Sempron 3400+ Review
By Gabriel Torres on September 17, 2005


Introduction

Sempron is the AMD processor targeted to the entry-level market and can be found in two platform versions: socket 462 (a.k.a. socket A), the same motherboard type used by Duron and Athlon XP, and socket 754, the same motherboard type used by some Athlon 64 processors.

The newest model in Sempron line, 3400+, is targeted to the socket 754 platform and is the fastest Sempron CPU to date.

Sempron 3400+
click to enlarge
Figure 1: Sempron 3400+.

Sempron 3400+
click to enlarge
Figure 2: Sempron 3400+, back view, you can see the socket 754 pinout.

Sempron 3400+ runs at 2 GHz, just like Sempron 3300+ and Athlon 64 3200+. The difference between these three CPUs is the amount of L2 memory cache. While Sempron 3400+ has 256 KB of L2 cache, Sempron 3300+ has 128 KB and Athlon 64 3200+ has 1 MB.

One thing you must keep in mind is that the AMD numbering system can only be used for comparison between processors from the same family. You could think that the 3400+ of this CPU means that it is faster than a Athlon 64 3200+, for example, which is not the case, since Athlon 64 3200+ has four times more L2 memory cache than Sempron 3400+. This number can be only used for comparison among other Sempron CPUs.

This new Sempron processor has the 64-bit instructions activated, which makes it a true Athlon 64 CPU but with less memory cache. So you can think of Sempron 3400+ as a “light version” of Athlon 64 3200+.

In this review we will compare the performance of this new Sempron to Athlon 64, Athlon XP, Duron, other Sempron models and also to Intel CPUs.

Main Specifications

The main specifications for Sempron 3400+ we reviewed are:

(*) OPN, Ordering Part Number, is the number that is written on the CPU body.

Sempron 3400+
click to enlarge
Figure 3: Technical specs from the reviewed Sempron 3400+.

The Palermo core introduced the 64-bit instructions to the Sempron family. However, there are Sempron processors based on this core that don’t have this instruction set enabled. The CPUs based on the Palermo core that have the 64-bit extensions use the letters “BX” as the two last letters on the CPU OPN.

Another feature introduced on Palermo core is the SSE3 instruction set, not available on Sempron processors based on previous core (Paris, based on 130 nm technology).

Processors Included in Our Benchmarkings

Below are the main features of the processors we included on our benchmarking.

Processor Internal Clock External Clock Data per Clock L1 Cache L2 Cache L3 Cache Socket
Athlon 64 3200+ 2 GHz 200 MHz 2 128 KB 1 MB No 754
Athlon 64 3400+ 2,2 GHz 200 MHz 2 128 KB 1 MB No 754
Athlon 64 3800+ 2,4 GHz 200 MHz 2 128 KB 512 KB No 939
Athlon XP 2800+ 2 GHz 166 MHz 2 128 KB 512 KB No 462
Athlon XP 3200+ 2,2 GHz 200 MHz 2 128 KB 512 KB No 462
Duron 1.8 GHz 1,8 GHz 133 MHz 2 128 KB 64 KB No 462
Pentium 4 3.2 GHz 3,2 GHz 200 MHz 4 158 KB 512 KB No 478
Pentium 4 3.4 GHz 3,4 GHz 200 MHz 4 158 KB 512 KB No 478
Pentium 4 Extreme Edition 3.2 GHz 3,2 GHz 200 MHz 4 158 KB 512 KB 2 MB 478
Pentium 4 Extreme Edition 3.4 GHz 3,4 GHz 200 MHz 4 158 KB 512 KB 2 MB 478
Sempron 2800+ 2 GHz 166 MHz 2 128 KB 256 KB No 462
Sempron 3100+ 1,8 GHz 200 MHz 2 128 KB 256 KB No 754
Sempron 3400+ 2 GHz 200 MHz 2 128 KB 256 KB No 754

Processors for the socket 754 and socket 939 platforms have two external clocks, one used by the memory bus and another used by the HyperTransport bus. The clock mentioned on the above table is the memory clock.

How We Tested

During our benchmarking sessions, we used the configuration listed below. Between our benchmarking sessions the only different device was the CPU being tested and also the motherboard, since the tested CPUs use different socket types.

Hardware Configuration

Software Configuration

Used Software

We adopted a 3% error margin; thus, differences below 3% cannot be considered relevant. In other words, products with a performance difference below 3% should be considered as having similar performance.

Processing Performance: PCMark04

PCMark04 measures the system performance by running a set of 10 tasks, like video compression, audio conversion, file cryptography, math calculations, etc, giving the final result in a proprietary unit. The results from this program we show on the graph below.

Sempron 3400+

Sempron 3400+ achieved a performance similar to Athlon 64 3200+. As we mentioned, the only difference between these two processors is the amount of L2 memory cache (256 KB on Sempron and 1 MB on Athlon 64), as both run at 2 GHz.

It was 4.01% faster than Athlon XP 3200+, 9.49% faster than Sempron 3100+, 10.27% faster than Athlon XP 2800+, 15.60% faster than Sempron 2800+ and 33.60% faster than Duron 1.8 GHz.

It lost to Athlon 64 3400+, which was 8.39% faster, Pentium 4 3 GHz, which was 16.78% faster, Athlon 64 3800+, which was 16.93% faster, Pentium 4 3.2 GHz, which was 22.19% faster, Pentium 4 Extreme Edition 3.2 GHz, which was 25.57% faster, Pentium 4 3.4 GHz, which was 28.11% faster and Pentium 4 Extreme Edition 3.4 GHz, which was 32.92% faster.

Overall Performance: SYSmark2002

We measured the overall performance of the tested processors using SYSmark2002, which is a software that simulates the use of real-world applications. Thus, we consider this the best software to measure, in practical terms, the system performance.

The benchmarks are divided into two groups: Internet Content Creation and Office Productivity. The programs used on Internet Content Creation batch are the following: Dreamweaver 4.0, Photoshop 6.0.1, Premiere 6.0, Flash 5 and Windows Media Encoder. The programs used on Office Productivity batch are the following: Word 2002, Excel 2002, PowerPoint 2002, Outlook 2002, Access 2002, Netscape Communicator 6, NaturallySpeaking 5, VirusScan 5.13 and WinZip 8.0.

The software delivers specific results for each batch and also an overall performance result, all in a specific SYSmark202 unit.

Sempron 3400+

On SYSmark2002 Sempron 3400+ achieved a performance similar to Sempron 3100+ and Athlon XP 2800+. It was 13.57% faster than Sempron 2800+ and 42.61% faster than Duron 1.8 GHz.

Athlon XP 3200+ was 4.38% faster, Athlon 64 3200+ was 11.16% faster, Athlon 64 3400+ was 24.30% faster, Pentium 4 3 GHz was 25.10% faster, Pentium 4 Extreme Edition 3.2 GHz was 27.89% faster, Athlon 64 3800+ was 28.29% faster, Pentium 4 3.2 GHz was 32.67% faster, Pentium 4 3.4 GHz was 33.07% faster and Pentium 4 Extreme Edition 3.4 GHz was 40.64% faster than the reviewed CPU.

Analyzing Office Productivity results, Sempron 3400+ achieved a performance similar to Athlon XP 2800+, being 3.24% faster than Sempron 2800+ and 22.44% faster than Duron 1.8 GHz. Sempron 3100+ was 3.14% faster, Athlon XP 3200+ was 7.85% faster, Athlon 64 3200+ was 8.90% faster, Pentium 4 Extreme Edition 3.2 GHz was 15.18% faster, Pentium 4 3 GHz was 16.75% faster, Pentium 4 3.4 GHz was 21.47% faster, Pentium 4 3.2 GHz was 24.61% faster, Athlon 64 3400+ was 26.18% faster, Athlon 64 3800+ was 26.18% faster and Pentium 4 Extreme Edition 3.4 GHz was 27.75% faster than Sempron 3400+.

On Internet Content Creation batch, Sempron 3400+ achieved a performance similar to Athlon XP 3200+, being 5.08% faster than Sempron 3100+, 6.43% faster than Athlon XP 2800+, 25.38% faster than Sempron 2800+ and 66.33% faster than Duron 1.8 GHz.

Athlon 64 3200+ was 13.29% faster, Athlon 64 3400+ was 21.75% faster, Athlon 64 3800+ was 29.61% faster, Pentium 4 3 GHz was 33.53% faster, Pentium 4 3.2 GHz was 40.79% faster, Pentium 4 Extreme Edition 3.2 GHz was 41.69% faster, Pentium 4 3.4 GHz was 45.62% faster and Pentium 4 Extreme Edition 3.4 GHz was 54.38% faster than the reviewed CPU.

3D Performance: 3DMark2001 SE

On the graph below we can see the results obtained on 3DMark2001 SE. This program measures performance by simulating DirectX 8.1 games. It continues to be a good program to evaluate the performance of games from previous generation, programmed using DirectX 8 API. This program was run in its default configuration.

Sempron 3400+

In this program Sempron 3400+ achieved a performance similar to Pentium 4 3.4 GHz and Pentium 4 3.2 GHz. It was 5.77% faster than Sempron 3100+, 6.28% faster than Pentium 4 3 GHz, 10.89% faster than Athlon XP 3200+, 16.49% faster than Athlon XP 2800+, 25.21% faster than Sempron 2800+ and 48.41% faster than Duron 1.8 GHz.

It lost to Athlon 64 3200+, which was 3.25% faster, to Pentium 4 Extreme Edition 3.2 GHz, which was 4.98% faster, to Athlon 64 3400+, which was 5.87% faster, to Pentium 4 Extreme Edition 3.4 GHz, which was 5.92% faster and to Athlon 64 3800+, which was 9.77% faster.

3D Performance: 3DMark03

3DMark03 measures performance by simulating DirectX 9 games, which are more recent games. The results you can see on the graph below.

Sempron 3400+

In this program, Sempron 3400+ achieved a performance similar to almost all processors included in our benchmark. It was 4.54% faster than Athlon XP 2800+, 5.40% faster than Sempron 2800+ and 8.38% faster than Duron 1.8 GHz. It was slower only to Athlon 64 3800+, which was 5.01% faster than the processor being reviewed.

3D Performance: Quake III

We used the demo four available on version 1.32 of Quake III to make our benchmarking with this game. We ran this demo three times at the game’s default configuration and we picked the middle value for our comparisons, i.e., we discarded the highest and the lowest values.

The video configuration was left on the factory defaults for a very simple reason: to measure the influence of the processor on the system 3D performance. If we raised the image quality settings, we would be measuring the influence of the video card instead.
Even though it is an old game, in this game the influence of a larger memory cache is clearly seen on the final results, being a good program to evaluate this feature.

Sempron 3400+

In this game Sempron 3400+ was 5.85% faster than Pentium 4 3 GHz, 7.63% faster than Sempron 3100+, 43.66% faster than Athlon XP 3200+, 53.42% faster than Athlon XP 2800+, 64.51% faster than Sempron 2800+ and 98.85% faster than Duron 1.8 GHz.
Pentium 4 3.2 GHz was 3.73% faster, Pentium 4 3.4 GHz was 7.95% faster, Athlon 64 3200+ was 10.29% faster, Athlon 64 3400+ was 17.12% faster, Athlon 64 3800+ was 21.63% faster, Pentium 4 Extreme Edition 3.2 GHz was 25.04% faster and Pentium 4 Extreme Edition 3.4 GHz was 26.95% faster than the reviewed processor.

Memory Performance: Sandra

Usually we do not include memory benchmarking in our CPU reviews because usually this test measures the chipset performance and not the CPU performance. But since on socket 754 and socket 939 processors the memory controller is integrated in the CPU – and not in the chipset as it happens in other platforms – we’d like to check the performance of the Sempron 3400+ embedded memory controller.

On the graph below you will find the results in MB/s. We also included on the graph the values for the maximum theoretic transfer rate of DDR300/PC3200 memories (3,200 MB/s) and also their maximum theoretic transfer rate when running at dual channel (6,400 MB/s).

Socket 939 processors (in our case, our Athlon 64 3800+) and the motherboards we used together with Athlon XP, Duron and Intel CPUs access memory using DDR Dual Channel mechanism, theoretically accessing memory up to 6,400 MB/s.

Sempron 3400+

Sempron 3400+ used 96.69% of the available bandwidth, which is an outstanding result, since it shows that its embedded memory controller is capable of using almost all bandwidth available.

Just to give you a better understanding, Sempron 2800+, which is a socket 462 processor – thus its memory access is controlled by the chipset, not by the CPU – achieved a 2,393 MB/s transfer rate, using 74.78% of the available bandwidth. But since this processor was installed in a nForce 2 motherboard, which has the Dual Channel feature, actually it used only 37.39% of the theoretic available bandwidth. So, our Sempron 3400+ was more efficient in single channel configuration than Sempron 2800+ in dual channel configuration.

Overclocking and Conclusions

Overclocking

We could overclock this CPU up to 225 MHz (2.25 GHz internally), a 12.5% increase over the default configuration of 200 MHz (2 GHz internally). That’s a great result for an AMD processor.

We didn’t play with memory nor voltage configurations, so you may be able to achieve even better results.

Conclusions

We were really impressed by this new CPU from AMD. Although it is theoretically targeted to entry-level market, in our opinion Sempron 3400+ is a mid-range CPU, which will please all users, especially gamers that want a high-performance system with an affordable price.

It was faster than Athlon XP 3200+ – the fastest CPU for the socket 462 platform and which was a high-end processor once – in almost all benchmarks, really great news.

Since this model already has the 64-bit instructions, which once was available only on Athlon 64, it is perfect for the user that doesn’t want (or can’t afford) to buy an Athlon 64.

Actually, as we mentioned, Sempron 3400+ is an Athlon 64 3200+ with less memory cache.

We also need to mention the terrific job AMD has being doing by integrating the memory controller inside the CPU. Sempron 3400+ is capable of using almost all memory bandwidth available, which doesn’t happen on Intel CPUs or AMD CPUs targeted to the socket 462 platform.

Our only constructive critic goes to AMD’s numbering system. Labeled as 3400+, you could think that this CPU is comparable to Athlon 64 3400+ or even to a Pentium 4 3.4 GHz, which is not the case – even though in some benchmarks the performance difference between the afore mentioned processors wasn’t so big. Anyway, this numbering system serves only for comparison between processors within the same family. However, if you think on Athlon XP, it makes sense, since Sempron 3400+ is really faster than Athlon XP 3200+.

Originally at http://www.hardwaresecrets.com/article/Sempron-3400+-Review/200


© 2004-14, Hardware Secrets, LLC. All Rights Reserved.

Total or partial reproduction of the contents of this site, as well as that of the texts available for downloading, be this in the electronic media, in print, or any other form of distribution, is expressly forbidden. Those who do not comply with these copyright laws will be indicted and punished according to the International Copyrights Law.

We do not take responsibility for material damage of any kind caused by the use of information contained in Hardware Secrets.