• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Intel CPU Roadmap and DDR5

Joined
Jul 9, 2016
Messages
1,089 (0.34/day)
System Name Main System
Processor i9-10940x
Motherboard MSI X299 Xpower Gaming AC
Cooling Noctua NH-D15S + Second Fan
Memory G.Skill 64GB @3200MHz XMP
Video Card(s) ASUS Strix RTX 3090 24GB
Storage 2TB Samsung 970 EVO Plus; 2TB Corsair Force MP600; 2TB Samsung PM981a
Display(s) Dell U4320Q; LG 43MU79-B
Case Corsair A540
Audio Device(s) Creative Lab SoundBlaster ZX-R
Power Supply EVGA G2 1300
Mouse Logitech MK550
Keyboard Corsair K95 Platinum XT Brown Switches
Software Windows 10 Pro
Benchmark Scores Cinebench R20 - 6910; FireStrike Ultra - 13241; TimeSpy Extreme - 10067; Port Royal - 13855
I am planning to upgrade to the next major Intel CPU release, but I have lost tracked and confused what is up next from Intel. Would be it 7nm or 10nm? Will it be utilizing DDR5? I Googled Intel CPU Roadmap and not sure I understand what the image means at all.

This is for work, not for gaming. No Ryzen. Intel only. I need stability and lots of cores + high single core speed. Current CPU is 7 years old 4930k and I don't want to go HEDT ever again.
 
The real question here

does Intel even know

Still, you should go Ryzen anyway. Just sayin. Things have changed. New Ryzen CPUs boost straight up to 4.6 Ghz and XFR is absolutely great for that single thread performance, they have more cores, actually secure SMT instead of HT that Intel recommends you turn off, they run cooler, use less power... should I go on? Oh yeah, they're cheaper per core/thread as well. You don't even have to OC.
 
10nm

It will, 10nm will come by 2021 for the desktop and AMDs zen ends 2020 and 2021 will have a new socket + ddr5, I don't see why it won't be the same for intel.

Hands in the air for those who think Intel will push high end, high frequency 10nm chips out. Not all at once... :D My two cents: they won't because they can't.

If they could, we'd have seen an ES already. We did see low power laptop CPU, which is I guess a sort of proof of concept that 10nm doesn't catch fire, but that's about it. I personally think Intel is going to act like they try real hard until halfway 2020, and then we get a news piece saying they will buy into 7nm capacity or switched over themselves.
 
Hands in the air for those who think Intel will push high end, high frequency 10nm chips out. Not all at once... :D My two cents: they won't because they can't.
They can't for the time being, 10nm for mobile is scheduled 2020 whilst desktop is 2021.
 
It will, 10nm will come by 2021 for the desktop and AMDs zen ends 2020 and 2021 will have a new socket + ddr5, I don't see why it won't be the same for intel.

This is an assumption that does not really match well with the statement about the purpose of DDR5, because the first wave is most definitely aimed at datacenter and not consumer at all. Don't forget that AMD is basically just glueing together datacenter CPUs and uses less glue for MSDT. Its one and the same thing. For Intel, not so much.

They can't for the time being, 10nm for mobile is scheduled 2020 whilst desktop is 2021.

Yeah and did you notice how short that 10nm period is in the scheme of things? It even overlaps with a 7nm bar halfway through for some segments. That right there is Intel's 'out' and they will probably use it. We have yet to hear a shred of real, verifiable success on 10nm. I think they'll stick to their volume segment - mobile - and that is all it will be.
 
The real question here

does Intel even know

Still, you should go Ryzen anyway. Just sayin. Things have changed. New Ryzen CPUs boost straight up to 4.6 Ghz and XFR is absolutely great for that single thread performance, they have more cores, actually secure SMT instead of HT that Intel recommends you turn off, they run cooler, use less power... should I go on? Oh yeah, they're cheaper per core/thread as well. You don't even have to OC.

I develop software. Years ago one of my developers compiled code using AMD FX CPUs and generated some weird errors that I had to spend days and found the issues with their errata. So I will never use AMD CPU again for development. Intel is the way to go, not to mention the compiled code will be running on Xeon boxes. I need 100% compatibility. Also, Ryzen had many issues with memory compatibility. I don't need the headache.

2021 might be a bit far away. So no new desktop Intel CPU at all for 2020? Will it just be 9900KS?
 
I develop software. Years ago one of my developers compiled code using AMD FX CPUs and generated some weird errors that I had to spend days and found the issues with their errata. So I will never use AMD CPU again for development. Intel is the way to go, not to mention the compiled code will be running on Xeon boxes. I need 100% compatibility. Also, Ryzen had many issues with memory compatibility. I don't need the headache.

2021 might be a bit far away. So no new desktop Intel CPU at all for 2020? Will it just be 9900KS?
Was/is that the Intel compiler that they got sued over for crippling performance on AMD CPU's?
AMD is a very different company today than it was then, and conditions have and are changing. Intel is not going to be competitive until 2021, it seems. I'd take a second look.
 
Take a pick...

 
Was/is that the Intel compiler that they got sued over for crippling performance on AMD CPU's?
AMD is a very different company today than it was then, and conditions have and are changing. Intel is not going to be competitive until 2021, it seems. I'd take a second look.

That was over 9 years ago and irrelevant to my question . Not sure what you are trying to say here.

Take a pick...


The Wiki pages don't give out much info. Not useful.
 
Last edited:
Not sure what you are trying to say here.

Compiled code should never produce any sort of error on any x86 CPU AMD or Intel. That is, unless the complier does some unorthodox optimizations meant for particular platforms.
 
Compiled code should never produce any sort of error on any x86 CPU AMD or Intel. That is, unless the complier does some unorthodox optimizations meant for particular platforms.

Compiler is all about optimizing code to execute fast and efficiently. This happens often in the past, especially when writing low level code. Not so much nowadays, especially I have since switched to writing higher level C# and JS from C and some C++.

The compiler was Microsoft VS, but the error also occurred with GNU C compiler. When switched to using my i7 2600k the problem went away. So the compiler(s) has some issue or optimization going on that was not compatible with the AMD.
 
The compiler is all about correctness first then about optimizations.

So the compiler(s) has some issue or optimization going on that was not compatible with the AMD.

I have never heard of any compiler producing incompatible code for any x86 target, given the same set of extensions of course. Sub-optimal yes, but incompatible, I find it hard to believe. But assuming this is true and it was the fault of the compiler I fail to see how this would translate to "Intel is the way to go, never touching anything AMD again".

Regardless, good luck with this narrow minded approach and also good luck finding something worthwhile form Intel right now and in the near future, as others have pointed out their prospects aren't the best.
 
So just as luck would have it, I was just informed that I have won an EVGA hosted competition. Perhaps I should start a new thread but I almost fainted when I see the price that Intel wants for their HEDT CPUs. A quick check of the $1k i9-7900x does not show drastic performance boost my 4930k, or it may even lost out to the new Ryzen. Really disappointed in Intel.

I won this guy and also an EVGA PSU. Thinking of selling the board as the cost to build is too high for gain/return.
https://www.evga.com/products/product.aspx?pn=131-SX-E295-KR
 
So just as luck would have it, I was just informed that I have won an EVGA hosted competition. Perhaps I should start a new thread but I almost fainted when I see the price that Intel wants for their HEDT CPUs. A quick check of the $1k i9-7900x does not show drastic performance boost my 4930k, or it may even lost out to the new Ryzen. Really disappointed in Intel.

I won this guy and also an EVGA PSU. Thinking of selling the board as the cost to build is too high for gain/return.
https://www.evga.com/products/product.aspx?pn=131-SX-E295-KR
Congratulations on the win.
 
The Wiki pages don't give out much info. Not useful.

Well, I'm sorry, but that's what there is. It's not like Intel offers detailed public roadmaps...
 
Some of the features of the LPDDR4X memory controller (like two channels per module) in Ice Lake sounds halfway towards DDR5.
One of Intel's recent leaked server roadmaps shows PCI-e 5.0 and DDR5 in 2021.

There is no definite confirmed ETA for CPUs or platforms with DDR5. FFS, Intel has no clear public and confirmed roadmap for now either :)
 
It's not like Intel offers detailed public roadmaps...
I think this makes sense too - regardless if Intel or AMD, or any high-tech company.

What happens if Elon Musk sets then misses a deadline to achieve some new technological advance in the Tesla?​
What happens if Apple delays the release of the next iPhone?​
What happens if Richard Branson set an aggressive Virgin Galactic deadline then some catastrophic event happens?​
What happens if Microsoft sets a deadline for a new feature update then slips the deadline back, or pushes the update out only to find it is flawed?​

One of two things always happens. (1) Haters and critics bash them for poor planning and incompetence for failing to meet their deadlines or (2) haters and critics bash them for poor planning, inadequate beta testing and incompetence for failing to move the deadline out until they have a finished product.

It is nearly impossible to accurately predict the exact date you will succeed at advancing the "State-of-the-Art" - especially when those advances are already up against the limits imposed by the "Laws of Physics" - as is the case when it comes to shrinking die architectures.

One of the big challenges is NOT shrinking the gaps in the transistor gates. Nor is it reducing the operating voltage low enough so it does not arc across those shrinking gates while at the same time keeping that voltage high enough to operate those gates. One of the biggest challenges is for manufacturers to refine and produce the raw materials (silicon, gold, etc.) that go into manufacturing those processors. When talking about gates approaching the size of individual atoms, even 1 foreign atom per billion silicon atoms could wreck havoc on the processor.

Of course, if a company sets a deadline then meets or even beats it with no after-release problems, that is great and the company can celebrate. But that's not "news".

So a road map showing the direction they want to go makes sense. Putting a timetable to that map is probably not a good idea - even though some critics and haters will bash them for that too.
 
I have never heard of any compiler producing incompatible code for any x86 target, given the same set of extensions of course. Sub-optimal yes, but incompatible, I find it hard to believe. But assuming this is true and it was the fault of the compiler I fail to see how this would translate to "Intel is the way to go, never touching anything AMD again".

Regardless, good luck with this narrow minded approach and also good luck finding something worthwhile form Intel right now and in the near future, as others have pointed out their prospects aren't the best.
When you deal with enough low level code, ultimately you run into this kind of problem. The error that my dev found was in the published errata, clear and simple. I don't work there anymore or else I would pull out more info for you. It is not narrow minded approach, it is a business need. When our IT guy was pulled into the director office and scolded for trying to save money by using AMD, the whole office takes notice. Wasting a day for several dev cost the company thousands of dollars. We all want to save money, but the trade-off in this case is not worth it.
Here is an example - https://www.theregister.co.uk/2012/03/07/amd_opteron_bug_dragonfly/

Well, I'm sorry, but that's what there is. It's not like Intel offers detailed public roadmaps...
Sorry, I didn't mean to be rude. What I was trying to say was the wiki pages are not helpful, not your post.

Congratulations on the win.
Thanks, but I am going to sell or trade it. I dig more into it and found the x299 is actually Skylake (the Skylake-x name is a dead give away) architecture. It was quite confusing to me as I thought it is the same architecture as the 9900K. My 4930k is fairly strong, even comparable to the Ryzen 2700x. The cheapest Skylake-x CPU is over $800, and AMD Blender benchmark shows the upcoming $500 3900x match or beat the $1200 9920x. I just feel there is minimal CPU advancement in the last 7 years. I am done with HEDT. Intel has made too much money off HEDT. If Intel is too late and my machine needs an upgrade really bad, I might even buy the 3900x if there is no reported compatibility issue
 
When you deal with enough low level code, ultimately you run into this kind of problem. The error that my dev found was in the published errata, clear and simple. I don't work there anymore or else I would pull out more info for you. It is not narrow minded approach, it is a business need. When our IT guy was pulled into the director office and scolded for trying to save money by using AMD, the whole office takes notice. Wasting a day for several dev cost the company thousands of dollars. We all want to save money, but the trade-off in this case is not worth it.
Here is an example - https://www.theregister.co.uk/2012/03/07/amd_opteron_bug_dragonfly/


Sorry, I didn't mean to be rude. What I was trying to say was the wiki pages are not helpful, not your post.


Thanks, but I am going to sell or trade it. I dig more into it and found the x299 is actually Skylake (the Skylake-x name is a dead give away) architecture. It was quite confusing to me as I thought it is the same architecture as the 9900K. My 4930k is fairly strong, even comparable to the Ryzen 2700x. The cheapest Skylake-x CPU is over $800, and AMD Blender benchmark shows the upcoming $500 3900x match or beat the $1200 9920x. I just feel there is minimal CPU advancement in the last 7 years. I am done with HEDT. Intel has made too much money off HEDT. If Intel is too late and my machine needs an upgrade really bad, I might even buy the 3900x if there is no reported compatibility issue
Well all the Lakes are basically the same. Just varying core and cache amounts.
 
I develop software. Years ago one of my developers compiled code using AMD FX CPUs and generated some weird errors that I had to spend days and found the issues with their errata. So I will never use AMD CPU again for development. Intel is the way to go, not to mention the compiled code will be running on Xeon boxes. I need 100% compatibility. Also, Ryzen had many issues with memory compatibility. I don't need the headache.

2021 might be a bit far away. So no new desktop Intel CPU at all for 2020? Will it just be 9900KS?

Fair enough. While this stance raises some questions I completely understand that a business has its requirements and own realities. I see them all the time... but it makes sense to question them from time to time.

Now would be such a time. As you have said, a lot of years have passed. If money is no object, a suggestion could be made to have one Zen test system up and running code. A suggestion - and one that may benefit you in several ways. Regardless, the Intel info's there... on MSDT the 9900K is the only real path. We're as anxious as you are to see what they'll pull out of the hat next.

Take a pick...


:roll:Intel skinny dipping in architectures. 'Hmm, does this feel right?'
 
The real question here

does Intel even know

Still, you should go Ryzen anyway. Just sayin. Things have changed. New Ryzen CPUs boost straight up to 4.6 Ghz and XFR is absolutely great for that single thread performance, they have more cores, actually secure SMT instead of HT that Intel recommends you turn off, they run cooler, use less power... should I go on? Oh yeah, they're cheaper per core/thread as well. You don't even have to OC.

And now they sport pci express 4.0 and higher ddr4 support and will be implementing ddr5 next year...
Intel’s ddr5 and pci express 5.0 is still late uptill 2021/22 and that even not to be desktop parts but server builds
 
Back
Top