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

[StinkyPinky]

Why?
What specific hands down is the useprofile to do such a change?

Why not? It's an enthusiast forum. Why do people get expensive toys for anything?

 

[dlerious]

To be honest, I'm thinking along those lines as well, even though the frugal (sensible?) part of me is telling me I have absolutely no need for an upgrade from a 8700K, being mainly a gamer. But like you said, we can easily sell the 8700K for ~$300 on the 2nd hand market and all of a sudden we are looking at a $200 upgrade to have the 'latest and greatest'. It's a nice option to have, but I'm perfectly happy with my 8700K at the moment and will probably stick it out until next gen parts arrive.

With the way Intel has been going, I wouldn't be surprised if you'll also need a Z390 motherboard.

 

[whm1974]

Why not? It's an enthusiast forum. Why do people get expensive toys for anything?

Not every enthusiast has the money or the willingness to keeping spending the money every time the next new expensive thing comes out. Or for that matter, is willing to deal with the trade-offs of using high TDP products.

Personally if I'm building a high end consumer(not HEDT) rig I want it to last me 10 years in case I have to keep that long.

 

[scannall]

I don't think AMD has to do anything really, except stay on their roadmap and continue iterating. The have a solid product at a very attractive price.

 

[PeterScott]

I don't think AMD has to do anything really, except stay on their roadmap and continue iterating. The have a solid product at a very attractive price.

Yes, that is primarily what most of the tech companies do. They advance their road-maps, that are planned years in advance.

The battles, and responses, argued on forums are mainly nerd inventions to pass the time.

 

[krumme]

Why not? It's an enthusiast forum. Why do people get expensive toys for anything?

Why?. At least a small difference imo. The latest and greatest is meaningless if there is no difference. Its what defines latest and greatest. It has always been like that. Or people get a new hobby or use their time other areas. Or get better gpu whatever.
8700k is for gamers and why its sold. Only. Lets be frank about it. Thats what the core line is about or amd makes far more sense be it 2700x or some threadripper 16c and soon 32c.
That why i have a 8700k and also others. With the newest bios patches it is bad at many office task because of 4k ssd transfers even at low que depths. It is a gaming tool and nothing else. Fine. Nothing wrong in that.
Its like many members is still stuck at pre zen time. Like it took years for some to adapt to the core2 and still thought the athlon64 was the fastest. A cpu that is bad at ssd transfers is hardly high end product and is reflected in the 8700k price. Its nice fairly cheap high end gaming cpu.

6c made a difference to prior kbl 4c in many games. A 4c8t eg tanked in bf1 64mp mins. But there was many other games where mins made a difference.
A 8c? Its getting darn close to no difference. But lets see. If we get new engines that might change. But as the next gen consoles is probably 2 years out i dont think we will see more cpu taxing games this year or next.

 

[JoeRambo]

With the newest bios patches it is bad at many office task because of 4k ssd transfers even at low que depths.

I had a good laugh, unless Your "office" work involved running 4K ssd benchmark all day, i doubt any sane person would even notice the difference in real world apps. Nice FUD spin tho. PRO tip, outside of random I/O bound servers no one would see a slow down of more than several percent.

 

[krumme]

I had a good laugh, unless Your "office" work involved running 4K ssd benchmark all day, i doubt any sane person would even notice the difference in real world apps. Nice FUD spin tho. PRO tip, outside of random I/O bound servers no one would see a slow down of more than several percent.

I use the same computer for work as i game on. It happens for all my computers. Have you considered office work can be different than you imagine? Back in the days i had office docs that took several hrs to print and often today i am waiting several minutes for the tasks to end. Eg. Excel db linked to word linked to outlook. And as a PRO tip its not st speed keeping it down. Sure not programs build for huge data handling and they are dog slow and unsafe but never the less it happens all over the world because they are flexible and there is a history of using them.

But hey if 7% doesnt bother you just get some cheaper gear. The 8700k with the bios patches is excactly slower where it counts most in my work and where it can be felt. Several percent 4k drop is bad. I hate it but not the same as i hate mins in bf1.

 

[JoeRambo]

Wasn't your original claim some load of FUD being spread about 8700K "being bad at SSD transfers"? Even with Spectre patches it is easily fastest at low I/O depths just by virtue of having great clock speed and best architecture in market. And Your Excel problems probably have nothing to do with I/O. More likely problems are elswhere ( and i also can make office documents that take minutes to open, but they are not I/O bound ).

 

[krumme]

"As reported by heise.de apparently the impact of the windows patches upon the performance of SSD's is even greater then the impact upon CPU operations.

https://m.heise.de/newsticker/meldu...Prozent-ab-SSD-I-O-deutlich-mehr-3938747.html

One relevant (google translated) section from the article reads


c't: Significantly lower SSD performance after security updates

c't has also seen significant drops in SSD performance after installing BIOS and security updates (Core i7-8700K, Asus Maximus X Hero, Samsung 960 Pro, Random Access with 4K blocks and 32 I / O-targets). The Samsung SSD only reached 105,986 I / Ops for reading and 79,313 I / Ops for writing compared to 197,525 / 185,620 I / OPs (previous BIOS, Windows with KB4054517). Further analysis, benchmarks on the effects of patches, and reviews of updates can be found in the upcoming c't issue 3/18.
"

https://linustechtips.com/main/topi...-shows-significant-impact-on-ssd-performance/

So much for the FUD rambo.
Its a very serious issue.

 

[JoeRambo]

Haha, so Your office now generates 100k IOPS ? Please stop embarrasing Yourself


EDIT: and one thing to note, is that early testing was done with horrible Linux vendor patches, once they were reverted and replaced by proper ones in mainline Linux kernel performance impact was lowered. Of course peak IOPS are still very much impacted, but that is for servers and in benchmark testing. For casual office/gaming workloads the impact is several percent.
Oh and those impacted tend to run trusted code only ( like dedicated DB node with PCIE flash cards providing million IOPS class storage ), and there is nothing to stop them from disabling patches.

It's complex topic and multifaceted topic, but claims about Coffee Lake somehow becoming bad at office tasks like You claim here - is pure FUD. Reverse is actually true, it is fastest at those tasks.

P.S. and the focus of my points is to dispel FUD claims about office workload impact, not "downplay" performance impact, which is obviously with each and every OS system call.

 

[krumme]

Read link instead of your stupid 100kiops strawman or linux kernel nonsense.

". Der Coffee-Lake-Prozessor Core i7-8700K ist im SYSMark 2014 SE unter Windows 10 nach allen eingespielten Sicherheitsupdates demnach insgesamt rund sechs Prozent langsamer. Intel erklärt allerdings, dass der Leistungsverlust bei bestimmten Nutzungsszenarien im Bereich Office und Media Creation durchaus höher ausfallen könne. Als Beispiel führt der Prozessorhersteller Web-Apps mit JavaScript an, die um bis zu zehn Prozent langsamer ablaufen können."

 

[krumme]

This is real office impact. Web aps and basic stuff.
Should we start talking server impact in io bound situations? Then its not just embarrasing but in many cases rendered useless. (And yes in many cases irrelevant as perf stays the same)
My sever cost benefit just took an average 20% nosedive. Meaning cost stays the same while perf tanks 20%. Thats only direct cost. Great.
Yes. No need to downplay this stuff. Even for heavy office use.

 

[Shivansps]

Yeah I guess it doesn't really matter since that Intel 8 core will cost as much as an Intel 8 core...which is a crap ton of money that no one wants to spend. That's a shame. Its a shame the CPU will cost at least $500. Why even release it? Intel would be better off simply telling us, "We could release an 8 core on mainstream, but NAH". Might as well not release it with how high they will price it. lol. Its fine though. Intel just does Intel.

So, AMD doesn't need to do anything except for continue selling a truck load of 2700X's.

Remember when i said it was wrong for AMD to try and charge $500 for a 1800X because it was increasing the roof on consumer sector pricing by a large margin? And people said i was trolling? well DONT COMPLAIN now if the Intel CPU is $500 or more.

 

[Arachnotronic]

Yeah I guess it doesn't really matter since that Intel 8 core will cost as much as an Intel 8 core...which is a crap ton of money that no one wants to spend. That's a shame. Its a shame the CPU will cost at least $500. Why even release it? Intel would be better off simply telling us, "We could release an 8 core on mainstream, but NAH". Might as well not release it with how high they will price it. lol. Its fine though. Intel just does Intel.

So, AMD doesn't need to do anything except for continue selling a truck load of 2700X's.

Why do you think it'll cost at least $500?

 

[whm1974]

Why do you think it'll cost at least $500?

Because it is an Intel i7 CPU? How much were the i7-8700/8700k processors at first release?

 

[french toast]

Remember when i said it was wrong for AMD to try and charge $500 for a 1800X because it was increasing the roof on consumer sector pricing by a large margin? And people said i was trolling? well DONT COMPLAIN now if the Intel CPU is $500 or more.

That doesn't make any sense, the 8 core amd processors were giving practically the same performance as a 1000$ 8 core i7 6900k...so 500$ was considered a bargain next to that.
If you didn't want to spend that much you could get virtually the same performance from a 1700x for around 370$...about the same price as Intel's top end 4 core consumer 7700k.
Of course 7700k had no equals in some applications (gaming) but the amd processors had no equals in other apps in that price range either..

Only problem for AMD was there was no need for 3 8 core SKUs with little performance difference between them when overclocked, you can blame the low power process for that.

 

[Vattila]

In the OP, I wrote:

Also, note that "Picasso" is expected as the 12LP refresh of the "Raven Ridge" APU. Presumably, "Picasso" is designed with 12LP libraries for optimal density and efficiency. If so, AMD will have to do the 12LP design work for the CCX anyway.

I think this is my best argument for a respin of "Pinnacle Ridge", but there has been little discussion about this point.

So to reiterate: If "Picasso" (the successor to "Raven Ridge") is implemented using 12LP libraries, i.e. is a true optimised shrink, which allegedly the current "Pinnacle Ridge" is not, then AMD will have 12LP blueprints for the CCX that are smaller, and presumably better optimised for the 12LP process and capable of clocking higher (at same power).

Why not reuse all this work to refresh the 2000-series?

 

[french toast]

In the OP, I wrote:



I think this is my best argument for a respin of "Pinnacle Ridge", but there has been little discussion about this point.

So to reiterate: If "Picasso" (the successor to "Raven Ridge") is implemented using 12LP libraries, i.e. is a true optimised shrink, which allegedly the current "Pinnacle Ridge" is not, then AMD will have 12LP blueprints for the CCX that are smaller, and presumably better optimised for the 12LP process and capable of clocking higher (at same power).

Why not reuse all this work to refresh the 2000-series?

Zen 2 is coming out in April, so what's the point?..unless you mean ryzen 2000 series should have used 12nm LP libraries in the first place for better performance..then in that case I agree, but pinnacle ridge fills the gap nicely until 7nm castle peak, they could save the top 1% of dice for a potential 2800x..which has 200mhz to turbos all round even if needed higher tdp (if achievable), could charge 399$ and keep somewhat competitive against 500$ coffeelake S.

I would seriously hope that we get a brand new 7nm die for Raven Ridge successor in Q3 2019, with zen2 + Navi + edram.
Perhapps there is not alot of capacity for 7nm mainstream chips even by then.

 

[PeterScott]

Why do you think it'll cost at least $500?

Didn't even the AMD Ryzen 1800X debut at $500? Isn't AMD is supposedly the "value" brand?

Does anyone really think Intel will charge less than $499 for a new 8 core CPU?

 

[Vattila]

Zen 2 is coming out in April, so what's the point?

Q4 holiday sales? 2018 revenue targets?

If AMD can refresh the 2000-series in Q3, and thus keep the pressure on Intel until the 3000-series is ready, it would contribute favourably to marketshare gains and revenue growth momentum. Assuming a gradual ramp and product rollout, the 3000-series probably will not contribute materially to revenue until 2019-H2, even if it they manage to release it in April on a yearly cadence.

 

[JoeRambo]

Does anyone really think Intel will charge less than $499 for a new 8 core CPU?

Not sure why would they charge more either. Upper bound is "professional" 8-core CPU in the form 7820x that has MSRP of $599. So quite easy to see that Intel will price consumer 8 core lower than that. MSRP of $499 sounds just right to me.

Or maybe Intel will decide to strike hard at AMD and price it @ $400, after all this same company was selling i7 920 for ~$300 when they felt they need to challenge AMD.

 

[LTC8K6]

7700K and 8700K were within $20 of each other at launch for two more cores.
The 8600K with 6 cores was nearly $100 less at launch than the 7700K.

I'm going with $399 for the 8 core chip.

 

[moinmoin]

Q4 holiday sales? 2018 revenue targets?

If AMD can refresh the 2000-series in Q3, and thus keep the pressure on Intel until the 3000-series is ready, it would contribute favourably to marketshare gains and revenue growth momentum. Assuming a gradual ramp and product rollout, the 3000-series probably will not contribute materially to revenue until 2019-H2, even if it they manage to release it in April on a yearly cadence.

I think you are expecting a little too much from AMD. Remember that AMD itself didn't make a big deal about the 2000-series whenever they lined out the update timeline from Zen 1 to 3 (even the moniker Zen+ came later and from outside). The 2000-series itself already is a refresh of Zen 1 by AMD's own positioning.

 

[moonbogg]

7700K and 8700K were within $20 of each other at launch for two more cores.
The 8600K with 6 cores was nearly $100 less at launch than the 7700K.

I'm going with $399 for the 8 core chip.

Hahaha!

Why do you think it'll cost at least $500?

Why?

Because Intel is still pissed that they can't charge us $1,200 for 8 cores anymore? Sounds like a legit reason to me. My god, they were really DEEP inside us weren't they? Of course I'm talking about their hands being deep inside our pockets when it came to their CPU pricing. They miss those good old days and will do anything for a little taste of the past, such as trying to get away with ripping us off just enough so they can get away with it, even if just barely.
They do things like take away hyperthreading on the i5 when really they should follow the AMD model and give us 6 functioning cores. They do things like take away PCI-E lanes on the high end, but really they should just do it like AMD and let us have them all. You can tell how they hate value for their customer. It absolutely makes their skin crawl. It makes them sick. They hate you for getting a good deal. They hate you for it. You can tell by how long its taking them to release an 8 core consumer chip...and even then they will probably try to sell you a version without HT, lol. They take forever to release an 8 core because it represents unprecedented value for the everyday high end consumer, the enthusiast gamer and typical power user. Its like pulling teeth trying to get an 8 core out of Intel that doesn't cost $1,200. It makes them want to throw up.

So yeah, all that is why.