der8auer proves that Intel’s LGA-1151v2, Z370 and Z390 are pointless

[Smffy]

Buying AM4 in 2020 is no different really than buying Intel any time, gotta change it with the next gen

Yeah completely true but just curious what the right cycle is ey... I had an 1800X for a bit but now still on the 8700k so might swap over next year we will see!

 

[Smffy]

It's actually expensive to keep the same socket for multiple CPU generations. AMD tried that with AM4 and thousands of customers bought Ryzen 1 motherboards because they were reduced in price but got a black screen when fitted with a Ryzen 2. This didn't affect customers who upgraded from Ryzen 1 because they could refit their old CPU to do the BIOS update then refit their Ryzen 2. However, first-time buyers had to contact AMD support and request an upgrade CPU which is actually a retail Athlon CPU so they could update their BIOS. AMD tried to save on return postage costs by telling people to keep the Athlon cooler after the BIOS upgrade and to send just the CPU back. AMD had to then send tons of chaser emails because people were slow to return the temporary CPU and some never returned. It would have been cheaper for AMD to change the socket. The whole fiasco also cost motherboard manfactures a lot of money in support calls as well.

This isn’t really any different from people doing the same on older skylake because they didn’t update the bios first... User error really but probably could be made more explicit on the box of next ten boards regardless and on websites to make them aware.

 

[pete910]

snip.

Which has nothing to do with the cost of the actually socket.

 

[Silent_Scone]

I'd imagine there are some legitimate reasons that nessessitate the change that unfortunately don't benefit enthusiasts. On the upshot, there are also legitimate reasons to change motherboard if pushing the envolope far enough (especially with memory), as well as cooling control. Whether one needs these features is entirely down to preference, though.

Not offering backward compatibility no matter the justification is still damning to the consumer, though.

 

[CuriousTomCat]

Which has nothing to do with the cost of the actually socket.

A new socket would have prevented the issue. An AM4 CPU should work in an AM4 socket. People with a lack of experience didn't realise their new CPU and new motherboard weren't compatible without a BIOS update. Clearer labling on the boxes and online retailers like Smiffy said would have helped a lot but some people would still get it wrong. I can see why AMD wanted to do this becasus the old Socket A days were so cool starting with the release of the Thunderbird 600MHz and lasting up until the 5th generation Athlon 3200. I never had any compatibility issues back then because tech wasn't as complex. If the CPU wasn't supported, the POST screen would be unable to display the correct CPU ID but everything was usable.

AMD dont even need to change the socket, just call it a different name like AM4v2 with the boxes and websites labled as such. Then it would appear under a seperate section in online retailers websites so 100% fool proof. Then obviously existing customers can do a bIOS update to update it from AM4 to AM4v2.

 

[Journey]

AMD dont even need to change the socket, just call it a different name like AM4v2 with the boxes and websites labled as such. Then it would appear under a seperate section in online retailers websites so 100% fool proof. Then obviously existing customers can do a bIOS update to update it from AM4 to AM4v2.

However they have gone one better than that, by asking manufacturers to try and ensure that the vast majority new generation boards support the BIOS flashing via USB without a CPU installed at all. Obviously the likes of the B350/X370 can't have that retrospectively added, but it is a feature that will be there now and forever more due to this issue. Keeping the same socket is far more friendly and cost effective for large OEM's who use 10,000's of boards, and it also means that boards do not have to be replaced or redesigned for certain models which is a huge expense for those that make their own boards, and where the end users couldn't care at all about the chipset etc.

 

[Minstadave]

A lot of people keep a CPU for at least 5 years these days so the board would need replacing anyway. Anyone who buys a CPU then replaces it after 12 months can DEFINITELY afford a new motherboard.

It's actually expensive to keep the same socket for multiple CPU generations. AMD tried that with AM4 and thousands of customers bought Ryzen 1 motherboards because they were reduced in price but got a black screen when fitted with a Ryzen 2. This didn't affect customers who upgraded from Ryzen 1 because they could refit their old CPU to do the BIOS update then refit their Ryzen 2. However, first-time buyers had to contact AMD support and request an upgrade CPU which is actually a retail Athlon CPU so they could update their BIOS. AMD tried to save on return postage costs by telling people to keep the Athlon cooler after the BIOS upgrade and to send just the CPU back. AMD had to then send tons of chaser emails because people were slow to return the temporary CPU and some never returned. It would have been cheaper for AMD to change the socket. The whole fiasco also cost motherboard manfactures a lot of money in support calls as well.

Just because people can afford to buy something doesn’t mean it’s legitimate to rip them off. What a strange attitude.

You seem to post opinion as fact too. Where is you source for the AM4 “fiasco” costing AMD more than introducing a new socket? What about all the people that benefitted from a clear upgrade path to Zen+ and beyond?

To me I’d much rather blow several hundred on a high end AM4 motherboard that I won’t have to junk come upgrade time.

 

[MartinPrince]

Is anyone honestly surprised by this? It's been this way for years with Intel.

I thought this to myself almost word for word. Though it doesn't make it any less galling.

 

[DragonQ]

A new socket would have prevented the issue. An AM4 CPU should work in an AM4 socket. People with a lack of experience didn't realise their new CPU and new motherboard weren't compatible without a BIOS update. Clearer labling on the boxes and online retailers like Smiffy said would have helped a lot but some people would still get it wrong. I can see why AMD wanted to do this becasus the old Socket A days were so cool starting with the release of the Thunderbird 600MHz and lasting up until the 5th generation Athlon 3200. I never had any compatibility issues back then because tech wasn't as complex. If the CPU wasn't supported, the POST screen would be unable to display the correct CPU ID but everything was usable.

AMD dont even need to change the socket, just call it a different name like AM4v2 with the boxes and websites labled as such. Then it would appear under a seperate section in online retailers websites so 100% fool proof. Then obviously existing customers can do a bIOS update to update it from AM4 to AM4v2.

All of this would be avoided by including a way to update the BIOS without a CPU installed. Some motherboards already allow this so it's clearly possible, and is far preferable to complicating matters with different comparabilities or simply changing socket every generation.

 

[CuriousTomCat]

All of this would be avoided by including a way to update the BIOS without a CPU installed. Some motherboards already allow this so it's clearly possible, and is far preferable to complicating matters with different comparabilities or simply changing socket every generation.

Yeah I know some of the expensive Asus ROG boards have that feature but I've never seen it on a low end board so I've always thought it was an expensive feature to implement. It would be the best way forward if it was a mandatory requirement for every motherboard.

 

[Dirk Diggler]

In my work in O&G I know we have design limits on currents in cables. Doesn't mean they stop working if you exceed them, just you're "stealing" the performance from some other aspect of the cable, be it insulation life, temperature operating envelope etc.

That's slightly misleading. The design limits on current doesn't just protect the cable, it protects the switchgear and also the load side apparatus. Obviously, the cable can cope with more insomuch as it won't melt etc..

 

[Jokester]

That's slightly misleading. The design limits on current doesn't just protect the cable, it protects the switchgear and also the load side apparatus. Obviously, the cable can cope with more insomuch as it won't melt etc..

But the principle remains the same though doesn’t it?

 

[Dirk Diggler]

But the principle remains the same though doesn’t it?

Yes, you can take an individual component to the brink of destruction, but there are other components to consider along the way. The same is true of this conversion (thread), it's not just about the CPU there's the memory, VRM's, PCI lanes etc..

 

[P4Clock]

On the launch of z370 msi went on the record saying the new CPU's would have worked in last get boards.

 

[easyrider]

I very much doubt my Asus Z170 plus would have clocked my 9900k to 5GHZ on all cores...hell, The Asrock s390 extreme 4 I returned coudn't

 

[DragonQ]

I very much doubt my Asus Z170 plus would have clocked my 9900k to 5GHZ on all cores...hell, The Asrock s390 extreme 4 I returned coudn't

Maybe, maybe not, but this isn't about overclocking.

 

[easyrider]

Maybe, maybe not, but this isn't about overclocking.

Well considering the 9900k is about overclocking your point is moot

 

[hominid]

The Pin itself can take 5 or so amps not a problem at all, that's all good. The problem is that pulling 2-3 amps the part that will be at risk is the tip of the pin that touches the CPU and the pad on the CPU itself. A curved surface of <1mm² will heat up fast when you ramp the current over 2-3 amps, the curved tip of the pin is what need to be tested more than the pin itself as it'll be the point of failure most of the time.

He also did no long term testing, one day doesn't cut it, and with a sample set of one - not very good testing methodology.

 

[JCLDJB]

Who knows what the long term consequences are of running higher currents into the CPUs are though? Though I guess probably negated by hardly anyone running CPUs flat out continuously.

In my work in O&G I know we have design limits on currents in cables. Doesn't mean they stop working if you exceed them, just you're "stealing" the performance from some other aspect of the cable, be it insulation life, temperature operating envelope etc.

If I remember rightly Der8auer does mention something about this (how he isn't sure what the max tolerances/loads are supposed to be on each pin, and what would happen over time).

But even with that unknown, for the CPU to work at all, after blocking the use of that many pins, goes to show that those pins are pretty tough.

It could also be due to the fact that they have a couple of, yet, unreleased CPUs they are saving that do require way more power, for that socket.

But either way it's looked at, for me at least, this is just Intel being Intel...just trying to please shareholders.
Think of it like this, for a second:

(If you had the choice to....tell an investor you have a new CPU and can guarantee X number of CPU sales // or // tell an investor you have a new CPU and can guarantee X numbers of CPU sales plus the same number of new boards) which would you choose?

Watching Der8auer's video does make AMD's AM4 socket just that little bit sweeter now..if it wasn't already sweet enough.