TI 2021 and drive performance
First I only Upgrade to the latest version of TI when new hardware or OS stops me from doing backups or restores with the rescue media I have. Reason being I have no use for most of what is installed by TI. All I want is rescue media creation, The ability to mount and validate images and the ability to make backups manually. I would be OK with that if TI did not kill system performance when installed. This has been a problem near forever, you can just feel it cripple your system. Slow booting and overall system performance is noticeably down . I have asked for a stripped down version for years. Well I needed another license for a new laptop for my wife so I decided to get a 3 PC perpetual license for TI 2021. I installed it on my main PC to be able to make the various types of rescue media so the story begins.
I had no problem installing build 34340 on my windows 10 64bit 20H2 PC . I also had no problem making manual backups or 3 different types of rescue media. I could also mount images without problems.Did TI slow my PC down? Yes! Note I did not enable the advanced security features that are really only a demo in the perpetual licensed version so I have no idea how bad performance could get. I run 3 Samsung 860 EVO SSD's on my PC so I have Samsung Magician installed to monitor my drives etc. Disk IO is non existent with TI 2021 installed. Same with 2020 unless active protection is turned off. Even with active protection off drive performance takes a small hit. It looks like a restore back to before TI was installed is in order as I said I really only need working rescue media. Sure would be nice If TI could be installed without crippling PC's though. Take a look at this screen shot.
As far as I can see TI was scanning nothing when I was doing my testing as there was no disk or CPU activity from any TI processes in task manager. I would say nothing can be done until Acronis wakes up and starts caring about system performance which from history will never happen. Same goes for limited option installs I would guess.
BTW care to share your Samsung Magician drive bench results with Ti 2021 installed compared to disabled?
WonderWrench,
Not sure how the Samsung Magician Performance Test is conducted as far as settings or what test engine is used. I am familiar with Crystal Disk Mark however and in my testing with that tool on a comparable system as yours I find some comparisons and some differences.
Crystal Disk Mark uses the MS System Internals Disk Speed test engine which in turn uses I/O Meter underneath to perform what in the testing world considered the gold standard. The system specs on which the test screenshots attached below are as follows:
- i5 4690K @3.50GHz
- Asus Z97 Deluxe USB 3.1
- 4x8GB G.Skil DDR3 2400
- Samsung 860 EVO 500GB
- 750 PSU
- Lian Li case
- Windows 10 Pro for Workstations 20H2
The tests conducted here used a 2GB file size. Test 208959 is the default test performed by Crystal Disk except I increased default file size from 1GB to 2GB.
Test 208961 is Crystal Disk default Peak Performance test again using the 2GB file. As you can see this test includes I/O for 4K file size.
This PC has TI 2020 25700 installed with Active Protection enabled.
Enchantech,
I was using Samsung Magician manly because it was handy. IMO a must for Samsung drives as it can flash firmware and report drive health more completely than other programs. I have used Crystal Disk Mark many times in the past. I downloaded the latest version and changed the file size to 2GB. There must be some setting different because I cant seem to get the IO tests to run like you show. Without reporting the difference with AP on and off it does not prove much anyway. Besides the 2020 version does not totally kill drive performance like 2021 does.
I do plan on installing 2021 again to do more testing before I image back to a clean image without TI ever being installed.
Click the Profile tab and you can access differing tests. One of those is Peak Performance which gives the I/O numbers.
Enchantech,
I ran some more tests with Crystal Disk Mark set as you said on TI2020 Build 25700 see results below. Looks about the same as Samsung Magician to me.
I also tested 2021 build 25700. Note I tested the perpetual version so the advanced protection is not enabled unless you want to demo that function (as in the first screen shot, upper left). Once advanced protection was enabled and the database files were updated I ran a full scan and followed by a reboot. FYI for people that do not know about Crystal Disk Mark higher values mean higher performance except for RND4K us which is time based so lower values are better. IMO the performance hit is not acceptable any time TI2021 is installed. TI 2020 with the AP disabled still takes a hit but I could live with it if I had to. For me I will just use the bootable media to do backups and restores. For others they need to decide how much you need the added functions and if you can live with the performance hit.
Note all my testing was done on the system in my sig. Before each test run the trim command was given to the drive under test and the system was allowed to idle for 15 minutes. Then the system was rebooted then again allowed to to idle another 15 minutes before the test. This was done to try and remove most variables. Also note this PC has a bios update and windows 10 build that corrects the Spectre and Meltdown vulnerabilities in my older Intel processor. These fixes can and do hurt system performance to some degree. Mainly drive performance, not a lot but measurable. Newer systems should already have these fixes in place or have processors engineered with the fixes built in. Download a free tester here to see if you are affected. https://www.grc.com/inspectre.htm
WonderWrench,
Yes the benchmarks are very similar. Where they differ is with the RND4K (µs) line which refers to latency and which Magician does not show. Latency can be caused by a number of things including hardware. I can confirm that latency does increase with AP on more than AP off with both being higher than if TI is not installed. If you have other processes running while ATI is running this can impact latency as well.
Note that latency in my screenshot is much greater than what your shows. In addition your read/write performance for RND4K is greater than mine and your IOPS are greater as well. Given those stats I would happy to trade with you. :-)
Enchantech,
I also tested my 2TB EVO while you tested your 500GB. Bigger SSD's are usually faster. My 500GB is 70% full or I would have tested that. Other factors, how hard do you hammer the drive you tested? Does it get trimmed often? Does it get a decent amount of idle time? BTW did you get the bios update for your motherboard that fixes the security issues with you processor?
So, the drive tested is not "hammered" as you say at all. If you look at my screenshot 208961 you can see that it contains only 8% data, the rest is free space. The drive is used as a repository for certain application installs. Do I trim it regularly? No as the data here rarely changes. Did I trim before the test, no I did not. As far as bios version, it has the latest yes and so are all drivers for my storage controllers.
None of that really matters though. I know why the performance is off of what you posted, the biggest factor is drive capacity. Your larger drive is and did produce better numbers. The other factor is that I have a full complement of drives in this machine. There are a total of 8 internal drives. Then there is a 5 drive 5 bay USB enclosure attached as well. The total number of drives installed is the biggest factor effecting latency.
Having said all that and just for kicks I ran a trim on this 500GB drive, let it idle, and ran the bench again. Screenshot attached. As you can see very little change. Since this is not my primary boot drive, I'm fine with the performance actually.
I have multiple PCs, two of which have multiple NVMe M.2 drives, as well as a few SATA SSD and HDD. I have not noticed any performance degradation. On two M.2 drives performance is well below max, but that is due to the connection being PCIe 2.0 rather than 3.0. The SATA SSDs all seem to be working at about rated speed.
Ian