This was a thing early in the lifecycle but didn’t know the slow boot is still a problem on some boards. My launch x670e taichi boots my 64GB kit just as quickly as the 32 it replaced, few seconds after the long initial one time training. This is with later BIOS, think they’re up to agesa 1.1.0.0.The 2 open box B650E-F ROG STRIX boards I bought appear to work fine with the 7900x. Both boards take forever to post (55 sec even after initial training) with the 64 GB kit and any of the 3 EXPO profiles loaded. It appears that the memory trains on every post attempt with this config. Seems like a pretty common thing on AM5 with 64+ GB kits after looking into it. I'm sure it was probably discussed in this thread at some point. Everything works fine after post but It's really annoying if you reboot a lot to play around with stuff in the BIOS. This is a non issue with my 32 GB kit running EXPO after the memory is trained. It's also a non issue if don't load EXPO on the 64 GB kit. I may return the 64 GB kit and run the 32 GB. I probably don't really need 64 GB of RAM anyway. I'll test the 64 GB kit for good measure in the 1st B650E-F I bought new a few weeks ago but I'm pretty confident it'll do the same thing and the open box boards are good to go.
The ROG Strix B650E-E from Amazon warehouse showed up today. I inspected it and it looks like it was never even installed in a system. None of the accessories are even touched. Fingers crossed it tests good so I can get my water cooling build back up and running soon. It's been colecting dust for over a year.
I vaguely remember reading about people having this problem early on. The latest BIOS I tested with is using 1.0.8.0. Perhaps there's hope for the 64 GB kit with another BIOS update. Thanks for the heads up.This was a thing early in the lifecycle but didn’t know the slow boot is still a problem on some boards. My launch x670e taichi boots my 64GB kit just as quickly as the 32 it replaced, few seconds after the long initial one time training. This is with later BIOS, think they’re up to agesa 1.1.0.0.
Just had this happen 5-10 times in a row, and it actually corrupted my windows bcd and I had to reload that from a macrium backup. I did a cmos reset and it seemed to improve (but not fully fix) it. Maybe this particular board is a dud and I should rma it, but it works fine once booted. I'm using the latest 7E12v14 bios and don't recall this being such an issue when I got the board, but they have that vsoc problem.I still get a lockup on warm boots 50% of the time, even after many bios revisions (I'm on the 1.0.0.7c bios). Not sure if the board is the issue. I kept the memory retraining on and am running it at stock speeds with expo. It gets stuck with the cpu and mem lights and usually the next cold boot fixes it, but sometimes it takes multiple cold boots to work. This is mainly a problem during windows updates and I rarely reboot the pc otherwise, but it's still annoying.
I think your problem is your running from an image. Try doing a clean install and see if that fixes your problems. Imaging software can create a ghost in the machine affect.Just had this happen 5-10 times in a row, and it actually corrupted my windows bcd and I had to reload that from a macrium backup. I did a cmos reset and it seemed to improve (but not fully fix) it. Maybe this particular board is a dud and I should rma it, but it works fine once booted. I'm using the latest 7E12v14 bios and don't recall this being such an issue when I got the board, but they have that vsoc problem.
If you think it's a bios issue. Then reset the bios to default settings. Then see if it boots normally after a few tries.The windows install works fine. The problem is something with the bios and post sequence, as it works fine if it gets to the post screen, and I get it even if no OS drive is installed. I never do a clean install since it's a huge amount of work.
Have you run Memtest86? It sounds to me like your memory might not be stable running EXPO. I always find it good practice to run memtest86 before I boot up into an operating system on a fresh system for the first time. It's not the be all end all of memory stability tests but it usually reveals blatant memory issues pretty quickly.Just had this happen 5-10 times in a row, and it actually corrupted my windows bcd and I had to reload that from a macrium backup. I did a cmos reset and it seemed to improve (but not fully fix) it. Maybe this particular board is a dud and I should rma it, but it works fine once booted. I'm using the latest 7E12v14 bios and don't recall this being such an issue when I got the board, but they have that vsoc problem.
I've got the same board, have been messing with it on my test bench for a few weeks now. I've got the 2x16GB G.Skill 6000C30 expo kit.I'll try that out, it's using the G.skill CL30 6000 pack. Maybe expo is not stable, but it doesn't seem to have any issue after it boots. I've used it extensively for games and other things.
I'm a newb on AM5 but this is 100% consistent with my experience on AM4. Running SOC voltage too high can most definitely cause instability on AM4 CPU's. I immediately turned down the SOC and mem controller voltages when I started playing with AM5 based on my AM4 experience. AIB's are generally way too aggressive with auto voltage IMO.These may or may not have anything to do with your intermittent issues, but it seems like on ryzen platforms, high voltages can cause instability just as low voltages can.
I think it is the USB not being bullet proof. I have 2 computers I work on in the office (1 Intel, 1 AMD) through el-cheepo KVM, and I get intermittent USB issues on both of them switching back and forth.It's not funny at all that AMD can't figure out perfect USB compatibility after this long.
And the first thing that happened after update was that my keyboard was not turning on, and just unplugging an reinsert didn't work. Had to reboot :/It's not funny at all that AMD can't figure out perfect USB compatibility after this long.
It does seem to have something to do with the expo speed or voltages. I got it to boot reliably without expo on as long as I don't set expo after the cmos reset. There is a new beta bios for this board, I may as well try it out, even though I generally avoid the beta ones.I've got the same board, have been messing with it on my test bench for a few weeks now. I've got the 2x16GB G.Skill 6000C30 expo kit.
I have it running 6400C30/2133FCLK with tightened timings and it is rock solid. A little slow to post, but that's the only complaint.
I did notice when you enable EXPO it sets some voltages a little aggressively, which I've toned down based on the values that have been working on my Gigabyte board. Particularly VSOC it set to 1.3V (!) I set to 1.2V for 2000/2066 or 1.25V for 2133. DRAM VPP it set to 1.95V, I set to 1.8V. And CPU VDDIO it set to 1.4V, I set to 1.1V. Lowering these voltages ended up causing the automatic voltages for VDDG and VDDP to lower themselves, as well.
These may or may not have anything to do with your intermittent issues, but it seems like on ryzen platforms, high voltages can cause instability just as low voltages can.
You could set it to manual and configure all of the timings and voltages yourself.It does seem to have something to do with the expo speed or voltages. I got it to boot reliably without expo on as long as I don't set expo after the cmos reset.
Is it possible to enable expo only in ryzen master or something, after it boots into windows, and keep it off by default?
There is a new beta bios for this board, I may as well try it out, even though I generally avoid the beta ones.
Hot take: It’s the user base that is at fault for that. There aren’t enough complaints about it and if you do bring it up you’ll just get gas lit over it. If 1 out of 8 customers building an AM5 system doesn’t have a trouble free experience and has to tinker to get things working that should be addressed. An issue like that would get solved if it were an Apple, Intel or Nvidia product.It's not funny at all that AMD can't figure out perfect USB compatibility after this long.
Guess AMD has put their internal validation team on the enterprise products and cheaply outsourcing client validation to third world countries. If people were returning their AM5 mobos in droves, they would take it seriously but apparently, the AMD loyal crowd is too patient and too excited to use their products to care about annoying quirks.To be honest the whole AM5/X670E platform feels like a beta product.