iXBT Labs - Computer Hardware in Detail






VIA C3 and Other Low-End Processors

July 7, 2002

It's quite difficult to give an exact definition to the concept of "low-end". It's possible to outline the idea, but even companies dealing in identical things (for example, production of processors and video accelerators) often have different views at certain points. Let's just consider that low-end means the cheapest models on the market, as a rule, with the worst technical characteristics.

It is considered that low-end products are meant for those who are short of money and who are guided primarily by prices of goods. However, the rational approach is to take all important factors into account and then choose a product at the minimal price not overpaying for megahertz or name. So, in the current situation of the minimal selling rates of processors around 1 GHz both those who are short of money and those who don't want to pay through the nose often buy the same. At the same time the second group of users are of course interested in comparison of speed and other characteristics of products referring to the low-end market and able to provide performance enough for their tasks. It also concerns the first group, because lack of money seldom has objective causes and is backed by knowledge.

Today we offer a comparison of all models of the modern low-end market, as well as some forecast. Probably the data we are providing will help you choose a processor line for your needs or show you that such comparison is useless. Well, read and make your own conclusions.

VIA C3 processor

VIA's processors do not keep step with the progress of megahertz, that is why they are not so often reviewed. The release of one more model of the VIA C3 line working at 1 GHz is a good occasion to turn to them again.

It is very possible that even our faithful readers have forgotten details of these processors - I recommend that you brush them up reading the reviews of the VIA Cyrix III (on Samuel), VIA Cyrix IIIa renamed into C3 (on Samuel 2) and VIA C3 (on Ezra). But it's enough to reread just the first review because the only architectural improvement in the Samuel 2 was an exclusive L2 cache located on a smaller die (because of the transition to the 0.15-micron fab process). Ezra, as the latest version of the Ezra-T core, is produced according to the 0.13-micron process, that is why it has better heat generation characteristics (the 1 GHz processor at the rated voltage of 1.35 V has it around 7 W (!) as typical, and around 12 W (!!!) as maximum). Anyway, today you will get enough information on the speed results of the VIA processors; a close examination of architecture of the processor line will be interesting only for fans and experts. But I should note that the C3 is compatible with any Socket 370 boards and will be produced at least up to 2004, that is why do not get rid of your old board - probably you have a chance to assemble a special-purpose computer.

Present and Past of Low-End Market according to AMD and Intel

Products of the two leaders of the processor market do not need any introduction. You will easily find their specs on our site, that is why we will just give a list of products available. At present, their low-end solutions are AMD Duron and Intel Celeron (Tualatin). The situation concerning the Duron is a bit complicated because it's easy to find models based on the old core but the performance of the new Morgan in some cases is a tad higher with the heat development being lower, that is why we think it's rational to take new models into consideration. For comparison we have taken the processors of the same frequency - 1.3 GHz.

Intel never kept back that the Tualatin's release was just an interim measure - a response to the high performance and frequency of the competing processors in expectation of the Pentium 4's growth. It's obvious that the moment came by May and Intel announced quick cut-down of production of the old line and refusal of the Socket 370. The new favorite of the company designed for the low-end market is based on the core of its senior brother (Pentium 4) and uses its form-factor (Socket 478), and, thus requires new (for low-end) and probably expensive boards. AMD is in a better situation in this respect: both the Athlon XP and Duron worked in boards on the same chipsets with a single processor socket. That is why now, when we expecting a new, reportedly, high-speed generation of processors AMD is preparing to move its Athlon XP line into the low-end market (and current Durons into oblivion), the infrastructure is perfectly in order. For comparison we have taken Intel Celeron (Willamette-128) 1.7 GHz and AMD Athlon XP 1700+ processors.

Performance tests


  • Processors:
    • VIA C3 (Ezra-T) 1 GHz (133x7.5), Socket 370
    • Intel Celeron (Tualatin) 1.3 GHz (100x13), Socket 370
    • AMD Duron (Morgan) 1.3 GHz (100x13), Socket 462
    • Intel Celeron (Willamette-128) 1.7 GHz (100x17), Socket 478
    • AMD Athlon XP (Palomino) 1700+ (1467 MHz, 133x11), Socket 462

  • Mainboards:
  • Memory:
    • 2x256 MBytes PC166 SDR SDRAM DIMM Tonicom, CL 2 (the memory worked at 100 MHz in case of the Intel Celeron (Tualatin) and at 133 MHz in case of the VIA C3)
    • 2x256 MBytes PC3200(DDR400) DDR SDRAM DIMM Samsung, CL 2 (the memory worked at 400 MHz in case of the ASUS P4S533 and at 333 MHz in case of the Acorp 7KT333)

  • Video card: ASUS 8200T5 GeForce3 Ti500
  • Hard Drive: IBM IC35L040AVER07-0, 7200 rpm


  • Windows 2000 Professional SP2
  • DirectX 8.1
  • Intel chipset software installation utility 4.00.1009
  • Intel Application Accelerator 2.2
  • VIA 4-in-1 4.38
  • SiS AGP 1.09f
  • NVIDIA Detonator v28.32 (VSync=Off)

  • CPU RightMark v1.01
  • RazorLame 1.1.4 + Lame codec 3.91
  • VirtualDub 1.4.9 + DivX codec 5.02 Pro
  • WinZip 8.1
  • WinAce 2.11
  • Discreet 3ds max 4.26
  • BAPCo & MadOnion SYSmark 2002 Office Productivity
  • BAPCo & MadOnion SYSmark 2002 Internet Content Creation
  • eTestingLabs Business Winstone 2001
  • eTestingLabs Content Creation Winstone 2002
  • MadOnion 3DMark 2001 SE
  • Gray Matter Studios & Nerve Software Return To Castle Wolfenstein v1.1
  • DroneZmarK

The VIA C3 and Intel Celeron (Tualatin) worked on the ASUS TUSL2-C board (each at its bus speed and respective memory frequency - the chipset is synchronous); the AMD Duron and AMD Athlon XP worked on the Acorp 7KT333 (each at its bus speed (DDR), the memory frequency was 166(333) MHz in all cases); the Intel Celeron (Willamette-128) worked on the ASUS P4S533 (at the respective bus speed (QDR), the memory ran at 200(400) MHz). We chose the fastest processors of the line, except the Intel Celeron (Tualatin) which ia already available as a 1.4 GHz version, the Athlon XP was selected as a pair for the Intel Celeron (Willamette-128), because this line is not shifted yet to the low-end market and AMD doesn't have clear plans on it. As you can see, for each processor we set the most comfortable conditions and the frequency was chosen so that we could compare exactly their processing power, though in real low-end systems it will be limited by the less efficient components.

As far as the C3 processor is concerned, the maximum frequency of the samples available is 1 GHz which is much less as compared with the competitors. In order to drag the tested model to the general level we decided to plot diagrams not only for the C3 1 GHz but also for the value increased by 30% - we, thus, make an attempt to estimate a performance of a hypothetical model - C3 1.3 GHz. Of course, we realize that the figures given are incorrect because none of applications is able to scale according to the processor's frequency by 100%. But it's also obvious that when we see diagrams of a processor lower in frequency than the others we try to "complete" them. That is why the figures given should be considered just as such tips, or you can ignore them if you wish. However, as you will see soon, we don't distort the general situation much ;).

Test results

Let's start with the CPU RightMark and estimate an overall performance of the processors before we turn to real applications.

As you can see, the C3 1 GHz and even the hypothetical C3 1.3 GHz are falling into the last positions (the results of the latter are very close to reality). It's not surprising: the VIA processors have a very weak floating-point unit. The other competitors work typically - we compared the Athlon XP and Pentium 4 a lot of times already with the CPU RightMark, and the tests is not sensitive to the L2 cache size which is a distinguishing feature of the Duron and the Celeron "Willamette-128". We can also see that the Celeron "Tualatin" is outscored by 40% by the integrated FPU of the AMD processors. But the future low-end Intel's solution based on the NetBurst has quite a moderate speed, and only at the expense of a much higher frequency and SSE2 instructions it shows the best scores.

This is one more demonstration of weakness of the C3 coprocessor, especially when compared with the results of its contestants obtained with the SSE instructions. The Celeron "Willamette-128" lags behind in speed of execution of x87 instructions and becomes faster again with the SIMD instructions.

The overall performance in the CPU RightMark test puts the processors in the order their test results are given in. The VIA C3 falls into the last position, in the AMD-Intel pairs the victory belongs to the former.

This is the first real test and the first surprise: the VIA fails to run. If you a man(woman) with nerve then look a the screenshot, if not just note that the Lame codec stopped working during the encoding process on the C3 because of errors. The rest have successfully passed the test. By the way, the Celeron "Tualatin" has outpaced the Duron.

Well, there is a great precipice between the C3 processor and the others. Neither the large cache nor the higher (as compared with the Celeron "Tualatin") memory frequency didn't help the C3 - the weak coprocessor and lacking support of at least the first SIMD instructions gave such poor results. Note that the advantage of the Duron over its competitor is greater than that of the Athlon XP over its one - on the one hand, it's because the SSE2 is supported by the Celeron "Willamette-128" and the DivX 5.0 codec, on the other hand, it's caused by the low throughput of the SDR SDRAM on the Celeron "Tualatin" platform.

The situation in the archivers is similar: the memory speed and the size and organization of the cache are much more important here, and it's the first time in these tests that the performance estimation for the C3 1.3 GHz processor (but it is far from reality - the test doesn't depend much on the processor's speed) gives the comparable results. The Celeron "Tualatin" shows perfect efficiency and fights against the Athlon XP 1700+, having the frequency lower by 200 MHz, probably because of the true synchronism of the chipset. The Duron and the Celeron "Willamette-128" go on a par in the archiving operations despite the difference in the frequency.

Rendering in the 3ds max brings in more unpleasant things for the C3 - although the gap is not so great as compared with the MPEG4, it is still very marginal. Unfortunately, the Celeron "Willamette-128" has severely fallen behind its competitors, which, taking into account a parity of the Pentium 4 "Northwood" and Athlon XP of the same rating in the 3ds max, can be caused only by decrease of the cache memory, which is crucial for this application, of the Celeron Pentium 4.

When looking at the poor results of the VIA C3, it's time to remember that the company doesn't plan on using its processors in computers of demanding users. The C3 is primarily designed for cheap office computers, and it makes very interesting to look at the office benchmarks' results. Unfortunately, there were some drawbacks as well. The SYSmark 2002 failed to finish on the VIA's processor - running of the script in the MS Word gave an error when we tried to open some text file. It's not clear why the script of the test reacted such a way only to the C3. To estimate speed in office applications we had to use the aging Winstones packet which often gives doubtful results. However, execution of the Content Creation test also resulted in hanging of the system in the Adobe Premiere. As a result, we can offer you a combination of the half-finished tests from both packets.

Well, the figures of the VIA's CPU look quite good, the twice lower scores in the content creation test mustn't confuse you - if we turn them into absolute values you will see that the real difference between the competitors is just several seconds, because implementation of some operations in these tests hardly exceeds 5-10 sec. (except a case with lengthy background processes, for example, encoding with the MS Windows Media Encoder). Note that the Internet Content Creation test which also includes Adobe Premiere was completed successfully on all processors.

The comparative results of the C3 in the 3DMark are predictable: our CPU RightMark test estimates a CPU's performance from the gaming standpoint, and we can see again that the coprocessor is too weak here. But let's take a closer look at the real operation in games.

Low resolutions and/or a low detailing level of the scene allow playing even on the C3 - without much comfort, but potential users of office computers based on the VIA's processor can count on Quake in the evening :). The other participants perform well in the tests, the Celeron "Tualatin" and the Athlon XP take their usual positions, and the Celeron "Willamette-128" fights against the Duron which works at a frequency lower by 400 MHz.


Low-end can be different. The VIA C3 working currently at 1 GHz belongs to the low-end sector for the cheapest and undemanding computers. These processors are meant for office applications operations which do not take a lot of time, and therefore, the speed difference is overlapped by the user's response time or makes several seconds. They do not suit games or tough applications, that is why the C3 will be bought by people who are going to assemble systems on aging boards based on the Socket 370, which must drop in price after Intel leaves this market. However, there is one more field of application for 1 GHz processors able to work without active cooling, which is desktop or wall solutions used as digital home entertainment systems. This subject will be developed further in our future reviews of VIA's special solution - the Eden platform.

As for the others tested models, I can say that they are useful in another low-end field, i.e. economical home or even semiprofessional computers. None of the contestants has any severe problems, the speed is low, as compared with the top processors, but they are good solutions for budget-minded people. I must note that in the tests the AMD Athlon XP was stronger than the Intel Celeron "Willamette-128", and the AMD Duron "Morgan" outpaced the Intel Celeron "Tualatin". The approximate price difference between them put the processors into the same order, except the Duron which costs cheaper than its performance. In future, when the old low-end wave will be actively replaced with the new one, the picture mustn't change much, and the choice will depend mostly on your own preferences and exterior factors.

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.