Question How ahead is Intel in CPU design compared to AMD?

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

Adonisds

Member
Oct 27, 2019
98
33
51
If we could get both companies to design a CPU core now on the same process and using the same number of transistors, who would be ahead and by how much? I'm assuming Intel would be ahead because Skylake has a similar IPC to Zen 2, but Zen 2 uses a more dense node, and Skylake is from 2015 while Zen 2 is from 2019.

The both companies using the same process scenario is just to illustrate the problem and I know it will never happen, but maybe there is a smart way to compare current CPU designs and find out who is ahead and by how much. Wikichip has die sizes and die pictures of Sunny Cove and Zen 2, but I'm not sure how to interpret and compare them.

Another question: If Intel had to design future cores using the same process and the same die size as Skylake, how much further could the Skylake design be improved? Also assuming it will have to have to be x86 too and have all the same functions.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
Zen 2 significantly improved gaming performance compared to Zen and Zen+ though, to the point that Intel's advantage is limited to some niche corner cases (low res and very high frequency) so Zen 2's high latencies seem not to be too bad (all while offering an obvious area for further improvements in the future).

I agree, but as I said earlier, AMD is going up against an Intel core that is almost 5 years old. AMD's recently achieved near parity in gaming in light of that truth isn't as impressive. When Intel returns to process node parity or superiority (just a matter of time), the tables will have turned once again. Intel's formula of a high clock speed, high IPC monolithic core with an IMC and SMT, is probably the best design for gaming workloads.

That's why I'm not so convinced that the Zen 2 chiplet design with massive L3 caches to offset the memory latency is the way to go if they want to get the upper hand on Intel in gaming in future.
 

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
I agree, but as I said earlier, AMD is going up against an Intel core that is almost 5 years old. AMD's recently achieved near parity in gaming in light of that truth isn't as impressive. When Intel returns to process node parity or superiority (just a matter of time), the tables will have turned once again. Intel's formula of a high clock speed, high IPC monolithic core with an IMC and SMT, is probably the best design for gaming workloads.

That's why I'm not so convinced that the Zen 2 chiplet design with massive L3 caches to offset the memory latency is the way to go if they want to get the upper hand on Intel in gaming in future.

And what happens when Intel inevitably gets rid of the monolithic core design, the ring bus, and the inclusive L3 cache? In gaming, anyway. I'll give you a preview, it's called Skylake-X.

I get the Core i7-7800X isn’t a gaming-focused CPU, but what’s surprising is that its performance continues to be lacking even when it is clocked higher than the 7700K. As seen in the benchmarks above, a Core i7-7800X overclocked to 4.7GHz is up to 5% slower than the stock 7700K.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
And what happens when Intel inevitably gets rid of the monolithic core design, the ring bus, and the inclusive L3 cache? In gaming, anyway. I'll give you a preview, it's called Skylake-X.

What makes you think Intel is going to get rid of the monolithic core design, especially when it has worked so well in its favor over the years? Especially for mobile and mainstream parts. They might for their HEDT and server CPUs, but desktop?

And I'm sure the mesh interconnect will be improved in subsequent generations, much like AMD's Infinity fabric.
 

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
What makes you think Intel is going to get rid of the monolithic core design, especially when it has worked so well in its favor over the years? Especially for mobile and mainstream parts. They might for their HEDT and server CPUs, but desktop?

And I'm sure the mesh interconnect will be improved in subsequent generations, much like AMD's Infinity fabric.

I'm not sure what to expect, but they will have to with HEDT and servers sooner than later. They certainly could keep making monolithic CPU's for mobile/desktop, but will the benefits outweigh the costs? I don't know what to expect from Intel these days. I think Intel might push Lakefield hard (which I think has a lot of potential) rather than continue with monolithic dies.

Also, the mesh will improve. I don't doubt that. But what many people seem to forget is the cache reorganization. Intel can't keep adding cores without moving to a non-inclusive L3. That inclusive L3 has done wonders for them since Nehalem. I want to see how well they transition before I get too excited about Intel's future.
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
Intel's formula of a high clock speed, high IPC monolithic core with an IMC and SMT, is probably the best design for gaming workloads.
That formula will end soon.

You are all deluded if you think that intel made their cpus to be very good at games. It's just an huge coincidence, intel designed the chips with servers/datacenters in mind or even with mobile.

Desktop gets the scraps of both, and believe me their next architecture will aim at datacenter first, if it's good for games or not it's not of any importance to them.
 
  • Love
Reactions: spursindonesia

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
And what happens when Intel inevitably gets rid of the monolithic core design, the ring bus, and the inclusive L3 cache? In gaming, anyway. I'll give you a preview, it's called Skylake-X.

Skylake-X is not as bad in games once you OC the mesh. With a decent overclock it should be faster than Matisse in games although slower than Skylake Client and friends. You would need to OC though.
 

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
AMD and Intel will have to go wider at some point, which pushes core size up. But still, real cores are better than extra threads on one core.
You are viewing this the wrong way, 1 Core with SMT is better than 1 Core without SMT, You get a speed up of 20-30% and up to 10-15% increase of power for nearly no added surface die area
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
You are viewing this the wrong way, 1 Core with SMT is better than 1 Core without SMT, You get a speed up of 20-30% and up to 10-15% increase of power for nearly no added surface die area
I think you are right i never saw tests where the main thread ran slower because there was another thread running there, contrary to the amd CMT where the main thread would go from 100% and slowdown to 80% with two.
 

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
I think you are right i never saw tests where the main thread ran slower because there was another thread running there, contrary to the amd CMT where the main thread would go from 100% and slowdown to 80% with two.
Really? I was not aware of that, I always thought that in an SMT and CMT the jobs were devided equally among available threads, for example when you run Cinebench in a 1C/2T processor(I was testing that way way back with a HT P4) so in a 1C/2T CPU the task evenly divided


As far as CMT, a single 2C/2T module has 80% throughput of a complete 2C/2T dual core
 
Last edited:

scannall

Golden Member
Jan 1, 2012
1,944
1,638
136
You are viewing this the wrong way, 1 Core with SMT is better than 1 Core without SMT, You get a speed up of 20-30% and up to 10-15% increase of power for nearly no added surface die area
1 thread of SMT is fine. That's the most bang for the buck. 4 threads of SMT? Uh no, you need to add considerable resources to support it, to the point just add real cores.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
That formula will end soon.

You are all deluded if you think that intel made their cpus to be very good at games. It's just an huge coincidence, intel designed the chips with servers/datacenters in mind or even with mobile.

So you honestly think that Intel and AMD don't do any testing on various common workloads/simulations throughout the design process for their CPUs? I think you're the deluded one.

PC gaming is huge, and highly profitable. It behooves Intel and AMD to design CPUs that are capable of performing well in games, as games are typically the most common intensive workload for consumers.

Also, AMD called the massive L3 cache on the Zen 2 CPU their "GameCache?" Maybe that too is a coincidence. And Intel and their unlocked K series for gamers and enthusiasts is also just a huge coincidence :confused_old:

AMD-GameCache-1.jpeg


Desktop gets the scraps of both, and believe me their next architecture will aim at datacenter first, if it's good for games or not it's not of any importance to them.

Apparently you can have your cake and eat it too.
 
  • Like
Reactions: Zucker2k

RetroZombie

Senior member
Nov 5, 2019
464
386
96
So you honestly think that Intel and AMD don't do any testing on various common workloads/simulations throughout the design process for their CPUs?
games are typically the most common intensive workload for consumers
Games are not common workloads, you can't design some cpu to be good at all games, each game have it's own characteristics, they do completely random workloads.

Contrary to some group of applications, be them all the file archiving software, all the image processing software, all the video processing software, all the cad software, all the database software, ... all the groups they all share the same kind of data workload.

You can predict those better when designing a cpu. Now you want intel waste time doing some cpu that will work very well with Doom and also be good with The Sims, they don't.


Also, AMD called the massive L3 cache on the Zen 2 CPU their "GameCache?"
Now you are confusing bad marketing with cpu design.

Apparently you can have your cake and eat it too.
Skylake-X is asking about you.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Games are not common workloads, you can't design some cpu to be good at all games, each game have it's own characteristics, they do completely random workloads.

Contrary to some group of applications, be them all the file archiving software, all the image processing software, all the video processing software, all the cad software, all the database software, ... all the groups they all share the same kind of data workload.

You can predict those better when designing a cpu. Now you want intel waste time doing some cpu that will work very well with Doom and also be good with The Sims, they don't.



Now you are confusing bad marketing with cpu design.


Skylake-X is asking about you.
A low latency memory subsystem is always desirable because it eliminates bottlenecks in every scenario you can imagine. If you can get zero delays going to main memory why won't you take it? Which apps or games won't benefit from faster processing times? You don't need to split hairs over this one.
 
  • Like
Reactions: Carfax83

naukkis

Senior member
Jun 5, 2002
702
570
136
I think you are right i never saw tests where the main thread ran slower because there was another thread running there, contrary to the amd CMT where the main thread would go from 100% and slowdown to 80% with two.

With CMT those two threads equal 160% of performance vs single thread. With SMT and normal 20-30% gain both threads execute at only 60-65% of single thread speed. CMT is SMT on steroids.
 

Glo.

Diamond Member
Apr 25, 2015
5,658
4,416
136
I agree, but as I said earlier, AMD is going up against an Intel core that is almost 5 years old. AMD's recently achieved near parity in gaming in light of that truth isn't as impressive. When Intel returns to process node parity or superiority (just a matter of time), the tables will have turned once again. Intel's formula of a high clock speed, high IPC monolithic core with an IMC and SMT, is probably the best design for gaming workloads.

That's why I'm not so convinced that the Zen 2 chiplet design with massive L3 caches to offset the memory latency is the way to go if they want to get the upper hand on Intel in gaming in future.
What if Zen 3 CPUs will give AMD the same gaming performance increase, over Zen 2, similar to what Zen 2 give them over Zen+?

How will Intel respond to that, considering they will be, in that scenario, 100% behind AMD, in all cases?
 

dmens

Platinum Member
Mar 18, 2005
2,271
917
136
When Intel returns to process node parity or superiority (just a matter of time), the tables will have turned once again. Intel's formula of a high clock speed, high IPC monolithic core with an IMC and SMT, is probably the best design for gaming workloads.

AHAHAHAHAHAHAHAHAHA

Even ignoring the faith statement on inevitable process leadership (which is 100% delusional), everything you said is utter bunk.

- High clock speed and high IPC work against each other, your statement is akin to saying the goal is "high performance", i.e. it is nonsense.
- Large monolithic dies are dead for obvious reasons.
- Gaming performance is at best an afterthought, if it is even considered at all.
 
Status
Not open for further replies.