Recent content by Catzilla

I understand FP4 is useful for AI software where it about doubles the number of calculations per second compared to FP8, only if the software is written to use it. Comparing FP4 to FP8 performance in general is misleading, and there is a much smaller increase from RTX 4000 to RTX 5000 series if...

If you are running AI programs, you should have an NVIDIA GPU, RTX 3060 minimum, a 16GB VRAM card is better. You can run quantized (GGUF) models on CPU only, but that's only for LLM, not StableDiffusion. and will be slowerAlso, more CPU cores or threads can be helpful if you are compiling...

I'm not buying any 5000 series card. I read the rumors and they were all consistent enough that I bought an RTX 4070 TI Super a couple months ago because I wanted more VRAM to run AI models. I also play some games at 4K (60FPS) with high quality settings and ray tracing and I'm happy with how...

I think AI is both interesting and worrisome. Worrisome because bad people can use it for questionable or illegal purposes like identity fraud, harassment, intimidation, election manipulation, etc. But that's not AI's fault. Much of this is possible today without AI, given enough time...

AI isn't the first technology that eliminated classes of jobs and it's not likely to be the last. Up until 2000 or so, if I wanted to travel somewhere I went to a travel agency to arrange tickets and reservations. Now I do it on the internet, places like kayak.com or orbitz.com. When I first...

My computer has an RTX 4070 and an RTX 4070 TI Super. (Gigabyte AI Top) I have both cards so I can run larger AI models and have my monitors hooked to the RTX 4070 TI Super for regular use. I was playing video games last night with GPU-Z running to monitor the RTX 4070 TI Super. GPU-Z reported...

I haven't installed it, so no personal experience. But I have not read anything about Windows 11 that convinces me I need it. I'm not interested in signing for online accounts just to use it. I don't think I need trusted boot, secure boot,or whatever it is called, for a PC that is just my...

I voted VRAM since I run AI language models and image generation. I have an RTX 3060 I bought a few years ago and an RTX 4070 I bought last year. I can run small to medium size language models with acceptable performance. Larger models run out of memory. Stable diffusion static image generation...

My computer dual boots Windows and Linux, but lately most often booted to Linux. I use windows when I am photo editing or playing video games. Other than that, Linux is booted. I play around with software development as a learning exercise, including computer graphics/ray tracing and AI stuff...

One category not listed is AI, running either Stable Diffusion image generators or large language models locally. I spent a lot of time last year running these trying to understand the AI stuff. I had a RTX 3060 that would consistently run near 100%. I was vram limited, 12GB vram is close to the...

Software developer for decades. One monitor is impossible. Started using dual monitors in 2005 or so. Hours of amusement trying to configure Linux/XWindows to use dual monitors. Now I have two 32" 4K monitors because I like to have a bunch of windows open at the same time, code editor, broeser...

Yeah, this looks like a joke. You do not want a heavyweight profiling tool running on the system you are trying to tune and you don't want it to become a bottleneck on your system. I have used vTune a few times and it's decent software

This is software that the software developers can use to optimize their code so it runs faster. As a developer, you want something to help you find where your software is spending most of its time and maybe give you clues what is going on inside the processor so you can make the code run faster...

You might be able to run the ComfyUI app for Stable Diffusion image generation, Stable Diffusion v1.5 only. Maybe run quantized 7 billion parameter language models, which I consider to be in the interesting toy category. I think the realistic entry level GPU for AI is the 12GB RTX 3060. Anything...

Hardware requirements aren't entirely NVidia's fault, but the large memory requirements don't hurt when pushing expensive cards. I'm not sure what models they include, since I have software to run these models already (Linux) but the smallest Mistral is a 7 billion parameter model which means it...