For what it’s worth, I captured a memory dump at the start of a frozen Greater rift and the windbg !analyze -v said the following:
*** WARNING: Unable to verify checksum for fmodex64.dll
I realise it’s not very interesting but without any debug symbols I can’t do much more. 
This effectively solves the freezing problem!!
Like the others, I see progress on the FMOD issues. I have been unable to reproduce them after the 8/6 PTR patch as I can on the live US server. However, at least for some of us this is not the only issue causing freezes. File sizes exceeded the limit for pastebin. The included WinMTR data is to the PTR server based on it’s IP within d3debug.
PTR d3debug.txt: PTR_ClientSocket_disconnect_8-7-21@2130
https://drive.google.com/file/d/1jOgESe3XmD7kIwIH8v9hEfXuM7KpIIox/view?usp=sharing
PTR d3debug.txt: PTR_ClientSocket_disconnect_8-8-21@0021
https://drive.google.com/file/d/1sK7pdUd-sXP_WGvgIrZPdtv3prSrig3c/view?usp=sharing
WinMTR_PTR_capture_on_disconnect_8-7-21@2130.txt
https://drive.google.com/file/d/1n_y1t2z0rWthbjE1RyP-SdXyoQucAWqH/view?usp=sharing
WinMTR_PTR_8-7_to_8-8_4hour_capture.txt
https://drive.google.com/file/d/1KHIfMQVkhcYqXrEqug0zHS21KheHL_sM/view?usp=sharing
TLDR for these two d3debugs is
|2021.08.08 00:21:04.269190800|ClientSocket Disconnect|
|2021.08.08 00:21:04.269327400|UIStartup_ClientDisconnected msg=The game connection has been lost: your client has been disconnected from the server.|
…a network disconnect. Watching WinMTR (output attached), latency spikes at the time of the freezing, starting with the first hop into the Blizzard infrastructure and becoming quite severe (I observed up to 900+ms worst case in my attached 4 hour sample, during which I experienced disconnect events twice) in the second hop.
Between me and the Blizzard/Equinix demarc, my worst sample over 4 hours was 22ms with an average of 3ms. During these issues, I’m able to continuously access resources from the rest of the Internet without issue, including continuously streaming media for which a disconnect would be immediately noticeable.
The connectivity issues that I’ve experienced can present in at least 3 different ways in d3debug depending on the length and severity of the spike:
No error logged at all.
Client side: Everything locks up exactly like the FMOD issue, then resumes after 5-15 seconds-ish.
recv() failed
OR
ClientSocket Disconnect
Client side: After locking up for a time, the client is forcibly disconnected after timing out.
I can reproduce these issues periodically just by standing in town on the PTR or on the live server and periodically preventing myself from going AFK in game. Be advised it can be 2 hours or sometimes even more between incidents. Other times it may be just minutes.
As an experiment, I tried running both the live and PTR clients at once. During the network issues I experienced, sometimes one was disconnected and the other survived. Other times, both were hit with a freeze or disconnect. Just depends if the client was lucky enough to survive the latency spike.
I really appreciate you guys finally taking this issue seriously and communicating with us. Please pass on our thanks to the engineers and devs. I hope this gives you some more items to consider.
Edit: Here’s another WinMTR demonstrating a 4399ms worst sample latency at that same device ae1-br02-eqla1.as57976.net
WinMTR_capture_on_disconnect_8-8-21@0225.txt
https://drive.google.com/file/d/1du-DQbiLAL16GQTdrjFhEGASDyB8xxDA/view?usp=sharing
Because I’m used to play like that, in previous seasons that wasn’t such problem for me, but in the season 23 and 24 it became worst thing to get…
The deep freeze seems to happen very frequently on me when playing (or playing with) a Firebird Mirror Image Wiz on wifi network
I have an identical PC as my wife’s and we play side by side; The only difference is that I use wifi and she uses wired network. when playing with another friend who uses a firebird wiz, I have significant lags and freezes while she doesn’t. I also had bad freezes very frequently when I was playing firebird myself.
Hope that helps a bit for you to reproduce this issue.
Hey everyone! This weekend, we received so much valuable information from the files and feedback on the Freeze issue. As of this morning, the team is preparing to jump into “Phase 3” of the PTR Pop-up and publish another build update with the intent to improve the current issues we’re reading into now.
At the moment, players can pause from sharing feedback and stay tuned for updates on the new build version coming. We’re very grateful for all the support and will keep you posted.
Happy Monday! Thank you for the update. Good luck with the new build, I’m excited for phase 3.
Finally, even if it was late, but I’ve done some testing (couldn’t done it earlier)… For now, it was acceptable lagging, FMOD sound ready after: was normally at 20ms, but at lag spikes it was from 100 to 200 ms. But for certain, those 200ms wasn’t able to kill my HC char, as I could still battle, but all those lag spikes was due to the Elites and other bosses (yes, I thought about that connections, as I usually died in such way), the more elites were at the same time, the bigger was lag spike, but in rare cases it could be due to the only one elite…
Edit: let see, what changes will be in the Version 3, as I’ve prepared for the full test (to the limit)…
The latest PTR pop-up build has been updated with improvements to address the freeze issue. We’re looking for more feedback from folks who encounter the issue and ask to share your D3Debug.txt files via pastebin on this thread.
Looking forward to hearing on how the build is performing and I’ll be chiming in for new updates to share. Hope you’re having a wonderful day!
you’re so close. 110 ms freezes (more like little jerks) and never more than 2 in a row.,…
Phase 3 ~100ms FMOD Error sound not ready - micro freezes Starting line # 1669
Whirlwind Barb PTR GR120
Phase 3 ~100ms FMOD Error sound not ready - micro freezes Starting line # 848
Whirlwind Barb PTR GR110 Freeze #2
Getting the same 100ms mini freeze others are. So close to fixed! Great job so far.
D3Debug - Inna Monk in Greater Rift
D3Debug - Whirlwind Barb in Greater Rift
D3Debug - Whirlwind Barb in Greater Rift #2
D3Debug - Roland Crusader in Greater Rift
Settings
Phase 3 ~100ms FMOD Error sound not ready - micro freezes - line # 967 & 2049
Monk Tempest Rush with Patterns of Justice
Two GR110’s back to back
Phase 3 ~100ms FMOD Error sound not ready - micro freezes - multiple freezes
Firebirds Wizard
Can’t run with Convention of Elements on at all have to remove before playing or game micro freezes all the time!
Phase 3 about 100ms FMOD sound ready several micro freezes.
Tal Rasha build Wizzard
Conventions of Elements is always on…
Much better then Phase 2 (link must be copied to browser)
https://pastebin.com/embed_iframe/ji9cCiCp
yes, same here.
a lot of FMOD errors now, but all in the range 100ms to 120ms. No hard freezes.
Even Hardcore is playable now.
Make a hotfix, let us ALL test this.
Now!
hi, your log says you are running Windows 7/64.
???
Make the step to Windows 11! take your W7 key and all will be fine.
@Boing, that would only be true if the computer has either TPM 2.0 or the firmware version of it. Along with a CPU that is on the list of CPUs that can handle Windows 11. Plus AFAIK Windows 11 isn’t out till the end of the year.
Phase 3 results
PTR_d3debug_8-9@1700_to_8-10@0330.txt
https://drive.google.com/file/d/1cX7qJjlJN_DqC6-YiCO95NW7OBvwEkDs/view?usp=sharing
WinMTR_long_capture_8-9_to-8-10.txt
https://drive.google.com/file/d/1LQ8C2_P6JBvU-WhukdAnQuKiAGxPnx4I/view?usp=sharing
D3PrefsPTR.txt
https://drive.google.com/file/d/15_nhyR7q4Sk2mElFtMiEMIcOVTLsGzdc/view?usp=sharing
I changed my channels from 16->32 and mode from mono->stereo sometime during today’s test.
Logged continuously for ~10 hours
“FMOD sound ready in” 3ms - 80ms
I noticed the sound stack appears to have reinitialized itself a couple of times in d3debug after message: Releasing FMOD sound with incorrect open state and loading screen disabled. In full disclosure, I had two clients open, the live US client had it’s sound disabled and the PTR client did not. Also made the above mentioned changes to my settings. Did not see this in round 2 despite the same setup.
ae1-br02-eqla1.as57976.net continues to barf latency >1000ms occasionally (worst sample in period 4718ms). Client performance suffered during the spike I watched happen firsthand around ~2:20am, but did not disconnect
Minor but non-zero packetloss today. Lost 1-2 out of 50,000. Lost 5 to above device (0.01%).
PTR client stayed connected for ~10 hours without a disconnect
…but so did a live US client, so I can’t offer any conclusions.
I did a few rifts without major issues, but spent most of the day background monitoring in town since my worst issues seem to be independent of what I’m doing.