Home
Creators
Search Recent 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

Telegrams for Guests: Dispelling Hardware Myths with a Software Developer (Patreon)

Published:
2021-01-24 00:07:44
Edited:
2021-01-24 00:17:19
Imported:
2026-06
Tags:

Content

The next episode of Broken Silicon will be a quite open discussion with an Anonymous Software Developer about anything and everything related to why certain decisions are made by Semiconductor & Gaming Hardware companies.  Planned subjects include:

Why wasn't RDNA 2 an MCM design?  Why is it taking so long to get a "Zen-like GPU" from AMD?

Is RDNA 2 going to age better at Ray Tracing than Ampere?

Why did the XBOX Series X go with a split memory system?

Multi-Threading Performance in Modern Games

Zen 4 and RTX 4000 launch possibilities based on how long it takes to go from tape-out to launch

MUCH MUCH MORE!

As you can see - the things we plan to discuss are very open ended, and that's because our guest has 25 years of professional experience in software and CPU design.  We aren't intending to get as technical as the last guest episode, but we can probably answer a little bit of everything.

Put your comments, questions, and thoughts below!  We just might get to them on Broken Silicon 85!  [NOTE: Questions will be considered till the end of Sunday]

Comments

Vipast

What is the legitimate situation concerning VRAM with the current generation of games and games within the next 2 to 3 years as those are the games that are being developed with current-generation graphics and consoles in mind. How long do you think the current console's RAM and VRAM of the current 3080 will hold up? Do you know if DLSS helps the VRAM with Nvidia as it's lowering the render resolution and then upscaling?

Deco

Does AMD's implementation of RT cores, that is scaling them on a per-CU basis instead of the monolithic approach that Nvidia uses by segmenting a separate area on the die for RT cores mean that with future generations of RDNA AMD will be able to scale performance, specifically with RT better than Nvidia (excluding gains that Nvidia can make by enhancing the RT cores themselves, purely in terms of a design choice).

Vipast

What types of incentives do the hardware companies provide for companies or programmers to "optimize" for their hardware? On the same topic, how much optimization is really needed at the game engine level over just allowing the hardware drivers do the optimizations of the API/Game Engine.

Tovmach

Because of the very slow start of populating the market with RTX 3000- series, does it mean that the launch of the RTX 4000 series will be postponed by a year extra into the future?

AdanFS

Is it possible for companies to order 1.5 GB memory chips? If so, then why are they not popular? A 192 bit card with 9 GB or a 12 GB 3070 would sell well.

evan .froelich

My question is about speed. With AMD seeming to be hell bent on almost apple levels of product launches combined with the fact that they dont seem to be slowing down their progress in how good these new designs are it got me thinking about lovelace and how it seemingly came to be after hopper was floating around but is rumored to come before hopper. is this nvidias hand being forced to follow amds lead when it comes to smaller but more consistent product refreshes much like how iphones are released? do you think thats what intel is also doing with rocket lake and alder late being only 6-8 months apart? Is this change in product cadence sustainable?

DeusNightFire

As AMD buys Xilinx and previously Intel bought Altera 2 years ago. How can AMD or Intel benefit providing their research or previous product to improve on the FPGA products? Can we see some sort of revolution of Zen-like on FPGA or even FPGA used on more consumer products like consoles, pc desktops, or laptops to improve workflow or gaming? My knowledge of FPGA is limited and I would like to understand them for work purposes like a small explanation on FPGA. Thanks!

Eleriam

What's going on with Horizon Zero Dawn on the PC? We have to waste some 10 minutes every time we launch the game for the game to "Optimize shaders", while it's clearly underutilising the CPU cores, RAM, GPU, or SSD. Then they added pillarboxing on ultrawide resolutions whenever there's a cut-scene, even though it's in-game (not pre-rendered or anything), so there's no logical explanation why they can't use the full 21:9 aspect ratio. And THEN, with the latest patch 1.10, they introduced a bug on all Radeon cards that randomly disappear whole forests and foliage, on and off! It's especially worse on Navi 21 cards! How can we not let this happen again in PC Gaming?

Jay

Has the semi-annual Windows 10 release schedule affected any planned hardware features or architecture level decision? These releases typically bring new features or at least api and compatibility changes. Something like hardware accelerated gpu scheduling comes to mind

10GHz Netburst

One of the main barriers to making significant changes in hardware architecture, such as changing ISAs or overhauling aspects like the memory hierarchy, is the fact that it requires significant adjustment by programmers. Existing code often needs to be rewritten, certain practices and techniques to obtain better performance need to be thrown out in place of new ones, etc. From the perspective of a programmer what do you think are the best ways to reduce or minimise how onerous such large scale overhauls are for the people writing code for our computers? Also, do you think that there are changes needed in the philosophies taken by programmers to be more flexible towards adopting new paradigms?

Conor Hoary

Do you think Ray tracing will cause this generation of graphics cards to age poorly over all? Currently there is a lot of debate about how relevant Ray-Tracing will be in the future. Reviewers hammer home the point that Nvidia cards can run X games at 40 fps with Ray-Tracing high while the equivalent AMD cards are 20 fps . In these examples both cards look bad to me especially given they are top tier cards. If Ray-tracing becomes much more relevant I would think this generation will actually age badly over all as they are essentially first gen, or do think it is possible Ray-tracing will be more efficiently implanted in the future in a way that may help keep this generation relevant. -- 2nd question Do you think Ray-Tracing will accelerate the ageing of current older graphics cards? an example is I know a few people (including myself) who have a GTX 970 a 7 year card that can do fine with most new games at 1080 60 depending on settings a lot are looking to upgrade but are happy enough to wait for the crazy prices to go away first. Do you think Ray Tracing could kick the chair from under these cards(1080,1680 etc) and push people to upgrade.

Screw Dropper

What use cases will there be for PCIe 5.0 and 6.0 for both SSDs and graphics cards? Do you think developers will start to take advantage of them, or do you think graphics cards will start to have only 4 lanes because they can't take advantage all the bandwidth?

David Cowden

What are the advantages of AMD moving to a new socket with Zen4?

David Cowden

What’s your take on what happened with Halo Infinite (and to an extent CP2077)? It almost seems like games attempting to be cross generation look worse than either a polished last gen game or what we expect a true next gen game to look like (Unreal 5 demo). Are these product desires to build generation-bridging games more than the industry can chew?

Salazar

when should we expect stock levels of new gpus Tom ?

Doomberry

I'd love to hear about the PS4/Xbox one release and specifically how underpowered their CPUs were. It certainly seemed to catch some devs by suprise, and there are a few games from around that time which are known for being downgraded at release. Some big examples that come to mind are: The Order: 1886, Watch Dogs and Dark Souls 2. What caused game devs to overestimate these consoles? Was the messaging from console manufacturers unclear?

Manuel Nascimento

What work and how much work is optimising a game to run better really involve? In the case of “poorly optimised” games, is it a case that it was just too much work to get it to run better or is it mostly pressure to release the game earlier rather than later? (Or maybe something else). Thanks!

Aaron Capell

I know this isn't traditional "gaming hardware" but do you know anything that you can share about AMD's team up with samsung to produce gpus for their phones?