Preparing lecture notes (or videos) sometimes brings you to revisit things you’ve known for a while, but haven’t really taken time to formalize properly. One of those things is fixed-point arithmetic.
This week, I’ll show my mirror assembly to reverse the image “in hardware” for the lightboard.
Last time, I gave the instruction on how to build a lightboard, but not much in terms of how you actually use it. Now I’ve been giving lectures from it (with graduate students as test subjects), I’ve started recording for the undergrad courses, and so I’ve tweaked my setup and learnt a few tricks. This week, I’ll discuss some of them
The C99 <stdint.h> header provides a plethora of type definition for platform-independent safe code: int_fast16_t, for example, provides an integer that plays well with the machine but has at least 16 bits. The int_fastxx_t and int_leastxx_t defines doesn’t guarantee a tight fit, they provide an machine-efficient fit. They find the fastest type of integer for that machine that respects the constraint.
But let’s take the problem the other way around: what about defines that gives you the smallest integer, not for the number of bits (because that’s trivial with intxx_t) but from the maximum value you need to represent?
As you may have noticed, a global pandemics got many of us working from home. While one can argue that you can do accounting from home, it’s a lot more complicated to teach from home. Pretty much everyone is trying to figure that one out. For my part, I decided that zoom and the virtual whiteboard is not very interesting. Like many, I decided to use a lightboard.
So, the problem is, where do you get a lightboard on short notice? Well, you can build one. Let’s see how:
Last week (at the time of writing, anyway), Ars Technica reported a serious bug in AMD’s implementation of rdrand, an instruction that helps you generate random numbers. Apparently, on (some) Ryzen 3000, 0xfff…ff is “random”.
I recently got an AMD Ryzen 9 3900x, and I wondered if I have the bug as well.
Remember ye olde dayes when we had to be mindful of the so-called “web safe palette“? Once upon a time, screens could display 24-bits colors, but only 256 at a time in some “hi-res” modes. But that’s not what I’m going to tell you about: I’d rather tell you about the encoding of the palette, and about a somewhat better palette. And also about using fractions of bits for more efficient encodings.
Sometimes, it’s the little things that make a big difference. For example, moving small, not machine-sized, pieces of data may be a problem. If it happens to be machine-sized (that is, the size of something like uint32_t), then the copy is readily done by a single (integer) copy. But what if it’s, say, 5 bytes?
The first solution that comes to mind is either a for-loop or a call to a library function like memcpy. The for-loop would give something like:
char * dest = ... char * src = ... for (std::size_t i=0;i<5;i++) dest[i]=src[i];
If you’re lucky, the compiler understands the small copy and optimizes. More likely, it will merely unroll the loop and generate something like:
14d4: 0f b6 07 movzx eax,BYTE PTR [rdi] 14df: 88 04 24 mov BYTE PTR [rsp],al 14e2: 0f b6 47 01 movzx eax,BYTE PTR [rdi+0x1] 14e6: 88 44 24 01 mov BYTE PTR [rsp+0x1],al 14ea: 0f b6 47 02 movzx eax,BYTE PTR [rdi+0x2] 14ee: 88 44 24 02 mov BYTE PTR [rsp+0x2],al 14f2: 0f b6 47 03 movzx eax,BYTE PTR [rdi+0x3] 14f6: 88 44 24 03 mov BYTE PTR [rsp+0x3],al 14fa: 0f b6 47 04 movzx eax,BYTE PTR [rdi+0x4] 14fe: 88 44 24 04 mov BYTE PTR [rsp+0x4],al
Let’s see how we can fix that!
This week, we’ll discuss a cool, but failed, experiment.
In the last few weeks (of posts, because in real time, I worked on that problem over a week-end) we’ve looked at how to generate well distributed, maximally different colors. The methods were to use well-distributed sequences or lattices to ensure that the colors are equidistant. What if we used physical analogies to push the colors around so that they are maximally apart?
Last week, we’ve had a look at how to distribute maximally different colors on the RGB cube. But I also remarked that we could use some other color space, say HSV. How do we distribute colors uniformly in HSV space?