iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Desktop x86 Processors in 2007:
Hopes and Accomplishments

Hopes to some, accomplishments to others.

January 4, 2008



Forgive me my lightweight beginning, but this joke is actually true, so why not? In brief, Year 2007 can be described as follows for the CPU market: Intel has been developing its Core architecture, while AMD has been trying to intimidate us with promises.... :) Now let's analyze the events.

Summary

Intel's herald reports no changes on the Western front

Having entered the year of 2007 with a quad-core processor (it was not ideal because of its "split" L2 Cache, but it still demonstrated decent results in most tests), Intel could afford not to launch anything new. According to results of our latest tests, it could relax until the very end - AMD Phenom based on the new K10 core, because even the Core 2 Extreme QX6700 (2006) was faster than any Phenom manufactured in 2007. But it's a purely technical outlook. Intel couldn't do it from a marketing point of view, of course. That's why Intel's year of 2007 passed under the slogan "let's make the series of new processors even more attractive and balanced!" There have appeared Core 2 Duo 6320 and 6420 to replace the "old" 6300/6400. No one was going to encroach on the 6300/6400, they just improved it. The reasons lie in the attempt to reach a balance in the family of products rather than to increase performance (in fact, it hasn't grown much according to our test results). Besides, they revived the Pentium trademark - Pentium E. Now we come to the main contradiction, which hasn't been solved by Intel in 2007. Moreover, the company only aggravated it. So it will have to be solved next year: too many products with vague positioning and overlapping segments.

Indeed: NetBurst is still active in three modifications: Celeron D, Pentium 4, Pentium D. On the other hand, even if we don't take Extreme Editions into account, the family of processors based on the new core counts four names already: Celeron (Series 4xx), Pentium E, Core 2 Duo, and Core 2 Quad. The Core 2 Duo family also includes two classes of processors (Core 2 Duo E4xxx and Core 2 Duo E6xxx), which differ much in performance. In our opinion, no one will explain to you positioning of these processors relative to each other, not even Intel. It's clear that Celeron D is an outsider. It fails to demonstrate performance on a modern level, it just executes x86 code correctly. It's a relief it does that, nothing else is required from it. Pentium 4 should have been in the wake of Celeron D, but it's not that simple: many Pentium 4 products (according to our tests) are outperformed by Celeron processors based on the updated core. Moreover: a low-end single-core Celeron 440 outperforms a low-end dual-core Pentium D! Although outdated and almost discontinued, top models of the latter compete well with the new dual-core low-end processor (Pentium E). It's hard to sort it all out, almost impossible, without testing absolutely all processors of all series. Nevertheless, at the end of the year Intel has announced Core 2 processors on the updated core (Penryn) — quad-core Core 2 Extreme QX9650 and 9770 (Yorkfield), the latter with a new 1600 MHz FSB.

As a result, Intel ends 2007 with a lot of new and old processors based on two different architectures. They are grouped into seven titles (if we differentiate between Celeron D and Celeron). Many of them are very good, some of them can even kick a little more, others have to be discontinued. Perhaps, the company will have to solve the problem of too many names and CPUs in 2008. It's good when a processor demonstrates good performance and low power consumption, of course. But when even advanced users cannot recite names of all processors from Intel by heart, they seem to be in need of optimization...

AMD reports
"coming in on a wing and a prayer..."

AMD had to cut down prices for the old processors on the K8 core to keep them afloat, because production-line processors on the K10 core were delayed until the end of the year. Besides, the company transferred a significant part of its products to the 65-nm fabrication process. That is the company tried to attract users by playing the popular "power consumption" tune. This "brilliant marketing idea" brought to life Athlon 64 X2 Energy Efficient, Athlon X2 BE, and Sempron LE. They produced an arbitrary impression, because some Energy Efficient processors demonstrated even worse performance, and their power consumption versus Core 2 Duo and Pentium E did not look as good as power consumption of usual Athlon 64/X2 processors versus Pentium 4 and Pentium D. The situation before the launch of the K10 was so tense that even the announcement of a top Athlon 64 X2 6400+ (3.2 GHz) went almost unnoticed: people were waiting for the K10 (Phenom).

Phenom X4 has become "disappointment of the year" for some people: constant problems with the fabrication process (they did not come as a surprise to AMD fans...) did not let AMD launch the first production-line samples operating at a high clock rate. Perhaps, these very problems did not allow to equip Phenom X4 with caches adequate to those of Intel Core 2 Quad/Extreme processors. As a result, performance of the Phenom X4 is inadequate (to say the least) to that of top solutions from Intel. So AMD had to grin and bear it, positioning processors on the new core as Middle-End solutions. The company was forced to do that, but this decision was ruinous both to the image and business. The image was ruined, because Phenom X4 was defeated by a number of quad-core processors from Intel - not only by the new 45-nm Yorkfield CPUs, but even by usual 65-nm Kentsfields. Business suffered because it became impossible to sell Phenom X4 processors at high prices even to AMD zealots - no one would have bought them after they had seen test results. The company had to dump from the very beginning, an alarming sign. Another fly in the ointment is the story with a bug in production-line Phenom processors, which drops performance by 20% (according to our test results). We had an impression that AMD was living through a number of misfortunes, which could finish the company off.

However, I still wouldn't use the word "failure". In my opinion, launching Phenom as it is was a forced move under a colossal pressure of expectations. Expectations of business partners, suppliers of ready solutions, probably even investors, and certainly a huge army of AMD fans. It was impossible to procrastinate any longer, so the company could only come up with a haywire solution and get ready for retaliation. Which followed immediately. AMD was ready for it, of course: it was easy to predict what reviews would appear after the first tests. But let's have a look at the situation from a different angle. First of all, Athlon 64/X2 is still popular for its price. Sales of GPUs and chipsets from the former ATI also yield some profits. Now what concerns Phenom. It will be perfected in time. The bug will be fixed, clock rates will grow, so it may be launched for the second time. Probably even more successfully than the first time...

Some of you may ask why I am so optimistic (or not so pessimistic) about AMD Phenom? The answer is simple: because I have powerful memory. Athlon 64 was born with similar problems, and it had trampled on NetBurst for almost 18 months before the launch of Core 2 Duo. The new Phenom has a low clock rate and small cache. And the old Athlon 64 had a low clock rate and worked slowly with memory (single-channel Socket 754). Contrary to the popular opinion among people with short memory (that Athlon 64 shot forward right after it was launched), this processor hit the road to the top with low results. Advantage over Pentium 4 was out of the question at first. Later on Intel Northwood failed to increase its clock rates, and Prescott did not come up to expectations. However, AMD fine tuned the fabrication process to increase Athlon 64 clock rates and added a dual-channel memory controller even to Mid-End systems. So I tell you: "let's wait and see." No matter how bad Phenom looks now, some key signs indicate that after proper polishing K10 can compete with the current Intel core. On the other hand, Intel will not rest on its oars either. In practice, bad cores happen in Intel as well...

Market Situation at Present

Back in 2006, AMD competed on a par with Intel in Mid- and Low-End segments. But in 2007 AMD lost its positions in the Upper Mid-End segment. At the end of the year there started a fight for the best price/quality ratio of Low-End processors. The launch of Celeron 4xx and the aggressive pricing policy of Intel for these products made Celeron 420 the cheapest solution in some price lists. Celeron 420 possesses a new core, even if only one. It consumes very little power, and offers performance sufficient for an office computer. Theoretically, a relatively inexpensive quad-core Phenom X4 might have improved the situation in the upper part of the Mid-End segment, but the bug in the first production samples scared off potential buyers, and the new revision with this bug fixed will appear only in March. Besides, we don't know whether AMD can manufacture this processor in sufficient volumes.

As a result, the most attractive offers from AMD are Athlon X2 BE — relatively inexpensive, energy efficient processors. Their performance is far from brilliant, but it's sufficient for a home multimedia center, which is occasionally used for gaming. Intel Pentium E belongs to the same segment, but its performance is on the same level, so the choice depends solely on personal preferences. Top solutions, like Athlon 64 X2 5000-6000+, are still alive and kicking as far as their price/performance is concerned. But they are under constant pressure from Core 2 Duo 6xxx. The situation will become only worse with the launch of dual-core Penryn processors. Besides, AMD is reluctant to switch top Athlon 64 X2 products to the 65-nm fabrication process, while 100 Watt TDP is rapidly growing unpopular.

On the whole, the situation on the market is hardly favorable to AMD. But the company will most likely keep up its current production volumes dumping on all fronts to stall the time. The question is how it will use the time...

Forecasts

From the point of view of market success, 2008 will certainly become the year of Intel. The company may take its time, as it has already fine tuned the 65-nm fabrication process and mastered the 45-nm one, possessing a fast and stable core, and almost 1.5-fold advantage over the competitor in top performance. Intel will most likely optimize its CPU product line: it will reshuffle core and FSB clock rates, the number of cores and cache sizes, remove some duplicate models, discontinue NetBurst processors. It's high time Intel do it, we've already mentioned it above. The company may come up with a new product at the end of the year. Fortunately, there are many options: more cores, built-in memory controller, shared L2 Cache on quad-core processors... Or there will be no innovations at all, because Intel does not need any new technologies in 2008. It's the competitor that must debug its new technologies. :)

AMD will carry on selling its K8 processors, GPUs, and chipsets at relatively low prices. They will be the main stable source of income in 2008. Besides, Opterons will bring their share of income — new processors from Intel didn't affect these processors as much as Athlon 64/X2. But this is a streamlined process, which requires no interferences or R&D investments. The only chance for AMD to avoid sliding down to the B-list and becoming a manufacturer of inexpensive solutions for those who cannot afford better products is to perfect Phenom. Let's hope that the company will indeed perfect the fabrication process and raise frequencies/cache sizes instead of spreading thin existing products by reshuffling core numbers and frequencies. The latter approach will bring no good. Apparently, AMD must raise top performance of its processors.

Stanislav Garmatiuk (nawhi@ixbt.com)
January 4, 2008

Write a comment below. No registration needed!


Article navigation:



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.