to -100C because the chip stopped utilizing higher volts. We thought we ran into this with the phase setup but just assumed it was the temps, but even with colder temps, the chip simply doesn't like much over 1.50v vCORE. We didn't have much LN2 left by the time we realized this so we simply put what we had into wPrime and got a solid 4.7GHz run done.

Like we opened with when we got to these overclocking sections, don't put all your eggs in one basket. The results shown here are from a single sample of each processor. A judgment cannot be made about the overclocking capabilities of these new chips based on a single sample. Heck, even if we were sent a dozen chips, we still couldn't even begin to draw conclusions on the overclockability of these new offerings from Intel. Anyone that likes to make general statements about overclocking is someone that doesn't know what they are talking about. What we can say is that unlock CPU multipliers definitely increase options for overclocking whether at the air, phase, or LN2 level, and that is what both the i7 875K and 85 655K offer. But as we have displayed with the i5 655K LN2 results, higher CPU frequencies aren't always equal to higher performance. Having an unlocked multi definitely makes it easier to find a processors performance maximum.</p>

 

[MAC]

Conclusion

Conclusion

Frankly, we can definitely thank AMD for the existence of the Core i7-875K. In our eyes, this is Intel's surprising counter to the Phenom II X6 1090T Black Edition. They could have just dropped the price of the i7-870 to $350 and lured away some potential 1090T buyers, but instead they went straight for the enthusiast market and offered the Holy Grail that is freely selectable CPU multipliers. This is a highly unusual, but exciting move on Intel's part.

Despite having two less cores, the i7-875K is faster than the X6 1090T in nearly every application, even highly multi-threaded ones thanks to Hyper-Threading. It is about $50 more expensive than the flagship Phenom II X6, but that is worthwhile premium when you consider the large performance advantage that the i7-875K has in lightly-threaded workloads. Whether somebody should pick an i7-875K over an i7-920/930 really depends on whether they envision ever upgrading to a six-core Gulftown processor, since those are exclusive to the LGA1366 platform for the foreseeable future. You can definitely have more fun and extract higher performance from an i7-875K right now, but future prospects seem limited. For though us, a $342 multiplier-unlocked Core i7 LGA1156 chip is still a winner, no doubt about it.

We are less enthused with the Core i5-655K though. We would have liked to see Intel release a similar (if lower clocked) chip in the $150 range. Such a processor could at least be considered a possible contender against AMD's extremely popular Phenom II X2 555 Black Edition, which retails for $99 and is also multiplier-unlocked. Yes, the Core i5-655K is significantly faster across the board than the Phenom II X2, but the X2 555 C3 revision processors are certainly no slouch in the overclocking department, and there is always the possibility of unlocking their two additional cores.

As we demonstrated in our Core i3-540 review, it's unbelievably easy to get very high clock speeds from Clarkdale-based chips, and the base clock (BCLK) is not usually the limiting factor unless sub-zero cooling is being used, so the unlocked multipliers don't really come into play much. We suspect that only extreme overclockers are going to find the $216 i5-655K attractive, which it is when compared to the $284 i5-670. For everyone else, the $195 Core i5-750 and $200 Phenom II X6 1055T provide much better Bang for the Buck.

Now our particular samples weren't exactly that great when we actually got down to overclocking. We certainly could have achieved similar results with default locked multipliers. We don't believe that Intel is binning these K-series processors for their overclokability, nor have Intel ever claimed to be doing so. Having said that, it will be interesting to see whether we other sites get better results, since perhaps we just got lame duck samples. Overclocking is luck of the draw after all. Nevertheless, we are extremely pleased that Intel has, at least for now, changed their stance regarding limiting unlocked multipliers to the costly Extreme Edition models and has actually embraced overclocking. We certainly hope that K-series models become common place in Intel's lineup...

http://www.hardwarecanucks.com/foru...nlocked-processors-review-comment-thread.html​