speedtest.net with KIS2019 [Closed]

[RockNLol]

hi, I recently noticed that youtube videos sometimes switched down to 720p, which made me do a speedtest on speedtest.net. This revealed, that of my 250/50 Mbit Internet I only got about 80/50, which still should be plenty for a 1080p youtube video. But the needle of speedtests "tachometer" also stuttered a lot and even froze my mouse for half a second during testing. This was caused by the CPU reaching 100% load. Speedtest should not do that, since my CPU is a Intel Core i5 2500K clocked to 4GHz (I know it is old, but not that slow). However disabling KIS completely during testing solves this issue. The CPU keeps at ~10% and I get my full 250/50Mbits. What is wrong with my KIS?

[Berny]

Welcome. Please contact the Technical Support https://center.kaspersky.com 

[kmscom]

Internet connection slowed down after installation of a Kaspersky Lab product

[giz02]

My speed drops from

911Mbit down/890 Mbit up

to

380Mbit down/640Mbit up.

 

 

I came here from [another antivirus] because of similar performance issues.

Not using hardware virtualization - it’s disabled. I’m investigating how to enable it.  I thought I had it enabled in bios - perhaps not.  Will check that, ensure it’s enable and showing as active in kaspersky before retesting.

 

Why is impact so high?
I believe my system to be equipped fairly well.

Ryzen 7 2700x

32GB Ram

1TB Evo Pro 970 nvme

 

[giz02]

I enabled hw virtualization but this did not improve my performance.

 

Disableing kaspersky resutled in an improvement, but  kaspersky seems to be still processing packets, even when shields are disabled.

 

Peak I could get with Kaspersky running is 500Mbit down, 800 up.

perfmon indicated that kaspersky was ‘processing’ during the tests, even when shield was enabled.

 

I suspect all av software gets in the way of performance, even when disabled.  A.v.st did the same thing.

[Berny]

@giz02 Welcome. Please see above suggestions.

[giz02]

HI @Berny - Sorry I’m new to the forums - I don’t see any suggestions above - am I missing something?  I see some links though - contact support and another thread - is that what you meant?

 

[kmscom]

@giz02  

Internet connection slowed down after installation of a Kaspersky Lab product

 

[giz02]

Ok - so essentially what this article implies is that my speedtest results  are not an accurate gauge of actual internet speed?

I’m not sure what to take from this?

[giz02]

Ok - assuming, as implied, that speedtest results aren’t real.  
Steam application downloads are:

I’ve downloaded ARMA3 4 times.

Procedure - configure av software, launch steam, download application, record results, delete steam downloading and arma3 folder, repeast

Attached photo shows steam downloads under the following conditions:

1. NO AV software installed
2. Kaspersky Secure Cloud Free - protection enabled
3. Kaspersky Secure Cloud Free -protection disabled
4. Other AV software protection enabled 

Results:

1. No av software installed, 99.5MB/s current 100.5MB/s peak (800Mbits/second)
2. Kaspersky SC protection enabled, 44.1 MB/s cur 46.3MB/s peak (370 Mbit/second)
3. Kaspersky SC protection disabled 59.3 MB/s cur, 59.9 MB/s peak (479.2 Mbit/second)
4. Other AV Protection Enabled 57.0 MB/s current, 59.1 MB/s peak( 472.8 Mbit/second)

My questions:
Why does having Kaspersky SC installed result in a 40-60% loss in throughput, EVEN when Protection is disabled.

I think it’s reasonable to assume that AV software has overhead, HOWEVER it’s also reasonable to assume that disabling protection would offset the performance tax.  Why is there still a high degree of overhead when disabled?

Does Kaspersky publish maximum throughput numbers?

Why Not?

Being transparent about maximum throughput as tested in Kaspersky’s labs would alleviate Kaspersky’s support requests.

I read the link tl;dr 
Download over 100MB
Open support case with screen caps.

 

[richbuff]

Speed test results for Kaspersky are Not real, due to technical factors. This has been the case for many years. I will leave this topic open for additional technical posting that may possibly happen, pending future closing. :)