Vijesti i novosti o Intel procesorima i čipsetima - Stranica 150

The Exiled · 31.03.2021., 20:10

To bude tek šou, jer sljedeće tri generacije im imaju taj hibridni dizajn. Ostaje da vidimo kud to vodi, ali zasad (i dalje) samo pričaju i obećavaju.

The Exiled · 20.04.2021., 19:05

U svakom slučaju zanimljivo, a Dr. Ian Cutress ujedno potvrđuje ulogu Jima Kellera (1 - 2) (ne samo u AMD Zen slučaju, već općenito diljem firmi u kojima se dosad nalazil) kao voditelja projekta.

Odnosno nekoga tko organizira timove koji onda konkretno rade na razvoju i dizajnu novih tehnologija i popratnih tehnoloških rješenja.

Ianova Alder Lake analiza je također u rangu s dosadašnjim očekivanjima.

EDIT:

Citiraj:

Citiraj:

Samsung remains formidable competitor for TSMC, says Morris Chang

Citiraj:

Samsung Electronics is still a formidable competitor for TSMC, according to Morris Chang, founder of the world's largest pure-play foundry. TSMC is based in Taiwan, which boasts a complete IC industry supply chain and a robust talent pool - both of which South Korea also has, said Chang. Both countries also have well-developed sea-air transport logistics, which are also critical to the local foundries' competitiveness worldwide, Chang added. As the largest chipmaker in South Korea, Samsung will be capable of competing against TSMC in the foundry segment, Chang indicated in a public speech. Similar corporate culture is another factor that makes Samsung the strongest competitor of the foundry he founded, Chang said.

Chang is also skeptical about Intel's ambitious bid unveiled earlier this year to regain its manufacturing lead. In its early years, Intel declined a bid to invest in TSMC, probably not thinking contract chip making would have a future, Chang disclosed. But foundry services have become so important that now Intel wants to play a part. Taking stock of the company's decision to enter the contract chipmaking sector more than three decades after its decision to not invest in TSMC, Mr. Chang described the situation as "ironic." The success of TSMC is not only because of the foundry model it has been embracing, but also because of the leadership of professional managers and its focus on R&D and innovation, according to Chang. For TSMC, it is of great importance to stay ahead of competitors, Chang added.

TSMC will be stepping up its 5nm chip output, scaling it up to as much as 150,000 wafers monthly starting in the second quarter through the end of this year.

Izvor: DigiTimes

Citiraj:

Intel Data Center sales slump stokes concern about market share

Citiraj:

Intel has released its figures for the first quarter of 2021 and presents a complex picture – some areas may continue the positive trend, while others may well be described as disappointment. The most important finding from the figures: Intel's net profit is shrinking sharply. Compared to the previous year, it has fallen by almost 50% to USD 3.4 billion. Its profit fell from 7 billion to 5.6 billion U.S. dollars. The margin fell from 50 to 23% and net profit fell from 3.5 billion to 1.3 billion U.S. dollars. Intel said sales of chips to cloud service providers fell 29% from the same period a year earlier.

Izvor: Bloomberg, The Next Platform i TechPowerUp

Citiraj:

Intel seeks $10 bln in subsidies for European chip plant

Citiraj:

Intel wants 8 billion euros ($9.7 billion) in public subsidies towards building a semiconductor factory in Europe, its CEO was cited as saying on Friday, as the region seeks to reduce its reliance on imports amid a shortage of supplies. The pitch is the first time Pat Gelsinger has publicly put a figure on how much state aid he would want, as Intel pursues a multibillion-dollar drive to take on Asian rivals in contract manufacturing. Gelsinger, on his first European tour since taking charge, met European Commissioner Thierry Breton in Brussels on Friday. The visit followed the launch of a plan for Intel to invest $20 billion in chip production in the United States. CEO Pat Gelsinger has also met with President Joe Biden in the White House to discuss the issue and hinted on being very open to the company receiving subsidies. On top of that, Gelsinger is prospecting for a location for a plant in Europe that he says would back Breton's goal of doubling the region's share of global chip output to 20% over the next decade. Gelsinger, who met Economy Minister Peter Altmaier and Bavarian governor Markus Soeder on the German leg of his European tour, was quoted as saying Germany would be a suitable location for a potential European foundry.

Izvor: Reuters

Citiraj:

Intel will spend billions to expand in New Mexico and Israel, seeks more government support

Citiraj:

Intel said it will spend $3.5 billion to upgrade its aging manufacturing facilities in New Mexico and confirmed it will spend $10 billion on a new factory in Israel, part of a broad push to expand the chipmaker’s production capacity under newly installed CEO Pat Gelsinger. After years of carefully managing its capital spending and returning money to investors, Gelsinger indicated the company is now budgeting for expansion instead. Intel’s spending spree follows severe deterioration of Intel’s manufacturing capabilities under its last two CEOs, Brian Krzanich and Bob Swan. The company lost its lead in advanced chip technology after successive delays to its 14-nanometer, 10nm and 7nm microprocessors caused by persistent manufacturing defects. As Intel struggled to adapt to new manufacturing technologies over the past several years, it spent little on new factories and instead committed billions of dollars to buying back its own stock. Stock buybacks typically have the effect of boosting a company’s share price, but they’re also an explicit admission that a business has more money than it knows what to do with.

Citiraj:

We have previously mentioned that we thought that Intel was looking for government help and maybe a handout which was touched upon in Pat Gelsinger’s interview, up front.
While certainly not directly asking for money, it certainly sounds like Intel wouldn’t say no. TSMC is clearly in the drivers seat and is not likely to change any time soon.

Izvor: The Oregonian i SemiWiki

Citiraj:

Intel’s Chief Revenue Officer: We have silicon, but shortages in Wi-Fi, substrates, panels

Citiraj:

Supply has been a topic since 2019, really since the end of 2018. Intel continues to increase our capacity by two fold over those years. We’re continuing to build more and more, amd we’re continuing to increase our investments in CapEx as you heard Pat talk about. The real thing here is that demand continues to be strong and with COVID and the pandemic it has gotten even stronger. But what we’re now seeing is that there are new industry and ecosystem challenges where the rest of the component ecosystem can’t keep up. Whether it’s Wi-Fi components, substrates, panels, those are now kind of the bottleneck to the next level of explosive growth. So you might be able to find a CPU, but you may not be able to find a panel, or a battery, or some other component to actually be able to finish that kit. We have plenty have silicon dies and I want everyone to know that those investments are absolutely paying off, and now we have got to go work on all the ecosystem pieces.

Izvor: AnandTech

The Exiled · 05.05.2021., 08:39

Exclusive info about Intel’s Alder Lake-S – Will the Intel Core-1800 be a threat to AMD?

Citiraj:

Intel wants and needs to set with Alder Lake as 12th generation a big mark in the mainstream market for desktop CPUs this year, if they don’t want to lose the connection (also from their target group’s point of view). And it is not only a repainted and further upgraded 14 nm CPU, but in addition to the smaller structure width of 10 nm (according to Intel’s reading), it is finally also a completely new architecture with a completely different solution approach. In the ES of the Core-1800 in the (not final) B0-stepping presented today, there are 8 large Golden Cove cores (Core) and 8 small Gracemont cores (Atom) on the die. While the big cores are also capable of hyperthreading, this is not the case with the Atom cores. In total, you’ll find 16 cores that can process a maximum of 24 threads. According to various leaks, Intel quantifies the performance leap compared to a conventional solution (unfortunately not specified) with up to 20 percent in pure single-thread scenarios and a doubling of the multi-thread performance in the ideal case, which seems quite optimistic when you consider a current 8-core of the 11th generation.

The processor, internally called Intel Core-1800, sits in the new Socket V (LGA 1700) and has a TDP of 125 watts. The screenshot below, which I modified slightly to protect the source, shows a time of 56 seconds for the PL1 and a peak time of 2.44 ms for the PL2 of a whopping 228 watts. That this up- and down-switching is well suited to aggravate the already nervous load change behavior of the current Intel CPUs even more doesn’t even have to be mentioned. This is where the voltage converters of the mainboards come into play. The revision B0 of this ES is still quite early and the base clock is also still quite low with only 1800 MHz. More interesting are the Turbo Boost limits of the big cores, where 1-2 cores are already supposed to reach 4.6 GHz, with 3-4 cores it’s still 4.4 GHz, with 5-6 cores 4.3 GHz and with all 8 cores then 4 GHz. The small Atom cores clock with up to 3.4 GHz with <=4 used cores, otherwise only with 3 GHz.

According to the initial information, the new desktop CPUs will support up to dual-channel DDR5-4800 memory and the DDR4-enabled motherboards will support modules up to 3200-MHz.
Supposedly, only the high-end Z690 motherboards will offer DDR5 support, while all cheaper variants will still rely on DDR4.

Izvor: Igor's Lab

Odinarius · 05.05.2021., 12:58

Dat ću samo svoje kratko mišljenje o malim jezgrama... Mislim da to nije loš potez.

Naime, igre rijetko kad pune sve jezgre, a dobitak od dodatnih threadova je marginalan. Dakle, kad se zadovolji potreba za glavnim threadovima preko velikih jezgri, dodatni threadovi nisu napravljeni tako da pune te dodatne threadove 100%. Zbog toga ima smisla da se u prostor i power budget jedne velike jezgre (2 threada) ubace 3 male. Ako svaki dodatni thread rastereti glavne za 5%, to je 5% dobitka povrh velike jezgre za isti floorplan i power budget. U nekim situacijama tu bi i 4t pristup možda osigurao isto...

Naravno, tu su problemi s Windows schedulerom koje treba riješiti i sve to, ali ako se softverski ispravno implementira, i naravno ne bude problema s latencijama i tko zna kojim drugim hardverskim izazovima, trebalo bi biti ok. Uz naravno sve prednosti power savinga koje to donosi u mobilnim uređajima.

Naravno, osim ako neka DX12 magija u novim engineima ne riješi te probleme multithreadinga. Plus, i dalje smo ovisni o toj prvoj glavnoj jezgri koja ne smije zaostati. Bit će zanimljivo to pratiti kad izađe i gdje je dobitak, a gdje gubitak. Prva značajna arhitekturalna promjena u Intelu u tko zna koliko godina (od Ringbusa vjerojatno)

The Exiled · 05.05.2021., 13:13

Nije loš potez i ARM-ovo orginalno big.LITTLE rješenje je dokaz da takva implementacija itekako funkcionira, pogotovo kad su mobilna rješenja u pitanju. IMHO ovo kaj Intel radi, ne mora nužno biti na tom putu.

Prvenstveno, jer eto čak i s novim 10nm procesom, opet moraju ići na visoke taktove, kaj ponovno podiže potrošnju, tako da na kraju mogu reći da imaju 20% poboljšanja u odnosu na prijašnju generaciju.

Alder Lake bi sasvim lako mogel ispasti kao Rocket Lake po pitanju utjecaja na aktualno tržište, a Intelu zapravo treba arhitektura na tragu ili bolja od konkurencije, tako da ne moraju posezati za velikim i malim jezgrama.

Odinarius · 05.05.2021., 13:37

Jedini je problem što je s tehnološkim vodstvom AMD jednako dekadentan kao i Intel na cjenovnom planu. Sad čekamo Nvidiu da s napravi nešto s ARM jezgr... Oh shit, a tek Nvidia

The Exiled · 05.05.2021., 13:41

Jebiga, AMD sad nema nekog prevelikog razloga za spuštanje cijena, možda ih korigiraju na jesen, ako bude Alder Lake (u igrama) blizu ili bolji od npr. R5 5900X. Kaj se nVidije tiče, to je sasvim drugi brod i ocean u odnosu na ovaj dvojac.

Odinarius · 05.05.2021., 14:16

Znam da nije x86. Ako će Nvidia ići u smjeru nekog gaming proizvoda s novim ARM portfeljem, neka konzola je prvi izbor. Zašto ne sad kad imaju sve in-house. Nešto moćnije od Shielda, naravno. GPU dio već imaju posložen što znači da je pola posla već obavljeno, igre bi morale prilagoditi infrastrukturu ARM-u, što nije nemoguće.

Ne znam tko bi na srednji rok mogao ući u CPU područje, osim možda Nuvie. Ako serveri počnu prelaziti na ARM ili u budućnosti Nuviu, to bi se moglo trickle downati na desktop jednog dana i razbiti duopol.

The Exiled · 05.05.2021., 14:27

nVidia hoće još veći dio serverskog tržišta, zato im tu ARM akvizicija dolazi ko naručena, dok je nVidijina GPU dominantnost na AI/ML/HPC području ujedno i razlog zbog kojeg Intel sad izmišlja (toplu vodu) i vlastita GPU rješenja.

Po uzoru na Tegra rješenja za Nintendo Switch, lako mogu složiti jače platforme bazirane na Ampere i budućim generacijama koje pak prate moćnije ARM jezgre isto tak bazirane na najavljenim Neoverse implementacijama.

Kaj se Nuvie tiče, kupil ih je Qualcomm, tak da smo tu opet na ARM modelima kak god se okrene. nVidia svoj dio sigurno ne pušta, tj. grabi još više, AMD više nije šrot kakav je bil, jedino ostaje Intel da se skocka ili nek opet nađu novog glavnog direktora.

Mladenxy · 05.05.2021., 16:49

Citiraj:

Autor Odinarius

Dat ću samo svoje kratko mišljenje o malim jezgrama... Mislim da to nije loš potez.

Naime, igre rijetko kad pune sve jezgre, a dobitak od dodatnih threadova je marginalan. Dakle, kad se zadovolji potreba za glavnim threadovima preko velikih jezgri, dodatni threadovi nisu napravljeni tako da pune te dodatne threadove 100%. Zbog toga ima smisla da se u prostor i power budget jedne velike jezgre (2 threada) ubace 3 male. Ako svaki dodatni thread rastereti glavne za 5%, to je 5% dobitka povrh velike jezgre za isti floorplan i power budget. U nekim situacijama tu bi i 4t pristup možda osigurao isto...

Naravno, tu su problemi s Windows schedulerom koje treba riješiti i sve to, ali ako se softverski ispravno implementira, i naravno ne bude problema s latencijama i tko zna kojim drugim hardverskim izazovima, trebalo bi biti ok. Uz naravno sve prednosti power savinga koje to donosi u mobilnim uređajima.

Naravno, osim ako neka DX12 magija u novim engineima ne riješi te probleme multithreadinga. Plus, i dalje smo ovisni o toj prvoj glavnoj jezgri koja ne smije zaostati. Bit će zanimljivo to pratiti kad izađe i gdje je dobitak, a gdje gubitak. Prva značajna arhitekturalna promjena u Intelu u tko zna koliko godina (od Ringbusa vjerojatno)

Kad su igre u pitanju sve što traži neke dodatne prilagodbe, programiranja... piši kući propalo. Pre malo će biti takvih korisnika da bi se to developerima isplatilo osim ako im Intel ne da direktnu paru za taj posao.

Sn4k3 · 05.05.2021., 19:00

31.03.2021., 20:10	#4471
The Exiled McG Moj komp Datum registracije: Feb 2014 Lokacija: Varaždin Postovi: 6,773	To bude tek šou, jer sljedeće tri generacije im imaju taj hibridni dizajn. Ostaje da vidimo kud to vodi, ali zasad (i dalje) samo pričaju i obećavaju.

05.05.2021., 12:58	#4474
Odinarius All-rounder Moj komp Datum registracije: Sep 2013 Lokacija: Zagreb Postovi: 1,160	Dat ću samo svoje kratko mišljenje o malim jezgrama... Mislim da to nije loš potez. Naime, igre rijetko kad pune sve jezgre, a dobitak od dodatnih threadova je marginalan. Dakle, kad se zadovolji potreba za glavnim threadovima preko velikih jezgri, dodatni threadovi nisu napravljeni tako da pune te dodatne threadove 100%. Zbog toga ima smisla da se u prostor i power budget jedne velike jezgre (2 threada) ubace 3 male. Ako svaki dodatni thread rastereti glavne za 5%, to je 5% dobitka povrh velike jezgre za isti floorplan i power budget. U nekim situacijama tu bi i 4t pristup možda osigurao isto... Naravno, tu su problemi s Windows schedulerom koje treba riješiti i sve to, ali ako se softverski ispravno implementira, i naravno ne bude problema s latencijama i tko zna kojim drugim hardverskim izazovima, trebalo bi biti ok. Uz naravno sve prednosti power savinga koje to donosi u mobilnim uređajima. Naravno, osim ako neka DX12 magija u novim engineima ne riješi te probleme multithreadinga. Plus, i dalje smo ovisni o toj prvoj glavnoj jezgri koja ne smije zaostati. Bit će zanimljivo to pratiti kad izađe i gdje je dobitak, a gdje gubitak. Prva značajna arhitekturalna promjena u Intelu u tko zna koliko godina (od Ringbusa vjerojatno)

05.05.2021., 13:13	#4475
The Exiled McG Moj komp Datum registracije: Feb 2014 Lokacija: Varaždin Postovi: 6,773	Nije loš potez i ARM-ovo orginalno big.LITTLE rješenje je dokaz da takva implementacija itekako funkcionira, pogotovo kad su mobilna rješenja u pitanju. IMHO ovo kaj Intel radi, ne mora nužno biti na tom putu. Prvenstveno, jer eto čak i s novim 10nm procesom, opet moraju ići na visoke taktove, kaj ponovno podiže potrošnju, tako da na kraju mogu reći da imaju 20% poboljšanja u odnosu na prijašnju generaciju. Alder Lake bi sasvim lako mogel ispasti kao Rocket Lake po pitanju utjecaja na aktualno tržište, a Intelu zapravo treba arhitektura na tragu ili bolja od konkurencije, tako da ne moraju posezati za velikim i malim jezgrama.

05.05.2021., 13:37	#4476
Odinarius All-rounder Moj komp Datum registracije: Sep 2013 Lokacija: Zagreb Postovi: 1,160	Jedini je problem što je s tehnološkim vodstvom AMD jednako dekadentan kao i Intel na cjenovnom planu. Sad čekamo Nvidiu da s napravi nešto s ARM jezgr... Oh shit, a tek Nvidia

05.05.2021., 13:41	#4477
The Exiled McG Moj komp Datum registracije: Feb 2014 Lokacija: Varaždin Postovi: 6,773	Jebiga, AMD sad nema nekog prevelikog razloga za spuštanje cijena, možda ih korigiraju na jesen, ako bude Alder Lake (u igrama) blizu ili bolji od npr. R5 5900X. Kaj se nVidije tiče, to je sasvim drugi brod i ocean u odnosu na ovaj dvojac.

05.05.2021., 14:16	#4478
Odinarius All-rounder Moj komp Datum registracije: Sep 2013 Lokacija: Zagreb Postovi: 1,160	Znam da nije x86. Ako će Nvidia ići u smjeru nekog gaming proizvoda s novim ARM portfeljem, neka konzola je prvi izbor. Zašto ne sad kad imaju sve in-house. Nešto moćnije od Shielda, naravno. GPU dio već imaju posložen što znači da je pola posla već obavljeno, igre bi morale prilagoditi infrastrukturu ARM-u, što nije nemoguće. Ne znam tko bi na srednji rok mogao ući u CPU područje, osim možda Nuvie. Ako serveri počnu prelaziti na ARM ili u budućnosti Nuviu, to bi se moglo trickle downati na desktop jednog dana i razbiti duopol.

05.05.2021., 14:27	#4479
The Exiled McG Moj komp Datum registracije: Feb 2014 Lokacija: Varaždin Postovi: 6,773	nVidia hoće još veći dio serverskog tržišta, zato im tu ARM akvizicija dolazi ko naručena, dok je nVidijina GPU dominantnost na AI/ML/HPC području ujedno i razlog zbog kojeg Intel sad izmišlja (toplu vodu) i vlastita GPU rješenja. Po uzoru na Tegra rješenja za Nintendo Switch, lako mogu složiti jače platforme bazirane na Ampere i budućim generacijama koje pak prate moćnije ARM jezgre isto tak bazirane na najavljenim Neoverse implementacijama. Kaj se Nuvie tiče, kupil ih je Qualcomm, tak da smo tu opet na ARM modelima kak god se okrene. nVidia svoj dio sigurno ne pušta, tj. grabi još više, AMD više nije šrot kakav je bil, jedino ostaje Intel da se skocka ili nek opet nađu novog glavnog direktora.

05.05.2021., 19:00	#4481
Sn4k3 Premium Moj komp Datum registracije: Jan 2008 Lokacija: Zd Postovi: 4,681