summary refs log tree commit diff stats
path: root/results/classifier/108/performance/589231
diff options
context:
space:
mode:
Diffstat (limited to 'results/classifier/108/performance/589231')
-rw-r--r--results/classifier/108/performance/58923129
1 files changed, 29 insertions, 0 deletions
diff --git a/results/classifier/108/performance/589231 b/results/classifier/108/performance/589231
new file mode 100644
index 00000000..8cade384
--- /dev/null
+++ b/results/classifier/108/performance/589231
@@ -0,0 +1,29 @@
+performance: 0.963
+KVM: 0.859
+device: 0.837
+graphic: 0.830
+other: 0.735
+network: 0.606
+semantic: 0.550
+PID: 0.494
+socket: 0.422
+debug: 0.416
+boot: 0.392
+files: 0.296
+vnc: 0.273
+permissions: 0.261
+
+cirrus vga is very slow in qemu-kvm-0.12
+
+As has been reported multiple times (*), there were a regression in qemu-kvm from 0.11 to 0.12, which causes significant slowdown in cirrus vga emulation.  For windows guests, where "standard VGA" driver works reasonable well, -vga std is a good workaround. But for e.g. linux guests, where vesa driver is painfully slow by its own, that's not a solution.
+
+(*)
+ debian qemu-kvm bug report #574988: http://bugs.debian.org/574988#17
+ debian qemu bugreport (might be related): http://bugs.debian.org/575720
+ kvm mailinglist thread: http://<email address hidden>/msg33459.html
+ another kvm ml thread: http://<email address hidden>/msg32744.html
+
+QEMU 0.12 is pretty much outdated - has this been fixed in a newer version of QEMU? I.e. do you think we can close this bug nowadays?
+
+[Expired for QEMU because there has been no activity for 60 days.]
+