Ah, good to know. I only had to worry about cache ops on my REDACTED when I was REDACTED it because the GPU used DMA, so if you swizzled a texture for faster GPU speeds (or otherwise modified it with the CPU) you had to call a writeback function on the specific memory range that you modified. Alternatively it had an uncached mirror of main memory starting at 0x80000000 IIRC.

Ah, good to know. I only had to worry about cache ops on my [REDACTED] when I was [REDACTED] it because the GPU used DMA, so if you swizzled a texture for faster GPU speeds (or otherwise modified it with the CPU) you had to call a writeback function on the specific memory range that you modified. Alternatively it had an uncached mirror of main memory starting at 0x80000000 IIRC.

Ah, good to know. I only had to worry about cache ops on my [REDACTED] when I was [REDACTED] it because the GPU used DMA, so if you swizzled a texture for faster GPU speeds (or otherwise modified it with the CPU) you had to call a writeback function on the specific memory range that you modified. Alternatively it had an uncached mirror of main memory starting at 0x80000000 IIRC.

Ah, good to know. I only had to worry about cache ops on my [REDACTED] when I was [REDACTED] it because the GPU used DMA, so if you swizzled a texture for faster GPU speeds (or otherwise modified it with the CPU) you had to call a writeback function on the specific memory range that you modified. Alternatively it had an uncached mirror of main memory starting at 0x80000000 IIRC.

Ah, good to know. I only had to worry about cache ops on my [REDACTED] when I was [REDACTED] it because the GPU used DMA, so if you swizzled a texture for faster GPU speeds (or otherwise modified it with the CPU) you had to call a writeback function on the specific memory range that you modified.