Overclock.net - An Overclocking Community - Reply to Topic
Thread: [Guru3d] AMD wants to increase performance per cycle by more than 7 percent annually Reply to Thread
Title:
Message:

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
01-09-2020 06:50 AM
KyadCK
Quote: Originally Posted by huzzug View Post
The point you tried to make was 2x increase in transistors becaise of Moores Law meant 2x increase in performance which never was the case and never will be.
I mean...

The 3950X is pretty close to dead on with 2x transistor count as compared to a 2700X and is roughly double the performance in highly parallel workloads. The logic can apply to node shrinks and core count boosts as adding more cores is, at least with AMD's design, largely linear.

But certainly never with IPC, no, and that was never Moore's law, which simply stated that transistor count would double every 24 months.
01-08-2020 11:37 PM
huzzug
Quote: Originally Posted by UltraMega View Post
That was actually not my point of you go back to what started this debate, though I'll admit I'm no hardware engineer and the finer points of how a CPU works are way beyond me.

In any case, I'm done with this thread. Pages upon pages of pointless debate.
Quote: Originally Posted by UltraMega View Post
It wasn't that long ago that we would see a 50% gain every 18 months.
Yea sure.
01-08-2020 11:17 PM
UltraMega
Quote: Originally Posted by huzzug View Post
The point you tried to make was 2x increase in transistors becaise of Moores Law meant 2x increase in performance which never was the case and never will be.
That was actually not my point of you go back to what started this debate, though I'll admit I'm no hardware engineer and the finer points of how a CPU works are way beyond me.

In any case, I'm done with this thread. Pages upon pages of pointless debate.
01-08-2020 08:14 PM
huzzug
Quote: Originally Posted by UltraMega View Post
Not really, the point I was trying to make was just that 7% annually isn't unreasonable.
The point you tried to make was 2x increase in transistors becaise of Moores Law meant 2x increase in performance which never was the case and never will be.
01-08-2020 07:29 PM
fastturtle Pipe down Peanut Gallery. I don't know where in Vulcan's Forge you all got the idea that AMD has to tape out new CPU's to improve IPC 1 percent per year as that's all they need to do each year to meet that goal. What I see them doing is refining the Microcode and improving the instruction sets of the CPU's.

Of course, there's always the fact that improvements in one place may slow things down in another and that's what has to be watched for. From what I can see right now though, by improving the microcode's efficiency, AMD has a real chance of hitting that 10 percent goal for a decade and if we are lucky, they may offer those improvements to all of us early adopters unlike Intel.
01-08-2020 11:53 AM
UltraMega
Quote: Originally Posted by huzzug View Post
So the sky is teal after all.
Not really, the point I was trying to make was just that 7% annually isn't unreasonable.
01-08-2020 05:57 AM
huzzug
Quote: Originally Posted by UltraMega View Post
I looked up some benchmarks and it looks like it was around 25% faster that times around. So OK, not 50%
So the sky is teal after all.
01-08-2020 03:22 AM
ToTheSun!
Quote: Originally Posted by warr10r View Post
Pictures of Jose Morinho at Chelsea FC?

I mean, we can talk about Jose's new stint at Tottenham if you guys want?
If you got any pictures of him moving goal posts at Tottenham to illustrate that that's what some people in this thread were doing, feel free to post them.
01-08-2020 02:57 AM
warr10r
Quote: Originally Posted by UltraMega View Post
I mean to say the Core 2 series to the first gen I series. I looked up some benchmarks and it looks like it was around 25% faster that times around. So OK, not 50%, but easily more than 7%, (or 10.3 for an 18 month period).
Several factors attributed to that jump in performance.
> Like Zen, Nehalem was years in the making and was a complete redesign of what came before.
> Like Bulldozer, Conroe had reached its design limitations so a redesign was required.
> Like Zen, Nehalem was designed to be a big bump in performance.
> Like Zen, Nehalem wasn't perfect and had room for improvement, so the Sandy Bridge leap was possible after it.

How in the world do you expect gains like that for every generation?

Subsequent designs were simply shrinks and tweaks of the original design. Shrinking stopped at 14nm as Intel had stubbed their toe on 10nm so Intel kinda had to make tweaks to Skylake and basically re-release the same CPUs every year on a more refined 14nm until 10nm works or 7nm is done. So pretty much no gains from Skylake onwards.
01-08-2020 01:09 AM
UltraMega
Quote: Originally Posted by WannaBeOCer View Post
Nehalem was no where close to being 50% faster than Yorkfield. If you said Phenom II, I could see that.
I mean to say the Core 2 series to the first gen I series. I looked up some benchmarks and it looks like it was around 25% faster that times around. So OK, not 50%, but easily more than 7%, (or 10.3 for an 18 month period).
This thread has more than 10 replies. Click here to review the whole thread.

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off