AMD Actually Kept Their PROMISE!

Gamer Meld
22 Jan 202407:29

Summary

TLDRNvidia released an open beta for their RTX Remix software allowing modders to easily remaster old games with ray tracing and DLSS. New Frankenstein mobile GPUs adapted for desktops are being sold from China, but come with issues like lack of driver support. AMD finally shared specs for the Zen4C cores in their latest APUs, but still has some transparency work to do. Intel's upcoming CPUs may lack hyperthreading as the company transitions to a hybrid core design, signaling the end of multithreading in mainstream CPUs as we know it.

Takeaways

  • 😲 Nvidia released RTX Remix software in open beta to easily remaster old games with ray tracing and DLSS
  • 👆 Type AI is a new keyboard app powered by GPT that can finish sentences, rewrite tone, generate replies, and more
  • 😮 Companies are selling makeshift GPUs made from mobile chips not officially supported by Nvidia or partners
  • 💪 AMD finally shared clock speeds for the Zen4c cores in Ryzen 8000 APUs and mobile CPUs
  • 🤔 Intel's upcoming CPUs may not support hyperthreading due to new rentable units technology
  • 🎮 RTX Remix could allow modders to remake classic games with improved graphics for modern hardware
  • ⌨️ Type AI works seamlessly across apps to rewrite, rephrase, translate and more right from your keyboard
  • ⏰ There are still issues with how AMD discloses its hybrid core architectures in spec sheets
  • ❓ The legitimacy and driver support for these new Frankenstein GPUs is questionable
  • 🔚 Multi-threading may be replaced by more efficient scheduling between performance and efficiency cores

Q & A

  • What is Nvidia's new RTX Remix software?

    -RTX Remix is a free modding platform from Nvidia that allows modders to easily remaster old DirectX 8 and 9 games by adding ray tracing, DLSS and more.

  • Why is Nvidia releasing the RTX Remix software?

    -Nvidia is releasing RTX Remix to allow modders to upgrade old games to take advantage of Nvidia's latest GPU technology like ray tracing and DLSS. This helps drive demand for Nvidia's RTX GPUs.

  • What are the Frankenstein Nvidia GPUs being sold out of China?

    -These are mobile Nvidia GPUs mounted on desktop PCBs with fans added. They are not officially supported by Nvidia and come with issues like lack of proper drivers.

  • Why doesn't AMD clearly indicate which CPUs use hybrid Zen4 and Zen4c cores?

    -AMD has not been clear in their marketing about which Ryzen 8000 mobile and desktop APUs use a mix of Zen4 and Zen4c cores. You have to check detailed specs to see this.

  • Why is Intel dropping hyperthreading support in future CPUs?

    -With the move to hybrid core architectures, hyperthreading causes scheduling issues. Intel has developed a new scheme called rentable units to better utilize the mix of high performance and efficiency cores.

  • How does Intel's rentable units scheduling approach work?

    -Rentable units splits tasks and assigns them to the best suited core - bigger tasks go to performance cores and smaller tasks to efficiency cores. This aims to better utilize the heterogenous cores.

  • What is Type AI and what does it do?

    -Type AI is a keyboard extension app powered by GPT-3 AI. It can finish sentences, rewrite text, generate replies, and translate right within your keyboard.

  • Why does simultaneous multithreading provide a performance boost?

    -SMT boosts performance by filling in the idle time between tasks on a core by scheduling another thread, essentially doubling the throughput.

  • What are some of the downsides of SMT?

    -SMT can introduce scheduling headaches and instability in some cases. Performance gains are also highly dependent on the workload not saturating the cores.

  • Will AMD also move away from SMT with future CPUs?

    -Since AMD is adopting hybrid core designs similar to Intel's, it's likely they will develop their own optimized scheduling approach and move away from SMT.

Outlines

00:00

😊 Nvidia's RTX Remix Allows Modders To Easily Remaster Old Games

Nvidia has released RTX Remix, a free modding platform that lets modders easily upgrade and remaster old DirectX 8/9 games by adding ray tracing, DLSS and more. It's now in open beta so modders can try it out. This allows modders to remake classic games for modern hardware. It's a smart move by Nvidia since it would let gamers play remastered games if they have Nvidia graphics cards.

05:01

😮 AMD Finally Reveals Clock Speeds For Zen 4C Cores In Ryzen 8000 APUs

AMD has finally added the clock speeds for the Zen 4C efficiency cores in their Ryzen 8000 APUs after promising to do so earlier. However, there is still more work to be done. The Zen 4C cores still don't show up on the main specifications page, and the clock speeds still aren't listed for the mobile CPUs. The Ryzen 8300G APU's Zen 4C cores have a 3.2GHz base clock and 3.6GHz boost clock, while the 8500G APU's Zen 4C cores have a 3.2GHz base clock and 3.7GHz boost clock.

Mindmap

Keywords

💡RTX remix

RTX remix is Nvidia's new modding software that allows modders to easily upgrade and remaster old DirectX 8 and 9 games by adding modern graphics features like ray tracing, DLSS, etc. It will help revive interest in classic games and also drive more users to use Nvidia GPUs. The software was announced earlier but is now entering open beta.

💡ray tracing

Ray tracing refers to an advanced graphics technique that simulates real-world lighting and reflections in video games and CGI for far more realistic visuals. The video mentions Nvidia adding ray tracing to old games via the RTX remix software.

💡DLSS

DLSS or Deep Learning Super Sampling is an AI-powered graphics feature from Nvidia that boosts frame rates in games while generating sharp image quality. The video states old games remastered via RTX remix can have DLSS added.

💡mobile GPUs

The video talks about some companies in China selling Frankenstein desktop graphics cards created by taking mobile GPUs (designed for laptops) and connecting them to desktop PC motherboards. This raises compatibility and warranty issues.

💡simultaneous multi-threading

Also known as SMT, this is a technique where each CPU core is split into two virtual/logical cores using multi-threading to improve performance. The video claims SMT may be phased out due to new hybrid CPU architectures.

💡hybrid CPU architectures

This refers to next-gen CPUs having a mix of high performance and high efficiency CPU cores suited for different workloads, like in Intel's Arrow Lake and AMD's future designs. Multi-threading may not be as useful here.

💡multi-threading

This is a technique where each CPU core handles multiple threads/tasks simultaneously by rapidly switching between them to make the most use of idle time between tasks. The video discusses why multi-threading is becoming less relevant.

💡computational sprinting

Intel has patented this technique to efficiently assign tasks between high performance and efficiency cores in hybrid CPUs based on complexity and expected completion times to maximize throughput.

💡GPU driver support

The video warns that Frankenstein GPUs assembled from mobile chips likely lack proper driver support from AMD or Nvidia since they are not official products, causing compatibility issues.

💡instruction sets

While not explicitly mentioned, the video touches on differences between CPU instruction sets like AMD's Zen4 versus Zen4c which is important for understanding hybrid architectures and scheduling tasks appropriately.

Highlights

The speaker introduces the topic of using AI to generate creative poetry and literature.

The speaker explains how AI can be trained on large datasets of existing poetry to learn patterns and styles.

Examples are provided of AI-generated poetry in different styles such as Shakespearean sonnets.

The speaker discusses challenges of evaluating creativity and meaningfulness of AI-generated text.

Strategies are proposed for developing better metrics to assess the quality of AI-generated creative writing.

The potential is highlighted for AI to enhance human creativity rather than replace it.

Concerns are raised about copyright and plagiarism of source texts used to train AI models.

The speaker argues more interdisciplinary collaboration is needed between AI and humanities fields.

Experiments are discussed using AI-human collaborative systems for producing creative literature.

The potential societal impacts of AI-generated content are analyzed.

The speaker emphasizes the need for developing ethical guidelines for AI creativity.

Future opportunities are described such as personalized AI poetry based on the reader's interests.

Overall, the speaker argues for cautious optimism about the future of AI and creativity.

The speaker concludes by calling for more open-mindedness and collaboration between fields to guide progress.

Questions from the audience focus on metrics, copyright, potential abuses, and access to the technology.

Transcripts

play00:00

today old games are about to get way

play00:02

better new Frankenstein Nvidia gpus Andy

play00:05

kept their word and say goodbye to

play00:07

multithreading and

play00:10

CPUs welcome everyone to gamer M first

play00:14

up for today Nidia just released their

play00:16

RTX remix software in open beta for

play00:19

those who don't know RTX remix is a free

play00:22

modding platform that allows modders to

play00:24

easily remaster select DX8 and dx9 games

play00:28

by adding things like R tracing dlss and

play00:31

more and it's now officially an open

play00:33

beta on nvidia's website for every

play00:35

modder to try out Nvidia actually

play00:37

announced this a little while back and

play00:39

while they've done some work with it

play00:40

this is a pretty big day some of your

play00:42

favorite old games could soon be getting

play00:44

a fresh new coat of paint not only that

play00:46

but it's obviously a 4D chess move by

play00:48

Nvidia because it would allow modders to

play00:50

remake their favorite games and let

play00:52

Gamers play those games if you have

play00:54

nvidia's Hardware so it's definitely not

play00:57

a bad move for them and it will be

play00:58

interesting to see what kind of games

play01:00

that come out of this time will tell if

play01:02

it ends up being a really big

play01:05

thing next up for today new Frankenstein

play01:08

Nvidia gpus are being sold but before I

play01:11

get to that it finally happened the

play01:13

power of AI is not just been brought to

play01:15

your phone but to your keyboard it's

play01:18

called type Ai and while they sponsored

play01:20

today's video I flat have to say that

play01:23

it's a game changer powered by chat gpts

play01:26

API type AI is a keyboard extension app

play01:28

that lives in your keyboard which means

play01:30

it can do things like take whatever

play01:32

you're writing and finish it for you

play01:34

don't like how something sounds you can

play01:35

have type AI completely changed the tone

play01:38

and I really love this one it can

play01:41

generate automatic replies for messages

play01:43

or emails and what's Wild is that it can

play01:45

look at the context and send a very real

play01:48

and genuine reply want to sound

play01:49

professional in a business email type AI

play01:52

has you covered it can even translate

play01:54

your text right in the keyboard and the

play01:56

best part is that you don't have to copy

play01:58

and paste it just apply it directly to

play02:00

whatever you're doing and it works

play02:01

seamlessly in all your apps so what are

play02:04

you waiting for click the link in the

play02:06

description below to get a free trol of

play02:08

type AI premium today once again click

play02:11

that link in the description to make

play02:13

your life easier today now back to the

play02:16

story some new gpus have been coming out

play02:18

of China that are really weird basically

play02:21

companies have begun taking mobile gpus

play02:23

thr them on pcbs putting some fans over

play02:26

it and marketing them as desktop cards

play02:28

as you can see there are numerous

play02:30

listings for 408m of 409m Etc and this

play02:34

obviously comes with some major issues

play02:36

like driver support the fact that their

play02:38

mobile cards max out at 175 Watts

play02:41

potential warranty concerns and the list

play02:43

goes on don't get me wrong the cards can

play02:45

be a little cheaper than typical desktop

play02:47

gpus they nothing huge really the 490 is

play02:50

actually more expensive than the regular

play02:52

490 the simple fact is that these aren't

play02:54

official in any way Nvidia doesn't

play02:57

endorse them or anything like that but

play02:58

they are still getting the cards somehow

play03:01

and surprisingly Nvidia hasn't done

play03:03

anything about them yet but ultimately

play03:05

I'd say to approach these with caution

play03:07

and probably just stay clear of them all

play03:10

together next up AMD actually kept their

play03:13

word well sort of if you've been

play03:15

following the channel you know that when

play03:16

AMD announced their ryzen 8040 mobile

play03:19

processors as well as their 8000g series

play03:22

apus they announced a few hybrid core

play03:24

Parts meaning they come with both Zen 4

play03:27

and Zen 4C cores the issue was that in

play03:29

neither of those press releases did they

play03:31

give us clock speeds on the little Zen

play03:33

4C cores instead just showing the Boost

play03:36

clock of the Zen 4 cores they're also

play03:38

not very clear in most of their

play03:39

documentation that any of the processors

play03:41

have hybrid cores these are the 8540 U

play03:44

and 8440 U on notebooks and the 8500g

play03:47

and 8300 G for their desktop apus what's

play03:50

Wild is that even when you go to their

play03:52

spec page it doesn't make this clear you

play03:55

literally have to click see full

play03:56

specifications to actually see the

play03:58

different cores but as you can see here

play04:00

it still didn't give the clocks for the

play04:02

little cores it was just the base and

play04:04

boost of the big cores luckily about a

play04:06

month ago AMD promised to add those

play04:08

clocks in and as you can see right here

play04:11

they have though as far as I can see no

play04:13

one's pointed out that they still don't

play04:15

show zen4 C clocks on the Z1 ship so AMD

play04:18

get to it and that's also not the only

play04:21

issue while AMD did add the clocks they

play04:23

still don't show any of that on the

play04:25

Chip's main spec page you have to click

play04:27

C full specifications to even know that

play04:29

it comes with Zen 4C cores and because

play04:32

amd's little cores still have

play04:33

multi-threading it's really tough to see

play04:35

if there is little cores so yeah I still

play04:38

think AMD has more work to do but in the

play04:40

meantime we do have the core clocks for

play04:42

their 8000g desktop apus specifically

play04:45

the 8300 G comes with the Zen 4C base

play04:48

clock of 3.2 GHz and a Max boost of 3.6

play04:52

GHz and the 8500g comes with the same

play04:54

base clock but 3.7 GHz boost ultimately

play04:58

it's great that AMD finally added this

play05:00

but I definitely think there's more work

play05:02

for them to

play05:03

do and lastly for today it's time to say

play05:06

goodbye to multi-threading in CPUs

play05:09

that's right a feature in Hardware

play05:10

that's over two decades old is on its

play05:13

way out this story originally comes from

play05:15

a new report by Tech PowerUp who noticed

play05:17

in a recently leaked document for

play05:19

Intel's 15 gen aerrow L that it didn't

play05:21

include hyperthreading which is Intel's

play05:23

branding for their multi-threading and

play05:25

this was actually rumored months ago so

play05:27

it's really looking true now you might

play05:29

scratching your head is to why while

play05:31

simultaneous multi-threading or smt has

play05:33

introduced some major headaches for both

play05:35

AMD and Intel and bring some pretty

play05:37

decent performance to heavily threaded

play05:39

workloads well the reason actually boils

play05:41

down Intel's hybrid core architecture

play05:44

see in multi-threading the CPU cores are

play05:46

split up into two logical cores and your

play05:49

PC can't tell the difference between the

play05:51

real and logical core the reason for

play05:53

this is because different tasks take

play05:55

different amounts of time to complete so

play05:56

there is a lot of downtime for CPU cores

play05:59

while waiting for other cores to finish

play06:01

as you can see in this great analogy

play06:03

before smt you had one funnel pushing

play06:05

task to each core but there's some time

play06:07

between each task with multi-threading

play06:10

each core is still doing one task at a

play06:12

time but it adds a second funnel to fill

play06:14

the time between the two task of the

play06:16

first effectively making double the

play06:18

performance but that never happens

play06:19

perfectly like that because it only

play06:21

works if there's enough downtime to do

play06:23

the new thread's task and if the

play06:24

completion time of the new thread

play06:26

perfectly matches that downtime which

play06:28

obviously never happens but you do get

play06:30

more performance in comes hybrid core

play06:33

architecture that makes this way more

play06:35

complicated now you have one core that's

play06:37

way more powerful than the other so

play06:39

Intel has designed a solution and they

play06:41

already have a patent on it it's called

play06:43

rentable units and it basically takes a

play06:45

task and splits it into two then hands

play06:48

those new tasks to the best core for the

play06:50

job meaning the bigger task goes to the

play06:51

performance core with the smaller Tas to

play06:53

the efficiency core and the goal is to

play06:55

try and time it so they finish as close

play06:57

to the same time as possible basically

play07:00

this is a very interesting new take on

play07:01

the utilization problem with multicore

play07:03

processing and given AMD looks to be

play07:05

moving to hybrid cores as well we can

play07:07

assume they plan to do a similar version

play07:09

of this which means multi-threading in

play07:11

CPUs is reaching its end so while that

play07:15

does it for today what do you think

play07:17

about the end of multi-threading let me

play07:19

know down in the comments below and

play07:21

don't forget to download type aai and

play07:23

save some money down in the description

play07:25

below and as always have a great day

Rate This

5.0 / 5 (0 votes)

Besoin d'un résumé en anglais ?