[ Print Article! ]

AMD Athlon 64 X2 4800+ 65-nm Brisbane Preview
December 19, 2006 Brandon Sandman Bell

Summary: AMD's released a flurry of new CPUs in the past few weeks. Up for eval today is the X2 4800+. This CPU is based on AMD's new 65-nm Brisbane core and runs at 2.5GHz. With the smaller manufacturing process, the X2 4800+ should deliver better thermals and power, but does it? Yes! In fact, you may be surprised by how much. What about overclocking? You'll have to read today's article to find out!


IntroductionPage:: ( 1 / 13 )


For a CPU manufacturer like AMD or Intel, the manufacturing process is actually one of the CPU’s most important features. This is because the manufacturing process can often play a large role in not only the final CPU clock speed, but most importantly, the manufacturing cost of the CPU itself: the smaller the manufacturing process, the cheaper the CPU is to manufacture.

It has been over two years now since AMD introduced their first 90-nm CPUs. At the time these Athlon 64 CPUs were based on the Winchester core, but AMD has since gone on to produce newer, more advanced cores such as Venice/San Diego for single-core processors, as well as Toledo and most recently, Windsor for dual-core CPUs. Each of these new cores brought different innovations to the table that have helped AMD remain more than competitive with Intel over the years, but the bottom line is that they were all still 90-nm processors.

With the CPU industry moving at a breakneck pace, two years is an awfully long time to be stuck on the same process, plus Intel’s been mass producing their own CPUs at 65-nm for roughly a year now. Fortunately for AMD, their 65-nm manufacturing process has arrived and today we’re taking a look at one of the first CPUs based on the new process: AMD’s Athlon 64 X2 4800+.

[image]

<% print_image("01"); %><% print_image("02"); %>

Same old CPU

Historically when breaking in a new manufacturing process, AMD’s been pretty conservative when it comes to new features. Rather than use the opportunity to increase cache size (as Intel often does when introducing a new process) or crank up the clocks, AMD typically plays it safe, sticking with a die shrink only.

For the 65-nm transition, AMD sticks to this strategy once again – if you were expecting a significantly improved core with new features, we’re sorry to say that you’ll be disappointed. AMD’s focus right now is on making tweaks and improvements to the process itself, then once AMD’s got everything down at 65-nm, they’ll introduce enhanced CPU cores based on that tweaked, more mature 65-nm process. Without a doubt, the most talked about of these improved CPU cores is Barcelona (also known as K8L), but we won’t see that CPU on the desktop until the second half of 2007.

In other words, since AMD’s move to 65-nm is only a die shrink, don’t expect 65-nm CPUs to deliver faster performance than their 90-nm equivalents. Transistor count between AMD’s 90-nm and 65-nm CPUs is also the same at 153.8 million transistors.

What you will get from today’s 65-nm AMD CPUs though is better thermals. This is because the new 65-nm CPUs boast lower power consumption figures. AMD lists a peak power rating of 65W across the board for all their 65-nm CPUs. In comparison, AMD’s 90-nm CPUs ran at 89W max, with energy efficient models at 65W and 35W.

With the smaller process, die size is also reduced, down from 183mm2 to 126mm2 on AMD’s 65-nm processors. The new 65-nm CPU’s are built on AMD’s “Brisbane” core and all announced models to date ship with 512K L2 cache per processing core (1MB total L2 cache). Model numbers announced to date include the 5000+, 4800+, 4400+, and 4000+. The following chart summarizes AMD’s current CPU lineup as it stands today, for simplicity we’re just listing X2 and FX CPUs:

AMD Athlon 64 CPU Lineup As Of 12/18/06
Model NumberClock Speed (GHz)L2 Cache SizeManufacturing ProcessThermal Design PowerPrice
FX-743.0GHz2MB90-nm125W$999/pair
FX-722.82MB90-nm125W$799/pair
FX-702.6GHz2MB90-nm125W$599/pair
FX-622.8GHz2MB90-nm125W$713
X2 5600+2.8GHz2MB90-nm89W$505
X2 5400+2.8GHz1MB90-nm89W$485
X2 5200+2.6GHz2MB90-nm89W$403
X2 5000+2.6GHz1MB90-nm89W$301
X2 5000+ Brisbane2.6GHz1MB65-nm65W$301
X2 4800+2.5GHz1MB65-nm65W$271
X2 4600+ Energy Efficient2.4GHz1MB90-nm65W$240
X2 4600+2.4GHz1MB90-nm89W$240
X2 4400+2.3GHz1MB65-nm65W$214
X2 4200+2.2GHz1MB90-nm89W$187
X2 4200+ Energy Efficient2.2GHz1MB90-nm65W$187
X2 4000+2.1GHz1MB65-nm65W$169
X2 3800+ Energy Efficient SFF2.0GHz1MB90-nm35WN/A
X2 3800+ Energy Efficient2.0GHz1MB90-nm65W$152
X2 3800+2.0GHz1MB90-nm89W$152


As you can see, AMD isn’t charging a premium for their new 65-nm Brisbane processors. Unfortunately though when it comes to retail availability we’ve been told that most of AMD’s Brisbane CPUs are going to OEMs right now, the retail channel won’t see Brisbane arrive en masse until the beginning of 2007. Despite this, we were eager to see how the chip stacked up to similar X2 CPUs as well as the Core 2 from Intel.



System SetupPage:: ( 2 / 13 )

System Setup


AMD Athlon 64 X2 5000+ (2.6GHz)
AMD Athlon 64 X2 4800+ (2.5GHz)
AMD Athlon 64 X2 4600+ (2.4GHz)

Intel Core 2 Duo E6600 (2.4GHz)

EVGA nForce 680i
ASUS M2N32-SLI Deluxe

2GB Corsair TWIN2X2048-6400C3

EVGA GeForce 8800 GTX w/ForceWare 97.44

Windows XP Professional with Service Pack 2

DirectX 9.0c


Benchmarks

LAME MT MP3 Encoding (MS Compiler)
DivX Converter
Windows Media Encoder 9
3DMark06
F.E.A.R. 1.08
Pacific Fighters 4.04
Quake 4 1.2
Oblivion
Company of Heroes 1.3
Cinebench 9.5




Media encoding and renderingPage:: ( 3 / 13 )

Microsoft Windows Media Encoder 9



DivX Converter



LAME MT MP3 Encoding



Cinebench 9.5



Valve Multi-core Particle Simulation Benchmark





3DMark 06Page:: ( 4 / 13 )

Futuremark 3DMark05








F.E.A.R.Page:: ( 5 / 13 )

F.E.A.R.







Quake 4Page:: ( 6 / 13 )

Quake 4







Company of HeroesPage:: ( 7 / 13 )

Company of Heroes







Pacific FightersPage:: ( 8 / 13 )

Pacific Fighters







OblivionPage:: ( 9 / 13 )

Oblivion










Temps and Power UsagePage:: ( 10 / 13 )

Power usage





Temps






Overclocking: GamesPage:: ( 11 / 13 )







Notes

With its new 65-nm manufacturing process, we were quite eager to see how far we could push our X2 4800+ chip. Unfortunately however we weren’t able to break the 3.0GHz mark during our OC’ing endeavors – the CPU wouldn’t go any further than 2.838GHz at 1.5625V, the max voltage supported by the motherboard. Also note that we used a Scythe Ninja Plus RevB for cooling.

[image]

<% print_image("03"); %>

We think armed with a little more voltage we could’ve possibly got the CPU a little further, as the chip would boot into Windows at 3.0GHz (240MHz HT) but would lock up when running apps. If running with a lower voltage, say 1.5V, the system wouldn’t even get past the Windows splash screen. Lowering the HT speed also didn’t help. Again, we tried using ASUS’ excellent CrossHair motherboard with the 4800+, but unfortunately it just wouldn’t run the chip at the proper clock speeds, even when we manually set the HT speed at 200MHz for instance the motherboard still insisted on running it at 240MHz. And at stock settings the motherboard ran the CPU at 2.4GHz instead of 2.5GHz.



Overclocking: Rendering and EncodingPage:: ( 12 / 13 )








ConclusionPage:: ( 13 / 13 )


With that being said though, there are some conclusions that we can take away from today’s 65-nm processors.

For one thing, they generate less power than their 90-nm counterparts. In our testing total system power consumption at equal clock speeds with our X2 4800+ was 11W lower under load than the equivalent X2 CPU at 90-nm. 11W doesn’t sound like much, but in practice our X2 4800+ CPU generated significantly less heat: temps were down up to 26 degrees Celsius under load.

That’s a pretty nice improvement, particularly if you’re an enthusiast that’s concerned about temps, or you’re looking for a good, cool-running CPU for use in a media center PC. The X2 4800+, and AMD’s 65-nm CPUs in general, are a huge improvement over the stock X2 (non-EE) for this type of application.

That’s the biggest story for end user’s today: lower power consumption equals less heat.

For AMD, the new 65-nm process has even bigger implications.

For starters there’s the manufacturing advantage the new process brings. With the new 65-nm process, the CPU’s die size is down from 183mm2 at 90-nm to 126mm2 on AMD’s 65-nm process. That’s a reduction of over 30%. In addition, for the 65-nm process AMD is moving over to larger 300mm wafers. By going from 200mm to 300mm wafers, AMD will be able to produce more than twice as many CPUs per wafer, as well as reduce energy costs. Back when Intel transitioned from 200mm wafers to 300mm wafers, they claimed to use 40% less energy and waste all thanks to the larger wafers. This should help AMD compete better with Intel on cost, which is particularly important since the two companies are currently locked in a bitter price war.

The other reason why 65-nm is so important for AMD is quad-core. Quite simply, AMD can’t affordably make a quad-core CPU at 90-nm. They need the smaller 65-nm process in order to make quad-core feasible from a production cost point of view.

All isn’t perfect out of the gate for AMD’s 65-nm process however. For instance, we noted a strange quirk in our performance testing with games: the 65-nm X2 4800+ CPU consistently ran slower than it should have based on its model number and specs, in fact it usually ran slower than even the 4600+. F.E.A.R. for instance was 7% slower than the 4600+ at 800x600, while Company of Heroes was 3% slower. (Pacific Fighters was the only game we tested that didn’t exhibit this behavior.) We’re not exactly sure what caused this, for instance it could have been caused by the motherboard, but the ASUS M2N32-SLI Deluxe with the 0706 BIOS has been tested and verified by AMD for use with their 65-nm CPUs. Whatever the case, hopefully AMD can get it resolved before these CPUs hit retail en masse next year.

UPDATE 12/22/06: We've received word from AMD on the surprising performance results we obtained in this article. It turns out that as a result of the half multiplier used on our 4800+ CPU, the DDR2 memory speed isn't running at the full 400MHz, while the X2 4600+ with its 12.0 multiplier is. As a result, the system memory on the X2 4600+ system ran faster than the memory on the 4800+. This is why the slower 4600+ was often able to outperform the 4800+ in our gaming tests. We will be examining this in more detail in a follow-up article.

Also, for those of you who were hoping to see massive overclocks thanks to the new process, you’re likely disappointed by our OC’ing results. In all honesty though this wasn’t too unexpected by us, there is a reason why AMD is limiting this release to CPU speeds of up to 2.6GHz, while their 90-nm CPUs are currently up to 3.0GHz. In this regard, we probably won’t see the real potential of AMD’s 65-nm process for many months, AMD needs time to gain more experience building CPUs at 65-nm. Things will finally get interesting for overclockers once they begin to perfect the process. Once this happens, you’ll likely see AMD producing 65-nm CPUs at higher clock speeds.

In the meantime, the greatest advantage AMD’s new 65-nm process brings is better power consumption, and thus cooler temps. If you’re in the market for a new AMD CPU, and you planned on build a media center PC, a 65-nm Brisbane CPU should be high on your wish list. For AMD enthusiasts looking for a real performance competitor to Core 2 however, you’ll have to wait until Barcelona debuts in 2007 before things get interesting.

© Copyright 2003 FS Media, Inc.
[ Print Article! | Close Window ]