Nadolazeći AMD socketi, ploče i procesori

[The Exiled]

AMD EPYC Rome 64-Core CPU brings the pain to Xeon Platinum in C-Ray benchmark demo

Zen has proven a capable architecture, and just as importantly, it's serving as a long-term foundation to compete with Intel across the board. The latest example of this is another demo AMD posted showing a single second-generation EPYC "Rome" CPU beating a pair of Intel Xeon Platinum 8180M CPUs in the C-Ray benchmark. The Xeon Platinum 8180M is a beastly server chip. It has 28 physical cores and 56 threads to throw at workloads, and is clocked at 2.5GHz to 3.8GHz. The burly chip also boasts 38.5MB of L3 cache. As for the EPYC processor, AMD did not share a whole lot of details, other to say it is a prototype 1P Rome platform. We do know it is a 64-core/128-thread processor, so even when combining the two top-shelf Xeon CPUs, the Rome CPU still flexes more cores and threads.

Obviously that presents an advantage, and we would expect the AMD chip to outperform the two Xeon processors, which it does. But what's impressive is that AMD crammed all of those cores and threads into a single socket while offering comparable performance per core. That is a big win in the server space, and it comes with certain advantages. One of those advantages is highlighted in the video at around the 59-second mark when the group of AMD demonstrators comment on the noise output of the Intel system. "I hope the Intel box isn't getting too loud," Amit Mehra, principal member of technical staff at AMD, says to the group. Bob Hershberger, server system architect, then points out that the Intel system "is getting a little louder, which can be heard in the video. The key here will be pricing. AMD's EPYC Rome processor is based on the company's 7-nanometer Zen 2 architecture and is not yet out, but will be relatively soon. AMD is very much facing an uphill battle. A recent report from DRAMeXchange noted that AMD managed to capture just a 2 percent share of the x86 market with its first-gen EPYC processors, and is expected to bump that up to 5 percent in 2019.

Izvor: HotHardware

[lobo64]

Autor Manuel Calavera

To je navodno od PS5 konzole, prokomentirano već na ovoj stranici

PS5 će imati AM4 socket a ne zalotani čip? Yeah, zvuči razumno sa stajališta rezanja troškova...

[MadBoY]

Gledaj plave znakove, ovo sto pise bijelim slovima je samo primjer, i mislim da se to odnosi samo na Zen i Zen+, za Zen 2 ne vrijedi.

[Manuel Calavera]

AMD 12nm mobility lineup is planned to last through 2019, 7nm to take over in late 2019 to Q1 2020

The second thing I wanted to talk about is the roadmap for 7nm. I have heard that AMD’s 7nm mobility lineup won’t hit the shelves till Q1 2020. That said, please keep in mind plans like these can change rapidly although I do not expect 12nm mobility lineup to be replaced before late 2019 at the earliest (even if we are dealing with an accelerated roadmap). Mobility lineups require a good-enough level of yield and achieved economies of scale so the later this ships, the better it is for AMD (and the partner foundry).
The good news is I am hearing very good things about the performance per $ levels of the AMD 7nm mobility lineup and it will be very competitive compared to what we have on the market right now. This means we are looking at a significant structural shift in cost structures of the laptop industry: things are about to get a lot faster – and cheaper! It is also worth mentioning that Intel plans to have its 10nm (which is ~equivalent to the pure-play foundry 7nm process) on the market before year-end as well so the race is very much back on.

[MadBoY]

Ako je istina da ce u Q3 doci desktop APU, onda je ocekivano da ce i mobile 7nm doci najranije tada, a vjerojatno malo kasnije. Desktop APU i mobile APU su isti, razlika je samo u taktovima i sto mobile APU nekad ima manje CU na GPU. Nece raditi razlicit APU za desktop i mobile. Znaci da bi mogli dobiti napokon 8c16t APU za laptope . A do prije par godina sve od i3 do i7 je bio dual core .

[The Exiled]

PS5 and next Xbox price and specifications leaked?

PS5 specifications and price

• CPU: Custom 8 cores / 16 Threads Zen 2 CPU
• GPU: Custom NAVI @8+ teraflops
• Memory: 12GB GDDR6
• Storage: 1TB SSD
• Price: $399

Next Xbox specifications and price

• CPU: Custom 8 cores / 16 Threads Zen 2 CPU
• GPU: Custom NAVI @12+ teraflops
• Memory: 16GB GDDR6
• Storage: 1TB NVMe SSD @ 1+GB/s
• DirectX Raytracing + MS AI
• Price: $499

Next Xbox streaming console specifications and price

• CPU: Custom 8 cores / 16 Threads Zen 2 CPU
• GPU: Custom NAVI @4+ teraflops
• Memory: 12GB GDDR6
• Storage: 1TB NVMe SSD @ 1+GB/s
• DirectX Raytracing + MS AI
• Price: $249

Some of these claims were confirmed by the insider at ResetEra.

Izvor: [H]ard|OCP

U prijevodu - tko će ga znati, ali kad se pogleda unatrag nekoliko stranica spomenuti Microsoft Scorpio Engine (1 - 2), neke specifikacije čak i nisu toliko nategnute/nedostižne.

[MadBoY]

Tu mi se jedino 1 TB SSD cini malo nerealnim, ali ocekivao sam da ce preci na SSD radi brzine ucitavanja igara. Ostatak HW-a skroz ok, jedino se PS5 cini malo slab po pitanju GPU-a.

[Manuel Calavera]

U najgorem slučaju nekakva sshd, store MI, intel optane fora, jer su samo igre na njima, neki cache za brži loading istih

[Sn4k3]

Škot strikes again

What is Zen 2?

[Manuel Calavera]

Zanimljivo da kaže da prema 2 njegova izvora infinity fabric više neće ovisiti o brzini rama...


Inače:

ASRock preparing nine X570 motherboards

The first PCIe 4.0 consumer motherboards have been spotted at EEC website. The leak with ASRock’s X570 motherboards features Phantom Gaming, Extreme4, Taichi and Pro 4 series.
The X570 is AMD’s chipset for Zen2-based processors. The Ryzen 3000 desktop series are expected around Computex. The leak from ASRock might be an indicator that X570 motherboards will be available sooner.

• ASRock X570 Phantom Gaming X
• ASRock X570 Phantom Gaming 6
• ASRock X570 Phantom Gaming 4
• ASRock X570 Extreme4
• ASRock X570 Taichi
• ASRock X570 Pro4
• ASRock X570 Pro4 R2.0
• ASRock X570M Pro4
• ASRock X570M Pro4 R2.0

[Eladio]

taichi je sad mid segment ploča?

[Manuel Calavera]

Čisto sumnjam, samo su ih tako poslagali

[Matta]

Fala qurcu da su se sjetili ubaciti koju microATX ploču !
S druge strane, fali ITX ploča. Za pretpostaviti je da to nije cijeli popis.

[Brko]

Autor Eladio

taichi je sad mid segment ploča?

Izgleda da je sada taj Fantom Gejming the shit od Asrocka uz te ploče će valjda dolaziti i kakva droga za bolje doživljavanje presijavanja RGB ledica.

I naravno, cijene barem 15-20% veće od X470 lineupa. Kod Asusa i više vjerojatno.

Sent from Usreo Mobara

[Manuel Calavera]

Što se tiče high enda, nije to po klasama poslagano, vidite da je "extreme4" stavljen iznad "taichi", a to neće biti po klasi. Ne kažem nužno da će "taichi" ostati najjača (kao što je kod x370 Professional Gaming), ali je sigurno i dalje high end

Evo na primjeru intela - Z390 taichi ultimate je skuplja od Z390 phantom gaming 9... Stoga ne vidim zašto bi phantom gaming 6 bila jača od taichi i na x570, eventualno jedino iznad nje možda može biti Phantom Gaming X, ovo drugo je sve ispod.

Autor Matta

Fala qurcu da su se sjetili ubaciti koju microATX ploču !
S druge strane, fali ITX ploča. Za pretpostaviti je da to nije cijeli popis.

Pa imao si ih i prije, očito su i ove entry/mainstreamuše ("pro4")

[Disco]

Autor Sn4k3

Škot strikes again

What is Zen 2?

OK, omda je rekao da su u ovom videu većinom njegova nagđanja, ali sviđa mi se dio gdje priča o vezi između čipleta na Ryzenima (Epyc ih navodno nema, nego samo preko I/O), pa da bi se tu moglo puno dobit na latencijama što bi se trebalo odrazit na gaming perf.
Sviđa mi se i pretpostavka da je AMD namjerno pokazao min dobitak u odnosu na i9 i da bi zapravo zen 2 trebao biti dosta brži od skylake refresheva.

Ugl. sad možemo samo prduckat u prazno, valja čekat prve rivjue i nadat se najboljem.

[Matta]

Autor Manuel Calavera

Pa imao si ih i prije, očito su i ove entry/mainstreamuše ("pro4")

Sa X*70 chipsetom od ASrocka nijedna, a i od ostalih proizođača slabo.

[MadBoY]

Od Asrocka imas X370M-HDV i X370M Pro4. Ali generalno je dosta slabiji izbor ploca za AMD u odnosu na Intel, skoro svi proizvodjaci imaju duplo vise modela za Intel.

[Sn4k3]

Autor Disco

OK, omda je rekao da su u ovom videu većinom njegova nagđanja, ali sviđa mi se dio gdje priča o vezi između čipleta na Ryzenima (Epyc ih navodno nema, nego samo preko I/O), pa da bi se tu moglo puno dobit na latencijama što bi se trebalo odrazit na gaming perf.
Sviđa mi se i pretpostavka da je AMD namjerno pokazao min dobitak u odnosu na i9 i da bi zapravo zen 2 trebao biti dosta brži od skylake refresheva.

Ugl. sad možemo samo prduckat u prazno, valja čekat prve rivjue i nadat se najboljem.

Tocno tako, mozemo samo nagadjati i bubati u prazno dok neke konkretne informacije ne izadju, al ako mene pitas 'my money goes to' 12c/24t max za AM4 socket, sve povrh toga ce biti rezervirano za TR. Naravno to je samo moje misljenje, uskoro cemo saznati hoce li to tako biti ili ne... Fingers crossed

[Manuel Calavera]

Ja bi na njihovom mjestu malo "muzao" situaciju tj. 12c24t sad, a 16 core bi izbacio sa zen3

Uglavnom da, uz infinity fabric neovisan o brzini rama spominje i neku "vezu (infinity fabric)" između samih cores chipleta osim veze cores - I/O

- mogućnost da prvo dođe 12 core, a 16 cores kasnije
- da I/O može biti svakakav i u njemu svašta (za potrebe APU-a), možda nešto poput ovog mog prvog crteža (manji I/O čip da stane HBM2 ili nekakav chache kao intel iris igpu unutar toga I/O die-a)
- da po njemu retail 8 core neće onako izgledati nego da će biti 2 chipleta po 4 jezgre + I/O die (pretpostavljam radi yieldova)

[Max Rockatansky]

Autor Manuel Calavera

Ja bi na njihovom mjestu malo "muzao" situaciju tj. 12c24t sad, a 16 core bi izbacio sa zen3
...
- mogućnost da prvo dođe 12 core, a 16 cores kasnije
...

Nadam se da neće krenuti Intelovim stopama "milkanja" .
Nemaju potrebe ništa čuvati kad za novi Threadripper lineup imaju od 16 do 64-core na raspolaganju a i sve skupa su ionako iste kockice (chipleti) pa ne bi baš bili u gubitku. Tko treba quad-channel RAM i više PCI lanesa taj će ionako odabrati TR a i bilo bi lijepo da 50-godišnjicu okrune 16-core mainstream procesorom.

[Manuel Calavera]

Pa nije to intelova stopa milkanja ako sa zen2 dobiješ 50% više jezgri (12c24t) i pitajboga 7+ % IPC i nešto clocka

Zen+ je bio veoma intel like release, pa im se nije gro zamjerilo, gdje će im netko zamjerit ako ne izbace 16 core pored onog svega što sam gore naveo...

[Gigi1]

intel im sigurno nece moci parirati do 2020 ako amd izbaci 16 jezgri, samo je pitanje zeli li amd cuvati 16 jezgri za zen2+/zen3 ili ce odmah udariti i zadat "killer blow" sa 16 jezgri na zen2

kratkotrocno bi vjerojatno bilo bolje ici na 12 jezgri(vece marze/zarada)
dugorocno bi bolje bilo odmah ici na 16 jezgri da svi...pa i mediji koje prenos dedu koji vjebža pricaju o amd-u, takav killer blow bi im bio strašan PR

[Manuel Calavera]

Inače neće biti zen2+ odnosno nikakvog refresha, odmah ide zen3 poslije ovog na 7nm+ nodeu negdje u 2020. Sam po sebi neće biti toliki pomak kao sad ovaj zen2 sa zen+

[Eladio]

Autor Gigi1

intel im sigurno nece moci parirati do 2020 ako amd izbaci 16 jezgri, samo je pitanje zeli li amd cuvati 16 jezgri za zen2+/zen3 ili ce odmah udariti i zadat "killer blow" sa 16 jezgri na zen2

kratkotrocno bi vjerojatno bilo bolje ici na 12 jezgri(vece marze/zarada)
dugorocno bi bolje bilo odmah ici na 16 jezgri da svi...pa i mediji koje prenos dedu koji vjebža pricaju o amd-u, takav killer blow bi im bio strašan PR

s obzirom na raspon cijena koji mogu postaviti, i kratkorocno bi profitirali od 16c, jer bi i na njima marža bila sexy

[The Exiled]

• AMD Amends Wafer Supply Agreement with GlobalFoundries: 7nm Freed, 12nm+ Targets Set Through 2021

The big news with the latest WSA update is that it allows AMD to contract any foundry to produce chips at 7 nm and thinner nodes without any one-time payments or royalties to GlobalFoundries. The net result is that as of this moment, AMD has slated all of their currently-announced leading-edge 7nm CPUs and GPUs to use TSMC's 7nm process, and AMD is further free to tap Samsung as well if they desire. Past that, the latest WSA update will see AMD continue to rely heavily on GlobalFoundries for any larger nodes, as the firm will remain AMD’s long-term strategic supplier for 12nm and above.
One of the important aspects of the new agreement is that if AMD fails to meet the annual wafer purchase target for 2019, 2020, and 2021, it will have to pay GlobalFoundries “a portion of the difference” between the actual wafer purchases and the planned target for that year. While this is not exactly a take-or-pay agreement that AMD used to have with GlobalFoundries several years ago, it will still have to use/sell 12LP and 14LPP chips in high volumes in a bid not to pay charges to its partner.

• AMD had an Epyc quarter: Server chip shipments 'more than doubled' Q3 to Q4
  
  
• AMD Earnings Report, Q4 FY 2018: Closing on a Profitable 2018

[Brko]

Autor Gigi1

intel im sigurno nece moci parirati do 2020 ako amd izbaci 16 jezgri, samo je pitanje zeli li amd cuvati 16 jezgri za zen2+/zen3 ili ce odmah udariti i zadat "killer blow" sa 16 jezgri na zen2

kratkotrocno bi vjerojatno bilo bolje ici na 12 jezgri(vece marze/zarada)
dugorocno bi bolje bilo odmah ici na 16 jezgri da svi...pa i mediji koje prenos dedu koji vjebža pricaju o amd-u, takav killer blow bi im bio strašan PR

Hoćeš reći da bi 16c/32t bio takav blow da bi porazio sve plave kuverte i "bonuse" za reklame koje lntel dijeli šakom i kapom svima, pa čak i čistačicama u Dellu npr? Hm, nadam se da je tako.
Sjetimo se samo kako su se svi držali ko pijani plota 2 stvari kad je Zen izašao: Segfault i Singlethread performance u scenariju 720P/1080P na low detaljima sa 1080Ti. Ali dobro, barem od onda potrošnja i zagrijavanje više nisu bitni, čak stoviše, to je nebitno postalo jer singltred FTW.

Samo me zanima što će sada biti kad Zen2 izađe, a lntelov 14 nm +++++++++ node bude spreman odma u Q2 2020, koju će onda dealbreaker "manu" naći kao pred 2 godine? Vjerojatno priču da AMD kupuje uslugu od Tajvanaca čiji radnici rade u nehumanim uvjetima bez prava na beneficirani radni staž i prekovremene sate.

Random blentavi Youtuber: "lako je Zen2 odlična arhitektura i procesor od 300 USD je u rangu ili bolji od 2.5x skupljeg lntelovog 8c/16t procesora, napomenimo to da za proizvodnju Zen2 procesora ljudi pate."

Tada više ni singltred FTW neće biti glavni forte, nego ljucka prava.

Ok, ne mora biti tako, ali neka nebitna stavka će se naći i hajlajtati gdje se god stigne

Sent from Usreo Mobara

[The Exiled]

Lako za AMD, oni su svoj dio napravili, sve ostalo kaj slijedi imalo bude itekak dobre temelje. Pitanje je kaj bude, ako se ispostavi da je Intelovih 10nm ipak čušpajz, mada sad naveliko pričaju (a opet nitko točno ne zna) kak je sve super i 10nm proizvodi samo što nisu. I sadašnja 10nm situacija im nije bajna, a 14nm ne budu mogli unedogled frizirati, a sve pizdarije koje su ovih godina skrivali - dolaze na vidjelo.

[Gigi1]

@brko mislim da da, ak je nesto hot stuff i kad se pokrene YT i ostala news mašinerija svi ce se htjeti slikati i stajati uz bok pobjedniku na postolju, takva je ljudska psiha.

ako single thread perf. i latencije budu na nivou intela, odnosno kak veliš 1080p gaming sa 2080ti bude jednak ili bolji od plavih, šansa je skoro pa nikakva da ih se zaustavi u vracanju dobrog dijela mindshare-a.

koliko intel moze i dalje obracati paznju na "detalje" sa placanjem cistaca u dell je upitno, odnosno kakvo je stanje u plavom taboru dosta govori da im nitko ne zeli doci na poziciju CEO-a vec ohoho


edit: ovaj novi ugovor sa GF im je odlicno dosao, sad vise nemaju penala za stancanje kod drugih.
sa 2% na 4% za epyc nije bas neka sreca, ocekivao sam min 5% ali izgleda da se ekipa zapalila za rome pa cekaju to.
prodaja ryzena ih je spasila, odnosno intel sa super konkurentnim cijenama zadnjih 4-5 mjeseci
q1 za 19 ce biti povuci potegni naspram q1 iz 18

[messiah]

Autor The Exiled

Lako za AMD, oni su svoj dio napravili, sve ostalo kaj slijedi imalo bude itekak dobre temelje. Pitanje je kaj bude, ako se ispostavi da je Intelovih 10nm ipak čušpajz, mada sad naveliko pričaju (a opet nitko točno ne zna) kak je sve super i 10nm proizvodi samo što nisu. I sadašnja 10nm situacija im nije bajna, a 14nm ne budu mogli unedogled frizirati, a sve pizdarije koje su ovih godina skrivali - dolaze na vidjelo.

Evo jedan od boljih postova na redditu koje sam pročitao zašto je Intel u problemima. Ukratko, nisu znali kako će se realizirati 10nm node pa su išli u 2 smjera:
1. Multi pattern ArF immersion - pokazalo se da ima hrpetinu problema koje još nisu uspjeli riješiti
2. EUV - za koje strojeve proizvodi doslovno jedna firma na svijetu te koštaju papreno. Vrijeme isporuke je bilo 2-3 godine, a nisu na vrijeme naručili.

https://www.reddit.com/r/Amd/comment...t_with/efbtmg2

The challenge for intel has to do with the lithography process. To resolve features below 32nm there are two forks: multi-pattern ArF+immersion and EUV. Multi-pattern is achieved by performing multiple exposures with matched photomasks on the same substrate layer before the wafer resist gets resolved. This was a viable path for sub 32nm, and was the basis for the double-patterning used for Ivy-bridge and Haswell (both 22nm processes). Intel had more difficulty shrinking to 14nm, because the random factors started to creep in at that scale.

To better understand this, let's do a thought experiment:

Imagine driving a car from Los Angeles to New York and then back again, and because this is a high-precision thought experiment, let's assume that there are no other cars on the road, and the entire drive is happening in a temperature controlled environment without any debris larger than a pebble to interfere with your journey. Got it? During the first trip you intentionally drive as steady as you can, because there is a twist: during the return trip, the car must drive exactly over the same path. Remember, there are no other cars in this experiment. It's just you, the open road, and your imaginary car driving for a few thousand miles.

Sounds impossible right? There is going to be some variance and traveling exactly over the prior path cannot be realistically achieved. I am not a monster, so I will allow a little bit of error. You can drive the car back, and as long as you stay in your lane, I will count that as a success. That is what a 22nm process looks like.

Now, 14nm is going to be a bit more of a challenge. Instead of only staying in your lane for the round trip, now you are only allowed about two tire's width of difference. You also need to make two round trips instead of one. This is still possible, and with a few tweaks to the steering and suspension you might improve accuracy a bit. Maybe you can slow down a little, drive less aggressively, etc. but you need to find a way to repeat that path. And you do. You make more mistakes and it takes several tries, but you manage to pull it off! Congratulations!

Let's talk about 10nm. You'll need to make the return trip with less than one inch difference. Oh, and you'll need to make FOUR round trips, not just one. Good fucking luck!

That's multi-pattering in a nutshell. The random errors that you almost did not notice at 22nm are now gigantic obstacles to completing a successful pass. Errors are multiplied while tolerances shrink. Even when your first round trip is perfect, you're only a quarter of the way there, and you're more likely to fail with each additional trip. You might get lucky every now and then, but it's not reliable.

Now let's compare that method with EUV. It's a newer lithography process, and it offers high resolution in a single pass (no multi-patterning penalty), but there's another downside: throughput. ArF+immersion tools can process 250~300 wafers per hour. EUV tools struggle to achieve even >100 wafers per hour (this is improving with better light sources, optimizations to photoresist, etc.).

Returning to the thought experiment, EUV is like riding a bicycle from Los Angeles to New York. It's a one-way trip, and it's much slower, but the bike tires are much thinner (higher resolution), so you get a little bit more wiggle room.

These are the two forks. One is fast but error prone, the other is slow but gives you sub 22nm features in a single pass. The immersion lithography process is about ten years old, meaning that it is a mature process, with lower operating costs. The tool footprint is comparable to older platforms, making it viable for high volume manufacturing. You can pump out 22nm chips with low defects and high yields. It works at 14nm but the yields are lower, and at 10nm you start to rely more on superstition than engineering. EUV has higher resolution, but the tools are MASSIVE, and can only be installed in fabs with high ceilings, and a ballroom layout (chase and bay clean rooms cannot handle these dreadnought-sized lithography tools). EUV tools are also considerably more expensive, and have higher operating costs. The light source is "soft x-rays" and only works under vacuum, which complicates maintenance - the entire chamber has to be purged of atmosphere before resuming production.

Intel invested in both of these platforms, without knowing which would be viable for each node. They bet wrong on 10nm multi-pattering but due to costs could not afford to switch to an EUV process. To complicate matters further, ASML is the only company in the world to offer a production capable EUV lithography tool. Intel would have needed to place their orders years in advance to meet their production needs.

TL;DR - semiconductor manufacturing is fucking complicated and mistakes are unforgiving. Even with excellent long term planning, you can easily fall behind.