right I remember that being a thing.No idea. It was just a guess, judging by the reported decrease in temps. The "secure" frame, as it is called, is supposed to mainly prevent the thermal paste from getting into the notches of the heatspreader but I guess it improves the thermal properties somehow by giving the paste a larger area to spread on.
I've no idea. I'm on a slightly older Intel platform myself that doesn't have this issue.Is it confirmed that this is an AMD vs Intel issue? If so, what's the reason AMD based PCs take longer to boot up? Seems strange that AMD PCs would take longer too boot up. What would be the technical root cause for that?
Or is it an AM5/Zen4 specific issue (so the problem does not apply to earlier AMD sockets or CPUs)? In that case, may BIOS updates be able to improve the boot-up times?
Have to get into Windows quick to waste time on these forums. Mission critical.the what now?
some people want to hop onto their chair first thing in the morning and get to 'puting.
The boot times are longer than Intel but so what. Go make a cuppa tea or coffee.
What MB, CPU and RAM are you using? Any other special tricks to get it down to 15 sec for cold boot from power on button until Windows login screen?After training my memory I've disabled memory training and boot times are around 15 secs.
It is all in my signature.What MB, CPU and RAM are you using? Any other special tricks to get it down to 15 sec for cold boot from power on button until Windows login screen?
So was all you changed to get the boot times down to 15 seconds to disable memory training on your ASRock MB? And you’re counting seconds from pressing power on button until Windows login screen is displayed, when doing a cold boot?It is all in my signature.
Those times are old and not comparable to the newer bios with agesa 1.0.0.7B which has a more extensive memory training and better support for high frequency memory.So was all you changed to get the boot times down to 15 seconds to disable memory training on your ASRock MB? And you’re counting seconds from pressing power on button until Windows login screen is displayed, when doing a cold boot?
Also, how many seconds did it take to boot up before you disabled that? For comparison the ASRock B650 MBs in the review I linked to in my previous post took between 37 to 43 seconds.
Oh it definitely helps. My chip rarely gets above 85C anymore. It also keeps the paste out of the notches. It replaces the existing socket latching mechanism, so I am curious if that has anything to do with it.No idea. It was just a guess, judging by the reported decrease in temps. The "secure" frame, as it is called, is supposed to mainly prevent the thermal paste from getting into the notches of the heatspreader but I guess it improves the thermal properties somehow by giving the paste a larger area to spread on.
You can enable memory context restore in BIOS settings. Just disable it and reboot if you need to change memory timings/speeds.What MB, CPU and RAM are you using? Any other special tricks to get it down to 15 sec for cold boot from power on button until Windows login screen?
Keep memory training enabled and always keep memory context restore disabled.So was all you changed to get the boot times down to 15 seconds to disable memory training on your ASRock MB? And you’re counting seconds from pressing power on button until Windows login screen is displayed, when doing a cold boot?
Also, how many seconds did it take to boot up before you disabled that? For comparison the ASRock B650 MBs in the review I linked to in my previous post took between 37 to 43 seconds.
Excellent. That means the chip is hitting the max frequency of 5.85 GHz without getting to TJmax of 95C. Very nice!Oh it definitely helps. My chip rarely gets above 85C anymore.
Hmm from what you've said my guess is that the template forces the heatsink to lay even more flush to the board and processor ihs through probably deflexing the board a little thanks to the template providing a more rigid area.Oh it definitely helps. My chip rarely gets above 85C anymore. It also keeps the paste out of the notches. It replaces the existing socket latching mechanism, so I am curious if that has anything to do with it.
Also, if its a tight fit, which I suspect it is, it could be transferring heat from the side of the IHS to the heatsink. There is your 5c alone.Hmm from what you've said my guess is that the template forces the heatsink to lay even more flush to the board and processor ihs through probably deflexing the board a little thanks to the template providing a more rigid area.
Keep memory training enabled and always keep memory context restore disabled.
Take solace in knowing a 40 to 50 second boot time will be stable for the hours you'll be using it afterwards.
Even if there is a gap the heat transfer would still happen due to how physics works. The z4 ihs is already a hefty piece of metal and that adds to the heft. it all makes sense. my memory of the launch products was that there was only a thin plastic plate that sat within the latch and covered the notched out areas to prevent leakage. Or my memory really isn't as good as I think it is.Also, if its a tight fit, which I suspect it is, it could be transferring heat from the side of the IHS to the heatsink. There is your 5c alone.
That is not true. Memory training can change between boots for various reasons.Why would you want to do that? Are you using some shaky/unstable RAM, or using extreme overclocking of it?
Otherwise, when the memory training has already been done successfully once, why not rely on the memory context restore function after that like others here suggested?
He probably wants to get on the anandtech forums fast.I should be asking you: Why is a sub 10 second boot time so important to you?
Is there any statistics to prove that a freshly trained system is more stable, than one using context restore?Keep memory training enabled and always keep memory context restore disabled.
Take solace in knowing a 40 to 50 second boot time will be stable for the hours you'll be using it afterwards.
The babes sectionHe probably wants to get on the anandtech forums fast.
Buildzoid:Is there any statistics to prove that a freshly trained system is more stable, than one using context restore?
If nothing has changed in the system, shouldn't the values for the memory be close to identical every time you train the memory? So what do get from training each time you boot?
very interesting. I did not know most of that and had presumed amd couldn't figure out why it was taking so long but I'd not heard many in the intel camp complain of longer than prior boot times either probably due to the low sales 13th gen has gotten. Would this ever speed up in the future or is this the new normal since we've waded into higher speed memory territory?Buildzoid:
And more reasons further in the discussion.
It's not something that is done for no reason.
Thx.Buildzoid:
And more reasons further in the discussion.
It's not something that is done for no reason.
Did we read the same comment by him?Thx.
He does says if the the training procedure is good enough you wouldn't need to retrain. So is the AGESA 1.0.0.7B good enough? Only way to find out is to enable context restore. And so far it seems perfectly stable for me
Further down:Did we read the same comment by him?
"The requirements for these may change slightly from reboot to reboot. So if the memory controller trained a set of settings that were borderline they may not be good enough if you try to apply them the next boot."
Although I am not sure how or what infuelnces minor changes from boot to boot that would trigger a retrain.
I should be asking you: Why is a sub 10 second boot time so important to you?
I don't have the immediate need to use my PC as I do my phone. I'd rather have the piece of mind knowing the memory training at boot is going to make whatever PC gaming session I have problem free. I can wait an extra 20 to 30 seconds.I should be asking you: Is a 60 second startup time ok for you on your TV, car, mobile phone, etc?
And why not have it quicker if possible? Why do you want the startup time to take longer than necessary?