Normally, I try not to write opinion pieces – I’m (admittedly) not an industry insider, but I do follow the tech news with some enthusiasm, often becoming disappointed at the “new”. It seems we are living in a climate where increasingly tech giants are stumbling, and falling – which might just open the door for a new group of smaller, more nimble companies to take their place. Now that we’ve just passed the mid-year refresh of tech and Computex, some of the press releases have inspired me to say what I’ve been thinking for ages.
The first company I will focus on is Advanced Micro Devices (AMD). AMD holds a place dear in my heart as the first CPU company that really gave Intel a run for their money (along with some that didn’t survive – Cyrix and Chips and Technologies (eugh) to name a few). I was a loyal supporter of AMD – it was a transaction where I got better performance for the dollar, and they got my dollars!
Their existence was considered paramount to ensure competition with the Goliath (Intel) to ensure pricing remained reasonable, while pressuring them to make progress. AMD really made a name for themselves around the time of the Pentium III and Pentium IV era where their CPUs were better performing (at a lower clock speed, power consumption and price) and were the pick of the enthusiasts. They also kept to sensible design principles and made upgrading a possibility – remember the long lived Socket A?
My first machine with an AMD processor had an Am486DX-40 CPU which was an ex-business machine. Things were simple back then (barring the lawsuits that AMD were involved in at the time causing delays), and you could plug these into almost any board which took the Intel i486 CPU and expect it to work. True competition at its best.
The next machine with an AMD processor that I owned was a Socket 7 machine with an AMD K6-2 300mhz CPU. Again, Socket 7 was very accommodating of processors of both camps – as well as Cyrix CPUs as well. There was lots of choice, and AMD made a bold move of trying to popularize their own 3D vector instructions named 3Dnow! Their floating point performance was weaker than the Intels of the day, but the integer performance was solid.
My ownership took another long gap – the next machine turned out to be an AMD again – this time, a Socket A Athlon XP Barton Core 2500+. I also ended up receiving some parts over the years including Durons (their challenger to the Celeron), Semprons (a better Duron …) and many Thunderbird and Thoroughbred based Athlon CPUs. In my eyes, this was AMD’s golden era.
Intel’s big mistake with the Pentium III to Pentium IV transition meant they were the butt joke of everyone. The Netburst architecture with its deep pipelines had large penalties when mispredictions occurred. This resulted in low performance initially – their top end Pentium III’s were faster than their Pentium IVs. To get this sorted, Intel pushed up the clock speed but quickly realized that the heat produced became a significant problem. If anything, their Prescott based CPUs were the first to hit 120w TDP – a figure unheard of at the time, resulting in “I can cook an egg on your CPU” jokes. Funnily enough, where power consumption mattered, Intel retained their Pentium 3-based architecture for their Pentium M CPUs.
AMD seized this opportunity well, around the year 2000 – their CPUs had TDPs around 60-80w, and their IPC figures were much better than Intel’s. This meant that clock speed was not so much a problem – lower clocks and higher performance was a possibility. In order to dispel the megahertz myth, they instigated a marketing campaign and performance rating figures (i.e. 2500+ meaning equivalent or better than a 2.5Ghz Intel CPU despite only running at 1.66Ghz). This resonated well with the technical community – enthusiasts were building AMD machines in droves, and their overclocking headroom was decent as well.
Enthusiasts were continually wowed by the fact that the Socket A setup was so longlived. Many times, older motherboards while not being able to reach the required FSB speeds, could still utilize newer CPUs to benefit from better performance and lower heat outputs. Sometimes, all that was required was a BIOS upgrade. AMD also capitalized on the workstation market by introducing the Athlon MP (multiprocessor) at the time, making SMP a reality for many (the first time since the Abit BP6 allowed Celeron Mendocino CPUs to unofficially work in SMP, and overclock wonderfully too). The earlier Slot A was much less memorable – having been inspired by the SECC connector used by the Pentium II and III’s Slot 1 turned around, they were not compatible and the limitations on heat dissipation became clear early on.
While this seemed like an eternity for me, having lived through it, it was only a brief glimmer before AMD sunk back into obscurity and oblivion. Intel had much to do with it – competing in anti-competitive ways with special discounts for OEMs, threats and deals behind the scenes that would take many years to resolve through the legal system. OEMs like HP were genuinely interested in providing AMD based systems only to have them pulled at the last minute. In that sense, AMD was prevented from seizing the opportunity to go mainstream just when it was most successful.
Then came the second punch – Intel’s Core architecture. Ironically, the story goes that the initial Core architecture came out of work on the Pentium III based architecture by the Israeli team, which was much improved over its lifespan. Of course, the Netburst architecture was done by the Americans [insert smirk here].
The Core architecture was the beginning of Intel getting off its laurels and making a genuine push. Their CPU IPCs made leaps and bounds, they introduced new instruction sets (successfully – the SSE series). Prices came down too, especially at the budget end.
AMD was floundering. They were confused. Gone were the days of simple upgradability, as they bought out Socket 754 which was quickly replaced with Socket 939 (and the incompatible server-workstation based Socket 940). These vanished too for a new 940-pin socket called AM2. At the same time, prices were falling too – if they couldn’t compete on performance, they decided to compete on price. That proved to be a nasty decision – as they were now relegated to the budget end of the spectrum, where the least money was made.
It was at this time, I loaded myself up with more than 6 Socket 754 based systems – single core CPUs could easily be had for $24, a motherboard for $42 and a bit of RAM for $50. Add in a generic case with PSU ($50) and a hard drive ($120) and you have a system! It was value (performance per dollar), but it was not upgradable anymore.
News of their demise became ever more persistent. They made a big acquisition of ATi (the graphics company) which cost them a lot of money, but ironically, they seem to be the last of what is keeping the company apart. Roadmaps for future products became more miss rather than hit, with products never appearing.
AMD scored a brief win when they proved that Intel had “ripped off” their 64-bit instruction set when they showed that two undocumented instructions were replicated in Intel silicon. This spawned a cross-licensing agreement which helped the company by allowing the use of Intel’s 32-bit set but apparently also stated that should either company go under, the remaining company gets to own all of the intellectual property. I’m sure neither one was banking on their demise when they signed that.
The Phenom series of CPUs also had their day when they were launched with an error in silicon in the TLBs causing a fairly large performance hit when corrected in microcode. Their power consumption never recovered to Intel’s new levels. Instead, they tried to sell more cores – six core based chips which, while having more cores and sometimes even more clock speed, failed to match the contemporary i5’s in performance at similar price levels. I still operate two of them – a 1090T and a 1055T . They also lost overclocking headroom – the enthusiast support base was starting to grow disenchanted, and their top end CPUs were all sitting near 95W or 125W TDPs – where it used to be unacceptable!
Then came the recent Bulldozer architecture which wasn’t a clear successor to the previous Phenoms when performance was considered. The impetus to upgrade is no longer there – especially when new AM2+, AM3, AM3+ motherboard compatibility issues are taken into account. Intel, on the other hand, with their tick-tock strategy, has churned out improvements and shrinks on time. Their close co-operation with their own fabrication facilities are a big boon, developing new 3D transistor structures and smaller process nodes for example.
Ironically, AMD earlier had separated itself from its fabrication facilities, spinning them off into a separate entity named GlobalFoundries. If you ever wonder how a computer chip is made, this dated video from them gives you a good idea.
While, as an idea to make some money and do fabrication for others, this is a great move, on the other hand, AMD shows its lack of finesse when they blame their own arm for their APU products failing to meet performance targets and rake in the profit.
Which brings me to the whole idea of the APU. Their APU platforms try to capitalize on Intel’s poor integrated graphics showing (which, remarkably, has improved significantly but is still lagging behind). They integrate a graphic core, based on a discrete GPU into the same package as the CPU, hoping this will appeal to end users.
Remarkably, this works – but only for users buying pre-made machines at the bottom end, or laptops – but their success was stymied by a lack of supply when the OEMs needed it most. Their choice of the new FM1/FM2 sockets which is incompatible with higher performance CPU-only packages also limit the upgrade potential. Where did the sensibility of the earlier AMD go?
That’s exactly my thought when I read this this morning … Of course, AMD knows the money is in large corporate workstations and servers – Opteron and FX lines specifically. But they’re trying to sell a 5Ghz CPU with a (likely, but unconfirmed) 220w TDP. In an age where power consumption and power efficiency are beginning to matter (see ARM servers for example from Calxeda), and with a performance that’s unlikely to be worthy of such increases in power (especially when compared with Intel’s newly released Haswell) – who the hell is this CPU being launched for? And what happened to those important metrics – i.e. smarter, not higher clock speeds – cooler, less power consumption. It certainly sounds like a Pentium 4 in AMD clothing.
AMD isn’t dead yet, but given their performance as of late – they need a miracle to regain the same prominence they once had. Their product lines in the x86/64 space just aren’t enticing, and I often will no longer recommend them as they are not value for money on performance nor power efficiency. Of course, management probably has a lot to do with it.
AMD has made some inroads to sales volumes having struck a deal with Sony on their Playstation 4 as their CPU and graphics processors – but the irony is that this may end up being less beneficial than first thought when you look at the other side – why Nvidia passed it up.
Microsoft and Nokia
I think the next biggest thing to mention would be the failings of Nokia and Microsoft – both separately and combined. We all know that Windows 8 is a flop – there’s just no denying it. People just don’t want change which is unnecessary and counter-productive. I personally have tried Windows 8 on multiple occasions – in many ways, it is unnecessarily getting in the way of getting things done.
Windows 8 itself has personality issues – does it want to be a desktop? or a touch-screen launcher? Does it want translucent rounded Aero like in Vista or Windows 7? Or should I look like Windows Basic with square edges? Why not caps for all your titles? Eugh. Then they’re trying to “Apple-ize” the system – everyone wants money, so lets start our own store and voila. There’s a danger of taking inspiration from your competitors – you merely emulate … and fail to differentiate.
I think the reason why I still run Windows is due to the compatibility of legacy applications and relatively unrestricted and unconfused way the OS works. If I could have it my way, Windows 2000 Professional would still be running most of my machines – it’s lean, efficient, compatible – and it doesn’t have a myriad of inefficient programs, inbuilt viewers, previews etc to slow things down. Square and ugly isn’t a problem when productivity is number 1. Unfortunately, security has prompted me to keep up with the times (and consequently, the bloat, although new hardware cuts through it just fine).
Once you start to try to take away the compatibility (ala Windows RT), a very compelling reason for using Windows is gone. I suppose this is one reason, amongst many, for the poor showing of RT devices. The other is the restrictive app store in terms of apps available. If anything, Microsoft is hanging on the thread of “it runs Microsoft Office because we’re not going to put it on anything else” in the hope that sells RT devices. Of course, that makes up for the fact that x86/64 based slates to date have poor/mediocre battery life and performance – but maybe Intel just has the cards to make this all go away … The other thing is the desire to sell at “high” Apple-like price points while offering hardware that is equal or lesser, and an ecosystem that is smaller. I suppose Steve Jobs made it work by selling a single-functioned iPod – but Microsoft has no Steve. Microsoft has Ballmer, and Ballmer can hardly be described as a ‘marketing genius’.
Unfortunately, while they start their store, they also want a slice of the cake – one that hasn’t been theirs ever before. This has driven Valve to move to Linux – after all, Windows is just a platform that can easily be replaced if the software moves along with it. I think the Open Source movement has shown that if we can relinquish control that proprietary formats (Microsoft Office Documents) have over us, we can level the playing field and thus an application or a platform will not hold a monopoly over us and this will open up competition.
So far, Windows 8 and Windows RT devices haven’t really sold well. OEMs are pointing the finger at Microsoft, and Microsoft points back. When blame games erupt – it’s clear that something has gone wrong and nobody really wants to solve it.
And then there is their Windows Phone platform, which, as far as I can tell, is as relevant as the ailing Blackberry (whose users are ashamed of using them). Nokia’s partnership with Microsoft literally killed the former leading phone manufacturer and has placed it firmly behind all the majors in a space of a decade. The hardware is okay – sometimes it’s even quite decent, and it’s always well built. But still, no dice. People don’t want it.
Just like people don’t really want IE. They use it if they have no choice. IE, on the back-foot, is now chasing the others – but why would we trust something that has eroded our trust in the past? Likewise, why would we want a newer Windows if every second edition is a bad egg?
I think it’s quite clear that Microsoft is struggling from a lack of vision. It’s not the only one. Sometimes we have to think – do we need something new, or is it just new for the sake of being new? Do we need to change the recipe? There’s a danger of being blind to competitors and ignorant – but there’s also an alienation which exists if you make radical changes. The value in Windows is its “freedom” from restricting software, the large library of software available, the compatibility of the software and its ubiquity. That can easily be lost, if they don’t respect the users which is really why they’ve existed in the first place.
Funnily Nokia has stood by Microsoft’s side. If someone dealt me such a mortal blow, I probably wouldn’t be loyal to them anymore. Manufacturers such as HTC bowed out early when they realized Windows Mobile wasn’t gaining, and Android was. While HTC can hardly be said to be in a pool of cash at the moment – they definitely made sensible choices here.
While I say this – on the other hand, Microsoft has been there for me. My first “smartphone” was a Windows Mobile 2003SE based device. I’ve used MS-DOS, and Windows from when I was small. They led the market, they had it first. And I liked them too. They even had Windows XP Tablet Edition with Tablet PCs – but they failed to make things happen. If anything, they deserve praise for pioneering many things – but in business, it’s very often a combination of conditions that make the opportunity exist. Be too ambitious, and you will fail. Be too unambitious, you will fail.
Which brings me to this week’s release of the Xbox One. While many have been waiting to see the release of the new generation of consoles, I’ve been one to believe the console is a relic of times past – and that PCs are the way to go, but that’s just me. Anyhow, the general consensus is that Microsoft has dropped the ball (again) on the requirement for the console to phone home periodically and the issue of used game reselling. And the Xbox One is a bit more expensive to boot. What are Microsoft thinking? Are they just another Apple?
If there was one company I have a bone to pick with, it’s gotta be Apple. But since Steve left the world, the world at Apple has been quiet – underwhelming to say the least. I’m sure things are still rolling along …
One of the big problem with large companies is inertia. I’ve been told at many entrepreneurial courses – big companies are resistant to change, the wheels turn slowly. Small companies are often much less encumbered by the masses of people within them, and are agile and reactive.
This week’s preview of the new iOS 7 really shows just how slow the wheels are turning. In fact, the wheels are turning so slowly that it looks like iOS is just finally getting features which we’ve seen in other OSes for a while already. Worse still, they have definitely made a hash of “changing for the sake of changing” by revamping all the icons and the display paradigm – from curvey buttons to flat stuff. Hmm.
The control panel that slides up is ironically quite similar in design to the early Android ICS toggles as well. The multi-coloured rainbow 7 of the iOS 7 logo reminds me of the multi-coloured 7 on the Nexus 7 as well.
Aside from that, I haven’t really seen anything truly interesting or revolutionary hardware wise from them. While the HTC One and Samsung Galaxy S4 are very over the top in terms of specs – I like their ambition. Apple doesn’t seem to have provided anything great to fawn over this WWDC. A refresh? Expected. A new version of the OS? Yeah – but it’s been pretty bloated as of late anyway. But no new Apple watch? No new iPhone? Aww.
They did bring a preview of a cylindrical Mac Pro, but honestly, who can afford that? The idea of dual GPUs, powerful CPU and SSD in a neat cylindrical container with a fan is nice but will the execution be? Or will it be like one of those old overheating NeXT Cubes? The problem is that most Pros like to use specialized expansion cards, large amounts of storage, etc. This hardly caters for them. It’s more like an iMac without the screen with a few more ports to make up for it. Of course, Thunderbolt could come to the rescue, but that would spoil the elegance of the one-piece cylinder wouldn’t it?
I think I’ve rambled long enough about the disappointing tech reveals of late. There is, of course, more to be said – but I’ll spare you all from it. While tech in general progresses very quickly – I can’t help but think that much of the progress nowadays is still just for the sake of hype. I’m definitely reaching a point of saturation – and disenchantment that there hasn’t been more exciting, fundamental, visionary developments.
I think this is quite endemic of large companies which are lacking vision – merely floundering and hoping to pass the time.
Interestingly, if we look around – there has been an explosion of ARM based devices while Intel sits on its laurels (having abandoned its ARM division named Xscale to Marvell), and other manufacturers in China (Rockchip, Allwinner) have seen a fair amount of market share. Other market competitors in the ARM space continue to strengthen – Samsung with their Exynos 5, Qualcomm (which, interestingly, has acquired a very large IP base by buying out other companies like Atheros (which bought out Attansic, ZyDAS) giving them advantages in highly integrated SoCs), etc. Intel has well and truly missed the first boat to the mobile/tablet revolution of the “post PC era”. They could still win yet …