PerfTCP & PerfUDP..

[WindBlade]

Do you feel that these 2 criterias should be given more weightage on Matousec's tests.. hmm.. For me, I do feel that these 2 are the msot impt criterias..

Leak tests are really not THAT practical as long as they perform resonably well..

Having said that, Mike mentioned that his PerfUDP performance has been fixed in the latest version of OA.. Can someone verify that? If so, I was jsut wondering why wasn't THAT version submitted when OA Pro was submitted for testing? If I recalled correctly, OA Pro was submitted for testing after the PerfUDP test was performed on OA Free?

[Fuzzfas]

-{ Quote: "Do you feel that these 2 criterias should be given more weightage on Matousec's tests.. hmm.. For me, I do feel that these 2 are the msot impt criterias.. " }-

I agree, but go tell that to the "leak army" that infests all firewalls with HIPS. In Comodo forum, there has been already an "alarm" thread, as soon as Comodo lost the "crown" at Matousec.

Performance? Who cares as long as you win the leak tests! ::)

-{ Quote: "
Having said that, Mike mentioned that his PerfUDP performance has been fixed in the latest version of OA.. Can someone verify that? If so, I was jsut wondering why wasn't THAT version submitted when OA Pro was submitted for testing? If I recalled correctly, OA Pro was submitted for testing after the PerfUDP test was performed on OA Free?" }-

From what i remember, after Matousec first released the first perfUDP test, there were 2 patches released for OA paid (none for the free), one of which were supposed to fix the low UDP performance. As i understand it, either Matousec didn't apply the patches or the patches failed to fix the UDP performance.

[alex_s]

-{ Quote: "As i understand it, either Matousec didn't apply the patches or the patches failed to fix the UDP performance." }-

I don't think 131 has any patches that affect performance. I tested current public beta (151) and got the following results:

perftcp, three tests

w/o OA with OA
18.125 18.312
18.015 18.652
18.015 18.468

perfudp, three tests
21.734 24.484
21.671 24.843
21.859 24.765

so, net total is :
tcp (18,125+18,015+18,015)/(18,312+18,625+18,468) = 0.9774 = 97.74%
udp (21,734+21,671+21,859)/(24,484+24,843+24,765) = 0.8809 = 88.09%

[Fuzzfas]

-{ Quote: "I don't think 131 has any patches that affect performance. " }-

Yes it has.

-{ Quote: "Release 2.1.0.131 (Patch Only) - 11th June 2008

This is a patch update only; The version number does not change.

Fixed: A bug in UDP handling which can cause a loss of performance
Fixed: User reported BSOD/Conflict in driver
Fixed: Keylogger detection issue

This patch is only available on the full versions of Online Armor , and can be obtained by pressing "Check for updates"

http://support.tallemu.com/vbforum/showthread.php?t=814
" }-

The free version wasn't patched though. So, if Matousec used the paid version, after June 11, he should have updated to get the patches. Or is OA automatically updated in the paid version? I have never used it.

If Matousec did update, well, seems that the UDP patch needs to be patched. ;D

[alex_s]

-{ Quote: "Yes it has.



The free version wasn't patched though. So, if Matousec used the paid version, after June 11, he should have updated to get the patches. Or is OA automatically updated in the paid version? I have never used it.

If Matousec did update, well, seems that the UDP patch needs to be patched. ;D" }-

It seems you you are right. Something has slipped there :)

According to the tests OA 131 patched (paid) performed tcp 96, udp 65, while free 131 version did tcp 91, udp 66. Hm .. Very strange result in case performance patch was applied :)

[Fuzzfas]

-{ Quote: "
According to the tests OA 131 patched (paid) performed tcp 96, udp 65, while free 131 version did tcp 91, udp 66. Hm .. Very strange result in case performance patch was applied :)" }-

Well, between 65 and 66, the difference may be within statistical error. However, if Matousec did use the patched version, it would appear that the patch didn't do much good...

Well, i decided to install Matousec's performance "king" in the meantime, Ashampoo. ;D

[alex_s]

-{ Quote: "Well, between 65 and 66, the difference may be within statistical error. However, if Matousec did use the patched version, it would appear that the patch didn't do much good...

Well, i decided to install Matousec's performance "king" in the meantime, Ashampoo. ;D" }-

Good luck. Will you share your experience ? :)

[WindBlade]

-{ Quote: "Well, between 65 and 66, the difference may be within statistical error. However, if Matousec did use the patched version, it would appear that the patch didn't do much good...

Well, i decided to install Matousec's performance "king" in the meantime, Ashampoo. ;D" }-

Maybe Mike_Nash would want to clarify on this point. :)

[MikeNash]

Checking now... it is possible I made a mistake here :(

[Fuzzfas]

-{ Quote: "Good luck. Will you share your experience ? :)" }-

If you use the forum search, you will find several posts of mine about it. It is very light firewall, nice GUI, but has many incompatibility issues with other security applications. The last time i used it, the tray icon in SP3 would disappear. I suspect it is an incompatibility with WinPatrol, because now that i restored an image and didn't use WinPatrol, it runs fine.

[WindBlade]

-{ Quote: "Checking now... it is possible I made a mistake here :(" }-

Any updates? :)

[MikeNash]

-{ Quote: "Any updates? :)" }-


It's more than possible :(

We fixed an (unrelated) bug in the driver;

It does technically increase UDP performance but only as a side effect - and not by much. The performance update is actually in the service component - which will be included in the next release.

Sorry for the inconvenience

Mike

[WindBlade]

hmm.,. So does that mean that technically,

1. The result on Matousec is based on the updated OA build 131, which had already fixed the bug "A bug in UDP handling which can cause a loss of performance"?

2. The next version of OA still suffers from UDP performance issues as shown on Matousec, since you mentioned "not by much"?

[MikeNash]

-{ Quote: "hmm.,. So does that mean that technically,

1. The result on Matousec is based on the updated OA build 131, which had already fixed the bug "A bug in UDP handling which can cause a loss of performance"?

2. The next version of OA still suffers from UDP performance issues as shown on Matousec, since you mentioned "not by much"?" }-

It means 131 did not actually solve the problem (by much) - but the beta builds solves it properly - so next official release (v3) will perform a lot better.