![[H]ardOCP](/images/blank.gif){kind=small}
- Date:
- Monday , May 09, 2005
- Author:
- Kyle Bennett
- Google +1
AMD Athlon 64 X2 Preview
A look at what AMD's new dual core 4800+ processor is going to bring to your computing table. We directly compare and contrast it with Intel's dual core Pentium Extreme Edition 840 in our basic benchmarks and real world heavy duty multitasking.
Test Systems
AMD
AMD Athlon 64 X2 4800+ (2.4GHz Dual Core Processor), AMD Athlon 64 4000+ (2.4GHz Single Core Processor), ASUS A8N-SLI Deluxe (BETA BIOS), GeForce 6800 Ultra (400/1.1) 61.45 & 71.89 drivers used, 2 X 512MB Corsair PC3200LL Pro DDR400 (2.2.2.5), 2 X Western Digital 36GB Raptor Hard Drives configured in a RAID0 array, Windows SP1 (single threaded benchmarks), SP2 (multithreaded benchmarks), nForce4 chipset driver v 6.53.
Intel
Intel Pentium Extreme Edition 840 (3.2GHz Dual Core Processor), Intel Pentium 4 3.2GHz, ASUS P5WD2(i955X w/BETA BIOS), GeForce 6800 Ultra (400/1.1) 61.45 & 71.89 drivers used, 2 X 512MB Corsair PC5400UL Pro DDR2-666 (4.4.4.5), 2 X Western Digital 36GB Raptor Hard Drives configured in a RAID0 array, Windows SP1 (single threaded benchmarks), SP2 (multithreaded benchmarks), Intel chipset driver v 7.0.0.1019.
Software Loaded
Norton Antivirus 2003 (all updates), Mirra Personal Server application, ASUS Probe, ASUS Update, NVIDIA nTune (on nForce 4 motherboard), Anapod Explorer with attached iPod, MS Office 2000, ICQ 5, MSN Messenger, Steam, and TIVO server.
Multi Threaded Benchmarks
While all of the benchmarks here are not multithreaded when run alone, we do couple them with other benchmarks in order to explore the depth of the processors in question.
Also, for our Multi Threaded Benchmarks we thought we would aquire a more “real world” approach by loading background applications on the system as listed above. There is no doubt that when you introduce these elements that all hopes of having “scientific” results go right out the window as some of the application above are utilizing system resources at different times and in differing amounts. The benchmarks below were run multiple times to assure that what we are showing below are the “most common” result.
Adobe® Photoshop® Elements
I tried to find a Photoshop Elements test that would truly capture the value of dual core computing, but I failed. The test above shows the length of time it took to Gausian Blur (10 pixel spread) a 14MB Jpeg that was 9372x9372 pixels. I think this is important in that it actually illustrates that the Photoshop Elements layman is likely to see little or no difference between the AMD Athlon 64 X2 4800+ and the Intel Pentium Extreme Edition 840. Should any of you have real world testing to share with us that would outline this, we of course would love to hear your suggestions.
Adobe® Premiere® Elements
Premiere Elements on the other hand is a totally different story. I pulled a 3.2GB section of digital video off of my camcorder and used Premiere Elements to encode it to a 130MB WMV file. The AMD Athlon 64 X2 4800+ and the Intel Pentium 840 come in at a draw in both single core and dual core numbers. Certainly if you encode large sections of video, this being only a 10 minute clip, you can see where the time saved would add up quickly.
We will look at Premiere Elements again below in a multitasking environment.
Windows Movie Maker
The Windows Movie Maker that comes with Service Pack 2 sees a huge benefit from a dual core processor as well. As opposed to the Premiere Elements results above, we do see a larger delta on the single core processors that is most likely due to the Intel Pentium 4 3.2GHz processor's HyperThreading abilities.
Multi Tasking Benchmarks
While video encoding is likely to be the “only” big desktop application we see currently in the dual core desktop processor arena, the true value of dual core processors will be found by those that utilize “heavy duty” multitasking. That is not to say that dual core benefits will not be seen by the “light duty” multitasking either, but simply not to the same degree. For this set of benchmarks we used the Video Stress Test that is included with the first person shooter game, Counter Strike: Source. To keep in line with more “real world results,” we set up Counter Strike: Source at 1600x1200 with 4XAA and 8XAF as you would play with on today’s top end systems.
Counter Strike: Source Video Stress Test
What you are seeing above is not a frame rate cap, but rather the limitations of the video card or better termed as a video card bottleneck. While some might argue this is not the way to go about benchmarking, they would be exactly right. What we are trying to show here is the real world multitasking impact graphed the best we could. If this were simply a CPU utilization test, we would just turn off the eye candy in the game and scale the resolution to 640x480 as we do in our single threaded tests.
CS:Source w/DivX Encode
What we did here was start a DivX encode and then run the CS:S Video Stress Test. So our results here from the CSS:S VST were gathered while our DivX encode was running in the background. Keep in mind that the DivX encode and CS:S VST are both single threaded applications. Another thing to remember is that our Intel Pentium Extreme Edition 840 has a total of 4 usable threads although two of those are virtual processors supplied by Intel’s HyperThreading technology. That said, this is the first benchmark we have seen the Intel 840 stand out in any way and even now the performance gains are likely to never be felt in real world usage. We will cover more about that on the next page as well.
What really catches your eye here is how poorly the Athlon 64 4000+ single core processor is performing. The Pentium 4 3.2GHz processor simply kills our trusty Athlon 64 4000+ in this situation. Again this is due to the Intel HyperThreading technology present on the Pentium 4 processor. While it is not doing hardly any more real work on our DivX encode, it is efficiently using the Windows scheduler to better prioritize what applications need attention, this one being our CS: VST. While this is not a proper technical explanation, it is a good way to look at it overall.
CS:Source w/WMV Encode
Here again we are running a video encode, this time using Premiere Elements, which is a multithreaded application. So now we have one multithreaded application and one single threaded application fighting for CPU cycles. You can see our performance hit compared to the DivX encode above. Certainly Premiere is utilizing more cycles from both cores on both processors. Our CS:S VST still in no way does poorly here with our dual core processors.
Our performance on our single core processors is vastly different though. Our Athlon 64 4000+ single core processor is basically rendered (pun intended) useless, while our Pentium 4 3.2 with HyperThreading still is able to prioritize its workload better giving us needed results in what would be considered by most to be the more important task at hand.