@ash_twtz Started with Jellybean, 1 Gb RAM, 16 GB ROM...
Felt like owning a computer in my hands.
Apps ran so smoothly.
Nowadays even a device with 6GB RAM and 128GB storage lags and feels sluggish sometimes π€£π€£
Imagine the bloat we have accumulated over time.
@datavorous_ Which NPU/chipset did you try?
I have a Radxa Dragon Q6A (QCS6490). Most LLMs either fail compilation or silently fall back to CPU. I've only managed to get YOLOv8 running because it was documented.
Any Qualcomm-specific profiling or debugging tools you'd recommend me to check?
@TheBongHead The best punishment this fool should get is that he should be made to install thousands of these in the city roads for free and repair all damaged ones even if he didn't damage those.
That way the city also benefits. This should be made a rule, whoever damages public property
@AnkurSingh What if some 3000-4000 students were common π€£
In both the coaching centers?
We are not assuming someone who'd enrol on Khan Sir's coaching would not enrol on Raushan Sir's coaching. And vice versa
@3mul0rd@_prophet_futu Because windows developers figured out the lazy option.. Package everything as a web app that runs its own web browser (Most apps are built on Electron framework nowadays)
which essentially embeds a full Google Chromium browser inside all these web apps nowadays
@scientifitt_ Breed a lot of white owls (or barn owls), feed them pigeons, encourage their growth, let them eat pigeons. Fight nature with nature. π€£
Pushing @RadxaComputer Dragon Q6A pretty hard lately
Built a custom lightweight monitoring tool:
β’ CPU/GPU/NPU thermals
β’ power rails
β’ container health
β’ network/storage stats
Very impressed by the Q6A for local AI and homelab workloads.
Repo in comments
#Radxa#DragonQ6A
@NVIDIAGeForce Me pretending this GPU is for #007FirstLightRTX while secretly planning a local AI setup powerful enough to concern both my electricity meter and nearby transformers
@rishibagree The authentication mode should be changed/updated for the blind users. Entering a pin like that poses a security risk as it is visible to everyone nearby, and bad actors can misuse it. There should be biometric/fingerprint authentication instead.
@Zai_org All great, I've been using GLM since September last year, built a couple of personal projects, was quite happy.
But prices of the coding plans went up 5-6 times since then
What was once extremely affordable and offered superb value for money is now gone, unfortunately. :(