Not sure if this is the right place, if not please let me know.
GPU prices in the US have been a horrific bloodbath with the scalpers recently. So for this discussion, let’s keep it to MSRP and the lucky people who actually managed to afford those insane MSRPs + managed to actually find the GPU they wanted.
Which GPU are you using to run what LLMs? How is the performance of the LLMs you have selected? On an average, what size of LLMs are you able to run smoothly on your GPU (7B, 14B, 20-24B etc).
What GPU do you recommend for decent amount of VRAM vs price (MSRP)? If you’re using the TOTL RX 7900XTX/4090/5090 with 24+ GB of RAM, comment below with some performance estimations too.
My use-case: code assistants for Terraform + general shell and YAML, plain chat, some image generation. And to be able to still pay rent after spending all my savings on a GPU with a pathetic amount of VRAM (LOOKING AT BOTH OF YOU, BUT ESPECIALLY YOU NVIDIA YOU JERK). I would prefer to have GPUs for under $600 if possible, but I want to also run models like Mistral small so I suppose I don’t have a choice but spend a huge sum of money.
Thanks
You can probably tell that I’m not very happy with the current PC consumer market but I decided to post in case we find any gems in the wild.
You must log in or register to comment.
You also could cool into high core count CPUs.
Using 7900XTX with LMS. Speed are everwhere, driver dependent. With QwQ-32B-Q4_K_M, I got about 20 tok/s, with all VRAM filled. Phi-4 runs at about 30-40 tok/s. I can give more numbers if you can wait for a bit.
If you don’t enjoy finding which driver works best, I strongly aginst running AMD for AI workload.
I didn’t know that. I thought just one ROCM binary to install, run Ollama and that’s it. Thanks for the explanation
I got it working with my 6800XT. I’m running deep seek r1 14b (somewhere around there) and the deep seek coder V2. I have a link to a blog with those instructions
https://gotosocial.michaeldileo.org/@mdileo/statuses/01JQA4M4Q33PMCADH9M2AWQSS8
Thank you. Are 14B models the biggest you can run comfortably?
The coder model has only that one. The ones bigger than that are like 20GB+, and my GPU has 16GB. I’ve only tried two models, but it looked like the size balloons after that, so that may be the biggest models that I can run.
Do you have any recommendations for running the Mistral small model? I’m very interested in it alongside CodeLlama, OogaBooga and others
I haven’t tried those, so not really, but with open web UI, you can download and run anything, just make sure it fits in your vram so it doesn’t run on the CPU. The deep seek one is decent. I find that i like chatgpt 4-o better, but it’s still good.
In general how much VRAM do I need for 14B and 24B models?
It really depends on how you quantize the model and the K/V cache as well. This is a useful calculator. https://smcleod.net/vram-estimator/ I can comfortably fit most 32b models quantized to 4-bit (usually KVM or IQ4XS) on my 3090’s 24 GB of VRAM with a reasonable context size. If you’re going to be needing a much larger context window to input large documents etc then you’d need to go smaller with the model size (14b, 27b etc) or get a multi GPU set up or something with unified memory and a lot of ram (like the Mac Minis others are mentioning).
Oh and I typically get 16-20 tok/s running a 32b model on Ollama using Open WebUI. Also I have experienced issues with 4-bit quantization for the K/V cache on some models myself so just FYI
Anything under 16 is a no go. Your number of CPU cores are important. Use Oobabooga Textgen for an advanced llama.cpp setup that splits between the CPU and GPU. You'll need at least 64 GB of RAM or be willing to offload layers using the NVME with deepspeed. I can run up to a 72b model with 4 bit quantization in GGUF with a 12700 laptop with a mobile 3080Ti which has 16GB of VRAM (mobile is like that).
I prefer to run a 8×7b mixture of experts model because only 2 of the 8 are ever running at the same time. I am running that in 4 bit quantized GGUF and it takes 56 GB total to load. Once loaded it is about like a 13b model for speed but is ~90% of the capabilities of a 70b. The streaming speed is faster than my fastest reading pace.
A 70b model streams at my slowest tenable reading pace.
Both of these options are exponentially more capable than any of the smaller model sizes even if you screw around with training. Unfortunately, this streaming speed is still pretty slow for most advanced agentic stuff. Maybe if I had 24 to 48gb it would be different, I cannot say. If I was building now, I would be looking at what hardware options have the largest L1 cache, the most cores that include the most advanced AVX instructions. Generally, anything with efficiency cores are removing AVX and because the CPU schedulers in kernels are usually unable to handle this asymmetry consumer junk has poor AVX support. It is quite likely that all the problems Intel has had in recent years has been due to how they tried to block consumer stuff from accessing the advanced P-core instructions that were only blocked in microcode. It requires disabling the e-cores or setting up a CPU set isolation in Linux or BSD distros.
You need good Linux support even if you run windows. Most good and advanced stuff with AI will be done with WSL if you haven’t ditched doz for whatever reason. Use https://linux-hardware.org/ to see support for devices.
The reason I mentioned avoid consumer e-cores is because there have been some articles popping up lately about all p-core hardware.
The main constraint for the CPU is the L2 to L1 cache bus width. Researching this deeply may be beneficial.
Splitting the load between multiple GPUs may be an option too. As of a year ago, the cheapest option for a 16 GB GPU in a machine was a second hand 12th gen Intel laptop with a 3080Ti by a considerable margin when all of it is added up. It is noisy, gets hot, and I hate it many times, wishing I had gotten a server like setup for AI, but I have something and that is what matters.
I don’t mind multiple GPUs but my motherboard doesn’t have 2+ electrically connected X16 slots. I could build a new homeserver (I’ve been thinking about it) but consumer platforms simply don’t have the PCIE lanes for 2 actual x16 slots. I’d have to go back to Broadwell Xeons for that, which are really power hungry. Oh well, I don’t think it matters considering how power hungry GPUs are now.
I haven’t looked into the issue of PCIe lanes and the GPU.
I don’t think it should matter with a smaller PCIe bus, in theory, if I understand correctly (unlikely). The only time a lot of data is transferred is when the model layers are initially loaded. Like with Oobabooga when I load a model, most of the time my desktop RAM monitor widget does not even have the time to refresh and tell me how much memory was used on the CPU side. What is loaded in the GPU is around 90% static. I have a script that monitors this so that I can tune the maximum number of layers. I leave overhead room for the context to build up over time but there are no major changes happening aside from initial loading. One just sets the number of layers to offload on the GPU and loads the model. However many seconds that takes is irrelevant startup delay that only happens once when initiating the server.
So assuming the kernel modules and hardware support the more narrow bandwidth, it should work… I think. There are laptops that have options for an external FireWire GPU too, so I don’t think the PCIe bus is too baked in.
I would prefer to have GPUs for under $600 if possible
Unfortunately not possible for a new nvidia card (you want CUDA) with 16GB VRAM. You can get them for ~$750 if you’re patient. This deal was available for awhile earlier today:
https://us-store.msi.com/Graphics-Cards/NVIDIA-GPU/GeForce-RTX-50-Series/GeForce-RTX-5070-Ti-16G-SHADOW-3X-OC
Or you could try to find a 16GB 4070Ti Super like I got. It runs Deepseek 14B and stuff like Stable Diffusion no problem.I am OK with either Nvidia or AMD especially if Ollama supports it. With that said I have heard that AMD takes some manual effort whilst Nvidia is easier. Depends on how difficult ROCM is
I know you said consumer GPU, but I run a used Tesla P40. It has 24 GB of vram. The price has gone up since I got it a couple years ago, there might be better options in the same price category. Still, it’s going to be cheaper than a modern full fat consumer gpu, with a reasonable performance hit.
My use case is text generation, chat kind of things. In most cases, the inference is more than fast enough, but it can get slow when swapping out large context lengths.
Mostly I run quantized 8-20B models with the sweet spot being around 12. For specialized use cases outside of general language, you can run more compact models. The general output is quite good, and I would have never had thought it was possible 10 years ago.
ETA: I paid about $200 USD for the P40 a couple years ago, plus the price for a fan and 3d printed shroud.
3090 has 24gb and rolls the 38b models beautifully.
I tried to run Gemma 3 27B Q4K and was surprised how quickly the VRAM requirements blew up proportional to context window, especially compared to other models (all quantized) at similar size like Qwq 32B.
Wait how does that work? How is 24GB enough for a 38B model?
It all depends on the size of the model you are running, if it cannot fit in GPU memory, then it has to go back and forth with the host (cpu memory or even disk) and the GPU. This is extremely slow. This is why some people are running LLMs on macs, as they can have a large amount of memory shared between the GPU and CPU, making it viable to fit some larger models in memory.
