summaryrefslogtreecommitdiffstats
path: root/tools/cuda-ab/run/memset_probe
diff options
context:
space:
mode:
authorArthur Hanson Rasmusson <arthur@vgpu.io>2026-10-21 10:58:58 -0400
committerArthur Hanson Rasmusson <arthur@vgpu.io>2026-10-21 10:58:58 -0400
commit9f68e5e304416a7d2148903eada9caffd7b0779e (patch)
treee4f0ce258497a38ba9235e8e5680a222373e8799 /tools/cuda-ab/run/memset_probe
parent7685d1b315d0a5435374c8c6b9ca1a3a4eab39fa (diff)
downloadopen-iov-9f68e5e304416a7d2148903eada9caffd7b0779e.tar.gz
open-iov-9f68e5e304416a7d2148903eada9caffd7b0779e.zip
tools/cuda-ab: memset_probe + wedge_recover forensics clients (session 10)
memset_probe: linear 8 MB cudaMemset + D2H with first-bad offset and a 4 KB page map (correct / other / malloc-prefill) - separates 'fill under-delivered' from 'readback delivered wrong pages'; standalone green, bisects the memcpy_roundtrip memset phase. wedge_recover: launch+verify loop that on failure sleeps N seconds and retries the same iteration (same context), reporting per-iteration attempts and launch times - tests whether an idle gap heals the per-boot no-execute wedge (r409 log shows a ~37 s client-side gap healed a wedged run). Also ships the missing inplace_scale run binary (its source landed in 7685d1b3).
Diffstat (limited to 'tools/cuda-ab/run/memset_probe')
-rwxr-xr-xtools/cuda-ab/run/memset_probebin0 -> 820840 bytes
1 files changed, 0 insertions, 0 deletions
diff --git a/tools/cuda-ab/run/memset_probe b/tools/cuda-ab/run/memset_probe
new file mode 100755
index 0000000..bd9df5e
--- /dev/null
+++ b/tools/cuda-ab/run/memset_probe
Binary files differ