@Gideon
Does it mean that ARM has taken over that market, or does it mean that cloud isn't buying as much hardware as expected? It sure isn't Ampere dominating.
Amazon has said that over 50% of their cloud capacity over the past two years has been Graviton processors.@Gideon
Does it mean that ARM has taken over that market, or does it mean that cloud isn't buying as much hardware as expected? It sure isn't Ampere dominating.
I thought maybe it had to do with per core software licensing.
Anad what were the other half ?Amazon has said that over 50% of their cloud capacity over the past two years has been Graviton processors.
I don’t think they said. The 50% number came from AWS re:Invent 2024 presentation by Mark Garman, AWS CEO.Anad what were the other half ?
Okay but that's only one (admittedly large) cloud provider. What's that in terms of the total market? Not 50%.Amazon has said that over 50% of their cloud capacity over the past two years has been Graviton processors.
Yea, this fits the overall picture. Due the price/perf of "AWS *g" it doesn't make much sense to run on x86. Even more so for AWS-managed services (RDS, Lambda, Kafka, etc.).Amazon has said that over 50% of their cloud capacity over the past two years has been Graviton processors.
Amazon has 30% market in cloud per one analysis so that is about 15% of total market.Okay but that's only one (admittedly large) cloud provider. What's that in terms of the total market? Not 50%.
Amazon has 30% market in cloud per one analysis so that is about 15% of total market.
The Chinese cloud providers are moving to RISC-V. Alibaba is on their third iteration of RISC-V.
Chips and Cheese did an article about the first one, the 910 and it didn’t fair well.Okay, thanks for putting a number on it. That kind of market squeeze would explain why Sierra Forest in particular is in trouble. It might also explain the rumours that Clearwater Forest may be the last of its kind.
Has any of this hardware been benchmarked?
Chips and Cheese did an article about the first one, the 910 and it didn’t fair well.
Intel announces Lip-Bu Tan as new CEO, effective March 18th.
Yup, I was just reading about that. It seems he was in favor of more layoffs. From August 27th, 2024 after he left the board:Pretty notable as this is the former board member who resigned after being critical of how Intel was/is run and butting heads with Gelsinger.
articleTo Tan and some former Intel executives, the workforce appeared bloated. Teams on some projects were as much as five times larger than others doing comparable work at rivals such as Advanced Micro Devices, according to two sources. One former executive said Intel should have cut double the number it announced in August years ago.
Tan has told people he believed Intel was overrun by bureaucratic layers of middle managers who impeded progress at Intel’s server and desktop chips divisions and the cuts should have focused on these people.
Well, looks like more pain ahead before growth, but at least Lip-Bu wants to cut out the unnecessary middle managers. Intel is simply bloated and cannot pivot or react fast enough.Yup, I was just reading about that. It seems he was in favor of more layoffs. From August 27th, 2024 after he left the board:
article
That's the easiest part to blame but Intel is generally very bloated.unnecessary middle managers.
Very funny development since the BoD didn't listen to him while he was part of it, so what will be the change when he's CEO? The BoD ended up firing Pat without base in reality, it will eventually do the same with LBT. Let's see what he manages to change until then (changes visible right away since we can't expect the Intel BoD to be able to look further than one quarter).Intel announces Lip-Bu Tan as new CEO, effective March 18th.
Board was on Pat's side until they realized that all Pat's predictions were BS and the company was going bankrupt. Now they are crawling with their tails between their legs back to Lip-Bu.Very funny development since the BoD didn't listen to him while he was part of it, so what will be the change when he's CEO? The BoD ended up firing Pat without base in reality, it will eventually do the same with LBT.
Pat was spending money he didn't have in hopes of beating TSMC. LBT is going to aggresively cost cut and make the company profitable, even if it means being uncompetitve for the forseable future. He's the Intel version of Rory Read.Very funny development since the BoD didn't listen to him while he was part of it, so what will be the change when he's CEO? The BoD ended up firing Pat without base in reality, it will eventually do the same with LBT.
lowkey wanna make bets on how much of NVL survives this yearPat was spending money he didn't have in hopes of beating TSMC. LBT is going to aggresively cost cut and make the company profitable, even if it means being uncompetitve for the forseable future. He's the Intel version of Rory Read.
16+32 is dead for sure.lowkey wanna make bets on how much of NVL survives this year
Well it's just a 2t config.16+32 is dead for sure.