I’m using a GeForce GTX260 with 1GB vram, on PCI-Express x16 slot. System is Win7x64 with drivers 197.45.

In my application I’m creating a lot of GL textures (I can’t do otherwise), so that it eventually reach the limit of VRAM and generate a GL_OUT_OF_MEMORY error. But I was pretty sure that when VRAM was full, it goes to RAM ? (and then when RAM is full, it goes to hard-drive).

If I’m right, that’s what TurboCache (introduced with GeForce 6200)/HyperMemory (introduced with Radeon x300) were meant for.

So what’s wrong ?

GPU’s still don’t feature strict virtualized memory, a technology introduced 25 years ago on the PC (just like task pre-emption for example).

So we have to deal with such shortcomings somehow. And yeah, this situation is frustrating a lot of people.