New Windows option in 7.9 Beta 1 to make Logos faster
Comments
-
My laptop automatically updates itself to Windows 10 Creative Update a couple days ago.
After machine and Logos restarted, Logos seemed to went to back to its crawling speed. I didn't know where / how to check whether the NGEN compile was wiped away. I simply reran the NGEN command line and let it compile. Logos is now back to the accelerated speed.
Can you confirm that the Win10 Creative Update would wipe out the NGEN compile, just out of curiosity?
Thanks,
Peter
0 -
PL said:
My laptop automatically updates itself to Windows 10 Creative Update a couple days ago.
After machine and Logos restarted, Logos seemed to went to back to its crawling speed. I didn't know where / how to check whether the NGEN compile was wiped away. I simply reran the NGEN command line and let it compile. Logos is now back to the accelerated speed.
Can you confirm that the Win10 Creative Update would wipe out the NGEN compile, just out of curiosity?
Thanks,
Peter
Yes. Updates to Windows, .NET, or Logos & dependencies can invalidate the compiled assemblies. Rerunning NGEN will update if/only if necessary. (fuslogvw should log failures that indicate an update is required)
0 -
Randy W. Sims said:
Rerunning NGEN will update if/only if necessary.
So is it ok to rerun it occasionally just to be sure no updates have knocked the speed back down?
Logos 10 | Dell Inspiron 7373 | Windows 11 Pro 64, i7, 16GB, SSD | iPhone 13 Pro Max
0 -
James Taylor said:
So is it ok to rerun it occasionally
Do it with each update of Logos until the issues raised in this thread have been resolved and this happens automatically.
Dave
===Windows 11 & Android 13
0 -
C:\Windows\Microsoft.NET\Framework64\v4.0.30319\ngen.exe uninstall Logos
It didn't work --> "the specified assembly is not installed". So I'll have to use the filename+full path, as I have two installations of Logos (haven't actually used ngen for the other installation).
EDIT: The uninstall with full path failed --> "the specified assembly is not installed". Possibly because "Note: Starting with the .NET Framework 4, the action
uninstall
* is no longer supported."?And installing the other installation (v7.8 SR-1) also failed ---> "Failed to load the runtime" Exception 0x80131700.
Dave
===Windows 11 & Android 13
0 -
What path? It should work if you point to ./Logos/System/Logos.exe
0 -
I excitedly tried this out, but to be honest, I barely notice any improvement. To try and be more scientific, I created a simple layout, and set it to load by default. I then ran Logos three times, and looked at the entry for "StartupUtility Finished startup". I then switched ngen off, and repeated the process.
- With ngen, startup times were 32.96, 34.24 and 32.98. Average is 33.39
- Without ngen, startup times were 35.34, 32.32, and 35.6. Average is 34.32.
So ngen improved performance by 3%, which is well within the margin of error for a test like this.
This is on an Intel i7-3770k with 32Gb RAM, and a Samsung 850 Evo SSD.
This is my personal Faithlife account. On 1 March 2022, I started working for Faithlife, and have a new 'official' user account. Posts on this account shouldn't be taken as official Faithlife views!
0 -
Hi Mark,
Did you find the app to be generally more responsive with ngen? To me, that's the biggest benefit.
Speeding up startup time is nice, but less crucial to me as I keep Logos open all the time, so it only needs to restart during system restart or upgrade installation. Otherwise, it's the click-to-click fast response that makes ngen most valuable to me personally.
Peter
0 -
PL said:
Did you find the app to be generally more responsive with ngen?
No, to be honest, although unlike many I've never found Logos particularly slow. Can you give a quick example of something that's much quicker now?
This is my personal Faithlife account. On 1 March 2022, I started working for Faithlife, and have a new 'official' user account. Posts on this account shouldn't be taken as official Faithlife views!
0 -
- Searches run faster
- Bible panels (usually with two translations paired up in multiview) initiate faster, and go to new BCV locations faster
- Bible Browser left-side list of filter facets now populates in < 5 seconds (as opposed to spinning for a minute)
- Library window more responsive
0 -
Mark Barnes said:
So ngen improved performance by 3%, which is well within the margin of error for a test like this.
This is on an Intel i7-3770k with 32Gb RAM, and a Samsung 850 Evo SSD.
From my post on page 1 of this thread, the improvement is c. 25% on a laptop with a 1 GB hard drive, 8 GB RAM & Intel i7-5500U (about 1 GHz slower than Mark's).
Dave
===Windows 11 & Android 13
0 -
Randy W. Sims said:
What path? It should work if you point to ./Logos/System/Logos.exe
Thanks Randy, I should have taken notice of your previous post.
Dave
===Windows 11 & Android 13
0 -
Dave Hooton said:
EDIT: The uninstall with full path failed --> "the specified assembly is not installed". Possibly because "Note: Starting with the .NET Framework 4, the action
uninstall
* is no longer supported."?And installing the other installation (v7.8 SR-1) also failed ---> "Failed to load the runtime" Exception 0x80131700.
Both these worked OK when I used the correct path! The Beta assembly folders were combined with those of the SR-1 assembly e.g. one Logos folder with two sub-folders, and uninstalling SR-1 left the Beta assembly (thankfully!).
Dave
===Windows 11 & Android 13
0 -
PL said:
- Searches run faster
- Bible panels (usually with two translations paired up in multiview) initiate faster, and go to new BCV locations faster
- Bible Browser left-side list of filter facets now populates in < 5 seconds (as opposed to spinning for a minute)
- Library window more responsive
I tried all this, and still see no real difference. A four-panel multiview takes around 5s, and Bible Browser about 3s, with or without ngen.
This is my personal Faithlife account. On 1 March 2022, I started working for Faithlife, and have a new 'official' user account. Posts on this account shouldn't be taken as official Faithlife views!
0 -
Dave Hooton said:Mark Barnes said:
So ngen improved performance by 3%, which is well within the margin of error for a test like this.
This is on an Intel i7-3770k with 32Gb RAM, and a Samsung 850 Evo SSD.
From my post on page 1 of this thread, the improvement is c. 25% on a laptop with a 1 GB hard drive, 8 GB RAM & Intel i7-5500U (about 1 GHz slower than Mark's).
Certainly you mean TB not GB right? (emphasis in quote is mine)
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
Dave
===Windows 11 & Android 13
0 -
Philana R. Crouch said:
In our efforts to improve the performance speed of the application we have added a new option that needs to be manually applied. This process will need to be repeated each time a new version of the application is installed.
To run the new performance command line:
1. Run a Command Prompt for `[logosinstallpath]\Logos.exe /ngen`
2. Click on Yes for changes to be made on the pop-up
Another window will pop-up and run some processes. It will look something like this:
To undo this process:
1. Run a Command Prompt for `[logosinstallpath]\Logos.exe /unngen`
2. Click on Yes for changes to be made on the pop-up
I'm seeing nothing like the results shown above. First, I don't get a pop up of any kind and I see nothing in Command Prompt. After hitting enter it just shows C:\Users\kevin>
Am I doing it wrong?
Dr. Kevin Purcell, Director of Missions
Brushy Mountain Baptist Association0 -
Here is the way I run ngen:
1) When I run CMD I right click on the shortcut and select "Run as administrator"
2) My User Account Control settings are completely off which I believe eliminates the "Do you want to allow this app to make changes to your device?" prompt. Depending on how your UAC settings are configured, you may or may not get the prompt.
3)The following command is pasted into the command prompt and then executed:
C:\Windows\Microsoft.NET\Framework64\v4.0.30319\ngen.exe install C:\Users\username\AppData\Local\Logos\System\Logos.exe
- Assumes you are using a 64 bit version of Windows
- Assumes you will insert the proper user name from your computer
- Can be run from any folder location in the Command prompt
As I mentioned on the first page of this thread, I registered a 42% improvement in startup speed (i7 5960X oc'd to 4.2 GHz, 32 GB DDR4 ram, Samsung 960 Pro ssd (1TB), EVGA Nvidia 970). Everything else certainly feels snappier but I cannot quantify it.
Hope this helps.
Instead of Artificial Intelligence, I prefer to continue to rely on Divine Intelligence instructing my Natural Dullness (Ps 32:8, John 16:13a)
0 -
I've noticed a speedier response time in some areas. One area that still astounds me for its sluggishness is opening a Bible. I've got a shortcut to my preferred Bible on the top of the screen. When I click it, it takes a good 3 seconds for the Bible to open. Why is it not instantaneous?
0 -
Kevin A. Purcell said:
After hitting enter it just shows C:\Users\kevin>
After the > type C:\Users\kevin\AppData\Local\Logos\Logos.exe /ngen
1. this is the default path. It could be totally different for a custom installation e.g. E:\Logos\Logos.exe /ngen
2. Your Logos folder name could be Logos4 or Logos5 (check C:\Users\kevin\AppData\Local for the name)
Dave
===Windows 11 & Android 13
0 -
Paul Strickert said:
I've noticed a speedier response time in some areas. One area that still astounds me for its sluggishness is opening a Bible. I've got a shortcut to my preferred Bible on the top of the screen. When I click it, it takes a good 3 seconds for the Bible to open. Why is it not instantaneous?
Probably because of visual filters.
This is my personal Faithlife account. On 1 March 2022, I started working for Faithlife, and have a new 'official' user account. Posts on this account shouldn't be taken as official Faithlife views!
0 -
Even if they're all unchecked, Mark?
0 -
Paul Strickert said:
Even if they're all unchecked, Mark?
Yes. https://community.logos.com/forums/p/133323/866528.aspx#866528
0 -
Dave Hooton said:Kevin A. Purcell said:
After hitting enter it just shows C:\Users\kevin>
After the > type C:\Users\kevin\AppData\Local\Logos\Logos.exe /ngen
1. this is the default path. It could be totally different for a custom installation e.g. E:\Logos\Logos.exe /ngen
2. Your Logos folder name could be Logos4 or Logos5 (check C:\Users\kevin\AppData\Local for the name)
Thanks Dave. I did check to ensure that Logos.exe is in the default location. I know I didn't install it in a special place upon installation, but double checked and it was there but also in the subfolder labelled System. So I ran it on both locations.
Dr. Kevin Purcell, Director of Missions
Brushy Mountain Baptist Association0 -
JRS said:
2) My User Account Control settings are completely off which I believe eliminates the "Do you want to allow this app to make changes to your device?" prompt.
FYI: You may want to read point #13 on this page: https://decentsecurity.com/#/holiday-tasks/
Donnie
0 -
Thank you, Bradley and Mark.
0 -
Donnie Hale said:
FYI: You may want to read point #13 on this page: https://decentsecurity.com/#/holiday-tasks/
Donnie
Thanks, Donnie, but I prefer to live 'dangerously' rather than have to listen to 6378.UAC Repeat.mp3 all day long. [;)]
Instead of Artificial Intelligence, I prefer to continue to rely on Divine Intelligence instructing my Natural Dullness (Ps 32:8, John 16:13a)
0 -
JRS said:
Thanks, Donnie, but I prefer to live 'dangerously'
As long as you realize that settings like that are what allow malicious actors to find hosts from which to launch their distributed denial-of-service attacks. You can disable a sound setting. FWIW, I only get the UAC elevation prompt about 2-3 times per month. And I'm a software developer / development manager.
-Donnie
0 -
Ran the utility and with an resource and tool intensive layout as default at open - less than 8 seconds to full open. Better with a less intensive layout.
Logos 10 - OpenSuse Tumbleweed, Windows 11, Android 16 & Android 14
0 -
Dave Hooton said:
QUESTION:
How do we get rid of previous assemblies as the new one doesn't remove the old one?
Dave, I'm looking into this and I don't see old assemblies being kept around. I'd like to know what you're seeing so we can correct the issue.
My steps (for comparison)
- Run logos.exe /ngen
- Run C:\Windows\Microsoft.NET\Framework64\v4.0.30319\ngen.exe display Logos
- Note that it has the current version on my system (7.9.0.6)
- Update to the new beta (for me this was 7.9.0.8)
- Run logos.exe /ngen
- Run C:\Windows\Microsoft.NET\Framework64\v4.0.30319\ngen.exe display Logos
- Note that it only shows a single assembly (7.9.0.8)
There is an update command that can be run. I'm not exactly sure what it will do since I cannot reproduce the issue you're seeing.
0 -
- Run C:\Windows\Microsoft.NET\Framework64\v4.0.30319\ngen.exe display Logos
- Displays 7.9.0.8 (from 17/8 4.50pm) and 7.8.0.37 (from 24/8 12:35pm)
- Update to the new beta (7.9.0.10)
- Run logos.exe /ngen
- Run C:\Windows\Microsoft.NET\Framework64\v4.0.30319\ngen.exe display Logos
- Displays 7.9.0.10 (from 24/8 1:04pm) and 7.8.0.37
When I checked C:\Windows\assembly\NativeImages_v4.0.30319_64 I realised I had been mislead by the assembly timestamps i.e folders dated 17/8 4.50pm had sub-folders dated 24/8 whilst the rest of the assemblies were in folders dated 24/8 1.04pm. Total folder size was 428 MB vs. 445 MB.
Dave
===Windows 11 & Android 13
0 -
Without ngen:
OurApp Starting application (7.9 Beta 4, version 7.9.0.0010 x64 on Microsoft Windows NT 10.0.15063.0, .NET 4.0.30319.42000) at 1.511 seconds.
7.8 was a little faster, but not much. maybe .1 or .2 seconds.
With ngen:
OurApp Starting application (7.9 Beta 4, version 7.9.0.0010 x64 on Microsoft Windows NT 10.0.15063.0, .NET 4.0.30319.42000) at 4.333 seconds
Second run was about the same: 4.2
Third was better: OurApp Starting application (7.9 Beta 4, version 7.9.0.0010 x64 on Microsoft Windows NT 10.0.15063.0, .NET 4.0.30319.42000) at 1.562 seconds
Fourth
OurApp Starting application (7.9 Beta 4, version 7.9.0.0010 x64 on Microsoft Windows NT 10.0.15063.0, .NET 4.0.30319.42000) at 1.115 seconds.
Is this expected behavior?
Initially thats a VERY significant slow down.
AMD Ryzen 1800+ (well 1700 overclocked to the speed of an 1800), 16gb ddr4 running at 3100mhz. Logos lives on a samsung SSD 960 EVO NVMe M.2 500gb.
Edit: more tests
1.146
2.419
1.121
3.466
Running some search speed tests. searched "The" in Everything. took .71 before to search the ESV 61460 results in 23789 verses. Now it takes 2.37 seconds.
Subsequent repeats of the same search were a little slower 2.41-2.9.L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
I suppose at my speeds an additional second is tolerable, but still represents a significant slow down. I hope this is something we can opt out of if it makes it into the product down the line. Guessing it has to do with processor architecture?
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
The times are going to be different for each user (and probably each run). It's very difficult to say whether it will always be faster because there are a lot of variables.
According to the Microsoft site the expectation should be something along the lines of a normal startup speed on the first launch after a computer restart. On subsequent startups (until the machine is restarted, or enough memory is used that Logos is taken out of memory) load times should be faster than if you hadn't ngen'd.
Essentially, ngening can be thought of as caching the loaded state of the app to the HD. In the non-ngen'd state the OS loads it every time. With ngen it only loads it once. (this is not a technical description and not actually what is happening but it is perhaps a simpler way to think about what's going on)
0 -
interesting. you would think on an M.2 with a theoretical max of 3.2gbps that sort of option would make a significant difference.
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
abondservant said:
OurApp Starting application (7.9 Beta 4, version 7.9.0.0010 x64 on Microsoft Windows NT 10.0.15063.0, .NET 4.0.30319.42000) at 4.333 seconds
I think it is more significant to know how long it takes to load the layout e.g.
Info StartupUtility Finished startup at 10.14s.
You can also see when it starts to load the layout (when the blank window appears) e.g.
Info OurApp Showing main window at 2.809 seconds.
They include the time for Starting application (0.268s in my case)
Intel Core i7 5500U (2.4 GHz with boost to 3.0 GHz), 8gb ddr3, 1000 GB hard drive.
Dave
===Windows 11 & Android 13
0 -
Finished Startup: 30.39, 19.18, 18.04, 19.27Dave Hooton said:abondservant said:OurApp Starting application (7.9 Beta 4, version 7.9.0.0010 x64 on Microsoft Windows NT 10.0.15063.0, .NET 4.0.30319.42000) at 4.333 seconds
I think it is more significant to know how long it takes to load the layout e.g.
Info StartupUtility Finished startup at 10.14s.
You can also see when it starts to load the layout (when the blank window appears) e.g.
Info OurApp Showing main window at 2.809 seconds.
They include the time for Starting application (0.268s in my case)
Intel Core i7 5500U (2.4 GHz with boost to 3.0 GHz), 8gb ddr3, 1000 GB hard drive.
Main Window: 21.9, 12.751, 11.9, 12.648
Ran /unngen
Finished Startup: 29.81, 23.71, 23.65, 24.22
Main Window: 19.337, 15.3, 15.6, 15.9
So it looks like it is consistently faster with ngen.L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
abondservant said:
Finished Startup: 30.39, 19.18, 18.04, 19.27
Main Window: 21.9, 12.751, 11.9, 12.648
Ran /unngen
Finished Startup: 29.81, 23.71, 23.65, 24.22
Main Window: 19.337, 15.3, 15.6, 15.9Your time to Main Window is interesting as thereafter it takes about the same time as mine to load the layout (c. 9s).
What is happening prior to Main Window?
Dave
===Windows 11 & Android 13
0 -
050870.Logos.log
Beware the fowl language. same log from my other post.
Maybe my computer doesn't like processing the F word.L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
Comparing log events before Main Window I found the reasons for the difference (12.7s vs 3.2s for me).
- Creating library catalog. 0.3s vs 2.0s ---> + 1.7s
- Creating New Sync Manager 0.2s vs 0.9s ---> + 0.7s
- Creating Resource Lists 0.34s vs 2.18s ----> + 1.8s
- Startup step: StartPostSetFactoryServicesWork 0.6s v 3.6s ---> + 3.0s
- Startup step: StartFirstSync 0.0s vs 2.3s ----> + 2.3s because your Starting layout is Most recent - Any
Total difference = + 9.5s
Faithlife can explain the items I highlighted, but:
- using Most recent - Local will save time by postponing First Sync until after your layout has loaded
- how many resources do you have?
Dave
===Windows 11 & Android 13
0 -
16000 or so. I can give a more precise figure if need be. but thats in the ballpark.
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
I have 1810 in Library, so that may account for the Library Catalog and Resource List differences. But I'd like some feedback/explanation from Faithlife.
Dave
===Windows 11 & Android 13
0 -
running this on my much slower laptop. will post results soon.
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
abondservant said:
AMD Ryzen 1800+ (well 1700 overclocked to the speed of an 1800), 16gb ddr4 running at 3100mhz. Logos lives on a samsung SSD 960 EVO NVMe M.2 500gb.<br><br>
Finished Startup: 30.39, 19.18, 18.04, 19.27
Main Window: 21.9, 12.751, 11.9, 12.648
Ran /unngen
Finished Startup: 29.81, 23.71, 23.65, 24.22
Main Window: 19.337, 15.3, 15.6, 15.9Laptop: AMD A10-7300 Radeon R6 6gb ddr3 64 bit 256gb ssd
Finished Startup (old version) 31.58
Finished Startup (Beta) 29.15
Finished Startup (ngen) 24.10
Main Window (old) 32.426
Main Window (beta) 15.406
Main Window (ngen) 15.523
5584.Logos - old.log
0218.Logos - beta.log
0218.Logos.logL2 lvl4 (...) WORDsearch, all the way through L10,
0 -
my desktop hardware should make more of an impact than that?
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
Is there a benchmark they prefer to show the speed difference?
L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
abondservant said:
my desktop hardware should make more of an impact than that?
You would think so. Was the layout the same?
BTW, ignore the (old) timing because you waited 16s to respond to the Startup window before the blank window was loaded (main window followed completion of startup!).
Dave
===Windows 11 & Android 13
0 -
abondservant said:
Is there a benchmark they prefer to show the speed difference?
Faithlife provide the log timings because loading is the area of improvement.
You should run your own benchmarks to determine the differences between laptop and desktop (CPU/HDD/Graphics and application benchmarks).
Dave
===Windows 11 & Android 13
0 -
I use them all the time, I have an idea of the speed difference, but without putting your hands on the machine its hard to know... So I thought perhaps it would help to illustrate that point. That the speed is significantly different... Is there a benchmark software they prefer? I forget what I used to use all the time, but it makes a composite score overall. I'll have to google it if they don't have a preference.
EDIT: A google search turned up novabench. Will have to run this on my laptop tomorrow... BUT the desktop came back as follows:
https://novabench.com/view/1319319
~1400
Should be a bit higher. the processor should be around 1400 all on its own. I'll have to look into that tomorrow.L2 lvl4 (...) WORDsearch, all the way through L10,
0 -
abondservant said:
EDIT: A google search turned up novabench. Will have to run this on my laptop tomorrow..
Passmark can compare your CPU to other user's Ryzen as well as different CPU's. Below average results usually point to the motherboard implementation.
Dave
===Windows 11 & Android 13
0