This is a quick note, but it will be a powerful one. I often get accused of "bashing" AMD (AMD) unjustly, but this is not my intent. When I see evidence in the company's products that the firm is bolstering its competitive position, I will be the first person to start whipping out the AMD-branded pom-poms. Unfortunately, while AMD's marketing and investor relations teams have done a superb job trying to "sell" the AMD story, cold hard facts tend to put a damper on that parade.
Here's Why AMD Is Losing In The PC Space
I always find it amusing that people blame AMD's shortcomings in the PC space on Intel (INTC) "bribes," but the fact of the matter is that AMD's products in the PC market just aren't competitive.
I'm not going to rattle off specs, TDPs, or other marketing nonsense to prove my point. I am going to show you some hard numbers from the perspective of somebody who has spent the vast majority of his life deeply entrenched in the PC enthusiast community.
Now, I'd like you to meet two processors. The first, from AMD, is known as the FX 8350. This is an 8-core, 4GHz pure-CPU (no integrated graphics) CPU from AMD:
Note the price: it's $199. Now, some more details that you won't find on the Amazon listing is - probably more interesting to investors - the die size, or the size of the actual silicon chip. According to AnandTech, the chip sports 1.2 billion transistors and a die size of 315mm^2.
Now, meet this little chip from Intel, known as the Core i3 4340:
This is a 3.6GHz, dual-core processor with an integrated graphics engine. The chip sports 1.3 billion transistors and a die size of 181mm^2.
Anyway, so you would expect that in just about everything, the 8 core, 4GHz 125W beast would completely outpace the dual core, 3.6GHz - right?
Well, no. And unfortunately for AMD, the one use case that it tries to market these chips for (gaming) is where AMD's parts fall flattest on their faces.
Gaming Performance - Yes, Folks, It's This Bad
In a recent review of the i3 4340 from Hardcoreware, the site did some gaming tests with all else equal in the systems to isolate the CPU performance. Here are the results:
Assassin's Creed IV
(1920x1080 resolutions, "high" settings)
Battlefield 4 (AMD's favorite)
(1920x1080, "high" settings)
(1920x1080, "high" settings)
You see the problem here, right? A lowly, cheaper to build (and cheaper to buy) Intel chip is meaningfully outperforming AMD's CPU in a use case that AMD is marketing these processors for (hardcore gaming enthusiasts)!
Wait, You're Not Being Fair!
So, one might say, "okay, that's just gaming. AMD's CPUs are probably better than that cheap Intel at something...right?"
Well, sure - anything that is well suited to take full advantage of 8 cores will perform better on the AMD 8 core chip than the dual core Intel. Here's an example - in this case, this is off-line 3D rendering using ray-tracing (this is a rendering scheme very well suited to many cores):
Yes, anybody who needs lots of cores "on the cheap" is probably going to be better off buying the AMD chip (assuming power consumption isn't an issue), but here's the rub:
- For people who "need" more cores, time usually equals money. In this case, these people will go and buy a $600 six core Intel chip (which can, without any doubt, wildly outperform any AMD desktop chip today) and they will get that extra $400 back in terms of productivity fairly quickly
- Most consumers don't need more cores for their applications; even the most demanding PC games would prefer fewer more powerful cores than a bunch of weaker ones
AMD's PC chips just aren't competitive, and while a big chunk of it is the fact that AMD's chips are built on inferior process technology (this is why the die size on the AMD 8 core is twice that of the 2 core Intel, which has the same number of transistors), another part of it is simply that AMD doesn't have the R&D resources to really be competitive in this market as well as all of the other markets that it's trying to tackle.
If AMD can just laser-focus on one or two markets and, I daresay, bet the company on just one or two really good strategic initiatives, then it can probably win big for its shareholders. But the fact that AMD is working on semicustom projects, two lines of CPU cores ("small core" and "big core"), X86 server chips based on both lines of cores, ARM servers, discrete GPUs, and so on...well, there's no focus there, and that means that AMD will probably do a lot of things, but it won't do any of them particularly well.