SiSoftware Logo
  Home   FAQ   Press   Download & Buy   Rankings   Contact  
New: SiSoftware Sandra 2014
DE DE FR FR IT IT JP JP RU RU 

Intel Icon

Benchmarks : Intel Mobile Ivy Bridge: Core Comparison (non-Turbo)


What is it?

Ivy Bridge (IVB) is the next generation Core CPU from Intel (v3 2012) replacing the current Sandy Bridge (SNB) (v2 2011) line-up in both desktop and mobile platforms. While socket compatible (LGA 1155 desktop), not all SNB laptops may be upgradeable - unlike desktops.

In this article we test CPU, (GP)GPU and memory performance with Turbo / Dynamic Acceleration disabled at the same speed in order to see the differences between cores. Since the introduction of IDA with Core 2 and then Turbo with Core i7 it has become more difficult to tell whether any performance increases are due to Turbo headroom (or better cooling) or the CPU cores themselves.

Why compare non-Turbo performance at all? Low-end CPUs (e.g. Core i3/Pentium) are not Turbo capable; even Turbo capable CPU/APUs depend on cooling system performance to stay in Turbo mode for as long as possible.

Many laptops are seen with insufficient cooling or fan settings which quickly becomes a limiting factor under load. Many users dislike fan noise and choose thermal profiles designed to limit fan speed and thus limit cooling performance. Under these circumstances Turbo cannot be sustained for long and the laptop may even throttle (reduce speed).

Hardware Specifications

We are comparing a SandyBridge i7-2720QM Mobile (2.2GHz) with the identically clocked IvyBridge i7-3620QM; note that its direct replacement IvyBridge i7-3720QM Mobile runs at 2.60GHz (18% faster). We wanted a clock-for-clock comparison.

Processor (CPU) Specifications Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Cores (CU) / Threads (SP) 4C / 8T 4C / 8T No change
Speed 2.2GHz 2.2GHz Same base clock, should show the differences between cores.
Power (TDP) 45W 45W No change in TDP for this model.
Voltage 1.051V 1.196V Lower maximum voltage - should result in lower power consumption .

Processing Performance

We are testing native, SIMD and cryptography performance using the highest performing instruction sets (AVX, AES, etc.). We are also testing virtualisation performance in .Net/CLR and Java/JVM.

Results Interpretation: Higher values (GOPS, MB/s, etc.) mean better performance.

Environment: Windows 7 x64 SP1, latest Intel drivers. Turbo / Dynamic Overclocking was disabled on both configurations. See here for Turbo Performance.

Processing Benchmarks Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Native CPU Arithmetic (Dhrystone/Whetstone) 86.2 / 54 (68.5 GOPS) [-0.6%] 88.6 / 53.6 (68.9 GOPS) (baseline) While IVB appears slower, the results are within the margin of error: there is no significant difference between the two cores' performance.
Java Arithmetic (Dhrystone/Whetstone) 100.3 / 27.7 (52.7 GOPS) [-0.6%] 96 / 29.2 (53 GOPS) (baseline) Java JVM performance is again unchanged, difference is within margin of error.
.Net Arithmetic (Dhrystone/Whetstone) 10.6 / 35.8 (19.5 GOPS) [+4.8%] 10.3 / 33.3 (18.6) (baseline) .Net CLR performance improves a little bit, ~5%. IVB finally manages to break away from its brother, but not enough to make a difference.

While with Turbo enabled we saw at least 25% better performance, with Turbo disabled, at the same speed (and same TDP rating) we see no appreciable differences. All the tweaks IVB brings, do not show any improvement - at least in these tests. If you intend to turn Turbo off, you should still see power efficiency gains (lower voltages) just not higher performance.

SIMD Benchmarks Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Native CPU Multi-Media SIMD 132.2 / 179.9 / 102 (154.2 Mpix/s) [+0.7%] 132 / 177 / 102 (153.1 Mpix/s) (baseline) AVX SIMD performance is unchanged.
Java Multi-Media 23.2 / 18.8 / 18.8 (20.9 Mpix/s) [+18.8%] 18.2 / 16.9 / 16.7 (17.6 Mpix/s) (baseline) While traditional Java code did not benefit on IVB, in this highly parallelised vectorised algorithm we see ~19% better performance. While the IVB cores may not be faster, other tweaks allow the JVM to execute code faster.
.Net Multi-Media 25.2 / 8.4 / 19.6 (22.3 Mpix/s) [+18%] 20.8 / 8.4 / 17.1 (18.9 Mpix/s) (baseline) Here we get confirmation that the performance improvement we saw with Java is also reflected in .Net. Whether .Net or Java, VM applications benefit from IVM even with Turbo disabled.

While we saw no improvement in native code, SIMD code, or even traditional .Net and Java code - computationally intensive (vectorised) code does improve by ~18% in both Java and .Net - even with Turbo disabled. The tweaks IVB brings over SNB seem to allow the JVM/CLR to execute code quite a bit faster clock-for-clock.

Cryptography Benchmarks Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Native CPU Crypto (Encryption/Decryption/Hashing) 6.24 / 0.75 (2.2) [~10%] 6.26 / 0.67 (2) (baseline) While AES crypto performance has not improved (memory bandwidth limited), AVX hashing performance higher - improving the overall score by 10%! Considering Turbo is disabled and memory the same - it is good improvement for just a few core tweaks!

Most Popular Processors, Video Adapters

Most popular Processors as benchmarked by users (past 30 days):   Most popular Video Cards as benchmarked by users (past 30 days):
1.6%Intel Core i7-3930KIntel Core i7-3930K599.00 USD
2.4% Intel Core i5-2500K 
3.4%Intel Core i7 920Intel Core i7 920249.00 USD
4.3%AMD FX-8350AMD FX-8350179.99 USD
5.3%Intel Core i5-3570KIntel Core i5-3570K209.93 USD
 
1.8%AMD Radeon HD 7900AMD Radeon HD 7900505.93 USD
2.4%AMD Radeon HD 7800AMD Radeon HD 7800484.66 USD
3.4%AMD Radeon HD 6800AMD Radeon HD 6800289.80 USD
4.4%NVIDIA GeForce GTX 560 TiNVIDIA GeForce GTX 560 Ti505.89 USD
5.3%AMD Radeon HD 7700AMD Radeon HD 770099.99 USD
For a complete list of statistics, check out the Most Popular Hardware page. For a list of more products, see SiSoftware Shopping.

Memory Performance

We are testing the memory bandwidth, cache bandwidth (L1D, L2, L3) as well as cache and memory latencies using all the access patterns supported by Sandra (in-page random, full random and sequential/linear access patterns).

Results Interpretation: Higher values (MB/s) mean better performance.

Base 2 Multipliers: 1MB/s = 1024kB/s, 1kB/s = 1024bytes/s, 1byte = 8bits, etc.

Environment: Windows 7 x64 SP1, latest Intel drivers. Turbo / Dynamic Overclocking was disabled on both configurations. See here for Turbo Performance.

Memory specifications Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Memory Size / Rating 2x 4GB DDR3 PC3-1600 2x 4GB DDR3L PC3-1600 Same memory was used
Timing 11-11-11-28 5-39-12-6 1T 11-11-11-28 5-39-12-6 1T Same timings were used
Voltage 1.5V 1.35V IVB laptops will hopefully support DDR3L properly and thus reduce memory power usage.

Memory and Cache Benchmarks Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Memory Bandwidth 20.5 GB/s 20.5 GB/s (baseline) No increase in memory bandwidth or efficiency. Memory controller seems unchanged.
OpenCL (GPCPU) Memory Bandwidth 10.2 GB/s 10.2 GB/s (baseline) No apparent increase in OpenCL memory bandwidth either.
Cache Bandwidth (L1D / L2 / L3) 320 / 173 / 103 GB/s 322 / 172 / 103 GB/s (baseline) Cache performance is unchanged at all levels. While we saw ~15% increases with Turbo enabled, it is now clear that they were all due to higher Turbo headroom and not arch improvements.
Memory and Cache Latency - In-Page Random (L1D / L2 / L3) 28.8ns 4 / 12 / 23.1 clocks (+7.5%) 26.8ns 4 / 12 / 18.5 clocks (baseline) While L1D and L2 latencies are the same, L3 and memory latencies have increased from SNB: we are told these are "phantom" latency increases due to the updated prefetchers and should not occur in "normal usage". Let's hope that's true!
Memory and Cache Latency - Full Random (L1D / L2 / L3) 69.9ns 4 / 12 / 30.7 clocks (-10%) 77.9ns 4 / 12 / 29.8 clocks (baseline) IVB does better out-of-page than in-page: either the page-walk hit is lower or the updated prefetchers are better in handling random accesses. In any case, software using large memory blocks but not large pages (due to Windows issues) should perform faster.

Memory performance is a mixed bag; there is no improvement in cache (any level) nor memory bandwidth; some latencies have increased and some have decreased. Overall the prefetchers should help but IVB cannot overtake SNB with Turbo disabled.

Graphics (GPGPU) Performance

Ivy Bridge comes with an enhanced internal DirectX 11 compliant GPU that, for the first time, includes GPGPU capabilities: OpenCL and DirectX ComputeShader. While Sandy Bridge's DirectX 10.1 compliant GPU did report GPGPU capabilities - they were actually emulated in software thus running on the CPU!

Ivy Bridge is effectively an "APU", going head-to-head with AMD designs, most likely "Llano" at the low-end and future "Trinity" (Bulldozer successor FX Series with integrated GPU) at the high-end. By harnessing both CPU and GPU processing power, it can outperform Sandy Bridge simply by using the processing performance of the GPU in addition to the CPU!

Here we test Ivy Bridge's GPGPU and GPAPU performance against Sandy Bridge's GPCPU performance.

Results Interpretation: Higher values (MB/s) mean better performance.

Base 2 Multipliers: 1MB/s = 1024kB/s, 1kB/s = 1024bytes/s, 1byte = 8bits, etc.

Environment: Windows 7 x64 SP1, latest Intel drivers. Turbo / Dynamic Overclocking was disabled on both configurations. See here for Turbo Performance.

GPU specifications Intel 4000 Series (IvyBridge Mobile) Intel 3000 Series (SandyBridge Mobile) Comments
Shaders / Compute Units 16SP / 16CU (DX 11) 12SP / n/a (DX 10.1) While the SP count has not increased dramatically, they are meant to be far more powerful and also GPGPU capable (OpenCL and DirectX). It may not be enough to match middle-range GPUs or even AMD APUs but still enough for general usage.
Speed (Min / Max) 350 - 650 650 - 650 Same rated speed but lower speed at the lowest power mode (thus lower power at minimum usage). By using the same clock we can see how the updated shaders do against the original ones.

GPU/APU Processing Benchmarks Intel 4000 Series (IvyBridge Mobile) Intel 3000 Series (SandyBridge Mobile) Comments
Video DirectX Shading 132.8 / 38.5 (71.5 MPix/s) [+2.47x] 77.5 / 10.5 (28.8 MPix/s) (baseline) IVB's series 4000 GPU's DirectX 11 shaders are at around 2.5x faster than SNB's DirectX 10.1 ones. Turbo does not seem to make a difference here, we see very similar performance with Turbo on or off.
GPGPU DirectX ComputeShader 248.9 / 51.6 (113.3 MPix/s) [+7.8x] 55.8 / 3.8 (14.5 MPix/s)* (baseline) IVB GPU performance does not change with Turbo disabled, while SNB's is lower as Turbo is disabled: IVB is thus almost 8x faster than emulated GP performance SNB can muster and 2x as fast as OpenCL CPU SNB.
GPGPU OpenCL 230.3 / 16.7 (62 MPix/s) [+4.3x] n/a IVB is still much faster even without native double (FP64), though only ~4x.
GPAPU (CPU+APU) OpenCL 262.8 / 45.7 (109.6 MPix/s) [+7.5x] n/a Harnessing both CPU and GPU increases performance even further, though with Turbo disabled the CPU cannot contribute as much. Still ~8x better performance than SNB OpenCL is a huge increase.

The integrated IVB GPU shows its power with Turbo on or off; with native double support (FP64) it brings significant performance increase at the same TDP/speed/cost versus the already very powerful SNB quad-core mobile CPU. In APU mode, IVB is over almost as powerful as SNB OpenCL CPU even with Turbo disabled!

While we have not tested the new dedicated GPGPUs that will be used on IVB high-end laptops, they will have a hard time matching the IVB GPGPU in power efficiency or cost efficiency (effectively being free!). Those requiring high-end mobile performance - where thus neither power nor cost matters - they will still use their dual (SLI/Crossfire) dedicated GPGPUs, but for the rest IVB's GPGPU will be more than sufficient.

GPU/APU Cryptography Benchmarks Intel 4000 Series (IvyBridge Mobile) Intel 3000 Series (SandyBridge Mobile) Comments
GPGPU DirectX ComputeShader 822 / 1000 (941 MB/s) [+3x] 124 / 757 (306 MB/s)* (baseline) Against SNB's CPU emulation, IVB's native GPGPU is 3x as fast; again with Turbo disabled IVB's improvement increases.
GPGPU OpenCL 751 / 1000 (883 MB/s) [+2.88x] n/a OpenCL is not as fast as ComputeShader - yet - but still 2.88x faster, again, a massive improvement from SNB's CPU.
GPAPU (GPU+APU) OpenCL 1090 / 2000 (1470 MB/s) [+4.8x] n/a In APU mode, with Turbo disabled, the CPU cannot contribute as much; however IVB still ends up ~5x faster, a huge increase.

The integrated GPU of IVB impresses whatever the algorithm even with Turbo disabled; in APU mode, IVB is between 5x to 8x faster in GPGPU mode than SNB's CPU - all for the same TDP, speed and cost.

GPU/APU Memory Benchmarks Intel 4000 Series (IvyBridge Mobile) Intel 3000 Series (SandyBridge Mobile) Comments
Video Memory Bandwidth 21.2 / 5.4 (10.7 GB/s) [+1.75x] 14.8 / 2.5 (6.1 GB/s) (baseline) Even with Turbo disabled, IVB improves both internal and transfer bandwidth (though not zero-copy).
GPGPU OpenCL Bandwidth 22.7 / 13 (17.2) [+43%] 12 / 12 (12)* (baseline) The increase in internal memory bandwidth makes even OpenCL overall bandwidth 43% better than SNB. GPGPU applications working with lots of memory can definitely appreciate these improvements.

Efficiencies

Because not all things in life are evaluated to their true value, the next measurements will take into consideration various efficiency aspects:

CPU Power Efficiency (this measures the efficiency of power design, or TDP) Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Native CPU Arithmetic Efficiency 68.5 GOPS / 45W 1.52 GOPS/W 68.9 GOPS / 45W 1.53 GOPS/W (baseline) No appreciable improvement in efficiency.
Native CPU Multi-Media SIMD Efficiency 154.2 Mpix/s / 45W 3.43 Mpix/sW 153.1 Mpix/s / 45W 3.40 Mpix/sW (baseline) No appreciable improvement here either.
Native CPU Crypto Efficiency 2.2GB/s / 45W 48.9 MB/sW [+10%] 2GB/s / 45W 44.4 MB/sW (baseline) Here, we see the first improvement, with IVB being 10% more efficient.
GPU/APU Power Efficiency (this measures the efficiency of power design, or TDP) Intel 4000 Series (IvyBridge Mobile) Intel 3000 Series (SandyBridge Mobile) Comments
GP APU/GPU Processing 109.6 Mpix/s / 45W 2.45 Mpix/sW [+78%] 61 Mpix/s / 45W 1.36 Mpix/sW (baseline) By harnessing the GPGPU, IVB is 78% more powerful and thus 78% more power efficient, even with Turbo disabled. If you ever wondered the point of an APU - here is why!
GP APU/GPU Crypto 1470 MB/s / 45W 32.6 MB/sW [+2.1x] 699 MB/s / 45W 15.5 MB/sW (baseline) Here the GPGPU makes an even bigger impact, IVB being 2.1x more power efficient, even with Turbo disabled, simply due to its GPU.

With Turbo disabled, IVB CPU performance matches SNB and thus power efficiency is the same; but when harnessing the power of its GPGPU, IVB can be up to 2x more power efficient - a big improvement.

CPU Cost Efficiency (this measures 'bang-per-buck', i.e. cost effectiveness) Intel i7-3620QM (IvyBridge Mobile) Intel i7-2720QM (SandyBridge Mobile) Comments
Native CPU Arithmetic Efficiency 68.5 GOPS / $378 181 MOPS/$ 68.9 GOPS / $378 182 MOPS/$ (baseline) No appreciable improvement in efficiency.
Native CPU Multi-Media SIMD Efficiency 154.2 Mpix/s / $378 408 Mpix/s$ 153.1 Mpix/s / $378 405 Mpix/s$ (baseline) No appreciable improvement here either.
Native CPU Crypto Efficiency 2.2GB/s / $378 5.82 MB/s$ [+10%] 2GB/s / $378 5.29 MB/s$ (baseline) For the same money, IVB is 10% faster, not a big deal but it helps.
GPU/APU Cost Efficiency (this measures 'bang-per-buck', i.e. cost effectiveness) Intel 4000 Series (IvyBridge Mobile) Intel 3000 Series (SandyBridge Mobile) Comments
GP APU/GPU Processing 109.6 Mpix/s / $378 290 Mpix/s$ [+80%] 61 Mpix/s / $378 161 Mpix/s$ (baseline) IVB is now cost-competitive with other APUs, even with Turbo disabled, while SNB does not have a GPU to help it. For your money, you're getting almost 2x performance with IVB!
GP APU/GPU Crypto 1470 MB/s / $378 3.88 MB/s$ [+2.1x] 699 MB/s / $378 1.84 MB/s$ (baseline) Again, IVB is cost-competitive with other APUs and leaves SNB in the dust, being over 2x more cost-effective.

SiSoftware Official Ranker Scores

Final Thoughts / Conclusions (Non-Turbo performance)

With Turbo disabled, we see IVB CPU performance showing no improvement in native performance (including SIMD AVX) but 18% faster in .Net/Java. The improvements across the board we saw with Turbo enabled seem entirely due to the higher Turbo headroom rather than core improvements.

Streaming memory performance, cache performance and latencies are also unchanged.

The GPGPU capable GPU finally allows IVB to improve performance over SNB and thus power and cost efficiency - up to 2x more efficient - which is significant considering the quad-core CPU is so powerful. If we were comparing dual-core CPUs, IVB would be 3x more efficient!

We need to consider that its direct replacement runs at higher clock (2.6GHz vs. 2.2GHz) thus is likely to be 18% faster.

Overall, non-Turbo performance is a mixed bag; for low-end non-Turbo capable processors and the very many laptops with weak cooling systems (or using profiles designed to limit fan speed) IVB won't be faster than SNB; an i3 IVB won't perform appreciably better than an i3 SNB in CPU tasks. It all comes down to the IVB GPU - unless you find the SNB GPU limiting and need GPGPU features but don't want a big laptop with dedicated graphics, there is no reason to upgrade from SNB.

News | Reviews | Twitter | Facebook | privacy | licence | contact