We have recently detailed a number of challenges being faced by Intel, mostly centred on ‘How do you build $7 billion fabrication plants when your average chip profit is $10?'. While there's no doubting Core's ‘King of the hill' for pure CPU performance credentials, KitGuru reads the A10 reviews and is left wondering ‘What about the mainstream market?'
Just so we're clear, whenever Intel gets around to launching the next generation of Core processors, they are going to be smashing world records all over the place – and anyone who buys a PC for more than £700 will have one. We'll go one stage further and say that, apart from Apple, you will find the same kind of ratio/line-up for laptops.
There. We're clear.
But what about the mainstream – the bulk of the market?
Well we can all see that the move to mobility has been strong and permanent. However, the flux in that market will make prediction hard – because no one can be certain about how far the iPad-led tablet incursion into Intel's mobile territory will go. We'll be working on this over the coming weeks and release our analysis closer to the launch date for Apple's ~£179 mini-iPad in the middle of October – on the shelves in November 2012.
So let's look at what we know with greater certainty. KitGuru has been in discussion with various sources to try and ascertain the size of the UK desktop market in the 12 full months following the launch of Haswell around Q2 next year.
The pricing on the left is in Great British Pounds – you can add 10% for Euros and 50% for dollars. This is the base unit only with no VAT, so £300 becomes £360 – which is the kind of price point where you can sell a low end Core i3 or AMD A10. That's the only place that Haswell will be able to maintain unit shipments against the existing Core product line-up.
After the £600 mark, you start to move into Core i5/i7 ‘K' territory and things are only going down. With little/no competing sales from AMD into the market as we get close to £1,000 – that drop will come entirely from Intel.
The challenge for AMD is, obviously, how can you make money when the market is geared around processors that you are selling for no more than £70?
But the Tick-Tock has been established and Intel is rocking its way forward to ever more powerful processors. Haswell is promising a significant bump in performance – certainly around graphics – but what is the likely result? One graph jumped out from KitGuru's +50 page analysis of the new A10 APU from AMD:-
This chart makes for unhappy reading if you're an Intel engineer gearing up for the next half a dozen launches. Intel's present i7 processors are on a different planet, in terms of performance, compared to the AMD APUs. There's simply no comparison. But that's important if you're doing professional video editing etc. If you're a gamer, then you're looking at frame rates with sensible resolutions and image quality settings. Looking at the way the market is dropping in terms of overall spend on desktops etc, that will increasingly mean integrated graphics ‘as standard' with the addition of a dedicated graphic card as ‘something only the cool kids do' [Yes, yes, we know that means KitGuru readers – Ed].
So if the fight moves to integrated, what will Haswell need to deliver in order to overcome AMD's APU? This chart shows the most powerful Intel desktop option available, HD4000 on the Core i7 3770 ‘K' processor against the AMD A10:-
So what about the comparative pricing? Looking online at places like Overclockers and Aria, we find a huge discrepancy between the pricing for the AMD A10 5800k and the Intel Core i7 3770k processors:-
So the basic maths, from where we're sitting today, is that AMD's best integrated solution is up to 300% faster than Intel's – while at the same time Intel's best integrated graphics option is almost 300% more expensive. We're not sure that 3x faster for 1/3 the price makes the AMD A10 5800k 9x better than the Intel solution, but that's one way of looking at it.
We'd like to go back and remind you, very clearly, that if you're aim is video processing etc using Adobe Premiere and other premium packages, then the Intel Core i7 3770k is going to be night-and-day better than the AMD solution. All we're looking at here is where Intel's best integrated graphics are now – and where they will need to be in the middle of 2013 to secure a win.
AMD's numbering system for GPUs would mean that a part named ‘5800' would have a 5850, 5900 and possible a 5950 or 5990 to follow. Given how easy the A10 5800k samples overclocked, we'd say that AMD can easily pull another 30% from its existing processors, simply by increasing the core clock when they ship.
In their own space, the Haswell replacements for 2500k, 2600k, 3570k and 3770k will – without doubt – reign supreme. But, as our market data shows, all markets are shrinking – so go get volume over the next 12-18 months, companies will need to hit the low end price points hard. That means integration. If the arguments stay with traditional CPU-heavy tasks, then Intel should hold its own very comfortably and AMD shareholders will be wondering what they're doing. If AMD can move the argument to graphics, then the ‘big change' can happen and ‘AMD shares for $3' will seem like a proper bargain.
KitGuru says: With a test like the Shogun benchmarks seen here, Intel's Haswell range will need to include a Core i3 processor that sells into the channel for around £69 – which is able to run games 250% faster than the existing Core i7 3770 k processor. That's a tough ask – even for a company like Intel. Any company that announced a ‘100% increase in performance' will be looking good – but that's at the top end. We're skeptical that Intel can manage this for processors under £70, but it would be cool to see. The real question then becomes, ‘Can AMD move the benchmarking argument across to gaming on integrated systems?' AMD's inability to move the argument may well outweigh any failing by Intel to move the technology forward fast enough.
Comment below or in the KitGuru forums.