asus confirms coffeelake could have worked in z270

[DarrenM343]

Wake me up for Z390+8 core and hopefully dump the igpu in the top tier CPU core but I doubt we ever see solder again.

I'm not an expert on this stuff but I question whether solder is needed and even optimal now. Aren't TR CPU's soldered? 7900, 7920 etc seem to have much larger overclocking potential with paste.

 

[Troezar]

I'm not an expert on this stuff but I question whether solder is needed and even optimal now. Aren't TR CPU's soldered? 7900, 7920 etc seem to have much larger overclocking potential with paste.

I would say that's a limit of TR design rather than paste now being better than solder. Historically Intel have usually had higher clocks than AMD.

 

[TallPaul1878]

I'm not an expert on this stuff but I question whether solder is needed and even optimal now. Aren't TR CPU's soldered? 7900, 7920 etc seem to have much larger overclocking potential with paste.

I'd say no, since once Intel CPUs have been delidded and some Thermal Grizzly Liquid Metal applied it effectively cures all of the heat related problems.

What is interesting is that there is a growing market for pre-binned and delidded chips from brand new. Surely Intel have noticed this but would they be willing to produce another SKU that is effectively a "delidded at factory" version and under manufacturers warranty?

 

[CAT-THE-FIFTH]

I would say that's a limit of TR design rather than paste now being better than solder. Historically Intel have usually had higher clocks than AMD.

Its a limit of the process node used. GF licensed a 14nm node optimised more towards mobile devices,and IIRC the optimal clockspeed for Ryzen is around 3.3GHZ~3.5GHZ it appears:

http://i.imgur.com/8Rch6JF.png

 

[DarrenM343]

I would say that's a limit of TR design rather than paste now being better than solder. Historically Intel have usually had higher clocks than AMD.

With the 7900, 7920's etc that seem to run hot capable of upwards of 4.5 (albeit not all of them), I do wonder how much further headroom there would be with solder rather than paste?

 

[Minstadave]

I would say that's a limit of TR design rather than paste now being better than solder. Historically Intel have usually had higher clocks than AMD.

Absolutely, TR is on a process not optimised for clock speed. It’s that which is limiting clocks rather than thermals.

 

[Troezar]

Absolutely, TR is on a process not optimised for clock speed. It’s that which is limiting clocks rather than thermals.

That's what's getting me very interested in future revisions of Zen. If they can compete with Intel on clock speed it could get very interesting. I'm under no illusion that Intel will likely always have the outright performance crown but I'll take 85% of the performance for 60% of the cost. (Figured pulled right out of somewhere, but you all know what I mean!)

 

[Minstadave]

That's what's getting me very interested in future revisions of Zen. If they can compete with Intel on clock speed it could get very interesting. I'm under no illusion that Intel will likely always have the outright performance crown but I'll take 85% of the performance for 60% of the cost. (Figured pulled right out of somewhere, but you all know what I mean!)

Zen doing even 4.5Ghz would be very tempting.

 

[mistersprinkles]

Am I right in saying that AMD have promised to maintain socket compatibility for the next few years at least? So Ryzen 3 (hypothetical example) will work with a motherboard released for Ryzen?

Ryzen 3 is already out and it works with AM4. Getcho faks straight

 

[Bluecube]

Ryzen 3 is already out and it works with AM4. Getcho faks straight

You mean the 3 series? It is indeed out. However in the context of the discussion it was obvious I meant the third iteration of Ryzen. I apologise most insincerely for not making that clearer.

 

[southernorth]

Watching Linus talk about it here:

https://youtu.be/0qRpRITk6k4?t=51m52s

Has made me think that the likelihood of Z370 supporting 9th gen seems more likely now. As he says, if they were truly just screwing everyone over, why would they both altering the pins/power delivery at all? Surely they could just push a BIOS change and nothing else. Also, Intel has supported 2+ generations of CPUs on the same motherboards for the past 10 years. This extra power delivery could mean the speculated 10nm 8-core 9th gen CPUs might work on Z370.

 

[CAT-THE-FIFTH]

Watching Linus talk about it here:

https://youtu.be/0qRpRITk6k4?t=51m52s

Has made me think that the likelihood of Z370 supporting 9th gen seems more likely now. As he says, if they were truly just screwing everyone over, why would they both altering the pins/power delivery at all? Surely they could just push a BIOS change and nothing else. Also, Intel has supported 2+ generations of CPUs on the same motherboards for the past 10 years. This extra power delivery could mean the speculated 10nm 8-core 9th gen CPUs might work on Z370.

Then he needs to ask the question of why there is a Z390?? So what exactly does the Z390 offer over the Z370?? Remember both have the same 300 series naming scheme.

Also why would a shrink to 10NM mean power consumption would go up?? The whole point of process node shrinks is not only for density improvements but to drop power per transistor. I can't see how only 33% extra cores would lead to much higher power consumption,with a node shrink when adding 50% extra cores on the same node,didn't increase power consumption by mahoosive amounts.

 

[Journey]

Then he needs to ask the question of why there is a Z390?? So what exactly does the Z390 offer over the Z370?? Remember both have the same 300 series naming scheme.

Also why would a shrink to 10NM mean power consumption would go up?? The whole point of process node shrinks is not only for density improvements but to drop power per transistor. I can't see how only 33% extra cores would lead to much higher power consumption,with a node shrink when adding 50% extra cores on the same node,didn't increase power consumption by mahoosive amounts.

Well then you could ask why did they release the Z77 if the Z68 supported the Ivybridge CPU's, and why release Z97 if the Z87 supported the Haswell Refresh CPU's.

They release chipsets to add features, and make $, so why would you not make a Z390, especially now that they have distanced it from the Z170, and Z270, both of which are compatible with Skylake and KabyLake CPU's.

 

[henson0115]

Watching Linus talk about it here:

https://youtu.be/0qRpRITk6k4?t=51m52s

Has made me think that the likelihood of Z370 supporting 9th gen seems more likely now. As he says, if they were truly just screwing everyone over, why would they both altering the pins/power delivery at all? Surely they could just push a BIOS change and nothing else. Also, Intel has supported 2+ generations of CPUs on the same motherboards for the past 10 years. This extra power delivery could mean the speculated 10nm 8-core 9th gen CPUs might work on Z370.

for the third time in this thread, there is no extra power delivery... according to asus.

 

[southernorth]

for the third time in this thread, there is no extra power delivery... according to asus.

If you read what Andrew Wu says, he doesn't outright say it has no difference. It could affect system stability, or might indeed have issues on budget boards. We need more sources to find out the full extent of the truth that a power delivery change was necessary

 

[CAT-THE-FIFTH]

Well then you could ask why did they release the Z77 if the Z68 supported the Ivybridge CPU's, and why release Z97 if the Z87 supported the Haswell Refresh CPU's.

They release chipsets to add features, and make $, so why would you not make a Z390, especially now that they have distanced it from the Z170, and Z270, both of which are compatible with Skylake and KabyLake CPU's.

Look at the naming though - its the same 300 series naming. In all those cases,Intel NEVER stuck with the same series naming for improved chipsets in recent history - Z60 to Z70 series,Z100 to Z200 series.

Each of those chipsets launched with a new CPU line. Using what you said,it the chipset should be named the Z470 and being launching with 10NM CPUs.Intel has NOT released a new HIGHER end chipset without a whole series name change for yonks.

This is apparently being launched for the same line as Coffee Lake. So two new high end 300 chipsets for the same line - that is not usual at all.Don't forget what ASUS said - the Z270 can work with Coffeelake. That means power delivery for the Z270 is fine. Why wouldn't it be fine with the fact a number have massively over-rated VRMs?

So that indicates to me if there is changes to power delivery in the new socket,its obviously NOT for Coffeelake,its for Cannonlake,and that increases the chance that the Z390 is the chipset that has that change.When you look at this logically,it makes sense. Coffee Lake is basically just Skylake/Kabylake with two more cores,released a few months before it was expected,so they adapted the current Z270 to the new socket,since it was quick to do.

Plus,another thing since the Z390 is technically the same series as all the other chipsets,Intel has not broken the two CPU cycle. It has happened before - it happened with the P4 and Core2 also,where the power delivery changed with chipsets which technically could run a Core2 CPU but were designed for the P4. AMD did the same with certain sockets too and so on. It could not be even a case of power delivery in terms of current itself,but the way it is delivered.

For all we could know,the Z390 might be literally the Z370 but with differences to the power delivery and regulation more suited for Cannonlake.

I mean did you even know the P67 chipset could work with P55 CPUs?? ASRock made a motherboard which worked with older CPUs. But the main change to P67 was things like power delivery,etc.

OTH,you might be right,its simply for moar money making opportunities.

But look at this post by Eurocom who are responsible for high laptop reference designs:

http://www.tomshardware.co.uk/intel-z390-chipset-coffee-lake,news-56789.html

A lot of those base reference designs are used by companies like MSI,etc for their gaming ranges.

Having said that only one company can clear this up - Intel. If they did that it would clear up any ambiguity.

TBH,none of us really know where this will go as we are simply speculating it will or will not be supported.

 

[henson0115]

If you read what Andrew Wu says, he doesn't outright say it has no difference. It could affect system stability, or might indeed have issues on budget boards. We need more sources to find out the full extent of the truth that a power delivery change was necessary

and yet they outright say it could work with a bios update; does that sound like it wouldn't work unless there was extra power delivery on the board?. how much extra power would 2 more cores require? 8700k is 95w, 6700k is 91w, i call bs on the extra power delivery, considering i know full well my cpu draws more than 91w oc'd to 4.6ghz.

 

[henson0115]

Look at the naming though - its the same 300 series naming. In all those cases,Intel NEVER stuck with the same series naming for improved chipsets in recent history - Z60 to Z70 series,Z100 to Z200 series.

Each of those chipsets launched with a new CPU line. Using what you said,it the chipset should be named the Z470 and being launching with 10NM CPUs.Intel has NOT released a new HIGHER end chipset without a whole series name change for yonks.

This is apparently being launched for the same line as Coffee Lake. So two new high end 300 chipsets for the same line - that is not usual at all.Don't forget what ASUS said - the Z270 can work with Coffeelake. That means power delivery for the Z270 is fine. Why wouldn't it be fine with the fact a number have massively over-rated VRMs?

So that indicates to me if there is changes to power delivery in the new socket,its obviously NOT for Coffeelake,its for Cannonlake,and that increases the chance that the Z390 is the chipset that has that change.When you look at this logically,it makes sense. Coffee Lake is basically just Skylake/Kabylake with two more cores,released a few months before it was expected,so they adapted the current Z270 to the new socket,since it was quick to do.

Plus,another thing since the Z390 is technically the same series as all the other chipsets,Intel has not broken the two CPU cycle. It has happened before - it happened with the P4 and Core2 also,where the power delivery changed with chipsets which technically could run a Core2 CPU but were designed for the P4. AMD did the same with certain sockets too and so on. It could not be even a case of power delivery in terms of current itself,but the way it is delivered.

For all we could know,the Z390 might be literally the Z370 but with differences to the power delivery and regulation more suited for Cannonlake.

I mean did you even know the P67 chipset could work with P55 CPUs?? ASRock made a motherboard which worked with older CPUs. But the main change to P67 was things like power delivery,etc.

OTH,you might be right,its simply for moar money making opportunities.

But look at this post by Eurocom who are responsible for high laptop reference designs:

http://www.tomshardware.co.uk/intel-z390-chipset-coffee-lake,news-56789.html

A lot of those base reference designs are used by companies like MSI,etc for their gaming ranges.

Having said that only one company can clear this up - Intel. If they did that it would clear up any ambiguity.

TBH,none of us really know where this will go as we are simply speculating it will or will not be supported.

spot on.

 

[Dirk Diggler]

It's a bit comical to think that people wanted to use the previous motherboard line up with the latest CPU in order to save money. If you really wanted to save money, you could've used the previous CPU along with the previous motherboard.

 

[kedge]

It's a bit comical to think that people wanted to use the previous motherboard line up with the latest CPU in order to save money. If you really wanted to save money, you could've used the previous CPU along with the previous motherboard.

Some of those high end Z270 boards are really expensive, it appears what was said in the bit-tech interview all that is needed is a bios update to accomodate the new cpu's.