Quoting Chris Wilson (2019-11-15 16:05:45)
> No good reason why we must always use a static ringsize, so let
> userspace select one during construction.
Do we have any news on whether userspace has materialised for this yet?
It's literally just
--- a/runtime/os_interface/linux/drm_neo.cpp
+++ b/runtime/os_interface/linux/drm_neo.cpp
@@ -182,6 +182,15 @@ void setNonPersistent(uint32_t drmContextId) {
ioctl(DRM_IOCTL_I915_GEM_CONTEXT_SETPARAM, &gcp);
}
+void setMaxRingSize(uint32_t drmContextId) {
+ drm_i915_gem_context_param gcp = {};
+ gcp.ctx_id = drmContextId;
+ gcp.param = 0xc; /* I915_CONTEXT_PARAM_RINGSIZE; */
+ gcp.value = 128 << 12; /* maximum ring size is 512KiB, or 128 pages */
+
+ ioctl(DRM_IOCTL_I915_GEM_CONTEXT_SETPARAM, &gcp);
+}
+
uint32_t Drm::createDrmContext() {
drm_i915_gem_context_create gcc = {};
auto retVal = ioctl(DRM_IOCTL_I915_GEM_CONTEXT_CREATE, &gcc);
@@ -190,6 +199,9 @@ uint32_t Drm::createDrmContext() {
/* enable cleanup of resources on process termination */
setNonPersistent(gcc.ctx_id);
+ /* Big rings for silly amounts of non-blocking work! */
+ setMaxRingSize(gcc.ctx_id);
with some justification.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
