iXBT Labs - Computer Hardware in Detail






Intel Core i5-660/661 Processors

<< Previous page

     Next page >>

Audio encoding

As soon as we managed to perform encoding in four threads instead of two, we immediately obtained a one third performance boost in relation to the Core 2 Duo. While it's obvious that Hyper-Threading isn't a fully-fledged replacement for actual processor cores, you can see the results. The new processors can already compete with some quad-core solutions in this and other areas. While the Core 2 Duo could only compete with triple-core products. However, we're talking the lowest-end quad-core solutions now. The Q9505 is still unreachable, performing on a par with the Phenom II X4 965 and outperforming even the Core i5-750. It's clock rate that is critical in this group of tests, given the architecture and the amount of cores are similar. And the AMD CPU architecture loses for technical reasons, given the clock rates are the same.

Video encoding

The high clock rate and HT support let the Core i5-661 and 660 perform on a par with the once top-end Q9550, not just the Core 2 Quad Q9300 and Q9505 -- with certain codecs, that is. Besides, the excellent results in the single-threaded Canopus pulls the Core i5-661 and 660 closer to the Q9650, from where the highest-end Phenom is not that far. However, the x264 codec, optimized for true multi-threading, cannot be deceived. Being able to use 8 or even 16 threads, it obviously prefers true quad-core processors. If you take a closer look at the summary table, you will see that the new Core i5 performs even worse than the Core 2 Quad Q8200. Remember that, if you're looking for a CPU for video processing. This fact doesn't mar the average result though, which precisely corresponds to the market positioning: between the Q9505 and i5-750.

3D games

The smaller cache logically decreased performance where it used to be decent. However, Hyper-Threading boosted results which used to be low. In particular, compared to the C2D E8600, the average fps in WiC grew from 37 to 44, while GTA4 set a new record (which was a priori expected) by having improved from 48 to 58 fps! However, HT still cannot replace actual cores, so FarCry2, which likes quad-core processors, performed at 44 fps. This is more than 42 fps of the 8600, but can be explained by Turbo Boost. The average values are worse thanks to reduced performance in older games. However, lower-end and mid-end Core 2 Quad CPUs, as well as all Athlon II processors, used to perform even worse in those games anyway. Though even in these tests the Core 2 Quad Q9505 performs generally on a par with the Core i5-66x, which is somewhat disturbing. Well, this actually depends on your point of view: on the one hand, the formal goals have not been achieved, but on the other hand similar performance at different prices is not bad at all. Anyway, we are not going to either overcriticize or overpraise the new processor in relation to games.

Total score

Many users thought that the Core i5-600 would be like Core 2 Duo, just better -- a very fast processor for single and dual-threaded tasks. Unfortunately, we have to disappoint them: multi-threading is meat and drink for these CPUs, they need it as much as higher-end processors do. As soon as older non-optimized applications become involved, it turns out that the Core 2 Duo E8000 series remains unconquered. Even the new architecture and Turbo Boost doesn't help, because 4MB of 2.13 GHz cache is not the same as 6MB of 3+ GHz cache. Besides, Hyper-Threading is a disadvantage under these conditions. By the way, this is also the reason the main character of the review cannot compete with the Core i5-750 in the same situation: the latter has more cache, no HT, and it boosts clock rate aggressively when only two cores are used. In other words, Intel did the right things, having spent the entire last year persuading software developers to move to true multi-threading. And the company was even quite successful, but it was actually AMD who enjoyed the results most.

This is not Intel's engineers' fault, they have made a very good processor (compared to Intel's older products). The 10% boost in the total score comparing to the Core 2 Duo E8600 is very good. It's even better considering manufacturability, because the Clarkdale is much cheaper to make than the Wolfdale. As we could see, the E8600 only catches up with the Athlon II X4 630, while the Core i5-66x outperforms that fair quad-core processor easily. The price is not beneficial though. Such competition should have been avoided. Who is to blame? Those responsible for long-term decisions about rolling out solutions to the market. Users just relaxed, they got used to thinking that the Core 2 architecture was competitive enough. Nice numbers advertised for the new architecture did their part too. As a result, the company took its time, considering there would be enough time for everything: cooperation with software developers, implementation of the new platform, mastering of the new process technology. The rollout of the LGA1156 was delayed by three months due to commonplace pressure from motherboard makers which accumulated lots of LGA775 sockets and chipsets. The step-by-step implementation of the new platform resulted in three sockets existing simultaneously in the Intel's share of the market. Moreover, the company gave much praise to wonderful performance offered by the future hexa-core Gulftown ("Hmm, so the high-performance segment is going to be owned by LGA1366," users thought), to the benefits of the new architecture which was being developed and expected to go live late in 2010 ("Hmm, so they're going to change the socket again, meaning the current ones are equally unpromising"). However, in stores, users with two hundred bucks or less could only see the same old LGA775 processors. Obviously, no one knew what to choose.

Then AMD joined in and declared that users should choose Socket AM3. They agreed with Intel that dual-core processors were morally obsolete, they also told about software developers' plans to move to multi-threading (referring to Intel every time) and hinted that running three or four threads wasn't that expensive. Thus AMD offered users new Athlon II X3 and X4 CPUs for less than $150, lower-end models being priced even less than $100. What users did was obvious. And since that "unpleasant surprise" happened right before the end of the year, when users were getting ready for massive PC purchases, Intel didn't have time to adjust its plans. There were certain rumors that Core i3/i5 solutions could be announced earlier, in December 2009, but those didn't work. And they couldn't. Rolling out products on paper only works in the high-performance segment, where it can be used to take a rival down a peg by promising something disturbingly incredible. However, this is useless in the mainstream market. The latter demands actual processors in quantities sufficient for real sales, at competitive prices at that. The prices are not very nice right now, as you can see. In terms of total score, the Core i5-660 fits between the Athlon II X4 630 and Core i5-750, closer to the former at that. It does outperform Athlon noticeably sometimes, but it also lags behind in many other applications. So it should cost adequately -- something like $150. Oh well, add a few bucks for the brand, and another few for novelty, which makes it $160-165. This is close to the price of the Core 2 Quad Q8200 which needs to be replaced. Oh well, they may even charge another tenner for the integrated graphics (and upset the users of discrete graphics), considering that H55-based boards have turned out to be suprisingly affordable. No matter how hard they try, the price shouldn't exceed $180. And we mean retail, not 1000-unit quantities. However, today, something like $195 is the best offer, which is closer to the Core i5-750. Considering the performance difference, there's more sense in buying a discrete graphics card instead, since an entry-level solution (performing at least as well as any integrated graphics core) can cost about $30. The only field of use that yields to the new processors completely is compact machines. The LGA775 CPU + chipset bundle generates much more heat than a single Core i3/i5 processor, and the latter has better performance (those processors have to compete with the CPUs with 65W TDP, so 95W plus the chipset's heat is too much). A fast mini-ITX rig is worth of paying an extra for it. The new processors will look good in laptops as well, since fully-fledged quad-core CPUs are not yet ripe for the mainstream mobile segment, and the Core 2 Duo could do better. However, our today's review is dedicated to desktop processors.

Anyway, the prices should obviously be cut. It is also obvious that the new processors have a greal potential for that, especially as their process technology matures. The second important benefit is original architectural features which let dual-core processors feel all right among quad-core ones (it even seems that Nehalem's Hyper-Threading is not called to boost the performance of higher-end models, but to provide an opportunity to release such lower-end ones). The third benefit is that the new processors are actually new. Their rollout will allow the company to introduce proper order in its market share, because three similar chipsets is too much. One mainstream and one "extreme" would be fine. The new processors do compete with the Core 2 Quad. The Core i5 solutions are not completely superior (this could be said about the Core i7-800 series or the Core i5-750 though), but, on average, they perform well, even considering the price. The fourth benefit is the simplification of the rig in general. Now even PCs with integrated graphics (which occupy two thirds of the market, mind you) can be built according to the dual-chip design: CPU + PCH. A basic machine doesn't need any other complex chips, just attach an audio codec, a network PHY controller, add some memory, a hard drive and you're ready to go. Considering that compact size is the trend now, this benefit in quite important. In general, if we leave aside market positioning, but assess the situation from a technical angle, Intel should be congratulated on a job well done. Congratulations.

Write a comment below. No registration needed!

<< Previous page

Article navigation:

Page 1: Introduction, testbeds

Page 2: Tests

Page 3: More tests, conclusions

blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.