I see this as worse for junior engineers than senior. Right now junior engineers don't have the experience to spot issues, ans if gpt-synopsys gives an answer they don't know to question it. It Will also be better than many junior engineers out of the box..
So junior engineers dont get a chance to learn and progress to senior. Senior engineers (for now) will interact with these agents and review the output, ans can direct them. Again, until it becomes better at that.
Interesting times. A lot has happened in a year..where will we be a year from now?
Yes, I work in hardware design ans the cost to code up the RTL for an idea and compare it to the existing code in terms of PPA, its very expensive in terms of engineering time. This week I had 5 different architecites coded, pushed through a functional test bench, bit accurate models created and then run through ppa analysis all done via llms. So its a task tbis project could not afford to do without llms, but can save us 10% in ppa
This is true, but a lot of the top positions are being replaced with unqualified loyalists. It's only a matter of time, if this continues, that the competent workforce gets eroded
My code wasn't written to be hard to decipher, and it wasn't a goal to get everything on one line by any stretch, I just didn't like an if with regext was 2 lines minimum in python, it felt inelegant for a language that is pretty elegant in general
I liked perl, it was the first language I used daily as a HW engineer. When I moved to python more recently what I missed the most was how easy it was to do a one liner if with regex capturing. That couldn't be done in python for a long time. I think the walrus operator helps, but it's still not quite as concise, but it's closer
I heard last year the potential future of gaming is not rendering but fully AI generated frames. 3 seconds per 'frame' now, it's not hard to believe it could do 60fps in a few short years. It makes it seem more likely such a game could exist. I'm not sure I like the idea, but it seems like it could happen
I didn't understand the background before the explanation, but afterwards I did. Inl walked me through the mathematical steps and each was logical and ok to follow if you have a basic calculus knowledge.
It was. I asked it to give more details on parts of the derivation I didn't quite follow and it did that. Overall it was able to build from the ground up to the solution and solve it both numerically and analytically (for smaller values of x)
I just had chatgpt explain that problem to me (I was unfamiliar with the mathematical background). It showed how to solve closed form answers for H(2) and H(3) and then numerical solutions using RK4 for higher values. Truly impressive, and it explained the derivations beautifully. There are few maths experts I've encountered who could have hand-held me through it as good.
This makes sense, but what does not make sense is who tested this 'ultimate mode', I mean they went to the trouble of adding a physical hardware switch on the motherboard for this, surely when testing there was some kind of benchmark or comparison to show this feature was an advantage. Maybe they don't test, or they have 'internal firmware' that is not what the user gets, but it's a serious fail either way.
It takes a long time to get form standard to silicon, so I bet there are design teams working on pcie7 right now, which won't see products for 2 or more years
Exactly this. If a junior dev is never exposed to the task of reasoning about code themselves, they never will know what the difference between good and bad code is. Code based will be littered with code that doe the job functionally, but is not good code, and technical debt will accumulate. Surely this can't be good for the junior Devs or the code bases long term?
No, this is just an a analogy. The reality is the data is heavily modulated, and also the video is encoded so at no point would something that visually looks like an image be visible in the fibre.
Well doesnt this go somewhat to the root of consciousness? Are we not the sum of our experiences and reflections on those experiences? To say an LLM will 'simply' respond as would a character in a sorry about that scenario, in a way shows the power, it responds similarly to how a person would protecting itself in that scenario.... So to bring this to a logical conclusion, while not alive in a traditional sense, if an LLM exhibits behaviours of deception for self preservation, is that not still concerning?
RISC-V is not immune from license fees, unless you want to design a high performance core from the ground up. If you want something as capable as an M4, there is years of R&D to get to that level. I'm sure a big player could do just that in house, but many would license Si-Five or similar. It will be interesting to see if Qualcomm and the like would make a move towards RISC-V, given their ARM legal issues
I have it in docker a d use supervised mode (which seems discouraged, but I want my machine for other uses also). The one thing I struggle with is updating, I'm concerned if I update it'll break. Is there a way to fully snapshot a container state and it's disk state, so I can 100% restore to it if something goes wrong ? I'm still running HA from 2020 because of this.
The other think I'm not a huge fan of is it's template language, it's clunky to say the least, but overall it's a great amd flexible system
Especially when doing the right/safe thing by default is at worst a minor performance hit. They could change the default to be sane and provide a backwards compatible switch to pragma to revert to the less safe version. They could, but for some reason never seem to make such positive changes
SystemVerilog is a much better language than verilog. You can get pretty strong typed behaviour now, if you use the language parts that allow that. It's like C, if you use it poorly/the 'verilog way' it's got some serious footguns, but you can use it quite safely if you use the more modern features.
That said, I'm all for better languages, of they really are better and as expressive
This is something that has been tried, basically constructing an analog matrix multiply/dot product and it gives reasonable power efficiency at into levels of precision. More precision and the analog accuracy leads to dramatic power efficiey losses (each bit is about 2x the power), so int8 is probably the sweet spot. The main issues are it is pretty inflexible and costly to design vs a digital int8 mac array, and hard to port to newer nodes, etc
I thought this was going to talk about how printf is implemented. I worked with a tiny embedded processor that had 8k imem, and printf is about 100k alone. Crazy. Switched to a more basic implementation that was around 2k, and ran much,much faster. It seems printf is pretty bloated, though I guess typical people don't care.
This is a very real issue for even expensive items. If you buy a car today, doesn't it have some sw component that requires phoning home to work properly. What if that company goes bust? Or if a war breaks out and that company is in a country on the other side? Both Tesla and byd seem like challenging purchases with all that's going on
How often do tabs really need to communicate, and when they do, does it really need to be as fast as possible. I would say slower and secure would be a better design philosophy, especially as tab interaction is generally rare, and low bandwidth