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 $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«441 »
  • Post
  • Reply
SwissArmyDruid
Feb 14, 2014



I mean, this has got to be performance art, right? No 10nm process, no 7nm process, and they're still claiming they might still be on track for 5nm in 2023?

Adbot
ADBOT LOVES YOU

Cygni
Nov 12, 2005

raring to post



SwissArmyDruid posted:

AMD had better make hay, Intel just announced that 7nm would be delayed.

Again.

https://www.bloomberg.com/news/arti...s-delayed-again

TSMCs "5nm" is still on schedule as far as I've heard, so AMD is likely going to only lose the outright process lead in laptops for about a year, and in servers for like 6 months (Intel says 10nm Ice Lake servers are still on track for this year), and not give up the process lead for desktops... at all.

(Intels 10nm and TSMCs 7nm are broadly similar in density, yes I know its not exact and depends on the product ok relax)

Cygni fucked around with this message at 21:42 on Jul 23, 2020

Kraftwerk
Aug 13, 2011


Maybe Intel doesn't want the desktop market anymore and is going the way of IBM when it comes to residential consumers. If any of this is true the Intel stock should come down a lot by 2022 and I can cash in on their comeback in 2025.

Kraftwerk fucked around with this message at 21:54 on Jul 23, 2020

SwissArmyDruid
Feb 14, 2014



So what you're saying is that Intel is going to sell its laptop division to China and then its fabs to TaiwanSaudi Arabia...

SwissArmyDruid fucked around with this message at 01:36 on Jul 24, 2020

NewFatMike
Jun 11, 2015



"Nobody got fired for buying IBM Intel"

E: their investors just think they're complete loving clowns for losing Apple and letting 10nm slip AGAIN within weeks.

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


SwissArmyDruid posted:

I mean, this has got to be performance art, right? No 10nm process, no 7nm process, and they're still claiming they might still be on track for 5nm in 2023?

theoretically the processes are independent but it's clear that intel's fab engineering is totally broken

SourKraut
Nov 20, 2005

POST QUALITY UNDER CONSTRUCTION


Boeing: Our engineering capabilities could not be any worse!

Intel: Hold our beer!

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!


SwissArmyDruid posted:

AMD had better make hay, Intel just announced that 7nm would be delayed.

Again.

https://www.bloomberg.com/news/arti...s-delayed-again

If there is one delay, there is always a second delay. Like, how does Intel catch up now? They're not going to have node lead for 4 years at this point in the best case scenario, if 7nm continues to trip up on meeting targets they're going to experience what it was like being AMD during the Sandy-Broadwell years. Like yeah they could backport core design to current process node, but every time they do that they waste a poo poo ton of money and time doing so.

Who was it that said uarch design was like playing Russian Roulette, except you pulled the trigger 4 years after loading the gun? Yeah, every time Intel has to backport because the process gets delayed, they load another bullet into the revolver.

MaxxBot
Oct 6, 2003

you could have clapped

you should have clapped!!


EmpyreanFlux posted:

If there is one delay, there is always a second delay. Like, how does Intel catch up now? They're not going to have node lead for 4 years at this point in the best case scenario, if 7nm continues to trip up on meeting targets they're going to experience what it was like being AMD during the Sandy-Broadwell years. Like yeah they could backport core design to current process node, but every time they do that they waste a poo poo ton of money and time doing so.

Who was it that said uarch design was like playing Russian Roulette, except you pulled the trigger 4 years after loading the gun? Yeah, every time Intel has to backport because the process gets delayed, they load another bullet into the revolver.

They announced that they're moving Ponte Vecchio, their HPC GPU, to TSMC to avoid any delays so I'm sure they will be doing more of that going forward.

Kraftwerk
Aug 13, 2011


Do we have any further info on when the 4000 series comes out?

MikeC
Jul 19, 2004

ACTUALLY IS MIKE CERNOVICH. WAS TOO STUPID TO FIGURE OUT THAT THE TRUMP LOVER AVATAR WAS THE STUPID NEWBIE AVATAR

Kraftwerk posted:

Do we have any further info on when the 4000 series comes out?

Nope, they just said consumer-grade Zen3 chips will be this year. Could be a paper launch.

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!


MaxxBot posted:

They announced that they're moving Ponte Vecchio, their HPC GPU, to TSMC to avoid any delays so I'm sure they will be doing more of that going forward.

GPU's maybe, but my impression is it's a big "lolfuckno" about moving their CPU designs to TSMC. I mean even if they wanted to, they'd have to be doing a CPU design targeting TSMC 3nmGAAFET because the time to port and design around TSMC 5nmEUV was...last year.

PC LOAD LETTER
May 23, 2005
WTF?!

Slippery Tilde

Cygni posted:

(Intels 10nm and TSMCs 7nm are broadly similar in density, yes I know its not exact and depends on the product ok relax)
They're supposed to be pretty similar in terms of power at the same clocks too though I believe.

EmpyreanFlux posted:

GPU's maybe,

Yeah some GPU's and chipsets will be made at TSMC.

I'd be real surprised if it turned out they went to TSMC to make their top of the line CPU's. I could maybe see them doing it for stuff like Atom cores though.

PC LOAD LETTER fucked around with this message at 12:56 on Jul 24, 2020

Khorne
May 1, 2002

Goonstone Champ x2

Kraftwerk posted:

Do we have any further info on when the 4000 series comes out?
Would expect leaks in ~3 months. There have already been engineering samples spotted but with no benchmarks or performance numbers.

Kraftwerk
Aug 13, 2011


Khorne posted:

Would expect leaks in ~3 months. There have already been engineering samples spotted but with no benchmarks or performance numbers.

I got my tomahawk ready to go I just want AMD to release this already so I can start building.
Alternatively if the gaming performance benchmarks are nearly identical and all they can offer are productivity improvements then I can just buy a 3700 and be done with it.

Anime Schoolgirl
Nov 28, 2002






Malcolm XML posted:

theoretically the processes are independent but it's clear that intel's fab engineering is totally broken
TSMC is going to enjoy becoming a monopoly soon because samsung has also been slipping in their EUV offerings lately

MikeC
Jul 19, 2004

ACTUALLY IS MIKE CERNOVICH. WAS TOO STUPID TO FIGURE OUT THAT THE TRUMP LOVER AVATAR WAS THE STUPID NEWBIE AVATAR

Kraftwerk posted:

I got my tomahawk ready to go I just want AMD to release this already so I can start building.
Alternatively if the gaming performance benchmarks are nearly identical and all they can offer are productivity improvements then I can just buy a 3700 and be done with it.

You are gonna be waiting for quite a while yet. The only reason to wait right now is for next-gen GPUs. GPUs are almost always the FPS limiting factor outside strange setups like a 2080 Super on 1080p. Zen 3 will likely be significantly faster in gaming but the question you have to ask yourself is whether you will be in that niche setup where the performance of a Zen 3 8-core chip vs a Zen 2 8-core chip will actually matter.

The usual caveat of don't buy poo poo unless you need to still apply though. So if in September when RTX 3XXX launches and ray tracing isn't a big thing for you AND your existing rig still plays what you want at an acceptable level, then don't buy.

edit: Just to clarify why I think you will be waiting for a long time.

AMD is going to service its business consumers first. We have no idea what the allocation at TSMC is like and how AMD will divy up its chips between the consumer, prosumer, and server markets. While Lisa has said consumer Zen 3 is coming out this year, tech companies frequently do paper launches where a product is released but in such limited numbers that it might as well not have been released unless you are willing to pay 2-3x MSRP from scalpers.

MikeC fucked around with this message at 15:45 on Jul 25, 2020

pixaal
Jan 8, 2004

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

Toilet Rascal

Anime Schoolgirl posted:

TSMC is going to enjoy becoming a monopoly soon because samsung has also been slipping in their EUV offerings lately

So TSMC is going to sit on their lead like Intel and let everyone else take 10 years to catch up?

Dr. Fishopolis
Aug 31, 2004

ROBOT

pixaal posted:

So TSMC is going to sit on their lead like Intel and let everyone else take 10 years to catch up?

It's not really that simple. Intel was trying to abandon tick-tock for good, they just wanted one 10nm process they could iterate into the sunset like they did with 14nm(plus plus plus plus plus)

As a result they overengineered the poo poo out of it to try and head off the problems they had with 14nm in advance. This required absurd multi-exposures and ludicrously expensive and slow testing equipment and the yield was still garbage, is garbage and will always be garbage.

TSMC meanwhile has just been tocking every two years with incremental improvements like intel did back in the day and hey presto whaddya know that poo poo actually works.

Kraftwerk
Aug 13, 2011


MikeC posted:

You are gonna be waiting for quite a while yet. The only reason to wait right now is for next-gen GPUs. GPUs are almost always the FPS limiting factor outside strange setups like a 2080 Super on 1080p. Zen 3 will likely be significantly faster in gaming but the question you have to ask yourself is whether you will be in that niche setup where the performance of a Zen 3 8-core chip vs a Zen 2 8-core chip will actually matter.

The usual caveat of don't buy poo poo unless you need to still apply though. So if in September when RTX 3XXX launches and ray tracing isn't a big thing for you AND your existing rig still plays what you want at an acceptable level, then don't buy.

edit: Just to clarify why I think you will be waiting for a long time.

AMD is going to service its business consumers first. We have no idea what the allocation at TSMC is like and how AMD will divy up its chips between the consumer, prosumer, and server markets. While Lisa has said consumer Zen 3 is coming out this year, tech companies frequently do paper launches where a product is released but in such limited numbers that it might as well not have been released unless you are willing to pay 2-3x MSRP from scalpers.

This is good advice. I might as well get my 3700 now.

sean10mm
Jun 29, 2005

Only dead doggos
follow the stream.



Yeah I am holding out for the RTX 3070 (or whatever they call it) before I do my next build, and if Zen 3 is out by then it's gravy.

gradenko_2000
Oct 5, 2010



Lipstick Apathy

I saw a post about someone that's going to try to import something called "coreboot" into AMD CPUs. Anyone know what that's about? A quick google tells me it's supposed to be an open-source BIOS, but it's not really clear what the advantages of that would be except maybe that it's supposed to be faster and/or more secure?

Some Goon
Jan 6, 2013

A golden helix streaked skyward from the Helvault. A thunderous explosion shattered the silver monolith and Avacyn emerged, free from her prison at last.




gradenko_2000 posted:

I saw a post about someone that's going to try to import something called "coreboot" into AMD CPUs. Anyone know what that's about? A quick google tells me it's supposed to be an open-source BIOS, but it's not really clear what the advantages of that would be except maybe that it's supposed to be faster and/or more secure?

If you were part of the target market that alone would be enough for you.

More helpfully, Anthony over on LTT just did a video on a laptop with coreboot, but tldw it's open source with all of the flexibility that brings.

peepsalot
Apr 24, 2007

††††††††PEEP THIS...
†††††††††††BITCH!



gradenko_2000 posted:

I saw a post about someone that's going to try to import something called "coreboot" into AMD CPUs. Anyone know what that's about? A quick google tells me it's supposed to be an open-source BIOS, but it's not really clear what the advantages of that would be except maybe that it's supposed to be faster and/or more secure?
Since all consumer motherboard BIOS are complete shitshow garbage blackboxes full of incomprehensible/undocumented options, I think the biggest advantage of any kind of open source BIOS would be making configuration much more transparent.
I've heard of coreboot before, but haven't been following the project very closely. Last time I looked at it there was pretty much no support for any "mainstream" platforms, so if someone is gonna try supporting AM4, that would be pretty neat.
Do you have a link to this post?

pedant mode: Not sure about "importing into a CPU", but maybe "porting to a chipset" (x570 ?) makes sense.

CaptainSarcastic
Jul 6, 2013

HAIL SATAN


Coreboot rings a vague bell, but mostly talking about replacement BIOS makes me think of how bad some proprietary BIOS were/are. I had a socket 939 motherboard years ago that was Compaq or HP or some other manufacturer branded, but that with a little research I found was actually a stock MSI board with only cosmetic differences. Flashed the BIOS to the stock MSI BIOS and suddenly I had a vastly more competent machine, including the ability to overclock.

gradenko_2000
Oct 5, 2010



Lipstick Apathy

peepsalot posted:

Do you have a link to this post?

https://old.reddit.com/r/Amd/commen...rting_coreboot/

It's this.

I guess my question is whether an open-source BIOS is going to allow things like overclocking where you shouldn't be able to, or allow for more CPU support on a single BIOS version (i.e. the AM4/B450 issue), or even that thing with Intel where only the Z490 chipset has fully unlocked memory speeds.

Of course, I also expect that being able to circumvent such things would get pushback from the manufacturers themselves, so this is also largely hypothetical speculation on my part.

Dr. Fishopolis
Aug 31, 2004

ROBOT

gradenko_2000 posted:

I guess my question is whether an open-source BIOS is going to allow things like overclocking where you shouldn't be able to, or allow for more CPU support on a single BIOS version (i.e. the AM4/B450 issue), or even that thing with Intel where only the Z490 chipset has fully unlocked memory speeds.

Motherboard vendors already take pride in making their overclocking options as good as possible, that's not what coreboot is for. Right now, it's mostly useful for journalists, activists, whistleblowers, really anyone with justifiable paranoia about evil maid attacks or intel AMT fuckery.

gradenko_2000
Oct 5, 2010



Lipstick Apathy

Dr. Fishopolis posted:

Right now, it's mostly useful for journalists, activists, whistleblowers, really anyone with justifiable paranoia about evil maid attacks or intel AMT fuckery.

ah ha, see, that makes a lot of sense. Thank you.

Crunchy Black
Oct 24, 2017

CASTOR: Uh, it was all fine and you don't remember?
VINDMAN: No, it was bad and I do remember.




gradenko_2000 posted:

I saw a post about someone that's going to try to import something called "coreboot" into AMD CPUs. Anyone know what that's about? A quick google tells me it's supposed to be an open-source BIOS, but it's not really clear what the advantages of that would be except maybe that it's supposed to be faster and/or more secure?

(Itís a hackintosh thing mostly)

BrainDance
May 8, 2007

Disco all night long!


I thought coreboot booted very, very quickly too because it lacks a lot of legacy nonsense, doesn't it?

Kazinsal
Dec 13, 2011




Coreboot always me think of open source firmware replacements that only run on a handful of laptops and no CPUs newer than Sandy Bridge because all the devs no true Scotsman'd each other down to accepting that era as the most power you can get on x86 while being acceptably Free (tm).

NewFatMike
Jun 11, 2015



Some folks are doing just fine with Coreboot on modern systems: https://system76.com/

The Oryx Pro which has 10th gen Intel CPUs and up to an RTX 2080 Super Max-Q has it.

Dr. Fishopolis
Aug 31, 2004

ROBOT

Kazinsal posted:

Coreboot always me think of open source firmware replacements that only run on a handful of laptops and no CPUs newer than Sandy Bridge because all the devs no true Scotsman'd each other down to accepting that era as the most power you can get on x86 while being acceptably Free (tm).

Basically everything built after 2008 or so has some kind of management engine in it with likely terrible security. I genuinely applaud the coreboot team for swimming against the tide, even though there's not much to be done about it. Maybe someday we'll have RISC V laptops that are open source down to the architecture.

Munkeymon
Aug 14, 2003

Motherfucker's got an
armor-piercing crowbar! Rigoddamndicu𝜆ous.





Nap Ghost

Dr. Fishopolis posted:

Basically everything built after 2008 or so has some kind of management engine in it with likely terrible security. I genuinely applaud the coreboot team for swimming against the tide, even though there's not much to be done about it. Maybe someday we'll have RISC V laptops that are open source down to the architecture.

Pretty sure the Intel Management Engine, at least, has been proven to be actually bad at security. AMD's offering is likely just as bad but hasn't been widely deployed so researchers don't give it as much attention.

mdxi
Mar 13, 2006

to JERK OFF is to be close to GOD... only with SPURTING



Wedge Regret

Munkeymon posted:

Pretty sure the Intel Management Engine, at least, has been proven to be actually bad at security. AMD's offering is likely just as bad but hasn't been widely deployed so researchers don't give it as much attention.
You'd think that, sooner or later, people would realize that non-updatable, non-verifiable cryptographic systems will be cracked and/or subverted 100% of the time.

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!


Zen4 as Genoa may be possibly 2021, not 2022.

https://www.notebookcheck.net/AMD-e...n.483288.0.html

According to those slides, RDNA3 is 2021 as well and it kind of looks like the APU's will skip RDNA2 in favor of RDNA3? Either way looks like a release date for Cezanne/Vega is likely 1H 2021, so maybe 7-8 months after Tigerlake. Such an aggressive timeline may indicate a Zen4/AM4/PCIE4 release with the I/O die from Zen2/3/4, and a Zen4/AM5/DDR5/PCIE5 with a new I/O die, which could be when the X670/B650/A620 launch.

Cadence could then be Vermeer September-October 2020, Cezanne April-June 2021, Genoa November-December 2021, ??? Zen4 March-April 2022, Rembrandt June-July 2022

Intel plans to release late 2022, early 2023 their first 7nm CPU's with Golden Cove and I have to wonder if those dates won't continue to slip

ConanTheLibrarian
Aug 13, 2004


dis buch is late

Fallen Rib

Great, now I can be underwhelmed by Zen3 just like I was with Zen2 and put off replacing my 3570k in anticipation of the next ryzen release.

MikeC
Jul 19, 2004

ACTUALLY IS MIKE CERNOVICH. WAS TOO STUPID TO FIGURE OUT THAT THE TRUMP LOVER AVATAR WAS THE STUPID NEWBIE AVATAR

ConanTheLibrarian posted:

Great, now I can be underwhelmed by Zen3 just like I was with Zen2 and put off replacing my 3570k in anticipation of the next ryzen release.

Server Zen 4 in 2021 according to the road map. No mention in the slides of consumer Zen 4 at all.


EmpyreanFlux posted:

Zen4 as Genoa may be possibly 2021, not 2022.

https://www.notebookcheck.net/AMD-e...n.483288.0.html

According to those slides, RDNA3 is 2021 as well and it kind of looks like the APU's will skip RDNA2 in favor of RDNA3? Either way looks like a release date for Cezanne/Vega is likely 1H 2021, so maybe 7-8 months after Tigerlake. Such an aggressive timeline may indicate a Zen4/AM4/PCIE4 release with the I/O die from Zen2/3/4, and a Zen4/AM5/DDR5/PCIE5 with a new I/O die, which could be when the X670/B650/A620 launch.

Cadence could then be Vermeer September-October 2020, Cezanne April-June 2021, Genoa November-December 2021, ??? Zen4 March-April 2022, Rembrandt June-July 2022

Intel plans to release late 2022, early 2023 their first 7nm CPU's with Golden Cove and I have to wonder if those dates won't continue to slip

Not sure where you are getting APUs skipping RDNA2 from those slides.

Craptacular!
Jul 9, 2001

Fuck the DH


I just want to say to the planners: the 3700X has been so cheap lately, especially if you want the next upcoming Assassins Creed or can find someone willing to pay you for the key, that the next chips would have to do a hell of a lot to justify their difference in price. They likely wonít close that gap.

Adbot
ADBOT LOVES YOU

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


I think they're going to go even lower and that may pose a problem for the new chips.

Would you buy a 4700X at $350 or would you buy a 3700X at $200?

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«441 »