AMD's Future Chips & SoC's: News, Info & Rumours.

Page 107 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
I understand their capabilities, what was the proposed price? If it is too expensive it won't sell compared to cheaper Ryzen 2000 CPUs. Or they are earmarked for OEMs or something.

The old Forbes article has it at $129, same as the 3200G. Current street price is $100... 2200G is $80, and the R3 1200 is sitting at $60.

I can't see shoving another SKU in there and expecting it to sell.
Right next or even slightly above (10 greenies?) their G counterparts would make perfect sense.

Cheers!
 

jdwii

Splendid
Nothing is more boring in the Ryzen 3000 lineup then the APU's without Zen 2 or Navi who cares its a OC 2400G that is soldered...

Makes me wonder why Amd even made those parts as their not even 7nm
 
Nothing is more boring in the Ryzen 3000 lineup then the APU's without Zen 2 or Navi who cares its a OC 2400G that is soldered...

Makes me wonder why Amd even made those parts as their not even 7nm
Because the markets for each CPU are slightly different, I'd say.

Someone willing to go for an iGPU knows that the overall performance will not be as good as the equivalent CPU with a dedicated GPU. That being said, I don't think AMD has anything lower than those price points, so in reality the 1400 and 1500 are like 60 greenies over their G equivalents if you take into account the need of a GPU.

Cheers!
 
That's a good and very honest take from him. I would imagine, a lot of us mirror that sentiment and words, even to the letter I'd say, about the state of affairs and the hope we have in AMD steering the ship the right way for the foreseeable future.

Now the clock is ticking for Intel, but AMD should keep the ship at full speed, as Intel has the means to catch up and surpass. Well, they should still have absurd amounts of money, that is. And their bullying shenanigans as well, I guess.

Cheers!
 
Now the clock is ticking for Intel, but AMD should keep the ship at full speed, as Intel has the means to catch up and surpass. Well, they should still have absurd amounts of money, that is. And their bullying shenanigans as well, I guess.
In the short term, I don't see Intel improving. They flat out fell behind on process technology, so they can't just stay ahead by virtue of die shrinks anymore. Core is just old, and it's time Intel junk it and come up with a better uArch.
 
Reactions: jdwii

jdwii

Splendid
In the short term, I don't see Intel improving. They flat out fell behind on process technology, so they can't just stay ahead by virtue of die shrinks anymore. Core is just old, and it's time Intel junk it and come up with a better uArch.
Agreed i simply don't think Intel can push the design much further.

They need to come up with a brand new design but the questions come up how?

We all know wider cores equal higher IPC when done right and i feel like Intel has done a great job in doing that. How can they squeeze more out? I guess for starters they should aim towards 4ghz instead of 5ghz and decrease their pipeline size in the newer design as they will struggle to hit 5ghz on 10nm and beyond.
 

Eximo

Titan
Herald
Well I think both AMD and Intel need to take a hard look at SMT and HT before doing anything drastic. We can't just keep patching out performance for security, basically going backwards. Any reasonable IPC gains are basically just getting everything back to where we were.

AMD already has their eye on 4 threads per core, lets just hope they don't build a huge vulnerability into it...
 
Reactions: remixislandmusic

jdwii

Splendid
Well I think both AMD and Intel need to take a hard look at SMT and HT before doing anything drastic. We can't just keep patching out performance for security, basically going backwards. Any reasonable IPC gains are basically just getting everything back to where we were.

AMD already has their eye on 4 threads per core, lets just hope they don't build a huge vulnerability into it...

So far i would be more worried about Intel as basically no security vulnerabilities have been found on Zen its Intel with those issues
 
Well, if you take a look at the fundamentals of "SMT", there's elements that you would catalogue as insecure by today's standards.

When you need to create hard boundaries for threads and data, it becomes expensive to do effective SMT and it's advantage becomes really reduced. I wonder how they'll design around the memory boundaries (registers) so that threads are really un-aware of other memory areas/locations when put in the same hardware "core".

Intersting challenge.

Cheers!
 
Well, if you take a look at the fundamentals of "SMT", there's elements that you would catalogue as insecure by today's standards.

When you need to create hard boundaries for threads and data, it becomes expensive to do effective SMT and it's advantage becomes really reduced. I wonder how they'll design around the memory boundaries (registers) so that threads are really un-aware of other memory areas/locations when put in the same hardware "core".

Intersting challenge.

Cheers!
Keep in mind the registers themselves aren't the problem, the shared resources between the physical and logical core are.

https://www.digitaltrends.com/computing/amd-graphics-card-nvidia-killer/

I never post Radeon news here but this seems big if Amd is calling their new cards "Nvidia killer"
They've been saying that since the 2900XT days. With the exception of that one period where NVIDIA just kept refreshing the 8800GT over and over, AMD hasn't really come close to beating NVIDIA.
 

goldstone77

Honorable
Aug 22, 2012
2,250
13
11,965
85
Moore's Law is not Dead 2019/08/14 by Godfrey Cheng, Head of Global Marketing, TSMC
https://www.tsmc.com/english/newsEvents/blog_article_20190814.htm?fbclid=IwAR27zBmVxM6iIa1s6dGMbR_BThoicycGzqUOfRAOOUxuBIsWfH2p9A4JPLc

To address this squarely, TSMC has recently announced our N5P node which further expands our leadership beyond the N5 node that will feature the world’s highest transistor density and offer the fastest performance. After being exposed to our technology roadmap, I can safely state that TSMC has many years of pioneering and innovation ahead of us where we will continue to shrink the individual transistor and continue to improve density. You will hear more from us in the coming months and years as we progress to new nodes.
Who knows what's coming down the pipe line in the future. Bold moves are being bet on I'm sure, but TSMC's is definitely saying we are here to pave the way to the future!
 

jdwii

Splendid
Keep in mind the registers themselves aren't the problem, the shared resources between the physical and logical core are.



They've been saying that since the 2900XT days. With the exception of that one period where NVIDIA just kept refreshing the 8800GT over and over, AMD hasn't really come close to beating NVIDIA.
Well to be honest i haven't heard Amd say that in years and i follow them quite closely also they haven't came out publicly saying that. Not that i would buy it anyways as i own a G-Sync monitor pretty much stuck with Nvidia lol(which is fine i like Nvidia too). Not to mention i care a lot about noise and performance per watt(heat).
 

jdwii

Splendid
Moore's Law is not Dead 2019/08/14 by Godfrey Cheng, Head of Global Marketing, TSMC
https://www.tsmc.com/english/newsEvents/blog_article_20190814.htm?fbclid=IwAR27zBmVxM6iIa1s6dGMbR_BThoicycGzqUOfRAOOUxuBIsWfH2p9A4JPLc


Who knows what's coming down the pipe line in the future. Bold moves are being bet on I'm sure, but TSMC's is definitely saying we are here to pave the way to the future!
Not to be a smart Xss but i can prove that they feel beyond like everyone else in terms of moore's law.

To be honest i doubt Moore's law was ever correct it was either we moved faster then Moore's law or slower

"Moore's law is the observation that the number of transistors in a dense integrated circuit doubles about every two years. "

Back in the 90's we probably moved much faster then that and after 2010's we moved much slower then that.
 

jdwii

Splendid

goldstone77

Honorable
Aug 22, 2012
2,250
13
11,965
85
Not to be a smart Xss but i can prove that they feel beyond like everyone else in terms of moore's law.

To be honest i doubt Moore's law was ever correct it was either we moved faster then Moore's law or slower

"Moore's law is the observation that the number of transistors in a dense integrated circuit doubles about every two years. "

Back in the 90's we probably moved much faster then that and after 2010's we moved much slower then that.
Yeah, over the decades since it the trend was first extrapolated, it was doubling every year at one point, and was changed later. You are right it was just an observation of a trend, which he extrapolated/guessed would continue.
Here is some background information on it.
https://t.co/GoF6CBzSgt?amp=1
But the main point I was trying to point out is that TSMC, AMD's foundry for 7nm and likely future nodes, released a marketing blog that suggests TSMC is very confident in there ability to lead the industry going forward.
 

jdwii

Splendid
I always felt confident in TSMC the greatest thing ever is that Amd moved to them(guess they had too lol).

I always had a hatred for global foundries as i always found them to be behind ever since the bulldozer days that limited Amd heavily in terms of frequency at launch.
 
Reactions: goldstone77
I always felt confident in TSMC the greatest thing ever is that Amd moved to them(guess they had too lol).

I always had a hatred for global foundries as i always found them to be behind ever since the bulldozer days that limited Amd heavily in terms of frequency at launch.
Some of us noted that the dream of >5GHz CPUs was as flawed in the 2010's as it was in the 2000's. The limiting factor was ultimately the laws of Physics. Bulldozer was simply a flawed design, period.

I'm still skeptical 4nm or smaller nodes will be economically viable, as I believe low yields will be an insurmountable problem. You are literally fighting physics at that point, and the margins for a "successful" part are tiny. While I do think some 4nm CPUs will ship, I expect them to be limited to the super-high-end tier (>$1000) with the bulk of chips remaining on the 7nm node. Regardless, there's only so much more you can shrink before physics stops you.
 
Reactions: Mandark

ASK THE COMMUNITY

TRENDING THREADS

Latest posts