17 articles with this tag
Decode AWS, Google Cloud, Azure and Oracle GPU names, check GPU counts and memory, then compare live AWS rates and dated Oracle rates with other providers.
guides / analysisCompare LoRA, QLoRA and full fine-tuning with worked 7B and 70B memory budgets, published quality results, GPU choices and live prices to estimate run costs.
guides / hardwareChoose how to split a model across rented GPUs, calculate training memory, and match tensor, pipeline and data parallelism to NVLink and cluster networks.
hardware / pricingNVIDIA's DGX Spark priced and dated against live cloud GPU rates: what its 128GB memory can run, its bandwidth limit, and when to rent instead.
guides / hardwareVerify your rented GPU's model, memory, power, PCIe and NVLink, then check errors and run a short server stress test before committing to a long training job.
hardware / guidesPick a precision for pre-training, fine-tuning or inference, and see which rentable GPUs accelerate FP8 and FP4 in hardware and which only emulate them.
pricing / guidesColab plans and compute units explained, a dated estimate of what a Colab T4 costs per hour, Kaggle's free tier, and live hourly prices to rent the same GPUs.
pricing / guidesHow on-demand, spot, reserved, capacity block and marketplace GPU pricing work, what each one commits you to, and a rule for picking between them.
pricing / guidesPick between a monthly dedicated GPU server, a GPU VPS and an hourly cloud GPU for an always-on job, with dated list prices and a break-even formula.
hardware / guidesVRAM, memory bandwidth and dense throughput decide an AI workload. How to find them on a datasheet, and why the headline TFLOPS figure is usually doubled.
providers / pricingLive in-stock A100, H100, H200 and B200 prices at GPU clouds comparable to Lambda, what Lambda does well, where it falls short, and a rule for choosing.
hardware / guidesA is Ampere, L is Ada, H is Hopper, B is Blackwell. A decoder table, a generation timeline and a four-step method for placing any NVIDIA GPU on a rental site.
pricing / hardwareVerified purchase listings for NVIDIA H200, B200, B300, GB200 NVL72 and A100 with seller and date, the misquoted figures corrected, and live rental prices.
hardware / guidesNVLink, PCIe and SXM explained for people renting multi-GPU servers: bandwidth per generation, what the H100 NVL is, and when the SXM premium pays off.
pricing / analysisPublished serverless GPU prices from Modal, RunPod, Replicate, Baseten and others, and the break-even utilisation formula for choosing serverless or dedicated.
hardware / guidesMIG splits one NVIDIA data centre GPU into isolated instances with their own memory. See which GPUs support it and when a slice beats a whole cheap GPU.
analysis / pricingA Tesla V100 costs $0.05/hr at one provider and $3.06/hr at another. Same chip, same VRAM, same performance. Here's what the data looks like across 27 providers.