A hands-on benchmark of GLM-4. 7-Flash on 2x RTX 3090 against qwen3. 5:27b: decode speed, prefill, VRAM, and one card versus two.
Source: [HackerNoon](https://hackernoon.com/glm-47-flash-on-2x-rtx-3090-my-hands-on-experience?source=rss)
1 points, 0 comments on Hacker News
1 points, 0 comments on Hacker News
1 points, 0 comments on Hacker News
1 points, 1 comments on Hacker News
Made a memory safe enough version of FFMpeg with <2% runtime overhead and negligible memory overhead so that no form of RCE is possible.
1 points, 0 comments on Hacker News