Speculation: AMD's response to Intel's 8-core i9-9900K

Page 20 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

How will AMD respond to the release of Intel's 8-core processor?

  • Ride it out with the current line-up until 7nm in 2019

    Votes: 129 72.1%
  • Release Ryzen 7 2800X, using harvested chips based on the current version of the die

    Votes: 30 16.8%
  • Release Ryzen 7 2800X, based on a revision of the die, taking full advantage of the 12LP process

    Votes: 17 9.5%
  • Something else (specify below)

    Votes: 3 1.7%

  • Total voters
    179

ub4ty

Senior member
Jun 21, 2017
749
898
96
What will make the whole computer building community much more relaxed is when DDR4 prices get cut in half.
I think everyone both AMD/INTEL centric builders can agree, this is the bane of their builds.
All of my builds are waiting to double their memory foot prints for when memory prices drop.
 

ub4ty

Senior member
Jun 21, 2017
749
898
96
I get that, I'm just saying if you couldn't wait a 9900k will serve you well for a few years.
If you have AM4 and some patience then waiting for 3xxx Ryzen is best bet, which is what I would do, 9900k would be good 8 months earlier either way...so long as it is priced right...$400 seems about right imo.
Yeah, $400/$450 is fine. Even w/ my critiques, I realize, given the price of other components and a build that an extra $100 or $200 isn't going to be a deal breaker for a number of people.. $280/$250 vs. $400. I think, only at $500+, you get a steep fall off in interest.
 
  • Like
Reactions: french toast

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,691
136
The irony here, is that in previous pages I did exactly this testing, and You are the one not reading the word. I have 1TB sam 970 PRO, sam 512GB 960 oem pro variant and 500GB external USB3. Intel is not limiting one drive, is not saturating m.2 to usb copy, is probably at least 500MB/s faster M.2 to m.2 than AMD desktop solution, and yet You somehow manage to spread urban bs about DMI being a limiting factor on desktop?

Kinda sad, that that there is no AMD guy with 2 m.2 drives in these forums, willing to repeat the M.2 copy tests

So we allow AMD to loose 8x on graphics and use PCI-E to M.2 adapter and do not allow using that same adapter on Intel? There are plenty of boards with 8 + 8 mode, including my low end Asus Z370-A, no need to sell a kidney for high end MB. This low end mobo allow 3 M.2 from CPU (and keeping GPU at 8x) and 2 more from chipset. All in full 4x glory mode and only disabling 2 sata slots.

You tested single drive-to-drive copying. Exactly where DMI link speed doesn't matter, because PCIe is full duplex. That means the link can transfer the same bandwidth, in both directions, at once. A good analogy would be a highway with traffic in both directions. Anyway, such a test will always be limited to the write speed of the slowest drive in the copy chain. The 970/960 both have good write speeds, but not that good.

And just for the record, you can't use a non-PCIe switch passive adaptor on a mainboard without bifucation support. You only get a single drive doing that.

I rest my case here. I don't think there is any point in continuing this debate.
 

JoeRambo

Golden Member
Jun 13, 2013
1,814
2,105
136
You tested single drive-to-drive copying. Exactly where DMI link speed doesn't matter, because PCIe is full duplex. That means the link can transfer the same bandwidth, in both directions, at once. A good analogy would be a highway with traffic in both directions. Anyway, such a test will always be limited to the write speed of the slowest drive in the copy chain. The 970/960 both have good write speeds, but not that good.

Wasn't that the problem I pointed out? The logic of Your post goes like this "Don't forget PCIe and DMI ( a carbon copy of higher speed PCI-E with marketing name ) is duplex, that reduces the bottlenecks even more. Then You go on inventing some magic scenario where You read from one drive at up to 3500, and write to two drives at up to 3500 total.

...

And that is somehow slower than on AMD where there is hard limit of 1300 (on that 2nd copy target and how is even the 3rd destination connected???"
 

ub4ty

Senior member
Jun 21, 2017
749
898
96
You tested single drive-to-drive copying. Exactly where DMI link speed doesn't matter, because PCIe is full duplex. That means the link can transfer the same bandwidth, in both directions, at once. A good analogy would be a highway with traffic in both directions. Anyway, such a test will always be limited to the write speed of the slowest drive in the copy chain. The 970/960 both have good write speeds, but not that good.

And just for the record, you can't use a non-PCIe switch passive adaptor on a mainboard without bifucation support. You only get a single drive doing that.
I rest my case here. I don't think there is any point in continuing this debate.

Wasn't that the problem I pointed out? The logic of Your post goes like this "Don't forget PCIe and DMI ( a carbon copy of higher speed PCI-E with marketing name ) is duplex, that reduces the bottlenecks even more. Then You go on inventing some magic scenario where You read from one drive at up to 3500, and write to two drives at up to 3500 total.

...

And that is somehow slower than on AMD where there is hard limit of 1300 (on that 2nd copy target and how is even the 3rd destination connected???"

I'll easily shut this debate down about what platform is better and more future proof'd.
NVME drive speed and capability is increasing not decreasing. It interfaces at PCIE 3.0 x4 officially. The trend is that they are moving the drive capability to full saturation and over saturation at lower latencies. At such a point, as intel even sees, NVME drives become first class cache storage. They already are in their current state. So, you will have constant flows to/from from all storage components through the NVME to System MEM and directly to the CPU.

You absolutely do not want such a drive interfacing through a friggin chipset with all of your other much slower components.
Caching software is now free and ubiquitous on Ryzen platforms. The PCIE 3.0 x4 link absolutely does get saturated. By gimping the 2nd NVME drive to PCIE 2.0 on Ryzen, they ensure it does not saturate the chipset connection which you do not want when you have multiple other I/O that is important. When you saturate or even during normal traffic, you're doing buffering/prioritization/throttling through the chipset which adds latency and results in unpredictable speeds. You want fair and sensible access. Chipset devices are secondary citizens which is why they all fairly share a PCIE 3.0 x4.. To be, no one citizen should be allowed to saturate the channel. First class citizens get their own dedicated lane.

Intel carried over a busted arse paradigm to gimp their platform and that's the end of this discussion. They have this junk even on their HEDT.
Meanwhile, in AMD land on HEDT, i have 3 dedicated PCIE 3.0 x4 lanes for direct nvme connection to CPU + x16/x16/x8/x8 that can be bifurcated to x4/x4 or x4/x4/x4/x4 allowing for insane I/O expansion. Meanwhile at Intel : DMI 3.0 ... The stunts intel pulls are as clear as day. You can ignore them and make excuses but no one with a brain is buying it.

Intel's strategy over the years they dominated was simple... One aspect of it was to gimp I/O so you have to buy more processors to scale your I/O. The way they have their I/O laid out on the desktop is still locked into this paradigm. It likely will change as it will have to in the years ahead but the current sockets and architecture are gimped to hell.
 
Last edited:
  • Like
Reactions: Insert_Nickname

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
You can't solve this PCI stuff with theoretical discussions about the hardware, a huge part of performance is how well the drivers work and how well the cores handle the drivers.
You can have all the bandwith in the world if the driver is crap you will loose speed.
You could also have very limited bandwith but not loose any speed because of the driver using compression,just like color compression in gpus.
 

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,691
136
I'll easily shut this debate down about what platform is better and more future proof'd.
NVME drive speed and capability is increasing not decreasing. It interfaces at PCIE 3.0 x4 officially. The trend is that they are moving the drive capability to full saturation and over saturation at lower latencies. At such a point, as intel even sees, NVME drives become first class cache storage. They already are in their current state. So, you will have constant flows to/from from all storage components through the NVME to System MEM and directly to the CPU.

You absolutely do not want such a drive interfacing through a friggin chipset with all of your other much slower components.
Caching software is now free and ubiquitous on Ryzen platforms. The PCIE 3.0 x4 link absolutely does get saturated. By gimping the 2nd NVME drive to PCIE 2.0 on Ryzen, they ensure it does not saturate the chipset connection which you do not want when you have multiple other I/O that is important. When you saturate or even during normal traffic, you're doing buffering/prioritization/throttling through the chipset which adds latency and results in unpredictable speeds. You want fair and sensible access. Chipset devices are secondary citizens which is why they all fairly share a PCIE 3.0 x4.. To be, no one citizen should be allowed to saturate the channel. First class citizens get their own dedicated lane.

Intel carried over a busted arse paradigm to gimp their platform and that's the end of this discussion. They have this junk even on their HEDT.
Meanwhile, in AMD land on HEDT, i have 3 dedicated PCIE 3.0 x4 lanes for direct nvme connection to CPU + x16/x16/x8/x8 that can be bifurcated to x4/x4 or x4/x4/x4/x4 allowing for insane I/O expansion. Meanwhile at Intel : DMI 3.0 ... The stunts intel pulls are as clear as day. You can ignore them and make excuses but no one with a brain is buying it.

Intel's strategy over the years they dominated was simple... One aspect of it was to gimp I/O so you have to buy more processors to scale your I/O. The way they have their I/O laid out on the desktop is still locked into this paradigm. It likely will change as it will have to in the years ahead but the current sockets and architecture are gimped to hell.

This was exactly what I was trying, and failing, to explain. You just did it so much better.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Essentially confirmed that 9900K is soldered, and single core boost is 5GHz:
https://videocardz.com/77356/intel-core-i9-9900k-confirmed-to-be-soldered

Intel-Core-9000-Main-Features-1000x563.jpg
 
  • Like
Reactions: lightmanek

Vattila

Senior member
Oct 22, 2004
799
1,351
136
Meanwhile I saw this in a recent blog post at GlobalFoundries:

While GF continues to support the 14LPP 9T library for 12LP designs, Lavigne said the 7.5-track library “offers the most bang for the buck” in both die size reduction and higher performance. “There is some redesign for customers to use that library.”

https://blog.globalfoundries.com/gfs-12lp-process-behind-covers/

That probably refers to a comparison between 14LPP 7.5T and 12LP 7.5T, and I presume 14LPP 9T still reaches higher clock speeds. In other words, my hypothesis about a better performing respin of "Pinnacle Ridge" on the 12LP 7.5T library still does not make sense. That said, it is unclear whether the 12LP 9T library (if there is such a thing) has worthwhile improvements.

Whatever is the case, perhaps AMD's tactic is to concentrate their forces in the mobile space, while Intel is defending on the desktop. Mobile is a bigger and more important market, and assuming the "Picasso" APU refresh will be using 12LP 7.5T and coming in late 2018-Q4, for a yearly cadence, I guess that would be a better use of AMD's resources.
 

ub4ty

Senior member
Jun 21, 2017
749
898
96
16 on the CPU, 24 on the chipset.
16 on the CPU + 4 more from the CPU that gets squeezed through the chipset and expanded out to all the rest of your IO : USB/NVME/SATA drives/etc.

Far less sexy. Far more honest and clear.
Marketing = lies to get you to buy things you otherwise wouldn't if you knew the truth.
aka propganda

Glad to see them not cheaping out with toothpaste and soldering the lid to the die like the competition.
Little by little, Intel will be made great again by competition.

I am also happy to hear them getting into the GPU business by 2020. Nvidia needs to be roughed up and thrown off its horse and AMD needs some fire under its feat.

ARM just announced it's full on attacking the desktop/server CPU market come 2020.
Like I said, were upon a new hardware era. Wall to wall competition which translates to better value for consumers.
 

ub4ty

Senior member
Jun 21, 2017
749
898
96
Meanwhile I saw this in a recent blog post at GlobalFoundries:

While GF continues to support the 14LPP 9T library for 12LP designs, Lavigne said the 7.5-track library “offers the most bang for the buck” in both die size reduction and higher performance. “There is some redesign for customers to use that library.”

https://blog.globalfoundries.com/gfs-12lp-process-behind-covers/

That probably refers to a comparison between 14LPP 7.5T and 12LP 7.5T, and I presume 14LPP 9T still reaches higher clock speeds. In other words, my hypothesis about a better performing respin of "Pinnacle Ridge" on the 12LP 7.5T library still does not make sense. That said, it is unclear whether the 12LP 9T library (if there is such a thing) has worthwhile improvements.

Whatever is the case, perhaps AMD's tactic is to concentrate their forces in the mobile space, while Intel is defending on the desktop. Mobile is a bigger and more important market, and assuming the "Picasso" APU refresh will be using 12LP 7.5T and coming in late 2018-Q4, for a yearly cadence, I guess that would be a better use of AMD's resources.
ARM owns mobile and embedded but is crossing over into the desktop/server cpu market. Their licensing model and overall structure ensured their dominance.
Neither Intel or AMD were able to properly penetrate mobile/embedded. Intel tried for years and it was an utter failure. It's a completely different beast than traditional CPU sales. I doubt AMD has the chops/capital/etc to mount significant competition against ARM but this may change in the years ahead. You're essentially seeing an open grab for all markets occurring. Intel is getting into discrete video cards by 2020. ARM has a server platform and is aiming for the desktop CPU market. Apple indicated they want to part ways w/ intel and are already cozy w/ ARM. As a consumer, I am rooting for no one. I want the tons of competition, progress, and great pricing. As a shareholder, its time to get disciplined as an all out hardware war is about to take place in every segment. I also feel there will be new entrants as hardware accelerators become a thing again. Nvidia just made the first step towards integrating a new range of hardware asics to accelerate their pipeline. There's lots of chatter in various circles and tons of startups centered on the same.

PCIE 4.0 needs to become a thing in the next wave of CPUs.
PCIE switches need to come down to the desktop.
I/O needs to be freed.

NVME drives are essentially storage accelerator chips.
You have quad core ARM processors accessing an array of storage through a caching system accelerating store/retrieval. This will move to octa-core and new exotic footprints.

HBM 2.0 needs to be married to CPU in some iterations ahead.
Exciting times
6634d1b3-3390-4f84-bf07-3739cc980b6e_screenshot.jpg
 

Spartak

Senior member
Jul 4, 2015
353
266
136
So much negativity around here regarding Intel being "stuck" at 14nm, but reality is never before has a manufacturer made such immense improvements on a node as Intel has with 14nm. Not even close.

if we compare two 14nm chips around 90-95W TDP we see intel went from 4c/8t 4.0B/4.2T (6700K) to 8c/16t 3.6B/5.0T (9900K). That's an incredible performance leap on one node.

I've always bought AMD and before that Cyrix, yes we are talking 20+ years. I held on to my Athlon II until early this year, replacing it with e 2400G. But my next PC is going to have an 8086K inside.
 

ub4ty

Senior member
Jun 21, 2017
749
898
96
So much negativity around here regarding Intel being "stuck" at 14nm, but reality is never before has a manufacturer made such immense improvements on a node as Intel has with 14nm. Not even close.

if we compare two 14nm chips around 90-95W TDP we see intel went from 4c/8t 4.0B/4.2T (6700K) to 8c/16t 3.6B/5.0T (9900K). That's an incredible performance leap on one node.

I've always bought AMD and before that Cyrix, yes we ate talking 20+ years. I held on to my Athlon II until early this year, but my next PC is going to have an 8086K inside.
There's zero negativity and strings of Facts/truth/critical analysis.
Negativity and 'hater' are typically associated with logical fallacies when someone has ran out of room to argue a point. AMD delivered 8 core on 14nm for affordable desktop CPUs. I haven't seen immense improvements from Intel over the years. I have done builds with just about every one of their "major" improvements. I've seen them milk leadership which AMD wasn't able to challenge. Intel had far more resources and let a company near bankruptcy beat them to the punch.

I've done AMD/Intel builds over the years. I have multiple Intel machines currently. I have multiple AMD machines currently. I don't share your sentiments and it has 100% to do w/ fact based analysis and sentiments. I have a range of 2 core/4 core/8core/16 core builds. I go to where the value is at not brand. I maintain a solid grasp on value by remaining impartial and not caring about brands whether it be AMD or Intel or Nvidia or ARM. They're in the business of maximizing their profits. As a consumer, I'm in the business of minimizing it (best value).

Next build is a dated intel 6 core? why...
 
  • Like
Reactions: scannall

Spartak

Senior member
Jul 4, 2015
353
266
136
That's quite a bizarre response...I'm well aware of Intel's practices, maybe you missed the part where I've been building Cyrix/AMD sytems exclusively for the past 22 years? That has a reason.

You haven't seen massive provements? Maybe let facts speak for them and compare performance / watt for the 6700K with the 9900K when benches are out. Jesus Christ some of the people around here.
 

Spartak

Senior member
Jul 4, 2015
353
266
136
Next build is a dated intel 6 core? why...

Another weird response. A few thoughts on that.

Do you even understand what 'dated' means outside your bubble? its the fastest regular desktop processor Intel has for offer now.

On a further note: I thought you just disputed Intel makes any advances?

Are you aware prices come down on these 'dated' processors once there is a faster alternative?

Do you even know what my workflow actually is? Try loading a 50ha 50x50 DTM in Civil3D and report back to me please.
 

ub4ty

Senior member
Jun 21, 2017
749
898
96
That's quite a bizarre response...I'm well aware of Intel's practices, maybe you missed the part where I've been building Cyrix/AMD sytems exclusively for the past 22 years? That has a reason.

You haven't seen massive provements? Maybe let facts speak for them and compare performance / watt for the 6700K with the 9900K when benches are out. Jesus Christ some of the people around here.
I've been building for around the same time frame.
I have been building AMD and Intel. So i have no bias. I switch whenever the platform is better. I've been on Intel for some stretch because AMD simply had nothing in comparative performance or value. That changed with Ryzen so thus changed my build. See how unbiased building works?

As I've actually been building Intel systems for the last couple of years and I can point to the facts of them stalling out on improvements... Yes, I can confidently say they have been holding back because they could (AMD had nothing to drive them to do anything more).

Performance/watt
Performance/$$$

Yes, both of these matter. What makes you think they can add 2 more cores to 8700k and somehow magic will occur? What makes you think the pricing of both 8700k/9900k aren't lower because of Ryzen competition. Meanwhile you're saying your next build is a dated 8060k? Please stick to the facts and reasoned projections in a pre-launch thread.

My responses aren't bizarre. They're quite sound.
I'm on a K variant as we speak.
It was the best at the time.
 

Spartak

Senior member
Jul 4, 2015
353
266
136
Again, its the fastest single threaded performance available. Autocad is still largely single threaded. Stop projecting your own expertise, and user needs and above ALL morals onto others. You say you have no bias yet you bring Intel's practices into the discussion?? Get a grip guy you need a time-out.

For MY needs the 8086K will bring the best possible performance I can find, for reduced prices once the 9900K hits the shelves.
 
  • Like
Reactions: ozzy702

ub4ty

Senior member
Jun 21, 2017
749
898
96
Another weird response. A few thoughts on that.

Do you even understand what 'dated' means outside your bubble? its the fastest regular desktop processor Intel has for offer now.

On a further note: I thought you just disputed Intel makes any advances?

Are you aware prices come down on these 'dated' processors once there is a faster alternative?

Do you even know what my workflow actually is? Try loading a 50ha 50x50 DTM in Civil3D and report back to me please.
A Simple reply on that
https://www.newegg.com/Product/Product.aspx?Item=N82E16819117932
$424.00

7nm is next year.
Intel's processors don't fall much in price when new products are launched and rarely go on sale.

Do you even know what my workflow actually is? Try loading a 50ha 50x50 DTM in Civil3D and report back to me please.

You've finally stated it...
One that is not common.