AMD Phenom X4 e9350 (2.0 GHz Quad Core)
AMD’s Phenom X4 processor has been around for almost a year, though it took AMD a few months to fix some issues with the first product generation. The current products are all fine and can be recommended, as AMD is delivering great bang for the buck despite the fact that Core 2 Quad processors are typically faster. The Phenom X4 9950 quad core, which runs at 2.6 GHz, costs only $220 and offers great performance for multi-threaded applications, while the 2.2 GHz Phenom X4 9550 can be found for less than $160. A Core 2 Quad processor typically costs at least 50% more, but the performance gains are nowhere near that level.
Phenom X4 e9350 is a 65-W Part
We were more interested in a low power version of the Phenom X4 quad core, because AMD has had versions that fit into the 65 W power envelope, and because that allows a direct comparison with Intel’s Core 2 Duo processors, which are also rated at 65 W. Considering that the Phenom processors are manufactured using the 65 nm process, and that there is no Intel Core 2 Quad processor that would be specified at less than 95 W—Extreme Editions are rated at 130 W or 136 W—this is an interesting product.
However, we need to point out that a processor’s TDP only represents the maximum thermal output specified by the manufacturer. It doesn’t mean that a particular model has to reach this power level, and there are many CPUs that do not even come close to their maximum thermal design points. In addition, a processor specified at 65 W TDP does not have to be more efficient than, say a 95 W part, when it runs under a low or medium load, because TDP does not specify idle power. All the TDP says is that the 65 W quad core Phenom X4 e9350 provides 2.0 GHz of four-core performance, while not exceeding its 65 W limit.
Aside from its low-power specification, the e9350 isn’t different from the other Phenom models. All have 64+64 KB instruction and data cache, as well as 512 KB of second level cache per processing core. In addition to that, AMD implemented a 2 MB third level cache as well, which is shared by all of the processing cores. Avoid the Phenom X4 models 9500 and 9600, as these are based on the B2 stepping that is known for issues with the so-called translation lookaside buffer (TLB). The B3 stepping Phenom processors are free of major bugs, and they all carry model numbers ending with -50, such as the e9350. All regular Phenom X4 processors are specified at 95 W, 125 W or 140 W TDP, while the e9100 and e9350 stick to the 65 W envelope.
All Socket AM2+ processors come with an integrated memory controller for DDR2 memory. Phenoms support memory clocks of up to 667 MHz, resulting in DDR2-1066 speeds based on double data rate technology. New 45 nm processors based on Socket AM3 with 6 MB L3 cache will support DDR3 memory, but will probably not be available before 2009.
Phenom X4 Dominates Multi-Threaded Benchmarks
As expected, the multi-core advantage is only on paper for the majority of benchmarks, giving the Core 2 Duo at its fast 3.16 GHz an advantage in many software titles that simply seem to perform best at a good balance between two cores and high clock speed. Thread-optimized titles such as AVG antivirus, Fritz 11 chess, the Mainconcept H-264 encoder and WinRAR 3.8 run faster on the 2.0 GHz Phenom X4 than they do on the 3.16 GHz Core 2 Duo. Many other titles, though, provide better performance on the Core 2 solution despite their optimization for multiple cores.
AMD Loses The Efficiency Battle
AMD loses both power consumption disciplines to Intel, which doesn’t come as a surprise: four cores simply require more power than two. This is very much like comparing a big V8 engine and a small 2-liter 4-cylinder, so we don’t want to complain about it. As the synthetic benchmarks prove, the Phenom X4 does provide the better performance, but it is not efficient in most of the application scenarios, including our performance per watt testing using SYSmark 2007 Preview and a 3D game cycle using Crysis.
"The Mainconcept 1.5.1 benchmark converts MPEG2 FullHD video into the H.264 format. Although the benchmark scales well with as many as eight cores—we used an Intel Skulltrail system to try this—the 2.0 GHz quad core isn’t enough to beat Intel’s 3.16 GHz dual core."
According to the graph you show us, it is...
Do you even have an editor anymore? This article is yet another nail in toms coffin. The graphs are wrong (two items on each graph and you still manage to swap them over), the words are wrong (eg 3.16 quad!!!) and worse still the article is pointless.
Here's another apples to oranges for you:
I own a motorbike and a car, both do 40mpg and both cost the same. But wait the car has more seats AND more wheels, its safer in an accident too so it must be better than the bike. oh no did i forget to look at the bikes good points never mind.
And why choose the low power version, you could have used a cheaper high power version and underclocked/undervolted to reduce power OR accept the fact that four cores should use more power than two but seeing as you didn't want to show the quad win anything i guess you can't accept that.
I guess whoever is in charge these days is only concerned with ad revenue not content or integrety
"Supreme Commander shows the same results: it runs much faster on the Intel dual core than it does on AMD’s quad core. Since the performance difference is 80%, the clock speed difference alone isn’t enough to account for the tremendous difference."
Wrong again. According to graph, Phenom is faster than C2D, not the other way around.
This artical is a little to bias to Intel for my liking. When applications that do support 4 cores are tested and unsuprisingly the AMD chip wins, they dont praise it, they just praise the intel chip instead for coming a close 2nd. While all the applications that dont support 4 cores get praise for Intel for winning and not to AMD for coming second.
The whole artical makes no sence about what it does compare.
My conclusion for the artical the E8500 3.16 GHz wins on all single/double core applications but when 4 cores are used the AMD Phenom X4 e9350 2.0 GHz wins. Which is what we should expect anyway.
This article is interesting from the standpoint of software, the main thing I see from this is just how little use applications currently make of extra cores.
Nehalem...sorry Core i7...yes, yes that's much better...will surely have an impact on applications use of multithreading. Or will it? with 'turbo-mode' perhaps there is no need for software to use those extra cores.
Annoyingly left out was the overclocking performance of these two processors. As we know Intel's current chips annihilate the competition in overclockability providing extra Hertz for just a few hours time. These E8500 are easily hitting 4GHz I do tend to wonder if the advantage the AMD had in some tests wouldnt be eliminated when both chips were fully OC'd.
What i was interesting in is,
having the benchmarks run with a current antivirus software "allways on", as it should, at least, be users default configuration.
thanx
Tom's, please replace your eeditor with any small child. I could have seen the mistakes here when I was 9.
Tom's, please replace your eeditor with any small child. I could have seen the mistakes here when I was 9.
However, I did make a typo when making that comment
A fast dual-core is best for games
An energy-efficient quad is good for use as a home/media server
^ I did that without looking at the article. Am I right? Hang on... yep, pretty much.
How'd I manage that? Well, it ain't cuz i'm psychic, that's for sure. It's because WE KNEW ALL THIS ALREADY TOM!
Games are more responsive to raw power and are less heavily threaded - most are threaded for dual-core, but as of yet relatively few can make good use of a massively multicore platform (except well-coded PS3 games, and that's a different subject entirely!). Modern applications, especially graphical, media (encoders!) and file-based (server/AV) are designed to split and combine threads on-the-fly and with Vista in tow really need a quad to crunch them efficiently in the background.
So what was the point of this article again exactly? Telling us what's very common knowledge?
I was hoping u did abit more the real workstation apps. particularly Virtualization. And running on Vista 64 bit.
For applications that are dependent on core speed, the AMD does not do so bad. Let's also remember that the AMD chip is being used in a chipset that offers HD playback and half-decent 3D game support..the same cannot be said for Intel-based chipsests. I would also go with AMD for a cheap, fast server..where the architecture comes into it's own (especially core-to-core and memory performance).
That's because Intel doesn't have a native quad core. It's also the reason why some of the world's fastest super-computers rely on AMD hardware. Intel may have caught up with AMD in the desktop sector (bar chipsets and graphics cards), but the server/cluster/super-computer sectors use AMD for a reason. The people knocking AMD should do well to remember that, perhaps.
you guys are so boring! i hate this website now all you dorks that are into computers need to get a life, play some basketball, and listen to some rock n roll like i am right now! im a seventh grader that is more cool than all of you put together!