cancel
Showing results for 
Search instead for 
Did you mean: 

Is This The Best Time to Upgrade?

Shenny
Level 11

Hi, guys

Currently, I am on Ryzen 9 5900X running on Asus ROG Strix X570-E Gaming. Next week I was planning to upgrade to ROG Strix X670E-E Gaming Wifi + Ryzen 7 7800X3D because now there is enough supply of those parts in my country and the prices dropped to quite reasonable levels, speaking of the mobo. I tend to upgrade my PC every 3 years since for me this is the sweet spot where I can do the upgrade and still get good money for the old parts, and pay off at least 30% of the new purchase. If  I wait for 6 months more, probably I will sell them for much less.

However, what I am reading in this forum frustrated me a little bit 😂 - slow or irregular boot times, unrecognized hardware, BSODs, memory kits from QVL list not working properly with EXPO enabled and so on. Still, this would not have stopped but now the news came about exploding X3D chips for which AMD EXPO profile is suspected to be the culprit, for example:

https://www.techpowerup.com/307742/asus-releases-official-statement-regarding-ryzen-7000-issues?cp=2...

This is an open forum so I would be grateful to hear your honest opinions whether you see risk in upgrading now or better wait for 1-2 months until the dust around this issue settles. I can live up with slow boot times as long as everything works stable. Thanks in advance for your opinion!

941 Views
13 REPLIES 13

Murph_9000
Level 14

Personally, I'd give it a week or two right now, until there's some clarity on the CPU meltdowns.  I think people are misinterpreting what ASUS said about it, as far as EXPO goes.  It's almost certainly a Vcore-related problem, as that's the only power bus that has the power needed to do the physical damage observed.  EXPO and SoC voltages don't really have enough power delivery behind them to do it, and I think that was just ASUS going a little broad in what they were saying to include plans to add additional safety to all of the power parameters (i.e. not just Vcore).  Given that the problem seems to be across all board vendors, it's likely either something in the CPU hardware or something in AGESA; unless all of the boards in question happen to use exactly the same VRM chips or something like that.

DDR5 is still relatively new, and almost certainly going to get better as it matures.  The slow boots (which seem to be a DDR5 thing) are kinda a bit of a nothingburger, in my opinion, and any issues with that are likely to resolve over the coming year with BIOS and AGESA updates.  A 60 second POST isn't exactly crippling the primary usability of the system.  Fast boots are a nice thing, but it's not a big hardship to hit the button and have a login screen in around 90s vs 30s, as you shouldn't normally need to reboot frequently.

Some of it, I think, is possibly the memory vendors being a bit overly aggressive/enthusiastic in setting their overclock speed and timings in the XMP/DOCP/EXPO profiles.  This is an area where DDR5 maturity comes in, as they will hopefully learn in time how far they can push the DRAM silicon while keeping broad compatibility.  That's not to say I think it's all on the memory vendors, just some of it; BIOS and AGESA updates will probably also improve things, as AMD and the board vendors learn how to get the most out of it as well.

We've just gone from DDR4 being a very mature technology, where one-click overclocking pretty much just worked with only some simple-ish rules (like you can't go super-fast with 4 DIMMs or 32GB DIMMs, and ultra-fast was really the domain of 2x8 configs; for easy one-click overclocking, at least).  Life was relatively easy for RAM towards the end of the DDR4 era.  Now we start over, with some significant architectural changes between DDR4 and DDR5, and it's going to be a while before it's as easy as it used to be.

Personally, I'm glad I built my system late in the X570 era.  I knew the new platform and CPUs were coming, but there's always something new coming.  As much as I'd love a 7900X or 7950X, I'm extremely happy with my 5950X-B2 on X570S for now, and will probably wait for X770.

Murph_9000
Level 14

I've been reading some of the latest analysis and information on the 7000 meltdowns, and it looks like I was both correct about Vcore doing the damage, but incorrect about it being the root cause.  It's looking like SoC voltage is the likely initiator of the failure (possibly driven by EXPO).  The failure scenario seems to be that SoC voltage gets too high and kills the CPU thermal management, leading to a Vcore-powered meltdown that produces the physical damage that's reported.  The new precision boost behaviour where it pushes itself up to 90/95C and holds it there by feathering the boost may be a contributing factor once the SoC is damaged, as the chip will just meltdown without the active thermal control from the SoC.

Either way, give it a week or so for the BIOS with the EXPO and SoC voltage limits to get out of beta, and you should be safe.  You'd probably be safe on the latest beta BIOS, or with all overclocking disabled (particularly XMP/EXPO).

7800x3d/asus x670e-e user here, and, them "pulling" some bios updates from the site, smelled really really squishy to me.  I think after Steve at GN gets done, ASUS will have a ton of explaining to do. Just me. And what im seeing. Calling a simple EXPO setting a "over clock" to save your ass, is bs, and also, having something in that bios that will bubble a cpu and pins on the mobo, from one simple click, is bs. But, im holding judgement till we learn more. But, i feel like, i already know where this may be headed. That said, it sounds like other board mfg's are in it, as well as not just stacked vcache chips, so, time will tell. But, the pulling of bios files from their site, and not even explaining why,..... already feels weird. Time will tell. 


@Koyote7667 wrote:

… Calling a simple EXPO setting a "over clock" to save your ass …


I don't think that's entirely fair.  The O in DOCP and EXPO is literally "overclocking".  All XMP, DOCP, and EXPO is overclocking.  The default JEDEC speed (with XMP/DOCP/EXPO disabled) of DDR DIMMs is 100% of the rated speed of the silicon as far as the manufacturers of the DRAM chips are concerned.

Pulling the older BIOS files after releasing one with additional safety fixes makes a lot of sense to me.  They can see the potential for a lot of RMAs, and quickly doing that is a simple step to try to limit the scale of the problem.

It's too early to properly apportion blame between AMD and the board vendors.  One thing is clear, however, it's not just ASUS, as the reports include the other board vendors too.  If ASUS did make some mistake in the BIOS, the other vendors also seem to have made the same mistake.

So, everyone in the ROG, and ASUS world, highly recommends using their "expo" setting. "its made for grandma to use". One click, and your done. Ill paste about 15 videos from the ROG boys themselves if you need, which im sure you dont.   Also, when you do, you get ZERO prompt of "hey this may ruin your ******".  Not like in tweakers paradise, nor in a ryzen master change or the like. They make you hit a "ok, im in" button. And dude, every swinging ****** on the planet, uses the expo setting.  The EXPO setting says "optimized by asus". Im one of the biggest ROG d-riders i know.. but come now. Also, if its on us,.. then why drop the bios files to use?  They could easily just say,.. welp, its "over clocking". Get over it. Come on dude. 

Also, pulling the older bios files, if they are going to ruin your gear, Hell yes thats a fantastic thing. (see where im going here?? )  But, NOT saying a ****** thing, a word, a anything from why? You wont find a person on this rock, that thinks "oh, well, ... ya, thats cool" . Not a one dude. Not a one. 

 

I am with you on the "opps, are bad" thing, asus, or anyone else. Im with you there. It also didnt even seem to be a thing, until the stacked variants dropped, Although Der8auer has non vcache chips that are bubbled, and your correct, from other MOBO makers.... I feel you there balls to bones. 

Oh yeah, XMP/DOCP/EXPO is certainly pushed by ASUS and all the other board vendors.  And yeah, it's supposed to be safe and easy (if your memory config is on the QVL).  It's certainly not supposed to have a significant risk associated with it, and you're correct about the complete lack of warnings associated with it.

By saying XMP/DOCP/EXPO is overclocking, I'm just trying to objectively say that's what it is, not that it's people playing with fire (or it's certainly not supposed to be).  I'd be very surprised if either AMD or any of the board vendors try to use overclocking as an excuse to deny RMAs over this; that would be potentially disastrous PR for them, given how common/standard it is.  Hopefully ASUS were just using the term as a statement of fact, rather than lining up an excuse for why it's not their fault or to push blame onto customers.  There might be the odd bump in the road, but I think they will do the right thing at the end of the day.

I just think it's healthy for the community at large to remember that it actually is a form of overclocking, no matter how low the risk should be.  That can be particularly important for people who need stability more than they need the last drop of performance, or professionals who need to worry about rare memory errors (and who should really be using a workstation with ECC RAM, but there's a trend for some of them to use high end gaming rigs because of the sheer power of the 9-series CPUs at a fraction of the cost of a workstation).

Sure. I feel you. Im picking up, what your putting down.  That said, im still running epxo1, pbo enhanced, and im still smashing frames (and still getting dunked on in games lol) 

For me,... this rig is running like a top. I think, like most things, it gets way over blown, everyone loves negativity, and uses it for "clicks" on their channel.  Cheers mate. 

Oh, and heres a funny one. (since you mentioned it)

Just because, i went into bios, disabled expo, and..... no post. Board showed Od. Just by disabling expo. Nothing more. I completely powered down, and restarted. Same. Cleared CMOS, and it posted. There is some weird ass ****** going on atm mate..... trust. Also, cheers. 

x670e-e/ddr5 6000 cl30/7800x3d bios 1202

Shenny
Level 11

Thank you very much for the detailed analysis! I can wait for a few more weeks to see how things will unfold, no rush.