4 ms·
4) Release a consumer GPU with 32-80 GB of VRAM.
by zirgs 3y ago
4) Release a consumer GPU with 32-80 GB of VRAM.
- ciupicri 3y agoWhat are you planning to play on that thing?
- freedomben 3y agoI think Hogwart's Legacy could benefit from it. It crashes on my 16GB system due to being out of video memory regularly. It might be lying about the cause but that's what it claims.
- Filligree 3y agoLlama 2.
- Tuna-Fish 3y agoIt sounds silly, but people would endure a lot of pain to fit significantly larger models into ram.
- fbdab103 3y agoIt could even be 2-3 generations behind and that thing would still sell.
- latchkey 3y agoOr at least make it so that you could rent the larger gpus by the hour. As far as I know, there isn't a single service that offers bare metal access to MI210/MI250's.
- codedokode 3y agoAMD has CPUs with built-in GPU with hundreds of ALUs and unified memory (GPU uses system memory). Cannot such CPU be used for ML tasks without purchasing expensive graphic cards? 64 Gb of ordinary RAM will be cheaper than 64 Gb of VRAM. Currently such built-in GPUs seem to be optimized for games only.
- dur-randir 3y agoVRAM speed is a major factor. Even ddr5 with 4 channels would be slow compared to current GDDRs.
- codedokode 3y agoThat's correct, but if you are, for example, multiplying two 1024x1024 matrices (typical for ML loads) then you need to perform roughly 1B operations on 1M numbers. There is no need for loading lot of data. So we can do without fast memory in this case, can't we? Hopefully the cache and registers of GPU can handle this.