AMD Ryzen Quad-Core 2+2 vs. 4+0 Core Distributions Compared

[btarunr]

With AMD readying quad-core variants of its Ryzen "Summit Ridge" processor, the question on everyone's minds is whether the chip features two quad-core compute complexes (CCX) with two cores enabled, each, or just one CCX, given that the L3 cache amount being advertised by the company is 8 MB (that of one CCX), in comparison to 6-core Ryzen parts receiving the full 16 MB (8 MB per CCX) available on the silicon. While we will be able to definitively answer that question on the 11th of April, a new UEFI firmware by ASUS for its Crosshair VI Hero motherboard lets users not just disable cores, but also the distribution of the disabled cores.

CPU cores on the Ryzen "Summit Ridge" processor are distributed in two groups of four cores, each, called the quad-core compute complex (CCX). Each CCX has an 8 MB L3 cache, and so the ideal way of distributing cores on lower core-count models would be to disable an equal number of cores per CCX. For 6-core chips, one core is disabled per CCX, resulting in a 3+3 configuration. For quad-core chips, however, you can either disable all four cores in a CCX (4+0 configuration), or do a purportedly more optimal 2+2 configuration, with two cores disabled per CCX. Hardware Unboxed took advantage of ASUS' new UEFI firmware to compare the 4+0 configuration to the 2+2 configuration. The results are somewhat surprising.



As you can see in the graphs above, there is practically no performance difference between the 4+0 configuration and the 2+2 configuration. In fact, the 4+0 configuration is mildly faster in some scenarios. AMD already advertised quad-core Ryzen parts to feature just 8 MB of L3 cache, and so it could make more sense to keep all 8 MB in one CCX, and disable an entire CCX to run the chip in a 4+0 configuration, than disabling 4 MB per CCX, and running it in a 2+2 configuration. This way, a single core can dump >4 MB of data onto the L3 cache it addresses (as opposed to being limited to 4 MB in a 2+2 configuration with just 4 MB per CCX). Inter-CCX communication may be fast, but not fast enough to make a core from one CCX address the L3 cache of another CCX (which by the way is not possible, according to AMD). This is what makes 4+0 a more desirable configuration for the upcoming quad-core Ryzen parts, than 2+2.



View at TechPowerUp Main Site

 

[ShurikN]

So it clearly doesn't matter.

 

[newtekie1]

The only issue I have with this test, is the when running 2+2, the full cache is still available. So 2 cores have access to 8MB of L3(or 4MB per core). In the 4+0 configuration, it is 8MB for 4 cores, or 2MB per core. In the final product, it is only going to have 8MB of cache. So the 2+2 configuration would only have 4MB per 2 cores. So this isn't a true test of what a 2+2 configuration would perform like, the 2+2 configuration is performing better in these test than it would in real life.

Still, I don't think the difference would be enough to write home about. Though I still think AMD is going to release the first generation of Ryzen 3/5 as 2+2 configurations, to get rid of the damaged dies. Then they'll relelase a "new and improved" Ryzen 3/5 with "much better performance" that is really just a single CCX quad-core and the performance comes entirely from eliminating the cross-talk between the CCXs and amounts to a couple % improvement.

 

[SKD007]

This looks good. Wish they release 4c/8t 4.3ghz to counter 7740k

 

[theGryphon]

The only issue I have with this test, is the when running 2+2, the full cache is still available. So 2 cores have access to 8MB of L3(or 4MB per core). In the 4+0 configuration, it is 8MB for 4 cores, or 2MB per core. In the final product, it is only going to have 8MB of cache. So the 2+2 configuration would only have 4MB per 2 cores. So this isn't a true test of what a 2+2 configuration would perform like, the 2+2 configuration is performing better in these test than it would in real life.

Still, I don't think the difference would be enough to write home about. Though I still think AMD is going to release the first generation of Ryzen 3/5 as 2+2 configurations, to get rid of the damaged dies. Then they'll relelase a "new and improved" Ryzen 3/5 with "much better performance" that is really just a single CCX quad-core and the performance comes entirely from eliminating the cross-talk between the CCXs and amounts to a couple % improvement.

1500X was revealed by Anandtech to have 16MB of total L3 cache, which means it has all the physically available cache enabled, which in turn makes this analysis relevant for 1500X.

1400 is said to have 8MB total cache, and being an inferior product we can expect it to perform worse than 1500X in worst-case scenarios (where L3 cache size matters), clock-to-clock, assuming it also comes in a 2+2 config as also claimed by Anandtech to have been confirmed by AMD.

Anandtech link: http://www.anandtech.com/show/11202/amd-announces-ryzen-5-april-11th

 

[atomicus]

So does this mean in respect to gaming that all else being equal, the 1600X and 1700X are going to be identical?

 

[TheLaughingMan]

So does this mean in respect to gaming that all else being equal, the 1600X and 1700X are going to be identical?

Probably the 1600X should come close to matching the 1800X since they have the same clock speed by default. I can't think of a game that would come close to needing the extra 2c/4t. That being the case, it could end up being the sweat spot for Ryzen Gen.1.

 

[atomicus]

Probably the 1600X should come close to matching the 1800X since they have the same clock speed by default. I can't think of a game that would come close to needing the extra 2c/4t. That being the case, it could end up being the sweat spot for Ryzen Gen.1.

Yeah that's what I was thinking... I think the 1800X occupies a space on its own for content creators etc. who really need those cores and want the ultimate performer. The 1700/1700X is more of the sweet spot in the current line-up though, but if you have a leaning towards gaming then it seems like the 1600X might be ideal if you still want some grunt and will be using those extra cores in some programmes, but find the 1700X slightly overkill (or over budget) for your needs. Certainly no game is going to be using more than 6C anytime soon. Of course, for someone who games and nothing else, a 4C will suffice.

 

[idx]

Do you guys remember the Core 2 Duo vs Core 2 Quad from 2007-2008 ?

This situation with Ryzen right now is so awfully similar to back then. I remember the Core 2 Duo was beating the Core 2 Quad in almost everything (specially gaming ) and people doing exactly like what they are doing with Ryzen right now, comparing core count. I still remember " It doesn't matter .. unless if you are using it for special applications..." .

I miss the days of UT2004.

 

[atomicus]

This situation with Ryzen right now is so awfully similar to back then. I remember the Core 2 Duo was beating the Core 2 Quad in almost everything (specially gaming) and people doing exactly like what they are doing with Ryzen right now, comparing core count. I still remember " It doesn't matter .. unless if you are using it for special applications..." .

Only that ISN'T what's happening here lol. Many games are VERY close, some are beaten by Ryzen, especially in the latest benchmarks with faster speed RAM. Plus FPS minimums are often higher with Ryzen.

 

[newtekie1]

1500X was revealed by Anandtech to have 16MB of total L3 cache, which means it has all the physically available cache enabled, which in turn makes this analysis relevant for 1500X.

Good to know, I forgot the 1500X had 16MB.

 

[Dimitris Papadopoulos]

Only that ISN'T what's happening here lol. Many games are VERY close, some are beaten by Ryzen, especially in the latest benchmarks with faster speed RAM. Plus FPS minimums are often higher with Ryzen.

It seems that my q6600 quad currently sporting a 4 year old SSD is relevant for most tasks as we speak. I am not sure I would have been be able to say the same if I opted for a core duo. In my experience cores count

Nonetheless my upgrade is long overdue and I am pretty much sold on ryzen 1700.

 

[Gasaraki]

Umm, this article needs to be rewritten.

"As you can see in the graphs above, there is practically no performance difference between the 4+0 configuration and the 2+2 configuration. In fact, the 4+0 configuration is mildly faster in some scenarios."

"This is what makes 4+0 a more desirable configuration for the upcoming quad-core Ryzen parts, than 2+2."


You just said that there was no performance difference and then your conclusion is that the 4+0 is better?

 

[newtekie1]

You just said that there was no performance difference and then your conclusion is that the 4+0 is better?

What part of "the 4+0 configuration is mildly faster in some scenarios" is hard to understand?

 

[SanchesS80]

What about TDP for both options? Will it be equal

 

[deu]

owning a 1080Ti and playing in 1080p = herpderp

 

[semantics]

owning a 1080Ti and playing in 1080p = herpderp

Some people want that 144+/165+/180+/240hz fps now, Granted i'd take 120hz ips at 1440p anyday some people always look for that edge for "gaming"

 

[Rivage]

Games is for the childs.

 

[eidairaman1]

Yet people will continue bitching even though it shows IT DOESN'T MATTER!

 

[sweet]

Umm, this article needs to be rewritten.

"As you can see in the graphs above, there is practically no performance difference between the 4+0 configuration and the 2+2 configuration. In fact, the 4+0 configuration is mildly faster in some scenarios."

"This is what makes 4+0 a more desirable configuration for the upcoming quad-core Ryzen parts, than 2+2."


You just said that there was no performance difference and then your conclusion is that the 4+0 is better?

His conclusion argues with his article, and that conclusion is also not correct.

In scientific application 2+2 configuration with 16M L3 is much more desired. Also from the base clock of 2+2 16M L3, it can go close to 4GHz much easier. So in the end you cannot just say 4+0 is more desirable basing solely on a simulated test on some games.

 

[Anggoro]

Nice theory but let's see the real benchmark and we're good to go. 1600x looks sweet to me, and at that pricepoint, if it beats at least last gen i5, I'll take it for my rig.

 

[GhostRyder]

Well thats good, it looks like its all within margin of error so nothing to write home about (Least yet).

 

[Totally]

So does this mean in respect to gaming that all else being equal, the 1600X and 1700X are going to be identical?

Yeah, looks like core count(having 4 of them) is all that matters

 

[medi01]

Surprising results.

 

[S@LEM!]

wonder when disabling 4 cores on Ryzen will that gives us an overclock advantage to exceed 4.2 GHz+? or we are just locked for good on this generation