Intel i7-3770K Ivy Bridge CPU Review

[MAC]

The Effect of Memory Bandwidth on IGP Performance

The Effect of Memory Bandwidth on IGP Performance

As we demonstrated previously in the A8-3850 and A8-3870K reviews, system memory frequency and overall memory bandwidth greatly affect an IGP's gaming performance. Thanks to its new dual-channel DDR3-1600 memory interface and 25.6GB/s of memory bandwidth Ivy Bridge already has a default advantage over every other processor with an IGP.

Futuremark 3DMark06

3DMark06 v1.1.0
Graphic Settings: Default
Resolution: 1280X1024

Test: Specific CPU Score and Full Run 3Dmarks
Comparison: Generated Score

The Futuremark 3DMark series has been a part of the backbone in computer and hardware reviews since its conception. The trend continues today as 3DMark06 provides consumers with a solid synthetic benchmark geared for performance and comparison in the 3D gaming realm. This remains one of the most sought after statistics, as well as an excellent tool for accurate CPU comparison, and it will undoubtedly be used for years to come.

Futuremark 3DMark Vantage

3DMark Vantage v1.0.1
Graphic Settings: Entry Preset
Resolution: 1024X768

Test: Specific CPU Score and Full Run 3Dmarks
Comparison: Generated Score

3DMark Vantage is the follow-up to the highly successful 3DMark06. It uses DirectX 10 exclusively so if you are running Windows XP, you can forget about this benchmark. Along with being a very capable graphics card testing application, it also has very heavily multi-threaded CPU tests, such Physics Simulation and Artificial Intelligence (AI), which makes it a good all-around gaming benchmark.

Futuremark 3DMark Vantage

3DMark Vantage v1.0.1
Graphic Settings: Performance Preset
Resolution: 1280X1024

Test: Specific CPU Score and Full Run 3Dmarks
Comparison: Generated Score

3DMark Vantage is the follow-up to the highly successful 3DMark06. It uses DirectX 10 exclusively so if you are running Windows XP, you can forget about this benchmark. Along with being a very capable graphics card testing application, it also has very heavily multi-threaded CPU tests, such Physics Simulation and Artificial Intelligence (AI), which makes it a good all-around gaming benchmark.

Crysis

Resolution: 1280x1024
Anti Aliasing: 0
Quality Settings: Low
Global Settings: DX10 / 64-Bit
Test 1: Benchmark_gpu demo
Comparison: FPS (Frames per Second)

Far Cry 2

Far Cry 2 1.02
Resolution: 1280x1024
Anti Aliasing: 0
Quality Settings: Medium
Global Settings: DX9 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 DX9 mode at 1280x1024 with all settings set to low.

Left 4 Dead

Left 4 Dead (Latest Update)
Resolution: 1680x1050
Filtering: 0X AA / 0X AF
Graphic Settings: Medium
Shader Detail: Medium
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 tested at 1680x1050 with in-game details set to low. For benching, we used a pre-recorded 20 minute timedemo taken on the No Mercy campaign during The Apartments mission.

World in Conflict

World in Conflict v1.010
Resolution: 1680x1050
Anti-Aliasing: 0X
Anisotropic Filtering: 0X
Graphic Settings: Medium (DX10)
Test 1: Built-in Benchmark
Comparison: FPS (Frames per Second)

One of the detailed and most visually stunning real-time tactical games in recent history, World in Conflict remains a staple in our gaming lineup . For this test we used the in-game benchmarking tool.

X3: Terran Conflict

X3: Terran Conflict 1.2.0.0
Resolution: 1680x1050
Anti-Aliasing: 0X
Anisotropic Filtering: 0X
Graphic Settings: Medium
Test 1: Built-in Benchmark
Comparison: FPS (Frames per Second)

The results here really shouldn’t come as much surprise. Naturally, higher memory speeds can increase IGP gaming performance. This is in part due to memory speeds affecting CPU performance and partially due to quicker access from the memory to the GPU’s shared L3 cache structure. In certain GPU intensive tasks however, the difference isn’t all that great since the integrated GPU continues to be a bottleneck.

 

[MAC]

Power Consumption / Temperature Testing

Power Consumption

For this section, every energy saving feature was enabled in the respective BIOSes and the Windows Vista power plan was changed from High Performance to Balanced.

For our idle test, we let the system idle for 15 minutes and measured the peak wattage through our UPM EM100 power meter.

For our CPU load test, we ran Prime 95 In-place large FFTs on all available threads for 15 minutes, measuring the peak wattage via the UPM EM100 power meter.

For our overall system load test, we ran Prime 95 In-place large FFTs on all available threads for 15 minutes, while simultaneously loading the GPU with OCCT v3.1.0 GPU:OCCT stress test at 1680x1050@60Hz in full screen mode.

First and foremost, in order to prevent the new Intel-provided Z77-based DZ77GA-70K motherboard from skewing the power consumption results, we installed the i7-3770K on a freshly flashed Intel DP67BG so that all the LGA1155 chips would be tested on the same motherboard. This way we can isolate the processor as the source for any increases/decreases in power consumption.

The Core i3-2120 is a dual-core 65W part, so it's no surprise that it features the lowest power consumption of this bunch. Given the fact that Sandy Bridge chips already power gates all unused cores, we aren't really surprised to see that there is no grand reduction in idle power consumption for Ivy Bridge. Thankfully, that's not the case when it comes to CPU load. Our shiny new Core i7-3770K needed 17 watts, or roughly 14%, less our i7-2600K sample at full load. That is a pretty solid reduction, and it further reinforces Intel's position as the unquestioned leader of performance per watt. However, we definitely expected to see even better results, perhaps to the tune of 20W-25W less than the previous quad-core 32nm SB parts. As it happens, we did experience a decrease in that range when using the IGP, which you can see below. When it comes to overall system load numbers there is no great revolution, the decrease is pretty much in-line with the drop in CPU load consumption.

Now let's ditch the discrete GPU and take a look at the power consumption when using the new HD Graphics 4000 IGP:

As always, we have to restate that the Sandy Bridge and Ivy Bridge chips in this graph have an almost unfair advantage in the form of the Intel DH67BL motherboard. This Intel-manufactured motherboard has unmatched idle power consumption, easily 10W less than comparable motherboards from the big three motherboard manufacturers. We couldn't do any Ivy Bridge IGP testing on our Intel-provided Z77-based DZ77GA-70K motherboard, since it has no video outputs. Nor would be want to, for the reasons explained above.

With that out of the way, the first thing that pops out is the fact that we did actually record a 26W decrease in CPU load consumption when compared to the i7-2600K. That is pretty much exactly what we were expecting from a quad-core part made on the new 22nm process. While that in itself is noteworthy, it's the very sizeable 32% drop in overall system power consumption that is the most impressive. Intel has almost managed to get this level of CPU and GPU performance under 100 watts...from the wall! We can't wait to see how the mobile parts are going to perform.

Temperature Testing

For the temperature testing, since we were not given a default cooler from AMD, we used a Thermalright Ultra-120 Extreme with a Thermalright TR-FDB-1600 fan. The ambient temperature was 21°C/69.8°F. The application used to monitor temperatures was AIDA64 2.30.1900. Keep in mind that the thermal sensors in most modern processors are not really accurate at measuring idle temperatures, hence the very small delta between the room temp and the idle results.

Idle CPU + Idle IGP: The system was left to idle for 15 minutes.
Idle GPU + Load IGP: OCCT v3.1.0 GPU stress test was run at 1680x1050 for 15 minutes.
Load CPU + Idle IGP: Prime 95 In-place large FFTs was run for 15 minutes.
Load CPU + Load IGP: Prime 95 In-place large FFTs and OCCT v3.1.0 GPU stress test were run for 15 minutes.

As mentioned above, you should obviously take the idle temperatures with a grain of salt, but either way the i7-3770K was exceptionally cool at idle. Athough it might look dire that the Ivy Bridge part experiences such a large spike in temps, given the minimal power consumption difference (67W vs. 64W) between both chips when fully loading the IGP it's not a surprise to see identical temps under this workload. When the CPU is loaded, the i7-3770K did indeed run hotter than the i7-2600K. However, overall we think it's fair to say that the i7-3770K is a cool-running processor at default settings. When overclocked though it's an entirely different story...check out the next page!

 

[MAC]

Overclocking Results

Overclocking Results

At IDF 2011, Intel revealed that Ivy Bridge would have a higher maximum CPU multiplier of 63X (up from 59X on Sandy Bridge), an up to 60X multiplier for the IGP, support for memory speeds up to DDR3-2800, and smaller 200MHz memory speed increments.

One of the really cool new features is dynamic overclocking, which allows users to change the CPU/IGP multipliers without rebooting. It worked flawlessly when using Intel's Extreme Tuning Utility (XTU). Overclocking an Ivy Bridge system is going to be identical as on Sandy Bridge, so you will want to buy an unlocked K-series processors if you want to increase the clock frequency in any significant way. We really don't expect to see any chips with a maximum postable multiplier lower than 50X, so that's already than on Sandy Bridge, but as you will see below that's really not the big concern this time around.

Highest Stable CPU Overclock

​

As you can see above, we managed to achieve a stable 4.8GHz core clock with about 1.35V. For comparison, we got 4.75GHz with 1.45V on our i7-3870, while our i7-2600K needed about 1.42V to hit 4.70GHz stable.

Ivy Bridge overclocks very well, generally better than mainstream Sandy Bridge at every voltage point up to 1.35V. Above that? We don't know. We don't know because we were told by an always reliable third-party that stress testing above 1.35V is a risk with IVB due to potential degradation, but we also don't know because we were scared that it would cause a house fire...

To put it simply, we think that the term "blazing hell inferno" might be short-selling Ivy Bridge's heat problem once you cross a certain threshold. Past 1.30-1.35V temperatures skyrocket, so we're going to recommend that everyone on air-cooling, even high-end air-cooling, stay in the 1.25V-1.30V range. Most chips shouldn't have any problems hitting 4.5GHz with 1.20V or less, which is awesome for your average casual overclocker, but scaling gets progressively worse as you go above these levels.

Our overclock was stable, we threw every app we could at it and had four 2-3 hour gaming sessions, but we aren't going to hide the fact that during the one hour AIDA64 stress test at least one core hit a truly absurd 101°C, while the others routinely reached into the 90's. At least that proves that these new chips are resilient to short-term heat.

It’s pretty clear to us that Intel is using very low leakage transistors with IVB, and when you combine extremely high transistor density with a die that has half the surface area of Sandy Bridge, the temperature situation gets out of control quickly when overvolting. Although our chip is an engineering sample sample, we know for a fact that retail chips will feature similar heat characteristics. We don’t think that it’s outside the realm of possibility that Intel might release a new stepping in 6-9 months, or at least that later chips might just be tuned for higher leakage as the manufacturing process improves, as was the case with the Bloomfield D0 stepping.

As mentioned above, degradation is a concern with these chips if you use too much voltage, so for stability testing we're going to recommend that cautious users switch over from IBT/LinX/Prime95 to the AIDA64 stress test or OCCT 4.2.0, both of which provide a much more realistic/real-world type of load on the processor.

With regard to power consumption, using the above overclock our system idled at 61W and topped out at 205W under the full load of Prime 95 In-place large FFTs. For comparison sake, our Core i7-2600 at 4.6Ghz with 1.42V had identical idle numbers, but higher 221W load numbers. This is actually a pretty decent result for IVB, it is certainly a heck of a lot better than the Bulldozer FX-8150 debacle, which when overclocked to 4.6GHz at 1.50V idled at 195W and pulled an immense 550W when running Prime 95.

Highest Stable BCLK Overclock

​

On the BCLK side of things we didn't notice any great leap forward either. Our result was just a little bit better than the 105MHz (without the Gear Ratio, of course) with hit on the Core i7-3820 and the 107MHz on the Core i7-2600K. It's all a coin flip though. Except for maybe the sub-zero crowd, voltage and heat considerations are going to be way more of a limiting factor than the BCLK on Ivy Bridge.

Highest Stable IGP Overclock

​

Intel's been pretty cocky with regard to touting the frequency headroom of the HD Graphics 4000 IGP, and sure enough we were able to achieve a lofty 39% overclock / 1.6GHz frequency with relatively little additional voltage (+0.223V). The associated performance gains were very impressive indeed, and would have been even more so had we jacked up the memory frequency instead of keeping it at stock.

Given the fact that Intel’s own documentation advised us to use up to +0.440V, we expect to see some insane numbers from those with a little better cooling and a lot more courage than us.

On the memory front we didn’t really set any new records since our 16GB G.Skill DDR3-2133 kit really doesn’t like anything above DDR3-2220 without an unhealthy boost in voltage. Having said, the Ivy Bridge/Z77 platform is absolutely amazing when it comes to memory clocking. Records are already being set, and there's a bunch of insane DDR3-2600/2800 32GB kits around the corner.

 

[MAC]

Conclusion

Conclusion

First and foremost, although we didn't really delve into it in the review these new Ivy Bridge chips will work perfectly on existing Intel 6-series chipset motherboards. We ran a few benchmarks and there is effectively no performance difference between Z77 and P67 based motherboards. A simple BIOS update is all that will be required for these chips to be supported by existing LGA1155 motherboards. We can't make any assertions with regard to the PCI-E 3.0 compatibility promises that various motherboard manufacturers have made, but it will be interesting to see how that plays out. Obviously, you won't be able to take advantage of the native USB 3.0 capabilities, but the third-party controllers on existing motherboards are no slouches.

This brings us to one of the most pressing questions: should those who currently have Sandy Bridge systems consider upgrading? Well as you will see below there’s relatively few reasons to do so at this point, at least for those who put an emphasis on performance instead of features or power consumption. Let's take a look at how the Core i7-3770K fares against some of its contemporaries:

First of all, obviously we would have liked to compare IVB with the i7-2700K, but we simply don't have one. Having said that, the flagship Sandy Bridge part was always overpriced in relation to the i7-2600K, so this is probably a more apt comparison for your average person. Overall, the i7-3770K is about 7% to 13% faster than the i7-2600K in real-life applications. This sounds decent, but you have to take into account the fact that the i7-2600K does have a 3-6% frequency disadvantage, depending on the workload and thus Turbo Boost level. For gamers, there is not much here to like. The performance 'gains' were well within the margin for error, except in the two games that are heavily CPU-centric.

When compared to the i7-3820, the results are generally much closer due its higher 3.6GHz default clock speed, larger 10MB L3 cache, and identical DDR3-1600 default memory speeds. The i7-3770K proved to be about 4% to 10% faster in most workloads. However, the i7-3820 actually has a suggested retail price that is about $20 less than the i7-3770K, although the other components like the motherboard and memory are pricier. Having said, as you will see below, we do think that it is a better purchase for enthusiasts/power users.

Overall though, from a performance standpoint it's good that we kept our expectations low. On average, Ivy Bridge is about 4% faster than the two Sandy Bridge variants on a clock-per-clock basis. Both the single and multi-threaded performance has improved about equally. 3D modelling and media creation are the areas that see the biggest boost, and although not demonstrated in the table above, minimum frame rates while gaming are consistently higher thanks to IVB's super-low latency and high-bandwidth L3 cache. Having said all of that, we really did not have any complaints about any areas of Sandy Bridge's CPU performance. There are no glaring shortcomings like there is on AMD's side.

One must also consider that Intel have achieved this level of performance within a much smaller TDP enveloppe than with the Sandy Bridge parts. To be honest though, we can't help but wonder how much extra performance could have been coaxed from this new process if they had set a 95W TDP, at least on the higher-end parts. Given the switch to 22nm we were, perhaps naively, expecting higher clock speeds than the previous generation, especially since we really haven’t made much frequency progress since the 45nm Nehalem days.

With a smaller TDP and the new 22nm manufacturing process, we were expecting to see a pretty sizeable decrease in power consumption. However, when we had a discrete GPU installed the drop in CPU load and total system power consumption was only about 17W when compared to an identically-equipped i7-2600K setup. However, when we switched to the IGP, the results were much more inline we with our expectations. Compared to the aforementioned Sandy Bridge chip, we registered a 26W drop in CPU load consumption and a 36W decline in total system power consumption. That is sizeable 32% decrease, and it tells us that Ivy Bridge is going to be phenomenal in the notebook sector.

Although Ivy Bridge runs quite cool at default clocks and voltage, all that goes out the window really quick when you start overclocking. While mainstream Sandy Bridge ran so cool that it gave users the courage to use more voltage than they should have, that's not the case with Ivy Bridge. The 22nm IVB die is so tiny and packed with transistors that the power density is off-the-charts, and the lack of surface area makes it really hard to cool despite the integrated heatspreader. As a result it will run really hot when overclocked and overvolted

For all practical purposes 1.25V-1.30V is going to be the highest useable voltage on Ivy Bridge chips, even on the highest-end air-cooling. Any more than that and you supposedly risk degradation if you run abnormal loading programs like IBT, LinX and Prime 95. Either way, we can fairly assertively state that if degradation doesn't scare you, the full load temperatures will. It’s not all doom and gloom though. Your average user should easily be able hit 4.5GHz with a tiny bit more voltage than stock, so the temperatures will be imminently manageable. It is just the enthusiast crowd that is going to suffer when trying to get above that level. We didn’t actually expect to be saying this, but those who take their overclocking seriously would be better served with a Core i7-3820. It might be a little slower clock-per-clock, and consume more power, but it overclocks as well, runs cooler when overclocked, and there’s a clear upgrade path to Ivy Bridge-E. For the LGA1155 platform, Ivy Bridge is the end of the road since Haswell is going use the LGA1150 socket. As a result, if you don’t like what you’re seeing now in this review, it might not get much better in the future...barring some awesome future stepping or miraculous improvement in the manufacturing process.

Obviously, we can’t summarize Ivy Bridge without discussing the IGP. The new HD Graphics 4000 is impressive, clearly superior to the HD Graphics 3000 found in select Sandy Bridge models. We found it to be about 86% more powerful in terms of raw performance, and it achieved 20 to 48% higher frame rates in our gaming suite. When compared to the A8-3870K, the i7-3770K’s enormous CPU performance advantage made it hard to isolate the IGP to compare it with the Radeon HD 6550D found in AMD’s flagship Llano A-series APU. Our selection of gaming benchmarks didn’t really help either, but it is clear that in GPU-limited games (like Crysis, for example), Intel’s new top-end IGP is still about 20-30% less powerful than then one found in the A8-series APUs. We definitely want to see what the equally new Graphics HD 2500 is going to perform, since processors that feature it will be the ones directly competing with the A8-3870K.

In summary, the best on the market just got a little bit better, but there is an asterisk as well. There is no debating that Ivy Bridge provides the best performance and the lowest power consumption of any quad-core processor. However, whether Ivy Bridge is right for you depends on your current computing situation, and whether you’re a regular user or an enthusiast. If you are still using an LGA1366 system, and it’s not overclocked to at least 4.0GHz, Ivy Bridge is worth a look if you want some new toys. Not only are you getting a very hefty stock performance improvement, but native support for USB 3.0, SATA 6Gb/s, PCI-E 3.0, better overclocking, and much lower power consumption as well. If you are building a system from scratch, especially an HTPC, Ivy Bridge is easy to recommend for most people. If you bought a cheap dual-core Core i3 as a temporary solution until Ivy Bridge was released, then by all means a quad-core IVB processor is going to be a terrific upgrade. However, if you own an i5-2500K or i7-2600K/2700K switching to their Ivy Bridge counterparts is not going to give you a lot of extra bang for your buck. The asterisk is obviously that overclocking hobbyists with air coolers won’t find Ivy Bridge particularly compelling, it simply runs too hot when you reach a certain level. A level which most Sandy Bridge chips have no problems achieving. But then again, we are talking about above 4.5GHz, which in the grand scheme of things is a realm that few users venture into anyways.

Intel Core i7-3770K Ivy Bridge CPU Review Comment Thread​