NOD32 9.0.318 slows down Samsung Pro950 per Magician Benchmark

[BJB 1]

Greetings.  I am not benchmark-crazy but just to make sure my new Windows 10 build's settings were correct on my motherboard, bios, etc, I ran the Samsung-supplied "Magician" benchmark.  Like a few others that have noticed this, the resuts are much slower with ESET NOD32 enabled.  When disabled results are normal. I am running this on a Samsung PRO 950 SSD drive.  Of course if it is just the benchmark, I don't really care, however, if it is slowing down the benchmark I sense it is slowing down performance also....

 

Note it also slows down a more widely used benchmark cyrstaldiskmark.

 

Is ESET aware of this issue?

BJB
​

Edited January 3, 2016 by BJB

[Marcos 5,072]

First of all, could you please re-test it with the latest version of ESS 9.0.349 which is downloadable from ESET's website? Let us know if the problem persists. Does temporarily disabling real-time protection or another protection feature / setting make a difference? If so, which one?

Also please run the test several times in a row to make sure that results are always basically same. Then post screen shots of both good and bad benchmark results.

[BJB 1]

I was logged in, but then it said I could not post.  I will try again.....

 

Just right clicking the ESET notification icon and choosing "pause" solves the issue. I ran it several times and it was definitive.

 

I will download the new version and re-test. Is this a live version or beta?  I just checked for updates within the program and it says my installed  version 9.0.318.0  is up to date. I am running Windows 10 64 bit.

I have no desire to run beta software.

 

Please let me know,
Thanks,
BJB

[stackz 112]

Version 9.0.349 is the latest release. i.e. it is not beta

[BJB 1]

Version 9.0.349 is the latest release. i.e. it is not beta

Okay, thanks.  I guess I'll ask the obvious question.  Why doesn't my NOD32 product update notify me of that?

I even manually clicked on the "check for updates" (for the program not the definitions) and it did not prompt me.

 

Curious....But understood.  I can download it and see if it helps.

BJB

[Marcos 5,072]

It takes some time until a particular build becomes available via a program component update.

[BJB 1]

I installed the new version although not sure why it is a month later and I still do not get notification of a new version.  I rely on those notifications.....anyway, the new version had no impact.

The sequential Reads are FIFTY percent less in Samsung Magician or Crystalmark with ESET on.  Same as before.  Iwas looking for a place to insert a snip of the results but I don't see how I can do that.

 

Again, if it is just a synthetic benchmark thing I could care less. However if my benchmark reads are 50% low, my sense is it must be impacting performance somewhat in real life.

Just test a Samsung 950 PRO M.2 drive on a Skylake motherboard.  I would think you would get the same results. 

 

BJB

[Marcos 5,072]

I was looking for a place to insert a snip of the results but I don't see how I can do that.

To add an image to your post, click "More reply options" and then Browse. Click "Attach this file" to upload it to the server. Move the cursor to the place where you want to add the image and click "Add to post".

[Peter Randziak 1,130]

Hello,

 

our QA was able to confirm lower benchmark results of Samsung Magician benchmark on Windows 10 (strange is that it does not happen on Windows 7)

However we were not able to reproduce the lower results from the CrystalDiskMark. :-(

 

Can you please provide us with:

1. exact version of the CrystalDiskMark tool you used

2. exact settings of the CrystalDiskMark you use to perform the test

3. screenshots from the tools with history of the results

 

Thank you in advance.

 

P.R.

[BJB 1]

Sorry, thought I went back and posted this. Here are results with NOD32 on/off for Samsung and Crystalmark.

I ran with default settings and you can see the version in the screenshot.  There are headings above or below each snip stating if it was run with ESET on or off.

Sequential read on crystalmark is where you notice it most.

 

Seems like it would be easy to replicate.  Download, run with/without NOD32.  Easy to see and many have noted this.

 

Thanks,

BJB
 

Benchmarks with without ESET.pdf

[sjpt4x 0]

Hi,

 

Has this issue been rectified in an update yet? I have this exact same issue so I have uninstalled nod32 for the time being until this has been fixed.

 

Can we get an update regarding this? I would like my PC to be fully protected again.

 

Also running Samsung 950 Pro on Windows 10 Pro 10586

 

Thanks

[Marcos 5,072]

Hi,

Has this issue been rectified in an update yet? I have this exact same issue so I have uninstalled nod32 for the time being until this has been fixed.

Can we get an update regarding this? I would like my PC to be fully protected again.

Also running Samsung 950 Pro on Windows 10 Pro 10586

 

Have you been observing some performance issues while working with the computer? We have reported the issue to Microsoft and awaiting a response.

[sjpt4x 0]

I can run some real world transfer test over the next few days with and without ESET enabled.

 

Once I have the results I will post them here.

[Marcos 5,072]

There will always be some impact on performance with an antivirus software installed. The point is the impact shouldn't be very big.

[BJB 1]

 

Have you been observing some performance issues while working with the computer? We have reported the issue to Microsoft and awaiting a response.

Hope some progress is made. I am not sure what ESET expects from Microsoft. The ESET product is slowing down Windows 10 when it is activated with this configuration. Microsoft is going to say if you uninstall ESET, it will run fine. Seems like the ESET software engineers have some work to do.

Interested to see what the other real world tests show.

As to the comment if we are seeing a slowdown. That would be very hard to say unless you have the innate ability to sense sequential read speed while running Premiere Pro or playing a game. Bottom line, some slowdown as noted might be expected. But not sure if this amount makes sense. Certainly the Samsung test shows a significant speed hit. So perhaps another program that interacts with NOD32 in the same way might have the same hit on performance.

BJB

[Marcos 5,072]

After discussing the issue with Microsoft, we assume the problem occurs with applications that open files in non-cached mode. Normally it shouldn't affect the performance as applications should open files in cached mode to take advantage of caching, however, some specific applications, such as benchmark tools may open files in non-cached mode deliberately.

This issue will be addressed in future builds of v9. We will appreciate feedback once new v9 builds become available.

[sjpt4x 0]

Hi,

 

Thanks for the quick update. i still plan to do some tests very soon so I will still post the results here.

 

Do you have an estimation on when a new build might be available?

[george-nt 0]

Hi there!

 

After installing the latest version of ESET smart security 9 (v9.0.375.0) i have seen a huge difference to my ssd performance! To be more specific with eset enabled the samsung magician gave me 410Mb/s sequential read. If i pause the protection and redo the test i've got a 551Mb/s. This is 140Mb/s degradation when eset is enabled.

 

The same goes on with CrystalDIsk , 120-130Mb/sec degradation :/

 

I have seen that some eset moderators are saying that its a windows fault or samsung magician fault but are they pretty sure about it?

[Marcos 5,072]

As already explained, this issue occurs only when an application accesses files in no-cache mode. That said, you should not notice any impact on performance while using standard applications as basically only benchmark applications have a reason to access files in no-cache mode.

[george-nt 0]

Thank you Marcos for your reply. Since I am not an operating system architect or developer could you please give me something that I could read about no-cache mode? I googled it with no luck.

[Mark Leyland 0]

I'm glad I found this thread because after hours of research into why my benchmark results were so low I was beginning to think there was something wrong with my system setup or 950pro SSD. After disabling Eset Smart Security 9 I was able to see the correct results in the Samsung magician software. I also hope this isn't having an impact on overall performance and look forward to the update as Eset is the best out there and would like to continue to using the software.

[Vladadam 0]

I run some benchmark with Samsung Magician for my 950 pro an I get twice less Read speed is Eset protection is enabled.

It runs at 2450 without and 1400 with.

Did you manage to fix that?

Best,

[montoya 0]

Performance issue's still not fixed in NOD32 (v10.0.369.1)...

[Super_Spartan 56]

Performance issue's still not fixed in NOD32 (v10.0.369.1)...

can you elaborate please?

[Aaron Stevens 0]

Same here, it's still affecting the Samsung Magician performance benchmark.This is well over 30% loss of speed and makes me wonder if this is a system-wide impact because of NOD32.

NOD32 10.0.369.0 on Windows 10 with a Samsung 850 EVO.

Edited December 21, 2016 by Aaron Stevens