AMD CPUs, SoC Rumors and Speculations Temp. thread 2

[jimmysmitty]

You both need to calm down.

Yuka, what he is claiming is not a first. There have been plenty of instances where a company labels a node different than what it truly is. I don't care but to me if it is true then it is GF trying to pull a "first" card which in the end will bite them in the arse.

I honestly don't think they will have a 7nm so soon. I don't care what experience they got from IBM or Samsung if Intel isn't near there yet almost no one will be. Even as a consortium before GF they still couldn't beat Intel to 14nm or 10nm.

I guess we will see in the next few years. Maybe GF gets lucky but it still seems suspicious that a foundry that has had issues with their own nodes and has had to license nodes from other companies all of the sudden getting a major leap in process tech all by their lonesome (not counting the fact that they got some of it from IBM)?

 

[ex_bubblehead]

Thread cleaned up.

That will be enough of the bickering. As all participants have been warned previously, on multiple occasions, there will be no further warnings. The next time there will be sanctions.

 

[ex_bubblehead]

I think my post may have been deleted by accident...

Not an accident. The entire conversation that was deleted was off topic in this thread. As was your last post.

 

[Vogner16]

I posted on the zen thread as well.

http://wccftech.com/amd-ryzen-desktop-zen-cpu-xfr/

 

[de5_Roy]

*yawn*

i wants to discuss the new Infinity Fabric. it seems to be the real interesting part of ry/zen launch.

oh and happy new year(!) (aka why is this thread still alive oh well...)

q/ed: my apologies and gratitude ([strike]brown-nosing[/strike]) to the hard working mods.

 

[8350rocks]

*yawn*

i wants to discuss the new Infinity Fabric. it seems to be the real interesting part of ry/zen launch.

oh and happy new year(!) (aka why is this thread still alive oh well...)

q/ed: my apologies and gratitude ([strike]brown-nosing[/strike]) to the hard working mods.

Infinity fabric on Ryzen is quite interesting...lots of good info out there, but it is still not completely defined yet.

Here is the most comprehensive breakdown I am aware of...from S|A: http://semiaccurate.com/2017/01/19/amd-infinity-fabric-underpins-everything-will-make/

 

[de5_Roy]

Infinity fabric on Ryzen is quite interesting...lots of good info out there, but it is still not completely defined yet.

Here is the most comprehensive breakdown I am aware of...from S|A: http://semiaccurate.com/2017/01/19/amd-infinity-fabric-underpins-everything-will-make/

If the level of granularity is as fine as was intoned, it allows a CPU core to pass info to a shader ‘directly’ regardless of the two being on the same silicon or across a system.

i wondered that halfway reading the article. imo this can directly affect vr, ar, ~r, ai/deep learning etc.
but it also seems a bit too good to be true since the vendor is amd.
btw, doesn't this make hsa redundant? there will be no need to abstract uarches as far as i can tell...

 

[8350rocks]

Infinity fabric on Ryzen is quite interesting...lots of good info out there, but it is still not completely defined yet.

Here is the most comprehensive breakdown I am aware of...from S|A: http://semiaccurate.com/2017/01/19/amd-infinity-fabric-underpins-everything-will-make/

If the level of granularity is as fine as was intoned, it allows a CPU core to pass info to a shader ‘directly’ regardless of the two being on the same silicon or across a system.

i wondered that halfway reading the article. imo this can directly affect vr, ar, ~r, ai/deep learning etc.
but it also seems a bit too good to be true since the vendor is amd.
btw, doesn't this make hsa redundant? there will be no need to abstract uarches as far as i can tell...

This is a part of their model toward HSA. In fact, it enables the next iteration of HSA, as does the Vega architecture designed to take full advantage of this fabric.

 

[eidolon171]

Has anybody else checked out the Bristol Ridge APU's that showed up over in Passmark's benchmark archive? Mature Excavator is showing 15% improvement over Steamroller in this single thread test. It's interesting because both the Haswell i7 4770k and the new Kaby Lake i7 7700k overtake Bristol Ridge IPC by just 25% and 27% respectively.

Considering we've seen the igpu in AMD's APUs post noticeable improvements to gaming framerates when fed by faster RAM, and since AM4 is DDR4 ready, we might get a chance to pair 4266MHz RAM with a Bristol Ridge APU (assuming someone builds us a capable board).

Now there is no consumer Bristol Ridge out on the market yet, or AM4 boards actually on sale yet either. But the boards should be out soon, and since OEM's already have several Bristol Ridge APUs, I feel confident a consumer SKU is on the way.

This could be a great product for entry level gaming, and HTPCs.