General Performance and Media Encoding
Winstones follow an expected pattern with Business Winstone 2004 performing about the same as a similar speed single core CPU. Multimedia Content Creation, however, has some components which benefit from the improved multitasking of the dual-core CPU and it performs better on the dual-core processor. Both Winstones scale very well with CPU speed, so we see a sizable boost in performance at the overclocked 2.7GHz setting.
PCMark2004 is astonishingly sensitive to the multitasking advantages of the dual core 4200+. At stock speed the 4200+ outperforms the higher speed 4000+ by over 35%. PCMark2004 has performed best on the Intel platform in the past, but that advantage appears gone with the launch of the dual-core Athlon 64. We were unable, however, to get PCMark2004 to work at overclocked speeds on the X2 processor. Tests would run, but PCMark04 refused to generate a PCMark2004 score even at modest overclocks. Since Futuremark is preparing to launch PCMark2005 shortly, we suspect this anomaly will be corrected in the new version.
Our AutoGK 1.6 Media Encoding benchmark dramatically shows the impact of Dual-Core on Encoding speed. Comparing the 4200+ to the higher speed single-core 4000+ we find encoding jumps to 68.6 FPS from 48.5 on the single-core. This is a 41% performance improvement. Overclocking the 4200+ to 2.7GHz raises the encoding speed to 82.8 FPS.
Media Encoding was one area where Intel continued to enjoy a small advantage, but it certainly appears the dual-core Athlon 64 will be an excellent choice for media encoding.
Facebook
Twitter
RSS
53 Comments
View All Comments
Qarl - Sunday, June 26, 2005 - link
As others have already posted, my two biggest questions are:Why wasn't the overclocked 4200+ benchmarked against a stock 4800+?
Why wasn't a 4400+ used instead of a 4200+? It has double the cache and is only slightly more expensive.
redhatlinux - Sunday, June 26, 2005 - link
Another great review from the BossDigitalDivine - Sunday, June 26, 2005 - link
What paper launch, the x2s have been in newegg for a weekval - Saturday, June 25, 2005 - link
paper launches sucks!AMD have no cpus
at80eighty - Saturday, June 25, 2005 - link
#45 Wesley:- "Specializing in certain review areas, as we do at AnandTech, makes you a lot less stupid and easily duped than you might imagine"a very nicely veiled jab there Wesley. Hope the recepients have skulls thin enough for it to trickle in!! Kudos!
at80eighty - Saturday, June 25, 2005 - link
wtf is up with these bitchy bitches bitching about the 'integrity' of AT these days?is it the replacement for the 'Soviet Russia.." cliche ???
boban10 - Saturday, June 25, 2005 - link
very nice review, im very happy that you tested this cpu and overclocked it. thanks.Icehawk - Friday, June 24, 2005 - link
Sheesh, I thought we'd put Anandtech's integrity to bed by now? I have no concerns.Wesley - I hope you can get a 4400+ as I am very curious to see what the results look like. The small price bump over the 4200+ makes it pretty appealing, especially if it can OC as well and provides a but of a performance bump.
I too would LOVE to see some 1gb vs 2gb RAM comparisons with various configurations.
Wesley Fink - Friday, June 24, 2005 - link
#42 & #35 - To bring the news to you first, our ONLY choice is manufacturer-supplied samples. When we test there is usually nothing available in the retail channel.In this case we had one one Retail 4200+ and one manufacturer-supplied 4200+. They performed within 5MHz of each other in overclocking, which is equal performance. Our performance with both processors is lower than sites that publish a screen capture of an OC speed and don't run any benches, so we stand by our results on air cooling.
The "Conspiracy" theory sounds good, and is usually spouted by the manufacturers who didn't do well in a roundup. In a truly competitive world like computer components there is no point to providing "cherry" parts to reviewers. If people buy a product due to a review that shows x performance and their part won't do the same they RMA the part. RMA's cost manufacturers lots of money. A high RMA rate will quickly kill any profits on a product.
Even memory - a business based on binning or hand-picking of parts for performance - has settled down on cherry parts. Manufacturers who tried that got burned on RMAs and came back the second time with representative parts.
There is always variation in overclocking results, but huge variation from reported results are someone who doesn't know how to overclock, a change in parts used (which is why overclockers are big on production weeks), or a change in binning (selection criteria). Specializing in certain review areas, as we do at AnandTech, makes you a lot less stupid and easily duped than you might imagine.
cryptonomicon - Friday, June 24, 2005 - link
*adding*here is the 3g on air
http://www.xtremesystems.org/forums/showthread.php...