AMD's K10: a "dead" product or not?
by Johan De Gelas on May 12, 2008 12:00 AM EST- Posted in
- IT Computing general
A few years ago it was fashionable to bash Intel's Pentium 4 as a braindead architecture. The fact that the Pentium 4 Northwood (533 MHz FSB) was the best performing processor from mid 2002 until late 2003 in many applications, and that the Pentium 4 Northwood remained competitive until early 2004 was conveniently forgotten: nuances do not make good headlines.
It is now trendy to bash AMD. One" PC doctor" at ZDNet goes as far to say that:
"When I look at AMD’s current product line, all I see is a forest of
deadness. Intel has products trump every category of products
going. Server, desktop, mobile, low-end, high-end, dual-core,
quad-core. Intel has all these markets stitched up."
Nuances, who needs them when you can make a sensational headline? And indeed, the lastest desktop CPU articles here at Anandtech show that Intel's midrange CPU have a significant lead over the fastest Phenom processors.
Like any design, the K10 is a trade-off. And most trade-offs were made in favor of the applications in the server and HPC market, at the expense of games and other desktop applications.
First take a look at this page which compares a Core 2 Duo 4400 (2 GHz, 2 MB L2 and 800 MHz FSB) with a slower 1.86 GHz Core 2 Duo E6320 (4 MB of L2 and a 1066 MHz FSB). One thing is for sure: games prefer the larger L2 cache. Some of the games were up to 10% faster on the CPU which was clocked 7% lower but with twice the L2-cache. The fact that games prefer a 4 MB L2 is not going to change when you run it on a AMD CPU with integrated memory controller. A L2 can deliver the necessary data in 12-20 cycles, an IMC needs about 100 cycles.
Now, take a look at the Cache architecture of AMD's K10/Barcelona. If your run a single threaded game on it, it gets a fast 512 KB L2-cache and after that a relatively slow (44-48 cycles!) 2MB L3. If you know that the same game can benefit from more than 2 MB cache, it is pretty clear that the 512 KB L2 is not going to cope, you'll end up using the L3 a lot. A dual threaded game might need a little less per thread, but the same problem will happen again: it needs to go to that slow L3 cache all too often. Run that same game on Intel Core CPU and each thread of your dual threaded game gets a low latency 4 MB (or 6 MB) L2.
Now let us now imagine that we run 4 threads of an HPC workload on it. Each thread has a very limited number of instructions, which perfectly fit in each of the L2 caches. You get 4 threads which gets a total of 4x the bandwidth of L2. In case of Intel, each two threads have to share the available bandwidth of the L2. The amount of data is huge, so caching the data is hardly possible. The fast IMC does wonders for the K10 chip.Data that is shared between the 4 cores remains in the L3-cache and all L2 caches are kept coherent at a incredibly fast SRI. So your cache coherency overhead does not increase with the number of caches, it increases per socket. Going from 2 to 4 sockets means that you double the amount of cache coherency traffic. Compare that to the Intel platform where all L2 caches need to be kept coherent.
It is just one example why we could never expect the K10 chip to be a super desktop chip. But how is Barcelona doing in the server world? Is it limited to an HPC niche market? Well, let us see what Intel thinks. First of all, where do most of the 45 nm chips go? Just a few weeks ago, Anand reported that Intel had no intention of flooding the desktop with 45 nm Core 2 chips quickly.
Those 45 nm chips are going to the server market. Why? Several reasons.
First of all, the server market might be only 20% of Intel's revenue. But look at this:
CPU
ASP
Profit margin (estimate)
Percentage of revenue Intel Server CPU >$400 >$300
+/- 20%
AMD Server CPU
$300-$400
$220-$330
+/- 16%
Intel Mobile/Desktop CPU
$100
$40-$50
+/- 80%
AMD Mobile/Desktop CPU
$50-65
$5-$30
>80%
Secondly, Intel needs those 45 nm to be competitive in the HPC market. A 2 GHz Barcelona is capable of keeping up with the best 65 nm Xeons in those applications.
It is pretty clear why AMD focused on the server market. Without a complete redesign it is not possible to beat Intel's integer crunching power and the fast and big L2-cache and that is exactly what a modern game needs. Barcelona built further on the K8 architecture and inherited the relatively inflexible integer pipeline. While Core 2 has sophisticated reordering of loads and stores, Barcelona does a limited reordering of loads. While Core 2 offers a 32 entry queue to the integer units, Barcelona has 3 rather inflexible separated 8 entry queues.
So the right way forward for AMD was to focus on HPC and server applications where it could leverage it's strong points. We can bash AMD for being so late, and coming up with relatively low clocked CPUs, but even a 2.8 GHz Phenom would not have raise AMD's ASP significantly in the desktop market.
We are almost done with our first round of quad socket benchmarking and we can tell you that we are having a lot more fun than Anand: it is a good old exciting fight between AMD and Intel. Don't believe us? Let Intel do the talking again:
Yes, projecting the bad performance of the desktop chip to say that "AMD's products are a dead forest" is ... just silly. If you have missed the previous entries of our IT blog, just go to it.anandtech.com
Facebook
Twitter
RSS
74 Comments
View All Comments
Plasmoid - Tuesday, May 13, 2008 - link
So I have my old 90nm Athlon 4400 X2 at home with it's juicy 2x1mb cache... and AMD are floundering around because they can only spare 512k per core now.Seems to me like the 65nm transition never paid out in cache size like they expected it to and like it did for Intel.
So i'm wondering if my 4400 X2 can give a K10 a run for it's money.
Visual - Tuesday, May 13, 2008 - link
AMD are still the better choice when it comes to low-budget builds - cheaper quad-core, plus they now have tri-core as well.Also, the best integrated graphic solutions currently are for AMD. Sure you can add a separate gfx card and beat integrated performance, but that means higher price (not good for budget builds) and a pci-express slot used up (not good for small builds, like the HTPC I'm planning with a case that fits only 1 expansion card via a riser)
Justin Case - Tuesday, May 13, 2008 - link
"leverage it's strong points"it's -> its
Other than that, good article. This is something that everyone doing memory-intensive multithreading knows (or at least should know, if they're doing their homework), but it's good to see it posted on a mainstream site.
AMD isn't competitive in the high-end single-threaded arena but when it comes to mid-range desktops they give about the same "bang for the buck" as Intel, and for HPC and some types of servers they offer more "bang" than Intel can deliver, period.
Now if only they could make me a couple of dozen quad-core S940 Opterons... :-P
Roy2001 - Tuesday, May 13, 2008 - link
Yes it is dead except niche HPC market. Fastest quad core won't even beat slowest Q6600. Northwood was fastest CPU when it was born but Barcy was NOT!MauriX - Monday, May 12, 2008 - link
I think the K10 architecture AMD was the first step towards what really is going to be revolutionary. AMD aims to create a CPU / GPU / PPU / xxx multi-core on a single silicon chip. K10 opened the door to an architecture of independent cores that perform different tasks. To realize the dream of AMD have a XPU, eg 4 CPU Cores + 3 GPU Cores + 1 PPU Core, it was necessary to posses an architecture where each core has full independence (energy, L2 cache, etc..) Between the cores.With the K10 failed to make significant progress in terms of performance (what everyone expected), but on an architecture basis for the next step: FUSION
The problem is going to have a future AMD will be to create the abstraction layer of software between DirectX 10 and their specific Cores. We have already seen as the last IPG (780G) AMD are coupled perfectly through hybrid-Crossfire. I think that AMD intends to use that same concept in their "GPU's Cores and external video accelerator. With FUSION, would no longer be necessary to add new set of micro instructions (as SSE4) to accelerate certain codes programs; Directly ONE Core can be added with all the specific sets of instructions that were necessary (either as support for facial recognition, CAD, Physics, etc.).
All this is pure speculation, but the fact of having broken the Karma Double Dual-Core / Double Double Dual-Core is an important step. They should now concentrate their energies on improving the CPU Cores and integrate new specific Cores.
(from "Mauricio Fernandez, Punta Alta, Argentina." translated by Google Translator;)
Makaveli - Monday, May 12, 2008 - link
Brand loyalty only hurts you the consumer! Wake up or shutup!And i'm posting this from my Opteron 170 that I love and still running well. However i'm not blind!
The saying Ignorance is bliss really does apply here!
Ensoph42 - Tuesday, May 13, 2008 - link
My reply to 7Enigma applies here as well.Crank the Planet - Monday, May 12, 2008 - link
AMD fanboy here and rightly so. Before C2D intel had crap. They haven't had a good proc since PIII. Netburst??? give me a break. Better internet experience- what a croc. Can you say "random restart because the chipset is crap?" Intel = Crap.That was my opinion before C2D. Now Intel has regained the crown in several areas where AMD had taken it away. So what else did you expect? All you intel fanboys need to stop talking trash. AMD is presently in a financial bind- everybody knew they would be if they bought ATI. Well guess what, there is only one year left and AMD will finish paying that off. After that they won't be cash strapped. They'll be able to put more into R&D etc. and then they will be able to put out amazing things. Fusion is going to be BIG!
What innovations has intel come up with the last 2 years? ZER0! Heck, they haven't come up with a good idea since Netburst. HA HA HA HA!!!
Even with all the cash, market share, yada yada yada. They can't come up with squat. Remember- a die shrink is not an innovation. All they can do with Nehalem is copy what AMD is already doing IMC etc., etc., and that is basically saying that the AMD engineers were right on that one. Intel has to do things like better interconnects between chips to stay competitive. Once you reach the speed barrier intel has nothing left. AMD will thrive because they are innovating and branching out.
Don't get me wrong C2D is a good product, it's just that Intel will not be able to keep one step ahead of AMD forever ;p
Angeloni100 - Wednesday, May 14, 2008 - link
I don't get this "fanboy" thing... I own a computer store in brazil, have been working with computers for about 20 years, owned just about every CPU ever created... For a long time I recomended AMD to my customers, especially during the Athlon rein, which lasted 7 or 8 years by the way (from athlon 500~ all the way up to the first Athlon 64's)... I loved my AMDs, had lots of them, K6-2 450Mhz and K6-3 550Mhz, Athlon 850, 1100 and the great 1400 that destroyed all P3s and P4s of its time... Had a couple of AthlonXPs, namely the 2100, 2400 and the 2800 with 333mhz fsb... Now I own a C2Duo 6600 (the first Intel I buy since 97).. It was the better CPU at the time, I've had it for a little over a year and love it too... It is unfortunate that AMD is no longer competitive with Intel, But being a "fanboy" is just not very smart, cause you end up buying products that not always are going to be the best... It is not easy to earn a living... we all have to work way too much in my opinnion... I just have too much love for my hard earned cash to throw it away at something simply because of its label... If AMD can come up with a better CPU then intel, I will buy it when I feel the need to upgrade or change my computer... I recently built a server with an Athlon 64 X2 6400 CPU in it for a customer based on my recommendation simply cause it was the best thing for his needs... I sell lots of AMD, but don't recommend them to everyone...I guess what I'm trying to say is that we need to be smart and faithfull to our money and not to multy billion dollar companies that have no idea of who we are and don't care to... and this goes to the apple fanboys too...
Be smart kids... honor your work and the fruits of it... not these companies...
Ensoph42 - Wednesday, May 14, 2008 - link
I agree 100% with what you're saying. However it's fun to cheer for you team, and take offense when it's disparaged. I'd consider myself an AMD "fanboi" but if I had the money to spend at the moment I'd weigh all possible options and build a system that gave me what I wanted.Also your pragmatism is couter-intuitive to how human nature works. When we buy cars we may at first research everything, but in the end it's that feeling we get when we sit behind the wheel.