A comparison of some CPUs using my favorite benchmark, md5 -t.

Dell CS24, Xeon L5450 @ 2.5GHz

Time   = 0.242135 seconds
Speed  = 412992751.977203 bytes/second

Thinkpad T430s, i5-3320M @ 2.6GHz (plus turbo)

Time   = 0.184372 seconds
Speed  = 542381706.549801 bytes/second

Thinkpad X200s, Core2 @ 1.8GHz

Time   = 0.325009 seconds
Speed  = 307683787.218200 bytes/second

Thinkpad X1 Carbon, i5-5300U @ 2.3GHz

Time   = 0.206281 seconds
Speed  = 484775621.603541 bytes/second

No name router, Atom @ 1.8GHz

Time   = 0.399222 seconds
Speed  = 250487197.599331 bytes/second

Sun T5120, T2 @ 1.2GHz

Time   = 1.809987 seconds
Speed  = 55249015.600665 bytes/second

BeagleBone Black, ARM Cortex A8

Time   = 1.373115 seconds
Speed  = 72827112.077284 bytes/second

EdgeRouter Lite, Octeon @ 500MHz

Time   = 2.198556 seconds
Speed  = 45484399.760570 bytes/second

Intel “Braswell” Celeron N3050 @ 1.6GHz

Time   = 0.334014 seconds
Speed  = 299388648.380008 bytes/second

A short note about my Dell CS24 to accompany the post about the Sun T5120.

more...

I’ve been looking for a sparc64 system for a while and noticed the Sun Enterprise T5120 models have become very affordable. They’re interesting machines and great for testing due to the built in virtualization support.

more...

Dammit, Apple, stop turning Bluetooth on after every iOS update. I turned it off for a reason.

Spotify had a blog post about how to shuffle songs, which included a link to earlier work on the art of shuffling music. The original algorithm uses a lot of both memory and CPU (in particular, a playlist containing a lot of loosies will be extremely memory hungry as each song is expanded). I think I understand how to implement the Spotify “dithering” algorithm efficiently, but there’s no pudding.

more...

The disclosure of the recent GnuTLS vulnerability forces me, as if against my will, to retread some tired ground. It’s been a busy week. For serious this time.

more...

I’m reading a CloudFlare blog post about serialization in Lua, and I’m thinking this might be useful. Then I scroll down to see what it looks like in action.

Err, that looks kinda weird. Now I’m thinking maybe this isn’t the serialization library for me.

Perhaps it’s just a mistake? View source.

<p>A sample data table looks like this:</p>
<pre><code>local data = {
    people = {
        {
            id = "123",
            name = "Alice",
            email = "<a class="__cf_email__" href="http://www.cloudflare.com/email-protection" data-cfemail="f3929f9a9096b3968b929e839f96dd909c9e">[email&nbsp;protected]</a><script type="text/javascript">
/* <![CDATA[ */
(function(){try{var s,a,i,j,r,c,l,b=document.getElementsByTagName("script");l=b[b.length-1].previousSibling;a=l.getAttribute('data-cfemail');if(a){s='';r=parseInt(a.substr(0,2),16);for(j=2;a.length-j;j+=2){c=parseInt(a.substr(j,2),16)^r;s+=String.fromCharCode(c);}s=document.createTextNode(s);l.parentNode.replaceChild(s,l);}}catch(e){}})();
/* ]]> */
</script>",
            phones = {

Ah, yes. The email protector has protected dear Alice’s email, but didn’t quite manage to get out of its own way. I’m assuming nobody typed that mess in on purpose, which means CloudFlare has some automatic protection injecting proxy magic. Too much magic for me.

Apple recently made a booboo, unlike any other booboo in the history of programming. Even though Apple’s bug is unprecedented, here’s a brief overview of some predecessor bugs.

more...

signify was (is) developed on OpenBSD-current and depends on a few features only found there. The first signed OpenBSD release, 5.5, will also be the first release to include signify. What to do if you’re running OpenBSD 5.3 or 5.4 and want to verify the 5.5 release you just downloaded is legit? You need the signify backport.

It’s all the necessary files packed into one directory. It should work on a few previous OpenBSD releases, and possibly other BSD systems, but beyond that it’s a lost cause. I don’t have much interest in making a true portable version.

This tgz also includes the set of public keys for 5.5 and 5.6. Building is simple. Run make and make install.

signify-backport.tgz

Several years ago, when Heroku was young and fresh, I had a question about their service. The only way to ask them a question was to create an account and file a support ticket, so that’s what I did. Some time passed and then suddenly I started receiving monthly Heroku newsletters. While I don’t recall there even being such a checkbox when I signed up, I am always fastidious about not opting in to receive such junk, even with fiendish “By not unchecking this box, you agree to not unsubscribe from our crap” phrasing, but nevertheless I find myself enrolled in their newsletter.

I have been ignoring the newsletter because unsubscribing is usually more trouble than it’s worth and my spam filter knows exactly where to put it. Eventually, somebody (or something) at Heroku caught on and a month ago I received this email:

“You are receiving this email because we do not have a current subscriber status on you for Heroku emails.“

You’d think no status would imply no newsletter, right? That email was followed an hour later by the January newsletter. The status email itself continued,

“If you would rather not opt-in to receive Heroku emails, you do not need to take any action. We will automatically change your status to opted-out within the next 30 days.“

Today I received the February newsletter.

Today I received the March newsletter. Looks like Heroku has decided to change my status to opt-in instead. Thanks.