Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
eames
May 9, 2009

Cygni posted:

Looks like those earlier slide leaks might be right. [...]

Coming back to this I do wonder where the 9900KS fits in here?

https://twitter.com/goerkemb/status/1182664034228801537

at 8C/16T 5.0 GHz all-core it may end up being faster than the 10C/20T 4.6 GHz SKU in many workloads but on an outdated socket and priced higher, unless the 10C also magically clocks to 5 GHz out of the box.

Speaking of which, shouldn't that CPU be out by now? I assume they're holding it back for the delayed 3950X release.

Adbot
ADBOT LOVES YOU

Khorne
May 1, 2002

Indiana_Krom posted:

Bringing DRAM closer to the die is for bringing data closer to the CPU so it can wait less and work more, which means it will also consume more power. Stacking everything together will definitely make cooling and power consumption worse in pretty much every way, because it sticks more power consuming and thermal dissipating components in less space than ever before and helps them all work harder besides.
AMD, and possibly Intel, have patents on through-die cooling for 3d stacks. Think tiny little heat pipes that go through the processor itself. Admittedly, it's for stacks because it goes through horizontally if I remember correctly.

I've no idea how effective they will be, but they thought it was a practical enough idea to patent it.

Khorne fucked around with this message at 16:44 on Oct 12, 2019

K8.0
Feb 26, 2004

Her Majesty's 56th Regiment of Foot
The next revolution in CPU cooling will be a waterblock integrated into the silicon itself.

in a well actually
Jan 26, 2011

dude, you gotta end it on the rhyme

Lube banjo posted:

I hope you don't think that they plan to make these chips 95w TDP. 7nm doesn't scale well with high power usage. They will try to target the optimal spot of the efficiency-curve. More likely we'll get similar performance of today's chips but at 15w or 7w so it can be passively cooled and tossed in a tablet or surface pro

Bruh, AMD is shipping a 280W 7nm part today.

necrobobsledder
Mar 21, 2005
Lay down your soul to the gods rock 'n roll
Nap Ghost
God, please let someone deliver a Core 2 Duo game changing level architecture to AMD to bring power efficiency back into computers so I don’t boil my nuts off with a small nuclear reactor on my lap that’s not furry, meows, and uses a litter box.

Setset
Apr 14, 2012
Grimey Drawer

PCjr sidecar posted:

Bruh, AMD is shipping a 280W 7nm part today.

yeah a 5700xt pulling 280w gets like 3 fps more than one using stock 180w. hence the poor scaling

Xaintrailles
Aug 14, 2015

:hellyeah::histdowns:
So, what are the chances of the 10700k being out before Doom Eternal, now it's delayed to March 2020?

Drakhoran
Oct 21, 2012

Lube banjo posted:

yeah a 5700xt pulling 280w gets like 3 fps more than one using stock 180w. hence the poor scaling

Even the Asus ROG Strix version of the 5700 XT doesn't draw much more than 220W under load. The Epyc 7h12 however has a 280W TDP

Malcolm XML
Aug 8, 2009

I always knew it would end like this.

HalloKitty posted:

Wouldn't it be neat to have a special board and case, where the CPU socket is on the back side of the board? Enormous amount of space around it for a giant, giant heatsink, and the components on the front of the board wouldn't need to be crammed in on little riser daughterboards and so on.. Maybe the space on the "front" could be used on my theoretical "mini ITX" sized board for a large chipset heatsink. I don't know, now I'm just talking nonsense.

p sure tr4/sp3 is too physically large for mini itx. The x299 went to SODIMMs. The x99/itx has a strange narrow ILM socket.

It's just kinda dumb now that you can get 12-16 cores in a small socket. It makes very little sense since a huge part of having the workstation/server boards is more memory channels which simply won't physically fit on the mini-itx board.


I want it because I'm insane.

karoshi
Nov 4, 2008

"Can somebody mspaint eyes on the steaming packages? TIA" yeah well fuck you too buddy, this is the best you're gonna get. Is this even "work-safe"? Let's find out!

K8.0 posted:

The next revolution in CPU cooling will be a waterblock integrated into the silicon itself.

An IBM patent, iirc.

in a well actually
Jan 26, 2011

dude, you gotta end it on the rhyme

Drakhoran posted:

Even the Asus ROG Strix version of the 5700 XT doesn't draw much more than 220W under load. The Epyc 7h12 however has a 280W TDP

Yeah that’s what I’m referring to. 30% higher base clock for 40% higher TDP relative to 200w part. 7nm may or may not scale well but its not going to stop vendors from putting as much in as possible.

Setset
Apr 14, 2012
Grimey Drawer

PCjr sidecar posted:

Yeah that’s what I’m referring to. 30% higher base clock for 40% higher TDP relative to 200w part. 7nm may or may not scale well but its not going to stop vendors from putting as much in as possible.

Original comment was referring to 3D stacked dies using 7nm. There would be no way to cool a 280w part (well, eight 35w parts) in a situation like that. So I was simply stating that Intel would probably try to hit peak efficiency curve rather than max out the clock speeds.

But yeah look at the efficiency on those 8-core chiplets on EPYC. From 65-95w tdp @ 3.6-w/4.4ghz boost to what, 25w at 2-w/3.3ghz turbo?

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

Malcolm XML posted:

p sure tr4/sp3 is too physically large for mini itx. The x299 went to SODIMMs. The x99/itx has a strange narrow ILM socket.

It's just kinda dumb now that you can get 12-16 cores in a small socket. It makes very little sense since a huge part of having the workstation/server boards is more memory channels which simply won't physically fit on the mini-itx board.


I want it because I'm insane.

Is it that much bigger than LGA3647?

priznat
Jul 7, 2009

Let's get drunk and kiss each other all night.

Paul MaudDib posted:

Is it that much bigger than LGA3647?

Ice Lake SP going to LGA4189 :haw:

SwissArmyDruid
Feb 14, 2014

by sebmojo

PCjr sidecar posted:

Bruh, AMD is shipping a 280W 7nm part today.

Yes, but it's a 280W 64-core/128-thread part. You know, where you can socket a pair of those into a board and achieve nuclear fusionthe kind of density that Intel cannot presently hope to achieve, at the cost of heat.

SwissArmyDruid fucked around with this message at 23:19 on Oct 12, 2019

Cygni
Nov 12, 2005

raring to post

TR4/SP3 is slightly larger than 3647 (which has an ITX board out there), so it would probably be physically possible with lots of engineering/compromises, but frankly the HEDT market is a small submarket and threadripper an even smaller sub fraction of that submarket.

there isnt a ITX TR4 board cause the big 4 dont think its worth the cost

BlankSystemDaemon
Mar 13, 2009



Isn't a nichel market like HEDT also more likely to go with a E-ATX motherboard to get four pci-ex for SLI or something equally silly?

coke
Jul 12, 2009
Maybe they are finally making this true

Kazinsal
Dec 13, 2011
Hmm. 4.6 GHz all-core on 20 threads is certainly interesting. 12T @ 5.0 GHz in Ghost Recon: Breakpoint bottlenecks my 1080 Ti (100% CPU usage and 85-90% GPU usage at 1080p), so depending on how much I really want to spend on a new motherboard this could be an interesting upgrade.

I'd rather go Threadripper tbh but I don't know if much of what I do other than a few really well-optimized multicore titles is going to eat up something nuts like 14C/28T @ 4.0 GHz.

eames
May 9, 2009

Kazinsal posted:

I'd rather go Threadripper tbh but I don't know if much of what I do other than a few really well-optimized multicore titles is going to eat up something nuts like 14C/28T @ 4.0 GHz.


I feel like there’s a good chance this we’ll see real-time software encryption DRM schemes become standard as high core count adoption improves. :smith:

Indiana_Krom
Jun 18, 2007
Net Slacker

eames posted:

I feel like there’s a good chance this we’ll see real-time software encryption DRM schemes become standard as high core count adoption improves. :smith:

Like denuvo, that makes games stutter, chop, freeze and average significantly lower performance because it steals so much CPU time away from actually running the game?

Potato Salad
Oct 23, 2014

nobody cares


in workloads where dram access speed matters, nobody is going to handicap performance so severely by adding yet more cycles to every operation with a cache miss

gradenko_2000
Oct 5, 2010

HELL SERPENT
Lipstick Apathy

D. Ebdrup posted:

Isn't a nichel market like HEDT also more likely to go with a E-ATX motherboard to get four pci-ex for SLI or something equally silly?

is SLI even still a thing nowadays? I keep getting the impression that it's no longer being supported or is at least on the outs with newer hardware

craig588
Nov 19, 2005

by Nyc_Tattoo

Indiana_Krom posted:

Like denuvo, that makes games stutter, chop, freeze and average significantly lower performance because it steals so much CPU time away from actually running the game?

Denuvo is great. It only hurts performance in situations where you shouldn't be running the game anyways. If you're planning to run the game at 720P or something I guess it's a miss, but at 1080P or higher it's no problem. It's literally the first DRM I've thought was good enough.

Arivia
Mar 17, 2011

gradenko_2000 posted:

is SLI even still a thing nowadays? I keep getting the impression that it's no longer being supported or is at least on the outs with newer hardware

It is definitely on the outs. Far less games bother with support, and nVidia is slowly pushing it out of their lineup for all but the bleeding edge overclockers/enthusiasts and professional products.

e: Crossfire is whatever, but Crossfire has always BEEN whatever.

latinotwink1997
Jan 2, 2008

Taste my Ball of Hope, foul dragon!


coke posted:

Maybe they are finally making this true



Where on this chart would “surface of the sun” be?

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

D. Ebdrup posted:

Isn't a nichel market like HEDT also more likely to go with a E-ATX motherboard to get four pci-ex for SLI or something equally silly?

E-ATX is wider, not longer :engleft:

Indiana_Krom
Jun 18, 2007
Net Slacker

latinotwink1997 posted:

Where on this chart would “surface of the sun” be?

~6300 W/cm2, higher than anything else but would still fit in the graph.

pixaal
Jan 8, 2004

All ice cream is now for all beings, no matter how many legs.


Do note that that is already on a logarithmic scale.

gradenko_2000
Oct 5, 2010

HELL SERPENT
Lipstick Apathy
https://www.techpowerup.com/260130/intel-scraps-10nm-for-desktop-brazen-it-out-with-14nm-skylake-till-2022

quote:

In a shocking piece of news, Intel has reportedly scrapped plans to launch its 10 nm "Ice Lake" microarchitecture on the client desktop platform. The company will confine its 10 nm microarchitectures, "Ice Lake" and "Tiger Lake" to only the mobile platform, while the desktop platform will see derivatives of "Skylake" hold Intel's fort under the year 2022! Intel gambles that with HyperThreading enabled across the board and increased clock-speeds, it can restore competitiveness with AMD's 7 nm "Zen 2" Ryzen processors with its "Comet Lake" silicon that offers core-counts of up to 10.

"Comet Lake" will be succeeded in 2021 by the 14 nm "Rocket Lake" silicon, which somehow combines a Gen12 iGPU with "Skylake" derived CPU cores, and possibly increased core-counts and clock speeds over "Comet Lake." It's only 2022 that Intel will ship out a truly new microarchitecture on the desktop plaform, with "Meteor Lake." This chip will be built on Intel's swanky 7 nm EUV silicon fabrication node, and possibly integrate CPU cores more advanced than even "Willow Cove," possibly "Golden Cove."

Khorne
May 1, 2002
How is that news though? We already knew that was the plan since at least late 2018. Although lots of people predicted 2021 for Intel's 7nm on desktop while that article claims 2022.

DrDork
Dec 29, 2003
commanding officer of the Army of Dorkness

Khorne posted:

How is that news though? We already knew that was the plan since at least late 2018. Although lots of people predicted 2021 for Intel's 7nm on desktop while that article claims 2022.

That they wouldn't offer 10nm on desktop was more or less an assumption, though, not hard knowledge. A more and more obvious assumption when no news for desktop Icelake was heard for months on end, but Intel up until now hadn't actually said they were going to skip desktop 10nm entirely, vs their previous "we're working on it" line.

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

Khorne posted:

How is that news though? We already knew that was the plan since at least late 2018. Although lots of people predicted 2021 for Intel's 7nm on desktop while that article claims 2022.

Ice Lake was always planned to be skipped but iirc Tiger Lake was supposed to be coming to the desktop.

That basically seals the deal on anything significantly surpassing Coffee Lake until 2022 as far as gaming is concerned. Zen3 will launch late 2020 and probably catch up but not significantly surpass it, that puts Zen4 in like late 2021 or more probably early 2022.

That's a bummer considering consoles are going to get a big increase in per-thread performance here. It's gonna be hard to hold 100+ Hz when consoles have a processor that's roughly on par with a 2700 or so and target that as their 60fps baseline.

eames
May 9, 2009

This also means that the 400 series boards are going to be 14nm for their full lifecycle and thus likely another refresh of Z170. :lol:

Previous rumors claimed that Comet Lake needs a new socket/chipset to be compatible with the 10nm arch coming down the road but clearly that’s not happening, so the new socket is probably there for mobo makers and to have less vdroop for those lovely unannounced 400W CPUs.

In a very odd turn of events I think I’d sooner buy into X299 than Z490. :chloe:

Otakufag
Aug 23, 2004

Paul MaudDib posted:

That's a bummer considering consoles are going to get a big increase in per-thread performance here. It's gonna be hard to hold 100+ Hz when consoles have a processor that's roughly on par with a 2700 or so and target that as their 60fps baseline.
But Nvidia's next arch on 7nm might push PC gaming ahead again, depending on the amount of gains they get and how does the PS5 GPU perform.

eames
May 9, 2009

https://www.tomshardware.com/news/intel-yes-there-will-be-10nm-desktop-cpus

New statement by Intel confirms that there will be 10nm desktop CPUs after all, though a 65W CPU in a NUC would technically qualify as a desktop CPU :haw:

redeyes
Sep 14, 2002

by Fluffdaddy
I got one of those I3 9400F processors for $90 bux. I was gonna go Ryzen but eh, that's the cheapest processor that doesn't suck i could find. Oh well, maybe next round AMD will win.

Twerk from Home
Jan 17, 2009

This avatar brought to you by the 'save our dead gay forums' foundation.

redeyes posted:

I got one of those I3 9400F processors for $90 bux. I was gonna go Ryzen but eh, that's the cheapest processor that doesn't suck i could find. Oh well, maybe next round AMD will win.

Where did you find this? Still available?

BIG HEADLINE
Jun 13, 2006

"Stand back, Ottawan ruffian, or face my lumens!"

Twerk from Home posted:

Where did you find this? Still available?

Micro Center with the $30 off promo on CPU/MOBO?

Adbot
ADBOT LOVES YOU

redeyes
Sep 14, 2002

by Fluffdaddy
https://www.amazon.com/gp/product/B07R7Q3JZH/ref=ppx_yo_dt_b_asin_title_o05_s00?ie=UTF8&psc=1

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply