Intel, Micron announce 25nm NAND flash production
You’ve probably glossed over the boring headline and aren’t even reading this, but it’s a lot more important than it sounds for the computer industry and computer users.
This is the tipping point for SSDs to become mainstream.
Currently, good SSDs (the bad, cheaper ones generally aren’t worth buying) are small and expensive. Intel’s excellent X25-M series, the gold standard, is about $450 for 160 GB.
SSDs based on this 25nm flash are likely to offer 160 GB in the $200 range and 320 GB in the $500 range.
It’s hard to overstate the performance gains that SSDs offer. It’s not the sort of incremental, you’ll-notice-it-5%-of-the-time gains that new CPUs usually offer.
If your computer feels slow, it’s almost definitely the hard drive’s fault.
If you’re waiting a little longer than usual for a popular website to render its Dashboard or show you an encyclopedia page or tell you which of your old high-school friends have gotten fat, you’re probably waiting for some hard drives in a server somewhere.
Nearly every slowdown of modern computer usage is caused by a very fast computer that’s sitting around doing nothing while it waits for its hard drive to move its heads.
Seek time is the delay required for the drive’s heads to move to the requested location on the disk, stabilize, and start reading or writing the data. Most hard-drive delays are seeks, not big sequential transfers1.
Ten years ago, the best consumer-class hard drives had seek times in the 12ms range. Today, most good drives are in the 8ms range. Got that? In a decade, we’ve made huge gains in nearly every other aspect of computer performance, and hard drives are much faster than they used to be. But seek time is still very high, and is still the bottleneck for everyday computer performance.2
SSDs — these tiny, laptop-hard-drive-sized boxes that have no moving parts and emit no noise and cost only $450 for 160 GB today — have an effective seek time of 0ms.
Zero.
Seeks become effectively free. (Technically, they do take some time. But it’s well under 1ms and close enough to zero, relative to hard drives, for the sake of argument.)
Imagine if nearly every computer slowdown vanished. That’s what it’s like using a good SSD.
And it’s very likely that, in 2010, SSDs will finally reach mainstream-friendly prices and capacities.
I’m incredibly excited about this. Say what you will about my geekiness or overenthusiasm if you’ve actually made it all the way through this post, but the first time you use a computer with an X25-M, you’ll be this excited, too.
-
Average people don’t realize how little of a hard drive’s performance is bottlenecked by the maximum transfer rate, which is why new interfaces always advertise their sequential-transfer rates. USB with 60 MB/s vs. Firewire 800 at 100 MB/s vs. SATA at 150+ MB/s. It doesn’t really matter — when you’re waiting for seeks, which is most of the time, you’re lucky if the drive transfers more than 10 MB/s. There’s a great car analogy here with the everyday relevance of top speeds if you want to make it. ↩
-
What makes ridiculously expensive, 15,000 RPM server-class hard drives worth their cost to server admins is the reduction in seek times down to the 5ms range. They’re worth a huge price premium just for that. ↩
Would U.S. troops obey presidential orders to deploy against the American people and take away our freedoms?
There is no doubt about it. Of course they would, especially if the president told them that our ‘freedom and national security’ depended on it, which he would.
Rumored Windows Phone 7 details surface: Zune-like UI, no multitasking
Perhaps just as notably, the OS supposedly won’t support multitasking, with applications instead simply pausing themselves when in the background (there will be support for push notifications, though). Also missing is Flash support (at least initially)…
Revolutionary.
Microsoft is confident that the first hardware will be ready by September of this year.
…the browsing experience is currently “better / faster” than the iPhone 3G, and that Microsoft is “aiming towards” the 3GS.
So Microsoft is trying to have a new phone OS and new (presumably third-party) hardware that’s not as fast or good as the iPhone 3GS only 14 months after its release, by which time Apple will likely have released its successor?
Ambitious.
David Lebovitz: Truffle Hunting
I always assumed that truffle pigs had a pretty miserable life. They spend all day looking for their favorite food, and as soon as they find it, it’s taken from them before they can eat it. The poor pigs have to keep finding them, but can never eat any.
This article claims that the human quickly takes the truffle and gives the pig a potato or apple instead, but I bet the pig still knows that it’s getting ripped off.
Optimal iPhone UI
I was asked via email:
It’s my job to research user interfaces and what makes a successful UI. My two conclusions I’ve come to so far are that UIs need to be invisible and familiar.
I was hoping you may have something to add, from your perspective, about what makes an optimal UI.
The “invisible” bit gets most of the way there for my taste.
As a user, I hate buttons and toolbars and sliders and panels and drawers and splash screens and instructional screens and settings screens. In short, I hate UI. I want to notice it as little as possible.
As a developer, I hate having to stop my work because I have to arrange a transaction with someone else (like a designer) to get something done. To minimize external-person dependencies, and because my graphical design skills are abysmal, I avoid needing icons, backgrounds, textures, or logos. The original Instapaper web layout only had one image.1 Since then, I’ve made extensive use of CSS for styling and Unicode characters for icons to minimize the need for images. Designing “invisible”, minimal UIs isn’t just a preference for me — it’s simply more practical.
But I look at Apple’s iPhone apps, and I don’t feel like my method is lazy or sub-par: all of their apps are the same way. They’re content-focused, with minimal UI. Think about it: what’s the UI to the Photos app? Messages? There’s almost nothing there. Even Safari is just two toolbars and a bookmarks list.
My design goal for Instapaper is for it to look and work like a hybrid of Safari and Mail. It’s not exciting and it won’t win much “design”2 recognition, but it’s how I believe most non-widget iPhone apps should be designed: just barely enough UI to get the job done, with the vast majority of the screen devoted to the content.
What’s particularly interesting about the iPad is that its screen is actually too large for this to apply as universally as it does on the iPhone. This is one of the reasons why Instapaper’s iPad edition can’t just be a recompile of the iPhone version: everything’s out of proportion and it just looks strange. It’s like maximizing a browser window on a 30” monitor. And, as I said last week, I suspect most initial versions of iPad apps will have this issue because their developers either didn’t have the time to do more complete redesigns or because they underestimated how different their iPad versions should be.
-
It was the “Read Later” bookmarklet. Jacob corrected its jagged edges for me, unsolicited via email, before I even knew him. In the new layout, these are pure CSS, but I’ve added a few small icons where it made sense to do so. ↩
-
On the internet, “design” usually means heavy use of rich graphics and textures. This isn’t how I think about it. ↩
The Truth About Agave Nectar: It’s All Hype
If you simply must have some sweets once in a while, a small amount of agave nectar every once in a while isn’t going to kill you. Just don’t buy into the idea that it’s any better for you than plain old sugar or HFCS.
The world needs another blog post about the iPad
Given:
- iPhones and iPods Touch in use today: 50-60 million1.
- iPhones sold in the first year (Q3-4, FY 2007): about 1.4 million2. For most of this time, the price started at $499, but nearly everyone bought the $599 model.
- iPhones sold in the second year (FY 2008): about 11.6 million, over half from Q4. Most sales in FY 2008 were the better, cheaper iPhone 3G.
How many iPads do you think will sell in 2010?
This is important for app economics. Many developers want to know how many potential iPad customers will exist to determine how quickly (or even whether) our apps need to be iPad-ready and how we should allocate our resources relative to the iPhone editions.
iPhone sales have largely followed the price drops. I expect the same to apply to the iPad, with the added caveat that I think the overall userbase will be smaller in the beginning. As I said in the first article:
Nearly everyone can justify having a phone — the only question is which phone, and it’s easy for people to rationalize spending a bit more money for the really nice one. The same rationale applies to iPod Touch owners: they’re already buying a portable music player, and the iPod Touch is just the medium- to high-end choice.
The national economic problems seem not to be noticeably affecting sales of high-end tech gadgets, so I won’t factor them in directly.
I’ll take a wild guess and say that I expect about 3 million iPads to be sold this year3.
So how will that translate to app sales from iPad owners? For simplicity, I’ll assume that the value of your app to users is approximately the same on the iPad as on the iPhone.
If my three-million-iPads prediction is accurate, and the iPhone and iPod Touch keep growing at steady rates, by the end of 2010, I predict that about 3-4% of app sales will be to iPad owners.4
But there’s another dilemma: it looks like there will be significant pressure, both political and technical, for universal apps that run on both the iPad and the iPhone. Like multiple-iPhone owners, this only counts as one sale, since one purchase of the app will run on both devices. So for any apps that use a universal edition, their entire existing customer-base will result in zero new sales. (This effect may not matter in the long run. For most iPhone apps, new sales outpace existing-customer upgrades by a large margin.)
A significant portion of iPad owners will probably also own an iPhone or iPod Touch, and you could argue that they would have bought the app for the iPhone anyway.
So not every iPad owner who uses your app will have been the result of another sale.
Economically, there doesn’t seem to be a huge incentive to spend a lot on an iPad edition.
I think, overall, this is a good thing: it should dampen the gold-rush effect, leaving most iPad development to those who care strongly about it and want to do it well.
Running the numbers like this helps clarify my opinion a bit on the iPad App Store timing question: since the initial gold-rush period won’t be anywhere near the magnitude of the iPhone’s, it’s unlikely to hurt your app much in the store if you’re not there exactly on day one.
Meanwhile, if the iPhone is likely to represent at least 90% of your sales for 2010, it’s probably not a good idea to slack off much on the iPhone versions of your apps, even if the iPhone is no longer the focus of the tech press.5
-
Apple has sold over 70 million, but I’m assuming that not every device sold is still in use. ↩
-
Via this Wikipedia image. ↩
-
Through Q1-FY2011 to include the 2010 holiday season. In case you want to claim-chowder me later when I’m way off, I’ll make yet another chowderable claim that my predictions will end up having been comically low, in which case much of this post will be irrelevant, overstated, and wrong. You can usually count on me for that. ↩
-
My math is intentionally fuzzy and hand-wavy. ↩
-
Normally, I dislike footnotes. But they made sense for this article. I’m sorry to anyone who is annoyed at them as I usually am. ↩
All original content is licensed under the Creative Commons Attribution 3.0 U.S. License except that which is quoted from elsewhere or attributed to others. In short, you may reproduce, reblog, and modify my content, but you must provide proper attribution.
