AMD Ryzen Chips 10% Smaller When Compared to Intel Skylake Dies

[cj09beira]

5 minutes ago, cj09beira said:

its more complicated than that as you can have the same design use more or less transistors depending on what characteristics the cpu should have 

edit:

and we don't know if the igpu is being counted or not 

the biggest diference probably comes from less PCIE lanes, and (big one) no quad channel memory,

[leadeater]

12 minutes ago, Curufinwe_wins said:

Actually it is pretty terribly discouraging information if true. IMHO

 

Look at the transistor standard size..... The Ryzen chips have to be far FAR less complex (and probably therefor worse at HPC type loads) in order to come in at the same total die size.

 

Obviously # transistors is no where near an all-encompassing stat for performance, but the fact of the matter is that Intel doesn't waste die space, so if AMD has a smaller die, and has a provably less dense process, they simply CANNOT house the same capabilities.

 

True but the claim is they have a more dense process(err design?), whether there is any problems with it will be shown during reviews like higher temps or bad OC.

 

The total die size won't be the same, as mentioned dual channel vs quad channel memory controller with less PCIe lanes and smaller area for cores and cache so it will be smaller. Not only that AMD will be intentionally making it smaller for cost reasons, that would likely have been a very early design goal.

 

I do however completely expect Ryzen on a per core and total package level to have less performance than Intel has, anyone thinking anything else is being a little naive. What AMD will have focused on is making their Ryzen design very competitive in a few areas most important to the general consumer and in common benchmarks. For the specialty workloads an Intel CPU will still be faster. 

Edited February 8, 2017 by leadeater

[Curufinwe_wins]

4 minutes ago, cj09beira said:

the biggest diference probably comes from less PCIE lanes, and (big one) no quad channel memory,

 

4 minutes ago, leadeater said:

True but the claim is they have a more dense process, whether there is any problems with it will be shown during reviews like higher temps or bad OC.

 

The total die size won't be the same, as mentioned dual channel vs quad channel memory controller with less PCIe lanes and smaller area for cores and cache so it will be smaller. Not only that AMD will be intentionally making it smaller for cost reasons, that would likely have been a very early design goal.

 

I do however completely expect Ryzen on a per core and total package level to have less performance than Intel has, anyone thinking anything else is being a little naive. What AMD will have focused on is making their Ryzen design very competitive in a few areas most important to the general consumer and in common benchmarks. For the specialty workloads an Intel CPU will still be faster. 

 

Intel doesn't have 4 core processors with quad channel memory.... You missed that boat haha.

 

And they have fewer PCIE lanes than Ryzen supposedly.

 

Also the SRAM cell is an EXTREMELY well optimized cell. The gap might not be as big as 37%, but there is NO FUCKING WAY IN HELL AMD is on a denser core process while also having a 37% larger SRAM 6T cell.

[leadeater]

1 minute ago, Curufinwe_wins said:

Intel doesn't have 4 core processors with quad channel memory.... You missed that boat haha.

 

And they have fewer PCIE lanes than Ryzen supposedly.

The main comparison has always been with Intel enthusiast chips, which have quad channel memory and more PCIe lanes....

[Curufinwe_wins]

1 minute ago, leadeater said:

The main comparison has always been with Intel enthusiast chips, which have quad channel memory and more PCIe lanes....

The graph doesn't compare Intel enthusiast parts....

 

Look again at it.

 

[leadeater]

Just now, Curufinwe_wins said:

The graph doesn't compare Intel enthusiast parts....

 

Look again at it.

It's not comparing with any full cpu die either. It's an AMD CCX and Intel equiv ONLY.

[cj09beira]

2 minutes ago, Curufinwe_wins said:

It isn't perfect, but it actually is pretty damn close to being right. KNL/Atom is fewer transistors per core because they are simpler, weaker, shallower cores. Same with ARM (higher performance cores means bigger). Every feature and process and register you put on a CPU means transistors being used, and it is quite foolish to assume that AMD has the capability to deliver all the same features (ALL, not just the standard stuff for normal consumer loads that aren't properly optimized) and the same performance at the same clocks with 10% reduction in area, WHILE using a process that takes 37% more space for a very well optimized 6T cell.

 

It is just not a realistic belief.

i agree but we know ryzen has some things that would help reduce space: only 2 channel memory and less pcie-express lanes,  and you could be right and ryzen doesn't have some of the extra instructions, but i doubt it, as they want to use ryzen in servers too, (where the big bucks are made)

[Curufinwe_wins]

4 minutes ago, leadeater said:

It's not comparing with any full cpu die either. It's an AMD CCX and Intel equiv ONLY.

No. It is comparing to Intel Second Gen 14nm (aka Skylake.) Thus it CANNOT have quad channel, or more pcie lanes. 

 

3 minutes ago, cj09beira said:

i agree but we know ryzen has some things that would help reduce space: only 2 channel memory and less pcie-express lanes,  and you could be right and ryzen doesn't have some of the extra instructions, but i doubt it, as they want to use ryzen in servers too, (where the big bucks are made)

 

That is NOT the excuse.

 

Quote

In a separate paper, AMD said its upcoming Zen x86 core fits into a 10 percent smaller die area than Intel’s currently shipping second-generation 14nm processor.

 

[cj09beira]

2 minutes ago, Curufinwe_wins said:

No. It is comparing to Intel Second Gen 14nm (aka Skylake.) Thus it CANNOT have quad channel, or more pcie lanes. 

 

 

That is NOT the excuse.

 

 

i see it now,

forgot it was comparing it to skylake :-p 

 

it seems wierd, i guess we will found out what have they done soon enough 

[leadeater]

25 minutes ago, Curufinwe_wins said:

No. It is comparing to Intel Second Gen 14nm (aka Skylake.) Thus it CANNOT have quad channel, or more pcie lanes. 

You missed the point, there is no memory controller in the compared die area, at all. The package design is to compete up to the Intel enthusiast chips hence my point about PCIe lanes and memory controller, 49mm² is not total package area.

 

http://www.anandtech.com/show/9582/intel-skylake-mobile-desktop-launch-architecture-analysis (scroll to bottom)

 

I already said for us what's been showing is nothing more than a meaningless spreadsheet of numbers, with critical pieces of information missing.

[Drak3]

19 minutes ago, leadeater said:

I do however completely expect Ryzen on a per core and total package level to have less performance than Intel has, anyone thinking anything else is being a little naive. What AMD will have focused on is making their Ryzen design very competitive in a few areas most important to the general consumer and in common benchmarks. For the specialty workloads an Intel CPU will still be faster. 

I expect Ryzen to hit between Haswell and Broadwell overall, with Intel taking small leads in some areas, and them seeing very similar performance, when software is compiled using a compiler that doesn't prefer Intel's architectures, (or AMD spoofing Ryzen's ID system to trick software into thinking it's an Intel CPU). Sky and Kaby will have small leads, but I can't imagine actual performance differences to dissimilar as a move from Haswell to Sky.

AMD's cutting costs by cutting specs that aren't exactly needed by the target audience, and might incur significant amounts of R&D money and man hours. They've got a lean product, and they know it.

[Dabombinable]

1 hour ago, patrickjp93 said:

The paste isn't the problem either. Both Intel and AMD use Dow Corning between the die and IHS (see AMD APUs). The problem is the glue Intel uses is so viscous the spacing between die and IHS is inconsistent and can be too large. That's the single biggest contributing factor to the temperatures. Just removing the glue and applying Dow Corning reduces temps by 15-20 degrees. The extra 5 CLU gets you is great, but you have to reapply CLU every 12 months or so.

That paste BTW, if my Pentium II 233 (Klamath) is anything to go by (aka, if it actually is the same), will definitely last practically forever. I recently pulled the heatsink off it because I want to replace it (unlocked multiplier FTW), and it seems to be the same TIM used with Haswell (except of course, spread evenly across the IHS without any bubbles and a minimal gap)

[Coaxialgamer]

16 hours ago, patrickjp93 said:

The only die we have is an 8-core, and the 6900K is a cut-down 10-core die. Seriously people, think...

 

And even if this is a quad-core module, Intel has the iGPU included and AMD doesn't. Either way, huge density DISadvantage for AMD.

Except here they are looking at how large the core/cache is.  Not how big the actual chip will be.  Doesn't take into account igpu,  memory controller etc. 

[Prysin]

18 hours ago, Ryan_Vickers said:

lol "Competitor" .... couldn't possibly be Intel 

Competitor A is VIA. Also that "ZEN" SKU is a XEON. 

 

[MageTank]

18 hours ago, themctipers said:

inb4 cannonlake destroys zen

Coffeelake*. Cannonlake won't be destroying much of anything, lol.

[MageTank]

17 hours ago, Curufinwe_wins said:

 

 

Intel doesn't have 4 core processors with quad channel memory.... You missed that boat haha.

https://ark.intel.com/products/92991/Intel-Xeon-Processor-E5-1620-v4-10M-Cache-3_50-GHz

 

https://ark.intel.com/products/92980/Intel-Xeon-Processor-E5-2623-v4-10M-Cache-2_60-GHz (Can find these on Ebay for $90 a piece, great throw-away chips for board diagnostics)

[Curufinwe_wins]

32 minutes ago, MageTank said:

https://ark.intel.com/products/92991/Intel-Xeon-Processor-E5-1620-v4-10M-Cache-3_50-GHz

 

https://ark.intel.com/products/92980/Intel-Xeon-Processor-E5-2623-v4-10M-Cache-2_60-GHz (Can find these on Ebay for $90 a piece, great throw-away chips for board diagnostics)

We talked this to death already. But it's skylake they are comparing to.

 

Aka no quad cores with quad channel.

[MageTank]

1 minute ago, Curufinwe_wins said:

We talked this to death already. But it's skylake they are comparing to.

 

Aka no quad cores with quad channel.

You said Intel has no quad cores with quad channel memory. I only skimmed, didn't see the context was about Skylake, my bad. Still, for anyone looking for a throwaway diagnostic chip, I totally recommend those little X99 quad cores. They get the job done for cheap, lol. 

[Unrealcpu]

AMD comeback!

[Vode]

This is why.

 

 

[Mira Yurizaki]

I'm looking at this going... is this comparing only the execution resources? i.e., the CPU cores themselves? Because Skylake's total die size is 122mm^2.

 

Granted they may have a denser or better packed CPU core, but what about the rest of the system?

[Sauron]

Well... that doesn't mean much

[Coaxialgamer]

42 minutes ago, M.Yurizaki said:

I'm looking at this going... is this comparing only the execution resources? i.e., the CPU cores themselves? Because Skylake's total die size is 122mm^2.

 

Granted they may have a denser or better packed CPU core, but what about the rest of the system?

They meant the actual cores + cache, as that would be around the 49mm² presented here. Obviously, there still is the igpu, memory controller, pcie controller etc, but apart from the igpu, those are quite small. 

 

But then again, this is amd's info. 

 

[Mira Yurizaki]

23 minutes ago, Coaxialgamer said:

They meant the actual cores + cache, as that would be around the 49mm² presented here. Obviously, there still is the igpu, memory controller, pcie controller etc, but apart from the igpu, those are quite small. 

 

But then again, this is amd's info.

Looking at a die of Skylake though...

 

The "Uncore", as Intel used to call it, takes up about a third of the die space. I wouldn't call it insignificant though.

 

Still, marketers at their finest.

[Vode]

TLDR

 

Part of why Intel has bigger chip because their Floating Point Unit is more complex/wide, which matters a lot in Enterprise workloads (AVX etc..) but not so much in Gaming and general Desktop use. Intel still has a superior node obviously. AMD has gone for a streamlined design which will pay off in Gaming and most workloads people on this forum will be using.