hajile

35 ms·

391.
by hajile 2y ago
Medfield was faster than A9 and Qualcomm Krait in performance, but not so much in power (see Motorola Razr i vs M where the dual-core ARM version got basically the same battery life as the single-core x86 version). Shortly after though, ARM
392.
by hajile 2y ago
X Elite is faster, but not enough to offset the software incompatibility or dealing with the GPU absolutely sucking. Unfortunately for Intel, X Elite was a bad CPU that has been fixed with Snapdragon 8 Elite's update. The core uses a t
393.
by hajile 2y ago
They killed StrongARM because they believed the x86 Atom design could compete. Turns out that it couldn't and most of the phones with it weren't that great. Intel should be focused on an x86+RISC-V hybrid chip design where they ca
394.
by hajile 2y ago
As a DM42 owner who loves RPL, this is an exciting project.
395.
by hajile 2y ago
It's a bit older (and focused on web), but you can try this https://github.com/pomber/didact At a high level though, you create a tree of nested objects. Each has a handful of special properties (children which co
396.
by hajile 2y ago
Some people here seem to have a hard time understanding that React !== DOM React is descriptive. It creates a virtual tree and tells how that tree should be changed. What that tree renders into is a completely different problem. On the web,
397.
by hajile 2y ago
If I'm not a professional and I hurt someone while trying to save their life by doing something stupid, that's understandable ignorance. If a doctor stops to help someone and hurts them because the doctor did something stupid, tha
398.
by hajile 2y ago
They aren't going to be using fp32 for inferencing, so those FP numbers are meaningless. Memory and memory bandwidth matters most for inferencing. 819.2 GB/s for M2 Ultra is less than half that of A100, but having 192GB of RAM ins
399.
by hajile 2y ago
Apple wants $1200 for a 4TB SSD. I'm sure a LOT of people would gladly pay $300 for a top-end SSD of the same size and pay someone to install it for $100 and still save $800 on the price of the machine.
400.
by hajile 2y ago
Intel supposedly put in a multi-billion dollar offer and got laughed out of the room.
401.
by hajile 2y ago
You are wildly mistaken here. Back in 2016, BOOMv1 (Berkley Out-of-Order Machine) had pipelining, register renaming, 3-wide dispatch, branch predictor, caches, etc. A quick google seems to indicate that it was started in 2011 and had taped
402.
by hajile 2y ago
The capacitors take time to charge and discharge. You can't do that more than around 400MHz with current materials. You are correct that it means you can't access the same bit of memory more than 400M/sec. This is the same wh
403.
by hajile 2y ago
Absolutely. There are nearly 2B smartphones sold each year and only 200M laptops, so Apple's 20M laptop sales are basically a rounding error and not worth considering. As 15-16% of the smartphone market, Apple is generally selling arou
404.
by hajile 2y ago
Oracle v Qualcomm would be epic.
405.
by hajile 2y ago
I don't believe any major company has done it. Even Intel failed numerous times to move away from x86 with iAPX432, i960, i860, and Itanium all failing to gain traction.
406.
by hajile 2y ago
Years ago. DRAM uses a capacitor. Those capacitors essentially hit a hard limit at around 400MHz for our traditional materials a very long time ago. This means that if you need to sequentially read random locations from RAM, you can't
407.
by hajile 2y ago
ARM already did the hard work. Once you've ported your app to ARM, you've no doubt made sure all the ISA-specific bits are isolated while the rest is generic and portable. This means you already know where to go and what to change
408.
by hajile 2y ago
Qualcomm pitched Znew quite a while ago. It mostly ditched 16-bit instructions and added a bunch of instructions that were basically ripped straight from ARM64. The idea was obviously an attempt at making it as easy as possible to replace A
409.
by hajile 2y ago
Qualcomm pitched a Znew extension for RISC-V that basically removes compressed (16-bit) instructions and adds more ARM64-like stuff. It felt very much like trying to make an easier plan B for if/when they need/want to transition f
410.
by hajile 2y ago
Graviton4 is based on Neoverse V2 which is based on X3. Neoverse V3 was announced Feb of this year. It should have an 80-85% performance increase and should basically be based on something very close to x925.
411.
by hajile 2y ago
Qualcomm doesn't have nearly as much to lose as ARM does and they know it. Qualcomm is almost certainly ARM's biggest customer. If ARM loses, Qualcomm doesn't have to pay out. If ARM wins, Qualcomm moves to RISC-V and ARM los
412.
by hajile 2y ago
Now that the Nvidia deal has fallen through, SoftBank is trying to ramp up profitability while continuing to search for another buyer.
413.
by hajile 2y ago
We're not quite there yet. A bunch of mission critical stuff like SIMD were only added in the last 2-3 years. As it takes 4-5 years to design/ship high-performance chips, we still have a ways to go. Ventana Veyron looks interestin
414.
by hajile 2y ago
AWACS can't lock onto a stealth plane from that far away.
415.
by hajile 2y ago
The only military commentators saying Russian troops are untrained are pretty ignorant and biased. Russia certainly sent untrained troops in the early part of the war, but most of them were by mistake and got recalled quickly. Russia recogn
416.
by hajile 2y ago
It's not a novel approach. Russia baked it into their aiming systems decades ago and they supposedly train on the maneuver. The software can detect when the correct pitch/direction are achieved so the barrage can be reasonably acc
417.
by hajile 2y ago
If the illuminating plane is not a stealth plane, then it'll get shot by the stealth plane long before it can detect the stealth plane. If the illuminating plane is a stealth plane, then you've just traded on stealth plane for ano
418.
by hajile 2y ago
Russia didn't defend Kursk because they didn't think Ukraine would be stupid enough to make that attack. That was a temporary PR win and a permanent strategic loss. The results have been the elimination of Ukraine's best troo
419.
by hajile 2y ago
A lot of sci-fi gets tossed around about 5th-gen fighters and stealth vs non-stealth is extrapolated onto everything. If you're communicating, you are broadcasting your position to the enemy. That means you can be shot down. If you&#x2
420.
by hajile 2y ago
In the case of HIMARS, we've seen what the Ukrainian version can do when there isn't jamming, so I don't see nerfing being a valid excuse.

More ›