ribit

18 ms·

151.
by ribit 4y ago
If I lose the key, sure, getting inside the house will be more difficult, isn’t that the entire point? Still, I don’t understand the relevance of the analogy. I have access to the entire system and can pretty much tweak things as I want. I
152.
by ribit 4y ago
Oh please. You are quoting an issue that happened two years ago and Apple has already acknowledged that their algorithms were crappy and promised to reengineer this stuff. This is like claiming that Linux does not allow you to install softw
153.
by ribit 4y ago
I feel like I pretty much have full control over my Mac laptop. Why would you expect to fully control a machine from inside a VM anyway? Sounds like a massive security risk too.
154.
by ribit 4y ago
What does Apple have to do with it? Their folder structure is fairly straightforward and transparent. Windows stuff is definitely the craziest I’ve encountered.
155.
by ribit 4y ago
I would almost view Proton as a risk factor for Vulkan as it discourages the devs to use cross-platform approaches. Much easier to write only for Windows and let Proton folks to figure out the compatibility problems. This also results in Vu
156.
by ribit 4y ago
Just a minor nitpick: Vulkan might have Mantle DNA but it severely limited some of the original functionality (no pointers in resource binding). Funnily enough the closest API that matches what Mantle could do is Apples Metal.
157.
by ribit 4y ago
Mixed precision won't do anything on Apple Silicon anyway since there is no performance advantage to using FP16 (aside from decreasing register pressure and RAM bandwidth which won't happen here as data is FP32 to start with).
158.
by ribit 4y ago
I think that about sums it up. The beauty of mesh shaders is that they don’t define how your input interface looks like, as long as you produce something in the end that the rasterizer can consume. Mesh shaders define two levels of compute
159.
by ribit 4y ago
I think OpenCL was a pivotal moment. Apple created the draft and donated it to Khronos, where is subsequently was stagnating before Nvidia effectively sabotaged the effort in order to push its own proprietary CUDA. Since then Apple has been
160.
by ribit 4y ago
Apple AMX is a private extension. It is not documented and not exposed to the developer. You have to use system-provided HPC and ML libraries to take advantage of the AMX units. This gives Apple the freedom to iterate and change implementat
161.
by ribit 4y ago
Long long time ago when Asus introduced the netbook I purchased an the Eee PC. I installed Linux on it and felt like one of the cool kids, sitting there in my university class typing on my tiny keyboard into the tiny vim window.
162.
by ribit 4y ago
Optane was a very interesting technology but I don't think it as revolutionary as the article makes it to be. First, Optane does not make the memory hierarchy simpler as you still need fast RAM to solve the write endurance problem. Sec
163.
by ribit 4y ago
> But the original article is completely, laughably wrong about one thing: an atomic increment or decrement is a remarkably slow operation on modern hardware, second only to pointer chasing. Very true. No matter how one looks at it, GC i
164.
by ribit 4y ago
I am curious about this statement: [OOP] might mean ... “modules, but using the word class” I would argue that "modules" is almost as useless of a term as "OOP". What does the author mean exactly?
165.
by ribit 4y ago
It's great that cross-vendor barycentric coordinate support is finally arriving in Vulkan. I've been using them on Apple platforms for a while now, the ability to do interpolation manually gives you a lot of flexibility in how you
166.
by ribit 4y ago
These days I tend to use R for these type problems. It’s slow, but vectors as first-class data types simplify things a lot readLines(file) |> strsplit(“ +”) |> unlist() |> table() |> sort(desc = TRUE)
167.
by ribit 4y ago
I am not quite sold on the dichotomy between “REST” and “RPC” as suggested by the article: if sending over a structured response that has to be interpreted by a client program is considered RPC then why is sending over a structured response
168.
by ribit 4y ago
The main advantage is that you can write a document, send it per email or check it into a VCS and the other party can immediately see the rich rendered version without having to install anything. I don’t see much value for websites, except
169.
by ribit 4y ago
I use it for documentation, notes and manuals and it’s great for that purpose. The fact that it renders out of the box greatly simplifies things, as you don’t need to deal with setting up and maintaining any infrastructure.
170.
by ribit 4y ago
I have been using an M1 Max MacBook Pro since last year. It builds code as fast as a desktop workstation, runs data wrangling workflows around three times faster than my previous Coffee Lake i9 Laptop, offers a Full day battery life (I have
171.
by ribit 4y ago
I have no idea about any of these things, but running memory bus via PCI-e sounds like a major regression in terms of power usage. Also.. what about latency? I mean, having all your RAM pooled into one coherent logical space sounds neat on
172.
by ribit 4y ago
My main headache with git — which is absent from the list in the blog – is that it tracks snapshots instead of tracking changes. It's probably less of an issue for software developers where a "version" is what this all is abo
173.
by ribit 4y ago
Now can someone do the same for web dev? I just started getting into it for a hobby project and the terminology is so incredibly idiosyncratic.
174.
by ribit 4y ago
I can’t help but think that the article is highly manipulative. It’s hard to argue with the author on the logical points, but they even to go out of the way to avoid addressing the elephant in the room: the fact that Google controls around
175.
by ribit 4y ago
First: we need to finally stop the harmful myth that floating point multiplication is slower than addition. This has not been true for a long while. Second: why are so many people insisting that the loop is auto-vectorised? Is there any evi
176.
by ribit 4y ago
But how does that apply to GPU ALUs? Looking at M1 die shots, they are comparatively tiny, and when comparing to other vendors, it doesn't seem like Apple is dedicated more logic space to the GPU. The M1 die is roughly 120mm2, an Nvidi
177.
by ribit 4y ago
Cache doesn't matter for pure ALU efficiency though. I mean, I did tests on long dependent chains of FMAs, the only memory touched there are two internal registers.
178.
by ribit 4y ago
And yet somehow Apples GPU ALUs are more efficient at 3.8 watts per TFLOP. Mind, I am not talking about specialized matrix multiplication units that have a different internal organization and can do things like matrix multiplication much m
179.
by ribit 4y ago
I entirely agree that Ive's design brilliance required Job's meticulous attention to details and taste to be really successful. He did make some questionable choices after Job's passing. And the article doesn't make much
180.
by ribit 4y ago
So far, Apple is the only vendor that makes really pragmatic hardware with true linear scaling. Cache sizes, RAM, power consumption, compute cluster configurations — everything makes sense and everything is balanced. It's a really refr

More ›