Two stories regarding chipmaker AMD in the news today. First, they missed their Q1 earnings target, after reporting losses of $180 million, or $0.23/share:
Revenue landed at $1.03 billion, down 26 percent on the year-ago quarter (statement) ...Non-GAAP earnings were a loss of 9 cents per share ...Wall Street was expecting a loss of 5 cents per share on revenue on $1.05 billion.
"Under the backdrop of a challenging PC environment, we are focused on improving our near-term financial results and delivering a stronger second half of the year based on completing our work to rebalance channel inventories and shipping strong new products," [said AMD chief executive Lisa Su].
Secondly, AMD is withdrawing from the high-density server business, reversing a strategy that began 3 years ago with the acquisition of SeaMicro:
AMD paid $334 million to buy SeaMicro, which developed a new type of high-density server aimed at large-scale cloud and Internet service providers ...The purchase was made under former CEO Rory Read, and has now been reversed by Lisa Su, who took over the CEO job last October ...AMD said the move is part of its effort to "simplify and sharpen" its investment focus. As a result, it is taking a charge of $75 million.
AMD still sees growth potential in the server market, but not from selling complete systems. It's returned its focus to x86 chips and to the development of its first ARM server processor, code-named Seattle.
(Score: 0, Disagree) by Anonymous Coward on Saturday April 18 2015, @10:46AM
AMD has an X86 license that, IMO, is pretty much is a license to print money, yet this company keeps finding a way to fuck up, I really want them to succeed so there is still some competition in the market but just don't understand how they can't turn a profit, their stocks have dropped 70% over the past 3 years and they haven't done a single thing except release a bunch of low-performance high-power HTPC chips, are their engineers really this retarded that they can't just dieshrink something that works, give it a small bump of performance and call it a day?
I've always been a fan of their products because they've always offered more for a lot less, especially in regard to the 2S+ tax, not to mention features such as AMD-V, but their IPC hasn't really improved at all since the Phenom II which was the last time they were even remotely comparable to Intel's offerings (an I7 920 and a Phenom II were pretty similar performance), what the fuck
(Score: 4, Insightful) by kaszz on Saturday April 18 2015, @11:07AM
Don't blame engineers for what management may be responsible for.
(Score: 0) by Anonymous Coward on Saturday April 18 2015, @11:49AM
One of the duties of an engineer is to work around management incompetence to provide a superior product.
(Score: 1, Informative) by Anonymous Coward on Saturday April 18 2015, @01:14PM
Engineers and Management suffer under the Bean Counters. They are the stifling influence here.
(Score: 0) by Anonymous Coward on Sunday April 19 2015, @02:12AM
Management are the bean counters. MBA = Master of Bean Accounting.
(Score: 2) by kaszz on Sunday April 19 2015, @12:19AM
Duty? If it's for profit.. well it better pay or the corporation can have their shit.
(Score: 5, Interesting) by wantkitteh on Saturday April 18 2015, @03:59PM
It was an engineering decision that hamstrung the performance of the Bulldozer, Piledriver and Steamroller core architectures. I can't remember where I read it, more's the pity, but someone from AMD said that they decided to share a single FPU between a pair of logical cores because server workloads were more dependent on integer performance than floating point, and that having more parallel cores made more sense than fewer cores with higher IPC. Nothing wrong with that thinking at all - whoever then made the decision to use the architecture line intended for server workloads in consumer CPUs really screwed up. Maybe the R&D budget made the decision for them?
AMD do appear to have made themselves some luck. Getting their slow 8-core APU design into the PS4 and XB1 has forced devs to optimise console games for multiple cores rather than high-IPC, something they've been slow to do in the past. Comparing similarly priced CPUs, AMD offer more processing power/£ than Intel, but it's spread over twice as many cores and is therefore only leveraged situationally. Those situations are proliferating, so maybe AMD was just too far ahead of it's time.
(Score: 0) by Anonymous Coward on Saturday April 18 2015, @07:35PM
> whoever then made the decision to use the architecture line intended for server workloads in consumer CPUs really screwed up.
Why is that? You state it like it's a given. But what does the typical consumer do with his desktop that is floating-point constrained? Maybe some filters in photoshop. But the typical workload is primarily integer - the vast majority of floats in games are on the GPU, web browsing is all integer, video decode/encode is handled by the GPU. I'm having a hard time coming up with a non-niche usage that is heavily dependent on floats. It isn't like we are all doing fluid-dynamics or signal processing.
(Score: 2) by wantkitteh on Sunday April 19 2015, @04:03PM
Check out this chart [cpubenchmark.net] showing AMD's market share since 2004 - the days of the old GHz races between Intel and AMD. There are three events whose dates are worth noting here. Firstly, Q2 2006 - Intel debuted the first Core architecture chips in mid-2006. Secondly, Q2 2011 - AMD debut the Bulldozer architecture with the release of the 8150. Thirdly, Q4 2012 - AMD debut the piledriver architecture with the release of the 8350.
The first two dates are marked by a marked drop in AMD's market share. That's not even slightly surprising when benchmark reviews like this [bit-tech.net] all over the web show the new architecture, that AMD crowed would outperform the i7's by 50%, having it's butt kicked by them - even losing out to the Phenom II X6 it's supposed to succeed in some tests. The third date shows an uptick in AMD's market share, mainly because they launched the 8350 at a considerably lower price point than the 8150, effectively surrendering the high end of the market to Intel. Today? AMD have their lowest market share on record at 23%.
The market spoke - specific usage cases not-withstanding, reviews showed Intel had the better chips and AMD overhyped Bulldozer, so people bought Intel. If that's not evidence of a mistake, I don't know what is.
(Score: 2) by Wootery on Sunday April 19 2015, @10:22AM
AMD has an X86 license that, IMO, is pretty much is a license to print money
This just in: it turns out CPU design is trivially easy.
Their recent strategy has been based around them conceding that Intel has them beat for single-threaded code, I believe.