Dota 2 micro lags.

[IgorHoruzha]

Description

There are several topics about micro-lags in the Ethernet(Like this one) or this one. I have the same issue since a long while. I managed to change two PCs from the first time I faced it and this is my third one. Each time, it helped a bit. But the issue hasn't gone away.

I decided to solve this bug again today. I haven't solved it again. But I managed to trigger it on purpose! And I'm going to share my experience with You. I hope It will help You make the DOTA 2 gameplay process even better than it is now!

My Pc Spec:

Processor 13th Gen Intel(R) Core(TM) i5-13600KF, 3500 MHz, 14 Core(s), 14 Logical Processor(s) Installed RAM 32.0 GB (31.8 GB usable) OS Name Microsoft Windows 11 Pro System type 64-bit operating system, x64-based processor Edition Windows 11 Pro OS build 22621.3007 Experience Windows Feature Experience Pack 1000.22681.1000.0 BIOS Version/Date American Megatrends Inc. 1604, 12/15/2023 SMBIOS Version 3.5 Motherboard Asus PRIME Z790-P WIFI Video Card NVIDIA GeForce GT 1030 Ram Kingston FURY DDR5-5200 32768MB PC5-41600 (Kit of 2x16384) Beast AM5 Black (KF552C36BBEK2-32)

Hyper-V - VM Monitor Mode Extensions Yes Hyper-V - Second Level Address Translation Extensions Yes Hyper-V - Virtualization Enabled in Firmware No Hyper-V - Data Execution Protection Yes

Mouse Xtrfy M42 RGB USB White Keyboard Xtrfy K4 TKL RGB Kailh Red USB White Monitor 27 DELL UP2716DA (210-AXWI)

Dota 2 settings: All graphic settings are lowest. Windowed 1024x768 Vulcan(But DX11 also has so).

To trigger the essential lag, not a micro one. You have to decrease the amount of cores to 4. The more cores You have, the less this lag lasts and feels. And then start walking for a while in Demo Hero mode. Once every 30 seconds or a minute, it usually happens for a short while. Sometimes several times in a short time.

This is a video that shows the issue, and how I managed to trigger it.

I haven't done it on video, but to trigger the lag even more frequently, it may be useful to set a low priority on the process.

The thing is that I feel this effect even if I set high priority and ran as administrator, on all cores, and even with an AMD RX 560 or AMD RX 590 video card. With the same or Intel Core I7-8700K processor. And even if Hyper-V is on. On 1920x1080 or even 2560x1440 resolution. The difference is just in the length of the lag, and it's frequency. But any way is not smooth. Except for some very rare, once a blue moon, exceptions.

If You need some additional information, I'm ready to give it. I hope You'll solve it, or help me understand where is the mistake if it is my mistake. Thank You in advance.

Example Match ID (and possibly Timestamp)

No response

Screenshots

This is the same video like in the description. Task manager processes view. Left click also do trigger it. Chat messages also trigger it. Task Manager settings to trigger the lag. Settings to trigger the bug in the video.

[jeffhill]

Could you try capturing an ETL file using UI for ETW, and sharing that? https://github.com/google/UIforETW/releases

ETL traces capture everything that's happening on your machine during the trace, so if you can trigger a performance issue, the ETL will show exactly where in the code it's getting lost.

It's also possible that the dashboard chat bugfix we just shipped also fixes this, so getting confirmation you can still see the problem would be helpful. Thanks!

[K-o-patra]

I have the exact same issue since 7.33. It was very frequent then. With the first few patches after 7.33 it was reduced in both frequency and severity by a lot, but it has never vanished completely. Last week I recorded my ethernet traffic during a game because I wanted to be sure that it's not my internet connection. I synced the wireshark recording with a screen recording of my game at the start, as shown in the first few seconds of this video. Then the clip jumps to a scene where the bug occured. It startet at minute 8:17. After respawning and shopping, I centered my camera on my hero and gave movement orders at 08:17, but they weren't processed until 08:20. It is the game with ID 7561285637, played on Jan 30.

In the vid it can be seen that at game start (in-game time 00:00), the wireshark recording is at second 155.5. When the bug occured, the in-game time (08:17 to 08:20) corresponds to the Wireshark timestamps from 652.5 to 655.5 roughly. I attached an excel sheet of the complete wireshark recording.xlsx. It is filtered to show only UDP. I didn't have any other online tasks active.

For most parts of the game, there are 60 packets per second (pps), 30 incoming and 30 outgoing. This number is relatively stable, maybe +/- 3 for 90% of game time roughly. But during the time of the bug (table timestamps 647 to 657, corresponding to in-game time 8:13-8:23) the pps-rate is significantly higher, up to 200 during second nr. 648. I made a table of the pps-rate during the bug:

It is also noteworthy, that for the whole game the packets are sent and received, taking turns. These pairs (one sent and one received packet) have a frequency of 33ms. (I assume this is the server tick-rate?). Meaning, every 33ms, one packet is sent and one received. But during the bug, up to 10 packets are sent or received straight one after another. And this happens all within a 33ms time-frame (roughly). I also have a screenshot of this:

Lastly, I wanted to attach the original wireshark recording as zip, but the filesize is 65MB and the limit is 25MB. If any dev is interested, I can send it to you if you tell me how.

Random thoughts: It is evident that the blackout-bug and the spike in network traffic are related. But it is not clear in which way.

One theory is, that poor network quality results in high packet loss, which in turn makes server and client believe they should re-send data, accumulating more packets and larger packets with each packet that is believed to be lost. This theory contradicts my understanding that UDP doesn’t offer receipts/confirmation about received packets, so it is only a plausible theory if the Dota 2 network protocol does any kind of packet loss estimation, or if dota attempts to track packets in any way.

Another theory is, that some code in some very specific scenarios results in an overwhelming amount of network data being created and sent simultaneously, which could cause input/output buffers to reach their traffic limits, or block CPU resources for some cycles.

Furthermore I ask myself, could this be related to a setting in the Dota 2 console? I vaguely remember some reddit post (loooong time ago) that said to set some cl_rate or bandwidth or packet limit to some high number (maybe it was 800.000?). I don’t know if I did that or not, but I remember it made sense in the context of the post so I might have changed it. But I definitely didn’t change it since my last fresh install of Dota 2 (last month), so I don’t even know if those changes would have persisted after a fresh install…

And my specs

Processor Information: CPU Vendor: AuthenticAMD CPU Brand: AMD Ryzen 5 5600X 6-Core Processor CPU Family: 0x19 CPU Model: 0x21 CPU Stepping: 0x0 CPU Type: 0x0 Speed: 3700 MHz 12 logical processors 6 physical processors Hyper-threading: Supported FCMOV: Supported SSE2: Supported SSE3: Supported SSSE3: Supported SSE4a: Supported SSE41: Supported SSE42: Supported AES: Supported AVX: Supported AVX2: Supported AVX512F: Unsupported AVX512PF: Unsupported AVX512ER: Unsupported AVX512CD: Unsupported AVX512VNNI: Unsupported SHA: Supported CMPXCHG16B: Supported LAHF/SAHF: Supported PrefetchW: Unsupported

Operating System Version: Windows 10 (64 bit) NTFS: Supported Crypto Provider Codes: Supported 311 0x0 0x0 0x0

Video Card: Driver: AMD Radeon RX 6900 XT DirectX Driver Name: aticfx32.dll Driver Version: 31.0.22011.4008 DirectX Driver Version: 31.0.22011.4008 Driver Date: 9 21 2023 OpenGL Version: 4.6 Desktop Color Depth: 32 bits per pixel Monitor Refresh Rate: 144 Hz DirectX Card: AMD Radeon RX 6900 XT VendorID: 0x1002 DeviceID: 0x73bf Revision: 0xc0 Number of Monitors: 1 Number of Logical Video Cards: 1 No SLI or Crossfire Detected Primary Display Resolution: 2560 x 1440 Desktop Resolution: 2560 x 1440 Primary Display Size: 23.50" x 13.23" (26.97" diag), 59.7cm x 33.6cm (68.5cm diag) Primary Bus Type Not Detected Primary VRAM Not Detected Supported MSAA Modes: 2x 4x 8x

[IgorHoruzha]

Could you try capturing an ETL file using UI for ETW, and sharing that? https://github.com/google/UIforETW/releases

ETL traces capture everything that's happening on your machine during the trace, so if you can trigger a performance issue, the ETL will show exactly where in the code it's getting lost.

It's also possible that the dashboard chat bugfix we just shipped also fixes this, so getting confirmation you can still see the problem would be helpful. Thanks!

I'm sorry for the delay. I have just checked the micro lags, they are still appearing. I'll try capturing an ETL file using UI for ETW. I need some time to investigate the tool.

[IgorHoruzha]

I have traced the microlags. In case I made a somewhere a mistake during tracing, I also recorded the video. To have some recommendations in this case. ETL looks like a powerful tool, but I haven't investigated it as well yet.

I uploaded etwtraces on the GoogleDrive.

Additional Information, during tracing I managed to trigger client hold for a long time, on purpose. More than one time in general. One one the video, two others are the tries that I had been trying to record before. On one of such crash I managed to see "System Interrupts" process (after had opened the Task manager) for a very short while that takes more than 50% CPU, But it disappeared in a second.

After a crash, I started the new Dota 2 client. But I haven't changed the available CPUs to the first 5. There were two moments where there were super mini lags that I managed to trigger even using 14 CPUs:

First one. Look attentive on the Draw Ranger between 17min50sec and 17min53sec. She lagged but on a very short time.

Second one Look attentive on the front top side of Draw Ranger's bow, at the moment when the bow got up and started getting back. Between 18min55sec and 18min58sec. It feels much less than the first one. Then I understood that I hadn't set only the first 5 CPUs available to the Dota 2 process, and fixed it.

Unfortunately, some why FPS looks bad on the video, but I found a small fall (from 60-61 to about 58) at the micro lags moments, while trying . If necessary, I can try to get more moments on 14 CPUs that may look even clearer than the first one.

If You need some additional information, I'm ready to give it.

[IgorHoruzha]

I have the exact same issue since 7.33. It was very frequent then. With the first few patches after 7.33 it was reduced in both frequency and severity by a lot, but it has never vanished completely. Last week I recorded my ethernet traffic during a game because I wanted to be sure that it's not my internet connection. I synced the wireshark recording with a screen recording of my game at the start, as shown in the first few seconds of this video. Then the clip jumps to a scene where the bug occured. It startet at minute 8:17. After respawning and shopping, I centered my camera on my hero and gave movement orders at 08:17, but they weren't processed until 08:20. It is the game with ID 7561285637, played on Jan 30.

In the vid it can be seen that at game start (in-game time 00:00), the wireshark recording is at second 155.5. When the bug occured, the in-game time (08:17 to 08:20) corresponds to the Wireshark timestamps from 652.5 to 655.5 roughly. I attached an excel sheet of the complete wireshark recording.xlsx. It is filtered to show only UDP. I didn't have any other online tasks active.

For most parts of the game, there are 60 packets per second (pps), 30 incoming and 30 outgoing. This number is relatively stable, maybe +/- 3 for 90% of game time roughly. But during the time of the bug (table timestamps 647 to 657, corresponding to in-game time 8:13-8:23) the pps-rate is significantly higher, up to 200 during second nr. 648. I made a table of the pps-rate during the bug:

It is also noteworthy, that for the whole game the packets are sent and received, taking turns. These pairs (one sent and one received packet) have a frequency of 33ms. (I assume this is the server tick-rate?). Meaning, every 33ms, one packet is sent and one received. But during the bug, up to 10 packets are sent or received straight one after another. And this happens all within a 33ms time-frame (roughly). I also have a screenshot of this:

Lastly, I wanted to attach the original wireshark recording as zip, but the filesize is 65MB and the limit is 25MB. If any dev is interested, I can send it to you if you tell me how.

Random thoughts: It is evident that the blackout-bug and the spike in network traffic are related. But it is not clear in which way.

One theory is, that poor network quality results in high packet loss, which in turn makes server and client believe they should re-send data, accumulating more packets and larger packets with each packet that is believed to be lost. This theory contradicts my understanding that UDP doesn’t offer receipts/confirmation about received packets, so it is only a plausible theory if the Dota 2 network protocol does any kind of packet loss estimation, or if dota attempts to track packets in any way.

Another theory is, that some code in some very specific scenarios results in an overwhelming amount of network data being created and sent simultaneously, which could cause input/output buffers to reach their traffic limits, or block CPU resources for some cycles.

Furthermore I ask myself, could this be related to a setting in the Dota 2 console? I vaguely remember some reddit post (loooong time ago) that said to set some cl_rate or bandwidth or packet limit to some high number (maybe it was 800.000?). I don’t know if I did that or not, but I remember it made sense in the context of the post so I might have changed it. But I definitely didn’t change it since my last fresh install of Dota 2 (last month), so I don’t even know if those changes would have persisted after a fresh install…

And my specs

Processor Information: CPU Vendor: AuthenticAMD CPU Brand: AMD Ryzen 5 5600X 6-Core Processor CPU Family: 0x19 CPU Model: 0x21 CPU Stepping: 0x0 CPU Type: 0x0 Speed: 3700 MHz 12 logical processors 6 physical processors Hyper-threading: Supported FCMOV: Supported SSE2: Supported SSE3: Supported SSSE3: Supported SSE4a: Supported SSE41: Supported SSE42: Supported AES: Supported AVX: Supported AVX2: Supported AVX512F: Unsupported AVX512PF: Unsupported AVX512ER: Unsupported AVX512CD: Unsupported AVX512VNNI: Unsupported SHA: Supported CMPXCHG16B: Supported LAHF/SAHF: Supported PrefetchW: Unsupported Operating System Version: Windows 10 (64 bit) NTFS: Supported Crypto Provider Codes: Supported 311 0x0 0x0 0x0 Video Card: Driver: AMD Radeon RX 6900 XT DirectX Driver Name: aticfx32.dll Driver Version: 31.0.22011.4008 DirectX Driver Version: 31.0.22011.4008 Driver Date: 9 21 2023 OpenGL Version: 4.6 Desktop Color Depth: 32 bits per pixel Monitor Refresh Rate: 144 Hz DirectX Card: AMD Radeon RX 6900 XT VendorID: 0x1002 DeviceID: 0x73bf Revision: 0xc0 Number of Monitors: 1 Number of Logical Video Cards: 1 No SLI or Crossfire Detected Primary Display Resolution: 2560 x 1440 Desktop Resolution: 2560 x 1440 Primary Display Size: 23.50" x 13.23" (26.97" diag), 59.7cm x 33.6cm (68.5cm diag) Primary Bus Type Not Detected Primary VRAM Not Detected Supported MSAA Modes: 2x 4x 8x

@jeffhill @K-o-patra IMO this topic may need some extra attention as a separate individual topic. I have some additional information about it that may be useful. A while ago, I have tried to solve may similar case by switching the wire Ethernet from 100mbps to 1000 mbps and pinging up to 40. I also experienced a different provider in another city by ~up to 75mbps using my phone connected to Wifi as a USB modem. It doesn't help. To be honest, it made it even more frequently.

BUT, I once tried my phone connected to LTE as a USB modem. With 100mbps. I was surprised, it helped a lot. It became so pleasant to play. The fights stopped be something like an almost a flash to fontain.

Phone specs: Samsung Galaxy A6 SM-A600FN/DS Android version 10

Also, I note that if my PC opens anything in the background, especially Google Chrome, it makes micro lags more frequent.

[goldjee]

Not sure if related to this particular topic, but I have encountered some sort of input lag after latest update. It seems like some of commands issued client side are slightly delayed or even not registered. This involves move command, skills, attack and so on.

It was a thing back at pre-reborn times, where players faced ~300ms input action delay, but now it is rather inconsistent. I am pretty confident that mya internet connection is fine (in game it shows about 60-70ms ping and 0 packet loss). Pre- "dragon's gift" everything was fine.

Shall I open another ticket for this issue? Also, don't hesitate to contact me if you need any details on the matter.

[K-o-patra]

In regards to my previous and extensive comment: I have now made a contract with a second internet provider which uses a different infrastructure. I want to compare the two over some time. I now have both (relevant) technologies that are available in Germany, cable and DSL:

1. Cable Internet (Vodafone is my provider) This is my original provider. I chose cable technology a long time ago because supposedly it offers more stable connections, lower ping values, less packet loss, and less interference through weather, whereas DSL can achieve similar values, it might not be as reliable.

2. DSL (MaXxim, part of the Drillish AG, is my provider) This is my new provider. Recently, a friend told me that he had heard the exact opposite: DSL was supposed to have lower ping, and less packet loss. And it "was known" that cable offers higher bitrates, but less reliable connections.

Internet research on this topic is futile, there is just no reliable information out there to confirm either statement. Most importantly, there seem to be zero representative measurements where the two network technologies are compared. It seems to be a reasonable assumption that local infrastructure makes the key difference. Factors could be:

• the physical signal transmitter mediums (aka cables under the streets) that are used in one's actual neighborhood,
• the routers and/or hubs that connect the neighborhood network to the house or block,
• the receiver (and its settings) inside a house or block, from where each apartment's individual connection is established,
• and lastly, the cables and the final outlet inside the house/apartment.

The only way to know which connection is more reliable is to actually compare them. I am in the fortunate position that both technologies are available in my apartment. A few hours ago, the technician was here to activate my second internet connection via DSL. First tests have shown what I expected: The values are veeeeeeeery similar. Ping fluctuates between 7ms and 15ms on both connections. Packetloss is 0% so far. I have to add here, that packetloss has been as high as 4% occasionally during the last couple of weeks on my original cable connection, which might be a sufficient explanation for the bug described in this ticket. For this reason, I will now use the new provider (using DSL technology) for some time to see if similar behavior occurs.

As I now have two modems (one for cable internet and one for DSL), I have to plug in/out my ethernet cable on my PC because my mainboard has only one ethernet port. Please @valve, don't ban me for account sharing or something during the time of my experiment, as I might change IP adresses frequently and fast in between games, or maybe even during a game (if I run into connection issues, and if that is even possible, I have not tried yet).

If the behavior described in this ticket does occur on my new internet connection as well, I would personally interpret that as strong evidence that this bug is caused by Dota itself, since (as explained above) the two internet network technologies are fundamentally different and don't share any physical or virtual mediums (only the major backbones, as far as I am aware). If on the other hand, the behavior does not happen over the period of two weeks, I interpret that as strong evidence that it was indeed caused by a faulty internet connection.

I will update this ticket in late February.

[K-o-patra]

Update to my previous comment:

I have now tested dota for a week with my new (DSL) connection. It turns out that the problem occurs on this connection as well (different ISP, different network hardware even, see previous comment).

Additionally, the problem discussed in issue 15559, happens on both my internet connections as well. Also seemingly random, at least I didn't recognize any patterns yet.

Reading the comment from @jeffhill on that issue made me think of my findings regarding this very issue, when I recorded my connection with wireshark, which I described in my previous comment. Can you do the same check for the match ID that I provided? Maybe we can at least establish if the issue is server-side or client-side...

[zenkovsan]

I'm having micro lags, but they seem to be FPS related. At the start of each game, my game stutters, going from 240 fps to 100 for a blink.

[IgorHoruzha]

Today, I conducted further experiments and have new data to share.

This time, I configured the BIOS to allow only 2 and 3 cores, respectively, and allocated 8GB of RAM on the same PC setup. I reduced the number of cores and the amount of RAM to make it easier to detect any small changes that might lead to lags. I used the Vulkan rendering API settings.

I was unable to intentionally trigger a lot of significant lags (such as those I might experience when using Task Manager to reduce the amount of cores). However, I did manage to trigger numerous micro lags. Interestingly, it appears these micro lags are not always related to CPU 100% load.

I've documented the experiments through various videos and moments.

The first video shows the setup with 2 cores and 8GB of RAM. Specific moments captured demonstrate different instances of micro lags and one instance of heavy lag potentially triggered by antivirus system checks, according to Task Manager observations.

The first moment The second moment The third moment The forth moment The fifth moment The sixths moment The sevenths moment The eights moment The nineth moment A heavy lag. Maybe due to anti virus activity system check. According to the Task manager. The tenths moment An interesting one. Here is the antivirus but also Dota 2 CPU usage increase. The elevenths moment The twelfth moment The thirteenths moment While I was changing the camera position it was lagging. The fourteenth moment The fifteenths moment 100% CPU but no lags was detected

The second video presents the scenario with 3 cores and 8GB of RAM, highlighting similar small-scale lags.

The first moment The second moment The third moment

I will showcase the most significant moments next, omitting minor lags.

I configured the system to use 2 cores instead of 3 via the Task Manager, which resulted in immediate, consistent lags for a period. However, there were no significant lags until I began extensively using the chat feature. For some reason, merely moving around did not trigger any substantial lags, only micro lags. Following this, I restarted Dota 2.

The forth momentI experimented with the Order Queue and noticed what appeared to be a micro lag while observing the Drow Ranger. I sensed a slight discrepancy when she began moving back to the top. The fifth moment

I began to suspect that background processes, especially the antivirus, could be one of the causes of the micro lags. However, it's possible that this is not the only reason. In some cases I had lags with CPU less ten 100% load. I don't have enough data yet to confirm or refute this theory. Another pressing question remains: how is it possible that I experience lags even when 14 cores and 20 threads are active?

It may be a placebo effect, but generally, the hero seemed to respond faster to my actions while moving. After one of previous updates. Also, when there were no lags, the hero's movements appeared clearer than before. In general, with 14 cores settings(and also with todays 2 and 3 cores settings test).

[IgorHoruzha]

Yesterday I conducted further experiments and have new data to share. This time, I focused on investigating micro lags on my optimized PC settings without reducing the number of cores or RAM. I permanently disabled Antivirus real-time protection and made slight improvements to the BIOS settings, which seemingly enhanced my PC's overall performance.

For a detailed observation of micro lags, I recommend viewing the clips at 0.25x speed. I feel like my PC started working essentially faster. IMO like if bought a new one.

I recommend to watch clips by 0.25x speed to see the micro lags clear.

In the first series of tests, I examined walking lags and captured four distinct moments:

1. First moment
2. Second moment
3. Third moment
4. Fourth moment

Additionally, I noticed lag issues with my camera movement. Despite V-Sync being enabled, it did not appear to alleviate the problem significantly. The camera movement, while free from freezes, still seemed laggy and noticeably blurry. I documented this in the next two videos, capturing ten specific moments where the camera movement exhibited two tips of lags and blurriness.

The second video.

1. First moment
2. Second moment
3. Third moment (notably, Drows' bows also lagged at the end)
4. Fourth moment
5. Fifth moment
6. Sixths moment
7. Seventh moment
8. Eighth moment
9. Ninth moment

10. Tenth moment

    Furthermore, I encountered side bugs, such as a bugged tower particle effect, and observed micro lags following various actions, like restarting the demo mode or shifting focus from the task manager window. I also noted a missing warning for the fountain attack.

In the third series of tests I used DX11, I captured various moments of lag under different settings, with and without V-Sync enabled. Additionally, I experienced interface issues, such as tooltips appearing under buttons and warnings in the console regarding the EntitySystem and Resource System.

1. First moment
2. Second moment
3. Third moment V-Sync is off
4. Fourth moment V-Sync is on
5. Fifth moment
6. Sixths moment V-Sync is off
7. Seventh moment V-Sync is on

Adjustments to the display refresh rate and Dota 2 FPS limit in Windows settings produced mixed results, with changes either leading to increased lag or slight improvements, which could be attributed to a placebo effect. Similarly, altering the camera speed and FPS in the game settings resulted in varied experiences with two main types of micro lags and and blurriness.

I adjusted the display refresh rate from 60Hz to 50Hz in Windows settings, which resulted in noticeably laggier camera movement.

I set the Dota 2 FPS limit to 30 via the console, but the camera movement remained laggy.

I adjusted the Dota 2 FPS limit to 50 through the console, yet the camera movement continues to be laggy.

I configured the Dota 2 FPS limit to 120 via the console; however, the camera movement remains laggy. There seems to be a slight improvement, but it could just be a placebo effect.

Some why the hero name and icon in the right panel are displaying incorrectly.

I decreased the camera speed to 1000. It appears that the micro lags have become shorter and less frequent, yet the movement remains laggy; this could be a potential placebo effect.

8. Eighth moment V-sync off
9. Ninth moment V-sync on

I changed the display refresh rate from 60Hz to 60Hz (Interlaced) in the Windows settings, but the lag persists, possibly even intensifying compared to before, which might be a placebo effect. Additionally, other windows, aside from Dota, started flickering, although this was not captured in the video.

I adjusted the display refresh rate from 60Hz (Interlaced) back to 60Hz in the Windows settings, but the lag continues.

I set the render quality to 100% and restarted demo mode, yet the lag persists.

I adjusted the camera speed to 500 and the FPS to 120, which unexpectedly caused an issue where the camera movement is consistently skewed towards the left bottom corner.

10. Tenth moment
11. Eleventh moment

The group of Drow Rangers positioned above and to the right momentarily displayed a blinking white line for some reason.

Moving the console may have caused a lag.

After restarting the demo mode, it appears that the bottom panel slid up accompanied by micro lags.

12. Twelfth moment camera speed 500

Changing settings and then returning to the demo mode possibly introduced micro lags. Could this be occurring during the saving process?

The sliding bottom bug reoccurred .

Changing settings and returning to the demo mode may have caused micro lags once again.

Adjusting the settings to 40 fps resulted in lag.

Reducing the settings to 28 fps resulted in increased lag.

Adjusting the settings to 61 fps did not significantly improve performance.

13. Thirteenth moment 61 fps

It is possible that the settings menu also experiences lag upon opening.

14. Fourteenth moment V-Sync off

Setting the FPS to 240 didn't significantly help; it may have slightly reduced the camera lag, but the issue persists, possibly due to a placebo effect.

15. Fifteenth moment up to 240 FPS
16. Sixteenth moment up to 240 FPS two more micro laggs
17. Seventeenth moment up to 240 FPS

Clicking the dashboard button resulted in some red text appearing in the console.

18. Eighteenth moment up to 240 FPS
19. Nineteenth moment up to 240 FPS
20. Twentieth moment up to 240 FPS

Overall, despite the adjustments and optimizations, Dota 2 still exhibited micro lags and camera movement issues on My PC.

[IgorHoruzha]

Yesterday, I completely reinstalled Dota 2 and collected additional data using UIforETW and console commands.

The first video is about UIforETW capturing.

Interesting moment launch video lagged. But I was not tracing yet.

Camera some why fell to the bottom a bit.

files.etl

Captured moments:

1. First moment
2. Second moment
3. Third moment on hero creating micro lags
4. Fourth moment
5. Fifth moment
6. Sixths moment I'm not sure whether this moment lags only on video or were also in game.
7. Seventh moment. The smallest file. The clearest lag. I do recommend to start with it.

The second video is about condebug capturing.

Launch settings:

-console -condebug +cl_ticktiming 300 detail

console.log

Captured moments:

1. First moment micro lags on hero creating. Especially on first four.
2. Second moment In the middle of the river.
3. Third moment
4. Fourth moment
5. Fifth moment. Very noticeable.

[gruianovac]

awesome work. this was mine https://github.com/ValveSoftware/Dota2-Gameplay/issues/13301 https://github.com/ValveSoftware/Dota2-Gameplay/issues/12865 .. not much value i guess compared to your work. but it definitely came after an update for me

[IgorHoruzha]

Today I decided to check Dota 2 Workshop Tools DLS. I removed all objects from the map, but anyway I got some micro lags. Example 0.10x speed:

https://github.com/ValveSoftware/Dota2-Gameplay/assets/29505510/18816cfd-e356-40a8-8195-08d14d30c394

Original video

[gruianovac]

did the new patch do anything for you ? it seems to have fixed mine

[IgorHoruzha]

@gruianovac

The Crownfall has significantly reduced the lag duration and frequency. At first, I thought the issue was resolved. When I was testing, I noticed them quite rarely.

1. The first moment
2. The second moment
3. The third moment
4. The forth moment
5. The fifth moment There were two lags a line.
6. The sixths moment First and the only one heavy lag.
7. The sevenths moment
8. The eights moment
9. The tenths moment
10. The elevenths moment
11. The twelfth moment There were two lags a line again.

12. The thirteenths moment

    After reviewing the recording of the testing session, I realized that I missed some segments either because they were too brief, I blinked or because it was unclear whether there was lag. Like this one Here were two lags in a line; there were quite a few such moments.

    Also I noticed some blurring and a different type of lags . First moment Second moment It seems like some threes on the left are blurring a bit. The interesting that not all just that are surrounded. Third moment The same trees in different time. The same behavior.

Dota 2 settings Monitor to 24.5" DELL G2524H 380Hz . Video Card RX460.

Source Video

[gruianovac]

yes, spoke too soon. it happens

[IgorHoruzha]

It seems that huge micro lags have become extremely rare, occurring roughly every 4-8 minutes or even less frequently. There's been a noticeable change since the start of Crownfall update and updates after.

Micro lags are still present, though. In my opinion, the micro lags were less noticeable a few updates ago compared to now.

I do recommend watching the source video at 0.1x speed. The issue isn't only apparent when I draw on the mini-map but also occurs frequently in other instances. They feel like the hero blinks or freezes for a very short while.

Some examples: 1) First moment 2) Second moment A long series of micro lags. 3) One more random moment

There are are a lot of this type of moments in the video. More than 10 in 43 seconds video.

@kaisenbaka Do You experience something the same to the video in Your issue?

Source video

[IgorHoruzha]

It seems that huge micro lags have become extremely rare, occurring roughly every 4-8 minutes or even less frequently. There's been a noticeable change since the start of Crownfall update and updates after.

Micro lags are still present, though. In my opinion, the micro lags were less noticeable a few updates ago compared to now.

I do recommend watching the source video at 0.1x speed. The issue isn't only apparent when I draw on the mini-map but also occurs frequently in other instances. They feel like the hero blinks or freezes for a very short while.

Some examples:

1. First moment
2. Second moment A long series of micro lags.
3. One more random moment

There are are a lot of this type of moments in the video. More than 10 in 43 seconds video.

@kaisenbaka Do You experience something the same to the video in Your issue?

Source video

I would appreciate a method to check or set up My PC to ensure that the issue is not on My side, such as processor-level exceptions. Could anyone provide a useful way to achieve this, please?

[kaisenbaka]

It seems that huge micro lags have become extremely rare, occurring roughly every 4-8 minutes or even less frequently. There's been a noticeable change since the start of Crownfall update and updates after. Micro lags are still present, though. In my opinion, the micro lags were less noticeable a few updates ago compared to now. I do recommend watching the source video at 0.1x speed. The issue isn't only apparent when I draw on the mini-map but also occurs frequently in other instances. They feel like the hero blinks or freezes for a very short while. Some examples:

1. First moment
2. Second moment A long series of micro lags.
3. One more random moment

There are are a lot of this type of moments in the video. More than 10 in 43 seconds video. @kaisenbaka Do You experience something the same to the video in Your issue? Source video

I would appreciate a method to check or set up My PC to ensure that the issue is not on My side, such as processor-level exceptions. Could anyone provide a useful way to achieve this, please?

sorry, but it seems that my issue is slightly different, or that i cannot reach out to someone else's case because it is indeed tiny glitch moment. Also there is patrol function for dota , you can try it, just use hotkey for patrol and click to some place ,and then your hero will automatically move around so that you can observe the issue more comfortably , i actually did the same thing as your video shows trying to record the choppy moment. I recently found that vulkan could produce me better game experience. or that i've decided to live with that. Maybe in the near future i'll swap my budget ryzen 5800+b550 motherboard just for the sake of it lol, Valve seems not caring anymore , last jeff response to this issue was two years ago and he went MIA in that post ever after he asked for ETW trace document. This is nonsense crap that we are experiencing with , same as crownfall patch drop.

[IgorHoruzha]

Today I reinstalled Windows 11(Activated) to a new M2 Samsung 990 PRO PCI-E 4 1tb. Also, I installed new Steam client and Dota 2 on the disk. Heavy lags returned back.

[IgorHoruzha]

Important that there were almost (99.9%) no heavy lags the previous SSD Samsung 960 Evo series 250GB M.2 PCIe 3.0 x4 3D V-NAND TLC and previous Windows 11 when I was playing yesterday.

[IgorHoruzha]

Today, after the update, I noticed that the amount of heavy lags significantly decreased.

Also I again permanently turned off Windows' real-time protection, and heavy lags became even rarer.

Micro lags are still present.

Also, a new effect was found after the update: the game is flickering. I recommend watching the video at 0.25x speed, especially if the game is set to 30 FPS. I'm not sure that I feel it essential if game FPS is the same or higher to monitor refresh rato(280HGz). I currently do not know how to, but I want to record the video with 280 refresh rato. Due to on this refresh rato I feel something wrong sometimes, but I'm not sure that it is the same to 0.25x speed of 60 fps video. But if I set FPS of the game to 30 I all the time see the same to the video 1x speed. View the video here.

Could anybody recomend how to achieve this?