|
本帖最后由 q3again0605 于 2023-12-20 11:59 编辑
有兴趣的朋友们可以看看这篇。https://www.tomshardware.com/new ... lculated%20by%20the,%3A%20330.4%20*%2016%20%3D%205%2C286.
This is calculated by the maximum compute for a given bit-depth, using TFLOPS (or TOPS for integer work) multiplied by the number of bits. For the RTX 4090, TPP is 660.8 * 8 = 5,286 for FP8 work running on the Tensor cores (sparsity doesn't count). Also note that the value is the same for FP16: 330.4 * 16 = 5,286.
目标是要在4800以下,所以可能就是简单屏蔽cuda数量到原来的90%左右就够了。我还是挺期待的,如果板厂能多上点2槽卡我就更开心了。4090对我们小公司来说还是很好的生产力工具,1块4090对于LLM还是限制不小,2块就好很多,但是风冷的4090除了风魔v2都太厚了,直接没法2块插消费级主板上。
游戏我主要就星际2,CS2,4k3060ti够了哈哈,挺想玩博得3考虑升级一下等等4070ti再降价或者4070Ti Super。 |
|