The Kaby Lake fraud

Do not buy a computer this year.

There it is. You can stop reading now. That’s the conclusion of this post, which I publish with outrage after seeing how the presentation of new Kaby Lake CPUs from Intel is an absolute disappointment. Marc Whalton expressed it well at the beginning of his analysis in Ars Technica (emphasis mine):

The Intel Core i7-7700K is what happens when a chip company stops trying. The i7-7700K is the first desktop Intel chip in brave new post- “tick-tock” world-that means that instead of major improvements to architecture, process, and instructions per clock (IPC), we get slightly higher clock speeds and a Way to decode DRM-laden 4K streaming video. Huzzah.

That first sentence is sadly true, and in fact you would not need to read much more than that. Just having a look at some of the graphs that show its performance in synthetic benchmarks makes everything clear again:

This is just one example of how Intel has managed to fool the entire industry and become a lazy company. One that lives on past glories. Like Apple, I’d say.

That review of the Intel Core i7-7700K at Ars Technica is by far the best you can read about this microprocessor. It is, in fact, the only thing you should read about it, and here I include reviews done in Spanish and in English that are a real shame. They are because they stick to the message that Intel wants to offer, and they close their eyes to the obvious criticism. Intel has become confortable with itself. I am especially surprised and disappointed by the short-sighted review at AnandTech, a tech site I normally revere but that in this case has published its review with a misleading headline and conclusions. They even criticized in the text the ability to overclock (“our CPU sample is somewhat average to poor in terms of overclocking performance“) and then end up with a section where they crowned it as “the new champion”. Wrong.

Why has Intel settled for this? I would say they did this for a simple reason: they could. At the moment there is no competition in the market, and unless AMD Ryzen tells us otherwise – and the leaked evidence does not point to it – it will not be there for quite some time. It seems that Intel will be able to continue in lazy mode, accommodated and without really innovating.

The situation is worrying and sad, especially beacuse Intel has been on this path for a long time. His old tick-tock philosophy amazed us and made us enjoy an astounding pace of innovation, but between the reality that physics imposes —Moore’s law has an expiration date, if it is not already expired— and self-imposed Intel, we end up with a regrettable strategy. The one of the delays and rehashes, that is what is Kaby Lake. A 14 nm rehash.

I would say more. It is a fraud.

Save your money and wait for Cannonlake. Maybe the 10nm are worth it, because Kaby Lake does not.

Intel: if we can’t defeat them, we’ll join them

intel

Intel lost a recent opportunity to save itself from the mobile disaster, but there’s still hope for them. Not as makers, though: the recent deal made with ARM will allow other microprocessor designers to take advantage of Intel’s resources in the production process.

That could be a good way to leverage the technology and experience Intel has accumulated through all this years, but it’s also another sign -we didn’t need much more of these- of an Intel that threw in the towel long ago in the mobile space.

AMD made a similar move when GlobalFoundries spun off, and it went pretty well for them. We’ll see what happens with this new strategy from Intel.

The  GeForce GTX 1070 is the video card you’ll want to buy

Nvidia’s GTX 1070 looks likely to be the best bang-for-buck graphics cards of its generation, stomping the GTX 970 and in many cases beating the Titan X.

These are good times for gamers: there have been nice price/performance solutions in the past, but I doubt there has been one as desirable as the new NVIDIA GeForce GTX 1070.

There is certainly a more desirable graphic card, of course: the GTX 1080 is the one you lust over, but as Jon Martindale has said at Digital Trends, the GTX 107 is the one you’ll buy.

The numbers at Techspot -they’ve got a complete review with lots of benchmarks- show us that the GTX 1070 performs even better than the current GeForce GTX 980 Ti and is really close to the Titan X. Both will suffer a lot from this new competitor, much cheaper and really efficient. The comparison with the previous generation, the GTX 970 -who has been set as a requisite for a good VR experience- is astounding. For almost the same price you get around 100% the performance.

Astonishing. If you are thinking about upgrading your PC, there’s at least one thing you should have in mind. Buy a GTX 1070 and enjoy it.

Now the next step is a GTX 1070M. Pretty please?

I want to be able to play in my laptop

I don’t play videogames as much as I used to, but that doesn’t mean that from time to time I don’t want to play them.

Unfortunately that’s the usual situation in my case: I’ve got a Dell XPS 13 (9343). A great machine for almost everything, but not for gaming. No laptop is suitable for this, in fact. Convertibles, Ultrabooks and even more resourceful notebooks can’t cope with really demanding games, and in this cases if you don’t have a good discrete GPU you’ll suffer a mediocre experience.

razercorfe

So if you want to really enjoy the gaming experience in a PC, you’re out of luck: you must have a desktop computer with a great discrete GPU, because even gaming laptops are not meant to enjoy the same visual detail and frame rates that a full desktop PC can.

We had a pretty inspiring view at the future of this market last CES in Las Vegas. Razer launched there its Razer Core external GPU enclosure, and the smart use of Thunderbolt 3 and USB-C connector allowed its new Ultrabook -the Racer Stealth- to become a chamaleonic machine: one that could work as a conventional ultrathin laptop but also as a full-fledged gaming PC when needed.

The idea seems to have inspired AMD as well: one of their marketing directors, Robert Hallock, mentioned a few days ago how this kind of solution could be the future for many users:

Gaming notebooks are great for gaming, but nobody in their right mind wants to carry one all the time. Ultrathin notebooks are awesome to carry, but nobody in their right mind would confuse one for a gaming notebook.
But there’s still a HUGE appetite for thin notebooks that can game.

External GPUs are the answer. External GPUs with standardized connectors, cables, drivers, plug’n’play, OS support, etc.

AMD is bound to try to democratize these kind of solutions, but once again hurdles will have to be avoided. Will AMD adopt Thunderbolt 3, a standard created by its arch rival? Will it work with NVIDIA in order to set a really universal standard? Will prices make us think this is a better solution than a full gaming PC?

I hope their answer is the right one. We’ll probably have more information in the next few days: GDC 2016 starts March 14th, and I’m pretty sure AMD will talk about this on that event. The PostPC era could open a new chapter after all.

Oh, and don’t forget: this systems will allow us to enjoy virtual reality experiences in laptops, something not possible (in almost all cases) nowadays according to requirements published by Oculus and HTC.

That’s what I would call an interesting turn of events for the demise of the desktop PC.

Update (10/03/2016): AMD XConnect has just been announced with the collaboration of Razer and the Intel Thunderbolt group. Promising. Very.