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
JohanAnandtech - Monday, May 12, 2008 - link
Yes, you are right, it was inaccurate, I rephrased it a bit. What I meant is that Intel was and is in no hurry to supply 45 nm CPUs to the desktop market, but at the same time launched Harpertown a few months earlier than planned. Almost a year will have gone by between the first 45 nm chips and really good availability of desktop parts in the channel. I believe that was quite different with 65 and 90 nm.Myrandex - Monday, May 12, 2008 - link
I have wondered many times if AMD uses their cache as efficiently as Intel does. I personally love AMD and my 5000+ BE is serving me plenty well, but I used to run a 1.8GHz. Dual Core Opty with 1MB L2 cache instead of 512K L2 with the regular Athlon64 X2 CPUs, and the reviews of the time didn't show a large difference in performance between the two...Maybe this should be revisted with current software?Jason
strikeback03 - Monday, May 12, 2008 - link
So AMD is competitive and possibly superior in servers. But when server sales only account for 16% of their revenue, can they sell enough server processors to make ends meet?WaltC - Monday, May 12, 2008 - link
Well, look at the breakdown above. Intel has to sell ~8 desktop cpus to equal the profits it makes selling 1 server cpu, and AMD has to sell ~11 desktop cpus to equal the profit it makes selling a single server cpu. What AMD is aiming to do is to ratchet up its share of the server market, because, per cpu, that's where the money is for both companies.Now that Barcelona has hit the B3 stepping I don't think that will be a problem for them with Opteron going forward. In the short run for the consumer desktop I think AMD will have a tiger of a cpu when the company introduces its 45nm Barcelonas with more L2. Meantime, I'm loving my Phenom 9850 on my MSI K9A2 Platinum desktop at home...;)
Sunrise089 - Monday, May 12, 2008 - link
Exactly. Users like Strikeback above need to use their arithmetic skills. Revenue numbers mean nothing, profit means everything (it's why a company like Microsoft is generally considered to be preforming better than a company like General Motors). If I have revenue of 100 trillion dollars, and profit margins of 0%, a company with $1000 in sales and 50% margins will walk away with more cash.What the chart Johan posted above shows is that the 20% sales tail wags the 80% profit dog, and that AMD and Intel are roughly even at the moment. Intel's success comes from a much larger slice of the 80% sales slice of the pie, which is an edge but nowhere near as important as the 20% servie slice.
Incidentally, the AMD vs. Intel situation is very close to the Nintendo vs. MS/Sony during the previous console generation. Forum poster after forum poster lamented Nintendo's small sales numbers, while a small minority of posters realized the importance of Nintendo's much higher margins of software and slightly higher hardware margins. Even without a runaway sales hit (which they now have) Nintendo made more profits than their rivals due to higher profit margins. Of course AMD isn't doing BETTER than Intel, nor are they guaranteed to combine margins with very high sales like Nintendo has done since the Wii/DS launch, but rumors off their death are greatly exaggerated so long as they hold on to 50% of the sub-market
that is responsible for 80% of industry profits.
@ Johan - great write-up. I have no knowledge of server computing at all, but i continue to like your articles more and more, especially for you offering such a clear overview of the market that anyone can understand.
Pirks - Monday, May 12, 2008 - link
Hey, it's not only Nintendo/Sony story, it's also Apple/PC story - Apple makes gobs of money because their computers are so much more expensive than mass market Wintel PC computers. Even though Apple sells much less units - they still make MORE PROFIT than a lot of cheapo PC making companies like Dell. That's something I love to hit Wintel PC fanatics with here on anandtech/dailytech forums.Griswold - Wednesday, May 14, 2008 - link
Would be a good example if apples profit mostly came from desktop and notebook sales - which it doesnt. And if Dell wasnt such a limping dog. Why not compare it to HP for example?Noya - Tuesday, May 13, 2008 - link
Of course Apple has higher margins...their customers have lost most of their higher brain functions- observe how they can only talk while their hands swing wildly through the air and they can't comprehend how to use a two-button mouse.(think of the Handyman from "In Living Color" with his one good arm and gimp voice):
"It just works for me!"
"Something you love to hit" PC enthusiasts with?
PC enthusiasts/hobbyists BUILD their own machine to the exact specs they want/need for market prices on the hardware with full upgradeability...they don't care about Dell/HP/etc.
If it makes you feel "special" to pay 3x the price for the same hardware that has an Apple logo on it...you're called a sheep and watch too many Apple commercials.
Pirks - Tuesday, May 13, 2008 - link
"3x the price" is an ancient urban myth that has been destroyed a long time ago, check your prices again.And while pc guys build their stuff you should not forget that vast majority of consumers buy brand name computers and they don't care what serial number is on their CPU and how much voltage it can take for proper OC. They care more about design, ergonomics, user friendliness and stuff like that - things that are totally alien to pc building enthusiasts. This is the area where Apple spits on Wintel PCs and rubs it in - 'cause their computers are head and shoulders more novice/non-techie (i.e. mass consumer) user friendly than any Windows PC. Just read all those reviews and horror stories about dumb Vista UAC, underpowered sllooowww PCs sold with Vista everywhere and stuff like that - they tell all the story the mass consumer needs to know.
And no other arguments like Crysis or overclocking or other enthusiast lingo can make any difference for mass consumer - it's like talking nuclear physics to them.
So, enjoy your niche PC enthusiast market, nothing wrong with that, but also do not forget about the masses and gazillions of computer illiterate people who don't give a thing about your PC enthusiastic overclocking watercooling blah blah glory - these are all future Apple clients, just look at how Mac sales grow all the time
Borreson - Tuesday, May 13, 2008 - link
Where the ancient urban myth still rings true is that Apple has chosen not to compete in certain segments of the PC market.For instance, you can't buy a $500 budget Core 2 Mac tower. You can't buy something with the iMac's specs that lets you swap out the video card or that allows you to choose what monitor you prefer. You can't buy a sub-$2000 notebook nor a 13" subnotebook with a discrete GPU.
The general level of hardware flexibility in expected by enthusiasts is only present in the Mac Pro, which starts at 3x the price of the white box that an enthusiast might consider building.
When you go and compare that Mac Pro to a competing Dell or a white-box Intel server/workstation, the Mac Pro's lofty price tag looks a lot more reasonable. It's really difficult to build a dual-quad-core workstation with gargantuan memory capacity and all the bells and whistles and have the price tag not balloon beyond reason. With the Mac Pro you get the signature Apple friendliness and hipness wrapped around a monstrous piece of hardware, and if you buy it immediately following a new release or a price adjustment, it's usually hundreds of dollars less than the competition's arguably inferior product (entirely depending upon your priorities and how much of Steve Jobs' pot you've been smoking). Oh, and they're whisper quiet.
But if you want a Mac with a PCI Express slot and don't want to pay $2300+, you're don't have *ANY* options, because for some reason having any expansion capability is bundled only with having an 8 core monster. The Mac Pro is complete overkill for most purposes and yet still has some significant hardware shortfalls in many areas (ie: memory performance for gaming, legacy hardware support, price/performance for upgrades) compared to an $800 white box tower.
(There's nothing like an AMD vs Intel article to bring out the Mac vs PC platform war. Woo?)