tails4e

11 ms·

91.
by tails4e 3y ago
We use confluence and jira, an d for some reason it seems perfectly usable compared to others stories here. It's not a small organization either, so I wonder if it's some misconfiguration that's killing performance for others
92.
by tails4e 3y ago
You're right, my mistake. Still cost and power per compute still favours AMD
93.
by tails4e 3y ago
When you say best performance on nvidia, do you mean against any other method of running this model an nvidia card?
94.
by tails4e 3y ago
Being 90% of a 4090 makes the 7900 xtx very attractive in a cost per compute perspective, as its about 65%, and power is significantly lower too
95.
by tails4e 3y ago
Nicely done, congrats. A really good reference, and the style and flow is very nice.
96.
by tails4e 3y ago
Absolutely. If anything ML is migrating to lower precision types, like fp16, mx9, mx6, int8, etc.
97.
by tails4e 3y ago
I recently tried to implement a 16bit 64k point fft in C using FP32. It thought it would work easily. Turns out it was very difficult, what worked fine in double gave huge inaccuracies in float. It was due to the large small issue, if you s
98.
by tails4e 3y ago
Exactly. I heard recently the Lumi supercomputer team converted loads of scientific code from CUDA to Hip to run on MI250, and apparently it was pretty seamless, so a strong sign this approach works
99.
by tails4e 3y ago
Is the RTL autogenerated from the chip images? I looked at one file 68k.v and the variable names remind me of a decompiler.
100.
by tails4e 3y ago
Was just on to apple support for an issue with my daughters phone, and I have to say it was a really nice experience. Good online chat first (may be a bot, but if so was a good one), and immediate call from a senior advisor when it needs to
101.
by tails4e 3y ago
Nutrition for me it a big issue. It seems like and ideal area for scientific and medical study to give us light, but seems so hard to get truly objective info. Fat is the devial/good/bad/OK, sugar is the devil/bad/o
102.
by tails4e 3y ago
The recent price reductions were seriously warranted to make it competive again. 88k for the current base model is pretty good
103.
by tails4e 3y ago
This is a very good point. The model s when it was 80k was a lot of car for the money. At 130k, the main selling point is the insane acceleration, which is fun, but not 50k worth. The other improvements are nice, but as you say many very hi
104.
by tails4e 3y ago
Excluding the investors that have literally paid for an advantage sounds fair to me (buying real estate closer to thr exchange because every ns counts).
105.
by tails4e 3y ago
I agree. HFT goes against the original ethos of a market open to all. Instead of trading on the merit of a given company, it simply reinforces the market is increasingly gamified and unfair. It's like futures, no ones cares about the c
106.
by tails4e 3y ago
I don't think in order is related to the ISA. There may be a guarantee of the order of operations when it comes to the view of memory the programmer sees, but that does not prevent the hardware from executing out of order, as long as t
107.
by tails4e 3y ago
I've been watching the videos of the ladybird browser development, and how it improves it's css compatability. I'm eqal parts impressed and aghast. The standards seem overly complex, and very hard to be compliant. It amazes m
108.
by tails4e 3y ago
Beautifully done. God I wish our fourier series lecturer in university motivated the class even a little with illustrations like this. It was 100% derivation and hard maths, no intuition at all
109.
by tails4e 3y ago
An RFSoC 2x2 kit [1] could handle the RF part (with an FPGA included!), it's 2k, still expensive. 1: https://www.xilinx.com/support/university/xup-boards/RFSoC2x...
110.
by tails4e 3y ago
Is there anything like this for ASIC design / Verilog?
111.
by tails4e 3y ago
Sounds like an open source/wiki for errata could be useful, at least then reports like yours could be published of the vendors do not
112.
by tails4e 4y ago
IMEC is govt funded in the EU and is very successful. Many of the advanced semiconductor techniques are trail blazed there before being adopted by industry world wide.
113.
by tails4e 4y ago
It's been around a long time, but a major contribution to the worsening of late is the US news media. When my family visited the us in the early 2000s we thought fox news was a parody at first. We could not believe it was being held up
114.
by tails4e 4y ago
I always interpreted moores law as having 2 aspects, the scaling part (cramming more per mm2), and the economic aspect (for minimum component cost). So when people say moores law is dead, it's often confusing as clearly things are stil
115.
by tails4e 4y ago
I guess the bit I was referring to as somewhat novel is when he referred to memory being combined with compute rather than separate - As 'compute in memory' or distributed memory and compute architectures, are recent enough. separ
116.
by tails4e 4y ago
It's incredible how in those few paragraphs he envisioned so many things that could be enabled by cramming more transistors in, that a half century later are still somewhat novel.
117.
by tails4e 4y ago
That's a stretch. We're literally talking about arid almost-wasteland. A few sq miles is a small portion, yet will provide clean power. The cost benefit is abundantly clear.
118.
by tails4e 4y ago
I reguarly use Vivado elaborated design view to browse RTL hierarchy. The schematic view, signal trace, etc is the best I've seen. You can also automate it with tcl, so I've used it to auto highlight clock domains in different col
119.
by tails4e 4y ago
Yes, it's crazy, and yet this is how it's supposed to function. Boggles the mind. Solid healthy profits have little value, and companies are forced to think short term rather than long
120.
by tails4e 4y ago
From what I've seen the CPU/ALU/decode at the center being ARM or x86 may make less difference than you think. The amount of circuitry and silicon area (correlated with power) for non core is significant. MMUs, vector instruc

More ›