Home
Creators
Search Recent Missing Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

Discussing SPR, GNR, Zen 4, i9-13900KS w/ Level1Techs (Patreon)

Published:
2023-01-21 05:08:19
Imported:
2026-06
Tags:

Content

Broken Silicon's next guest will be Wendell from Level1Techs.  We plan to discuss:

Intel Sapphire Rapids

AMD Genoa, Genoa-X, and Bergamo

Intel vs AMD server strategies over the next 3 years

Intel Granite Rapids

Intel Sierra Forest

i9-13900KS vs Zen 4 X3D

Anything else you guys want!

Wendell is able to speak to just about any subject related to silicon, so don't be shy! Put your thoughts, questions, and comments below!  Just be sure you are concise, use good grammar, and be respectful to have your submissions considered.

You ~36 Hours (Till Noon 1/22/23 US Central Time) to submit below!

Wendell's YT: https://www.youtube.com/channel/UC4w1YQAJMWOz4qtxinq55LQ

Previous Episode: https://youtu.be/MYJ_9zfqWUg

Comments

Reginald Aryee

Will operating systems ever stop using/requiring a page file/swap? Ram is not really a limiting factor like it was back in the day. The last time I disabled it on Windows I eventually ran into a compatibility issue with a video game and had to re-enable it.

Prior is the theme 999

Hello Tom and Wendell, with the 13900KS being intel’s recent cpu launch that is truthfully disappointing, it looks like they finally ran out of levers to pull. Previously they had the node shrink advantage that they could always rely on until they couldn’t, increasing the tdp gave them the extra juice (pun intended) to reach this point. The recent KS I’d argue isn’t better than the regular K at all, what can intel do going forward, server, mobile, and soon to be desktop just seems one sided. Are there any other levers they can pull? Or can they just keep increasing TDP? It seems the next few years will be AMD dominating and I really don’t want only one company dominating for any period of time. Healthy competition is always good.

Xavbeat03

Will Server APU's become the future? Or will they continue to use separate CPU's and GPU's?

panic582d

Hi Tom and Wendell, I have a couple of datacenter-specific questions: 1. What upcoming datacenter-specific trends and technologies are you personally most excited by? 2. Is there anything coming down the pipe (can be technological, geopolitical, social etc) that you think will play an outsized role in defining the datacenters of the future? 3. What do you presently see as Intel's competitive edge in datacenter CPUs? What about AMD's? Where do you see nVidia's new CPU's (Grace) niche being?

10GHz Netburst

Many people seem to complain about the lack of aggression on AMDs side in the consumer markets, particularly the GPU side. I think this comes mainly down to a failure to recognise the internal competition of these consumer product lines against the far higher revenue per wafer of AMDs server products. What are your thoughts on this regard? Do you think AMD should be more aggressive in their consumer product lineups? Or do you think consumers should just accept that the consumer products at the end of the day are of secondary importance?

10GHz Netburst

It's been 3 months since the A770 and A750 launch, what are your thoughts on how these products have evolved since initial reviews? Anything that surprised you?

Svein Sæther Jensen

When do you think the 7900xtx will be in stock(eu), or what are your thoughts on waiting for the 7980xtx? When will they launch it and what will the price/performance be?

Aluveitie

Hello Tom and Wendel. Do you know where Ice Lake Xeon D is at? Intel announced it February 2022 and the only thing I can still find are the launch reviews from back then. But actual products are nowhere to be found, did Intel silently “unlaunch”?

Timo H

Hi Wendell Are you going to get MI300 once that launches in retail, to your home/office workstation?

QuickJumper

Analysts and investors predict that by the end of 2023 amd could reach 30% of server market share. Could AMD reach 40% or even 50% before intel gets competetive again? Also Tom you keep mentioning that AMD is like 1/5 of Intel. In reality AMD's expected revenue is 23 billion and Intel's 63 billion. So AMD isn't small anymore even compared to Intel.

BaconHouse Collective

When do we project server grade CPUs will break the 6.0 GHZ clock speed barrier? With this in mind, do you feel we will see a shrinking or widening clock speed gap between server and consumer as nodes improve? I'm analyzing the future of cloud based gaming models and whether the enthusiast market moves towards "home" DIY cloud based gaming servers as a replacement for these services.

shea

How likely is it for ASrock's X670 expansion card to get a consumer release? Glad to see Wendell back!

Chris Rijk

The accelerators built into Sapphire Rapids are certainly useful for a number of tasks, but I feel that Intel has made a mistake by deciding to charge extra for them with their "On Demand" programme. I think this will put two doubts into customers' minds - about whether these accelerators will continue to exist in future and whether better alternatives exist for the money. In fact, I'd expect customers who would particularly benefit from them to already have alternatives in the form of PCIe cards and the like. Are Intel shooting themselves in the foot?

Chris Rijk

Hi Wendell, do you personally know of any companies who have started buying AMD based servers instead of Intel in the last few years? Have they only bought AMD servers since then and if not do you have any idea why, such as AMD servers lacking something? Would you expect the release of Sapphire Rapids to tempt such customers back? If we consider the next two years, which do you think would be a more significant product, Bergamo (which doubles down on AMD's lead in cloud) or Siena (which potentially opens up a new market for AMD)? Does AMD need to offer 4 socket and larger servers?

Bart

Hi Wendell and Tom, I have two questions on a similar topic. First, what accelerators for new cpus does Wendell think the server market can make good use of/which is he most looking forward to? Second, with the integration of fpgas into cpus in the future, are people already building tools/products to help program them for different use cases? I think we can't expect most server users to get the expertise to do that themselves.

Dr Forbin

It seems to me that Intel had to have an answer to Zen 4 X3D and according to reviews thus far its a total crap show. The 13900ks is a space heater that barely in the best case has 4 to 5 percent improvement from the 13900k. Intel shouldn't have ever marketed this CPU!

Chris Rijk

While the MI300 is obviously a successor to the MI200 series, might it be better thought of as a new parallel platform to EPYC? For example, the reference design that AMD showed had 3 sets of CDNA compute and 1 set of Zen 4 compute with 24 cores, but based on what AMD has said before they could just as easily make a configuration with 4 sets of Zen 4 cores giving it a total of 96, the same as Genoa, but with 128GB of HBM3 memory. Paired with CXL based memory, that could make for an interesting server for a number of scenarios. Do you see any other interesting uses for MI300 or the packaging technology behind it? (*) the reference MI300 design: https://cdn.videocardz.com/1/2023/01/AMD-MI300-1.jpg

PCDog

Hey Tom and Wendell. Linux gaming has come a long ways in the past few years. The introduction of the Steam Deck has helped increase the Linux user base. Do you see Linux gaming reaching feature parity with Windows in the near future (e.g. HDR support, Direct Storage, equal developer support, etc.)?

dj5k

Question for Wendell: Do you have any recommendations or experience to share for a good Linux laptop as a sysadmin / developer workhorse? Would you go AMD or Intel and any specific manufacturer, model or other suggestions to look into? (Must be x86) I recently made some sobering (read: disappointing) experiences with running Ubuntu on a Thinkpad p14s Gen3 with an alder lake i7 1270p. Battery life is meh, cooling fan is noisy even at low load and overall I am not impressed with it at around 2000 € (coming from an Intel MacBook Pro).

Nicholas Buckner

As a follow up, does Microsoft's Dzn mesa code hitting 99.75% vulkan 1.0 compliance make this easier to develop for through WSL?

Nicholas Buckner

How useful are accelerators if they aren't on every model that generation, and possibly require an upcharge? Consumer seems to require a hardware feature to be present several generations before developers target for it. Enterprise can be more nimble, but would they?

Max Eliaser

Hi Tom, and hi Wendell -- Love your work! Wendell, you have often described yourself as a "computer janitor," but what does that look like in practice? What would be a typical work assignment or contract for you? I get the impression that tech companies hire you to implement, profile, and optimize complex workflows involving multiple software and hardware components, including having you apply patches to open source components to enable or optimize the customer's use case. How accurate is this perception, and how often do these patches get upstreamed? How does one break into the computer janitor biz?

Max Eliaser

Another question for Wendell: When Alder Lake came out, you reported that the heterogeneous core microarchitecture was causing scheduling problems on Linux. This was surprising to me, because Android (which uses the Linux kernel) has primarily targeted big.LITTLE ARM devices for over a decade. I had expected that these codepaths in the Linux scheduler would have been pretty much plug-and-play with Alder Lake. Can you talk about how Alder Lake and Raptor Lake differ from big.LITTLE on ARM, how that affected the requirements for the kernel, and what the obstacles were (or are) to applying CFS to ADL/RPL? Also, what is the current state of play, is this working better in Linux now? Do you think Linux will have good scheduling on the heterogeneous 7950X3D on launch day?

Durmij

Hello Wendell and Tom, Can you talk about AMDs GPUs in workstations, particularly and insights/predictions about near future viability and competitiveness? It's really rough out here with no alternative to CUDA,

AWS Danny

Hey Wendell and Tom! Very excited for this podcast and I hope you both are well. AWS is pushing Graviton instances both externally to customers and internally to service teams with increasing incentives to get everyone on board. Do you ever hear anything from AMD or Intel engineers worried about Amazon putting this kind of pressure on the server market with their own ARM based chips in a way Apple never could?

Robert S Barnes

Hi Guys, how close are we to a situation where we have a generic substrate and can order custom heterogeneous compute "chips"? For example, I'd like 16 cores of Zen4, 32 workgroups of CDNA3, an FPGA and an AI inference accelerator and 64 GB HBM3e please, all on the same substrate in the same socket. With the SuperAPU that was shown off at CES that seems really close.