Summary: NVIDIA G80. ATI R600. AMD K8L. Intel Yorkfield. All of these codenames are attached to upcoming products that will be hitting shelves shortly. In this article Brandon goes over the roadmap for both CPUs and GPUs from AMD, ATI, Intel, and NVIDIA. Whether you plan on upgrading soon, or you're waiting until next year, this is one article you won't want to miss!
First of all, in order to keep things simple, we’re going to stick with what we widely regard to be true and accurate when it comes to rumors. That means we may not always go into specifics, but instead provide an overview of what to expect, and when to expect it. AMD, ATI, Intel, and NVIDIA like to keep as much information about their upcoming products out of the public eye as much as possible so as not to give away their plans to the competition.
AMD and Intel have to provide a little more information to their partners than ATI/NVIDIA, simply because their partners need to know what’s coming in advance so they can ensure that their products are compatible with the upcoming parts. For example, for Intel’s upcoming quad-core Kentsfield launch, motherboard makers needed to know what kind of power Intel was shooting for in terms of voltage/power consumption, so they could provide the proper voltage regulators and other power circuitry needed as well as being able to ensure that their boards were ready for the task thermally.
Determining these figures isn’t exactly rocket science, as Intel and motherboard manufacturers discovered during the latter stages of the Pentium 4 era with Prescott and other CPUs: these newer chips required increasing amounts of power and dissipated more and more heat and thus Intel and their partners went through a spate of new chipsets and motherboards to support them. But Intel and AMD do their best with their guesstimates in order to ensure that the transition process is as seamless as possible.
From what we’ve been told by manufacturers, NVIDIA is at the complete opposite end of the spectrum, it hasn’t been unheard of for NVIDIA to brief their board partners in detail just days before the press is briefed under NDA. NVIDIA can get away with this because their board production is often farmed out to Flextronics, who will frequently manufacture the first-generation high-end and midrange graphics cards for all of NVIDIA’s board partners. All the board partner then has to do is take the card and slap it in a box and it’s out the door.
We’re also not going to be looking ahead too far in the future. The next 12 months should suffice for this article. Let’s start with the CPU manufacturers first, beginning alphabetically with AMD.
Let’s take a look at our projected roadmap for AMD for the next year:
Note: The 5000+, 4800+, 4400+, 4000+ are listed in italics as they’re based on AMD’s 65-nm process
Looking over the above table, you’ll see that AMD’s got quite a busy quarter ahead for Q4’06. Not only will AMD be introducing new FX CPUs, the FX-74, FX-72, and FX-70, but they’ve also got a trio of new X2 CPUs on tap for Q4’06 as well. The FX CPUs in particular are the chips grabbing the most headlines, as they’re based on AMD’s 4x4 technology.
4x4 is AMD’s two processor, dual-core solution for hardware enthusiasts. The CPUs will all ship with 1MB of L2 cache per core (2x1MB L2) and will be clocked at 3.0GHz, 2.8GHz, and 2.6GHz respectively. AMD has stated that they plan to make the 4x4 platform less expensive than previous dual-processor solutions in the workstation/server space, but we do know that AMD’s 4x4 CPUs will not be based on AMD’s AM2 socket, relying on their server-oriented 1207-pin Socket F instead. This could end up making 4x4 motherboards more expensive than the comparable Intel-based solution, as Intel’s first-generation quad-core CPUs will be compatible with many of today’s existing motherboards.
Let’s take a closer look at AMD’s upcoming X2 lineup:
To compliment today’s X2 lineup, which tops out with the ultra rare 5200+, AMD plans to introduce the 6000+, 5600+, and 5400+. All of these chips are based on AMD’s 90-nm Windsor core, which is in use today on AMD’s current Athlon 64 FX/X2 AM2 CPUs. With their larger caches, the 6000+ and 5600+ were originally intended to be sold as the FX-64 and FX-62, but AMD has decided instead to re-task them as X2s.
All of the aforementioned CPUs, including the 4x4 FX processors, are rumored to be debuting in November.
One additional chip which may also be introduced worldwide in Q4 that’s not listed in the table above is the X2 3600+. This CPU is actually shipping today in select regions of Europe and Asia, where it’s used in low-cost dual-core systems from OEMs, but it hasn’t hit North American shores just yet, and perhaps it may never will. Some analysts expect AMD to release it worldwide, and some don’t. In quarterly conference calls AMD has admitted to “walking away” from some business in order not to erode the ASP (average selling price) of their CPUs, so it’s possible that they may not release this CPU for the rest of the world, as AMD’s current low-end CPU, the X2 3800+ already officially lists for $152.
The X2 3600+ is built on the exact same Windsor core as the rest of AMD’s Socket AM2 X2 CPUs, and it runs at the same 2.0GHz clock speed as the 3800+. The only difference is that half its L2 cache is disabled, leaving it with 256KB L2 cache per core (2x256KB).
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.
AMD actually plans to begin sampling their first 65-nm chips by the end of this year, with the official introduction set for Q1’07.
As has become their tradition when introducing a new manufacturing process, AMD’s playing it conservative with their first wave of 65-nm parts. Rather than use the new process on an unfamiliar, untested core, AMD’s 65-nm Brisbane CPUs are essentially die-shrunk derivatives of today’s Windsor CPUs. In other words, AMD’s introducing no new features with these chips, and they’ll ship with the same 2x512KB L2 cache configuration used today on the 5000+ and other CPUs, as well as similar clock speeds. Take a look at the table below:
As you can see, AMD’s 65-nm variant of the 4800+ is a little different than today’s 90-nm 4800+. The 65-nm chip is clocked 100MHz higher, at 2.5GHz, versus the 2.4GHz of today’s 4800+. In exchange though the 65-nm chip ships with a smaller 2x512KB L2 cache, in comparison to the 2x1MB L2 cache used on today’s 4800+ model. The same applies for the 4400+ and 4000+ chips, they gain an extra 100MHz in clock speed but ship with smaller 2x512KB L2 caches.
Thanks to the smaller process the CPUs will consume less power, AMD lists a TDP (thermal design power) of just 65W for them, in comparison to the 89W of today’s 4800+ and 5000+.
While the aforementioned chips will definitely be nice performers, AMD’s true answer to Intel’s Core 2 CPU won’t come until next year, in the form of K8L. K8L brings with it a number of improvements. Here are the highlights:
HT 3.0 runs at 2.6GHz, providing up to 20.8GB/sec of peak bandwidth. In comparison Athlon 64's HT 1.0 tops out at 1GHz, yielding up to 8.0GB/sec peak bandwidth. According to DigiTimes, the new Altair chips will reside in AMD’s Socket “AM2+”. In all honesty though, we’re not 100% confident in the codename of this new CPU, as it breaks with AMD’s current naming convention. Back in June of this year, AMD announced a collaborative effort with Altair Engineering, which may have led to some of the confusion. In any case, the DigiTimes article mentions two Altair lines, one high-end SKU for enthusiasts that will sell under the high-end Athlon 64 FX brand and a desktop variant of Altair that will go under the Athlon 64 X4 designation. The DigiTimes article also says that these Altair chips are expected to be compatible with AM2 motherboards, but if you want the added bandwidth provided by HT 3.0, you'll want to get an AM2+ motherboard.
Also launching around the same time as Altair is AMD's dual-core variant of K8L, this CPU is codenamed Antares. Antares will support Socket AM2 and AM2+, just like Altair.
Update 10/4/06: This news article from HKEPC is similar to the DigiTimes report, yet it provides more info on Altair and Antares. According to the HKEPC article the Altair CPU intended for the FX line will be built on a new "F+" socket, presumably F+ being AMD's 1207-pin F socket that has been adapted to include support for HT 3.0. This FX CPU would be AMD's first quad-core 4x4 CPU: two quad-core CPUs running in one PC for a grand total of eight processing cores between the two CPUs! The desktop X4 Altair chip would run at the same clock speeds as the FX Altair (2.7-2.9GHz), only it would be used in AM2 and AM2+ systems, with just 1 CPU in the system.
The HKEPC article also gives more guidance on Antares. Motherboard manufacturers told HKEPC that Antares will be clocked between 2.0-2.9GHz and have a 512KB L2 cache and 2MB L3 cache.
Finally, in Q4'07 AMD will introduce two additional K8L variants according to HKEPC, Arcturus and Spica. Arcturus will be contain a 512KB L2, but won't include support for an L3 cache, and will come in speeds ranging from 2.1-2.3GHz. Spica is intended for the value segment, with just a single processing core. Presumably it will ship at slower clock speeds and contain less cache, but HKEPC doesn't provide any specifics.
Fresh off Core 2’s introduction earlier this summer, Intel’s ramping up from dual-core processing to quad-core, a transition Intel had originally planned to begin in 2007, but has now been bumped up to the end of 2006. The first of these quad-core CPUs will be the Core 2 Extreme QX6700, which was previously codenamed “Kentsfield”:
As we noted on the previous page, Intel’s first-generation quad-core chip, the Core 2 Extreme QX6700 won’t be a true quad-core chip, rather it consists of two dual-core processor dies integrated onto one package. This is a little different than Intel’s first dual-core Smithfield core, which consisted of one large die with two cores. The newer approach used for Presler and now Kentsfield helps Intel improve their manufacturing yields.
The downside of course to all of this is that the two dies in Kentsfield are still forced to share available front-side bus (FSB) bandwidth with each other. A simple analogy for Kentfield is that it’s basically two Core 2 Duo E6700 chips crammed into one package, linked via the FSB.
Kentsfield will be clocked at 2.66GHz, just like the Core 2 Duo E6700, with each of its two cores containing 4MB of L2 cache, for a grand total of 8MB L2 on the CPU itself. At IDF last week Intel CEO Paul Otellini confirmed that the Core 2 Extreme QX6700 would ship next month in November, so Intel will be first to ship quad-core. In fact it was later disclosed that Intel plans to ship 1 million quad-core CPUs before AMD ships their first quad-core chip.
With its “Extreme” moniker, it goes without saying that the QX6700 is targeted at the high-end enthusiast segment.
In the table above we’ve also listed a “Core 2 Extreme Edition X6900” chip. We put a question mark at the end because this proposed chip hasn’t been discussed in quite some time. Basically ahead of Core 2’s July launch, Intel disclosed a few forward-looking statements on where they planned on going with their Conroe core powering the Core 2 Extreme and 4MB Core 2 Duo CPUs. One such statement was that the Core 2 Extreme would hit 3.2GHz by the end of the year. The Core 2 Extreme Edition X6900 listed in the table above is that 3.2GHz chip.
Of course, since Intel made that statement, Kentsfield’s release date has been pushed up from Q1’07, to Q4’06. With Kentsfield getting pushed up one quarter, it’s possible that Intel may have shelved their plans to hit 3.2GHz by the end of the year.
They may have even ditched the idea of a 3.2GHz Core 2 Extreme Edition CPU completely.
As performance enthusiasts, we certainly hope that’s not the case, as a 3.2GHz EE chip would easily be the fastest CPU on the planet for gaming, which are typically single-threaded in nature today, but Intel may not feel as compelled to release a 3.2GHz part as they did a few months ago. After all, unless AMD’s got a trick up their sleeve no one knows about, today’s Core 2 Extreme Edition X6800 will likely retain the gaming performance crown for the remainder of the year. The only reason Intel may want to continue with their 3.2GHz EE plans is so they can retain the overall clock speed crown, as everyone expects AMD to hit 3.0GHz by the end of the year.
It’s believed that the Core 2 Quad will run at 2.4GHz and contain 8MB of L2 cache, just like the Core 2 Extreme QX6700.
It will be interesting to see how well these quad-core CPUs overclock once they’re released. Obviously Intel’s Core 2 CPUs have proven to be marvelous overclockers, with most enthusiasts OC’ing their CPUs by 30% or more with little or no effort, and many overclocks eclipsing 60% or more. Intel was demonstrating some nicely overclocked quad-core systems at IDF last week, including the Falcon Northwest system that was used to demo Alan Wake, but we’ve seen Intel show off massively overclocked systems at IDF in the past so you can’t put much weight behind this.
Shipping around the same timeframe as the Core 2 Quad Q6600 is the CPU enthusiasts on a budget are already drooling about: the Core 2 Duo E4300. The E4300 is the first in a line of Core 2 CPUs that will ship supporting Intel’s older, slower, 800MHz FSB. Today’s current Core 2 CPUs all share the same 1,066MHz bus speed.
Besides the slower FSB, the E4300 will also lack support for Intel’s virtualization technology, but it will support 64-bit and execute disable bit, as well as Intel’s Enhanced Speedstep technology. The general consensus seems to be that the E4300 will be based on Intel’s Allendale core currently used on the E6300 and E6400, with the CPU sporting a 2MB L2 cache. The Core 2 Duo E4300 will be clocked at 1.8GHz.
Here we should also note that Intel plans to introduce the Pentium D 935. The Pentium D 935 will be based on Intel’s Presler core, supporting an 800MHz FSB with 4MB L2 cache (2x2MB), and run at 3.2GHz.
DigiTimes reports that the Core 2 Duo E4300 will launch in January priced at $163, while the Pentium D 935 will come in at $133.
In Q2’07, Intel plans to ramp up the transition from Pentium to Core 2 significantly with the introduction of the single-core variant of today’s Core 2 processors, codenamed “Conroe-L”.
When it debuts, Conroe-L will essentially replace the Pentium 4 from Intel’s low-end lineup. Besides being single-core, Conroe-L will sport a 1MB L2 cache, making it cheaper for Intel to produce. Like the E4300, Conroe-L won’t support virtualization, but it will support 64-bit, execute disable bit, and Enhanced Speedstep technology. Some variants of Conroe-L will also sit on an 800MHz FSB like the E4300.
The Yorkfield listing in our table also comes courtesy of DigiTimes, who was the first to report that Intel plans on introducing this part in Q3’07.
Because of this, there isn’t much left to talk about in terms of either company’s immediate roadmap, but that doesn’t mean that their upcoming releases aren’t significant. In fact it’s quite the opposite.
The upcoming GPU ATI enthusiasts are eagerly awaiting is R600. R600 is ATI’s next-generation DirectX 10 GPU. Word leaked out over a year ago that the chip would sport up to 64 shading units and 32 ROPs, but after the G71 fiasco where sites were reporting different pipeline counts for the GPU on practically a weekly basis, we’re not going to put much faith behind those kinds of numbers until we hear official word from ATI, and you should to.
In the meantime ATI’s got one more refresh product to introduce: RV570. Word on this GPU has been out for quite some time now, but we’ll quickly go over the basics.
RV570 is ATI’s replacement for today’s Radeon X1900 GT. It’s expected to be built on TSMC’s 80-nm manufacturing process and feature 36 pixel shaders and 12 ROPs, just like the X1900 GT. It will be clocked a little higher than the X1900 GT however, with its core clock expected to be 600MHz (25MHz higher than the X1900 GT), while its memory is rumored to run at 700MHz (100MHz higher than the GT). Like the X1900 GT, it will have a 256-bit external memory interface, and for the first time in an Radeon graphics card, ATI’s compositing engine will be built-in to the GPU, giving the chip native CrossFire support. In other words, you won’t have to buy a CrossFire master card, all RV570 GPUs will support CrossFire technology out-of-the-box.
Graphics cards based on the RV570 GPU will be known as the Radeon X1950 Pro, and will be offered with 256MB of memory.
The GPU NVIDIA enthusiasts are pining over is codenamed G80. There’s actually less concrete info on G80 out there than there is on R600, even though it’s expected that G80 will debut first, possibly as early as sometime next month, but most likely definitely before the end of the year. NVIDIA’s PR/marketing group does a really good job of keeping info about their upcoming GPUs out of the public eye for the most part; they’ve even been known to use misdirection from time to time, so you’ve got to take the rumors about their upcoming products – particularly the earliest rumors that have the least chance of being credible – with a grain of salt, if that.
What is known is that G80 is NVIDIA’s next-generation part, and that it will support DirectX 10. Believe it or not, that’s the only thing that we really consider to be concrete info at this point.
For most of its development cycle it was expected that G80 wouldn’t sport a unified shader architecture, simply because every time NVIDIA was asked about unified shaders they’ve always downplayed the importance of them. This was most apparent during the PS3/Xbox 360 launch at E3 over a year ago, where both consoles were introduced to the public for the first time. Most recently in an interview with ExtremeTech, NVIDIA’s Tony Tamasi again downplayed the important of a unified architecture:
Yowch. In PR speak, that’s about as brash as it gets, and as a result, if there was one aspect about G80 that all the rumor sites had a consensus on, it was that NVIDIA would definitely not be employing a unified shader architecture for their upcoming part.
Then came some pretty shocking specs from hardware site VR-Zone. The specs were quickly taken down by the site, but chief among them was that G80 would sport a unified shader architecture and an unconventional 384-bit memory interface, as well as be available with both a conventional heatsink/fan cooler, as well as a hybrid water/fan cooler. Then low and behold a few pictures pop up of just such a G80 card with a weird memory configuration and cooler in the last week!
Coincidence or not? Who knows. It has been said that the best way to tell a lie is to mix in a little truth, perhaps that’s exactly what NVIDIA’s doing? After all, as we stated, disinformation in a tactic NVIDIA has used to throw off the competition. Perhaps that’s precisely what NVIDIA’s doing with the latest round of G80 rumors.
Hopefully the whole world will know in a matter of weeks.
So there you have it, our look into the crystal ball for CPUs and GPUs.
|© Copyright 2003 FS Media, Inc.|