Jump to content

RAM Diminishing Marginal Returns

Go to solution Solved by RONOTHAN##,

It depends on the specific software and game, including what settings you're gonna be playing at. For the most part for games that tends to be 5600 being the point where it starts to level off a bit, but for some applications it scales upwards of 7000MT/s+, and some it just never matters. Simple rule of thumb is to look at the applications where the 5800X3D pulls ahead, if the 5800X3D is 10% or more faster than its other AM4 counterparts, get as fast RAM as your motherboard can support (I wouldn't expect anything faster than 6200MT/s to work in that board though), and if it's in line with the 5800X, 5600MT/s is more than adequate. 

 

That said, any kit rated at 5600MT/s will be able to easily do 6000MT/s with a bit of overclocking, probably getting up usually getting up to 6200MT/s as well, so if you're willing to put in the bit of effort to do so, you can max that board out very easily with a 5600MT/s kit. 

It depends on the specific software and game, including what settings you're gonna be playing at. For the most part for games that tends to be 5600 being the point where it starts to level off a bit, but for some applications it scales upwards of 7000MT/s+, and some it just never matters. Simple rule of thumb is to look at the applications where the 5800X3D pulls ahead, if the 5800X3D is 10% or more faster than its other AM4 counterparts, get as fast RAM as your motherboard can support (I wouldn't expect anything faster than 6200MT/s to work in that board though), and if it's in line with the 5800X, 5600MT/s is more than adequate. 

 

That said, any kit rated at 5600MT/s will be able to easily do 6000MT/s with a bit of overclocking, probably getting up usually getting up to 6200MT/s as well, so if you're willing to put in the bit of effort to do so, you can max that board out very easily with a 5600MT/s kit. 

Link to comment
Share on other sites

Link to post
Share on other sites

13 minutes ago, RONOTHAN## said:

It depends on the specific software and game, including what settings you're gonna be playing at. For the most part for games that tends to be 5600 being the point where it starts to level off a bit, but for some applications it scales upwards of 7000MT/s+, and some it just never matters. Simple rule of thumb is to look at the applications where the 5800X3D pulls ahead, if the 5800X3D is 10% or more faster than its other AM4 counterparts, get as fast RAM as your motherboard can support (I wouldn't expect anything faster than 6200MT/s to work in that board though), and if it's in line with the 5800X, 5600MT/s is more than adequate. 

 

That said, any kit rated at 5600MT/s will be able to easily do 6000MT/s with a bit of overclocking, probably getting up usually getting up to 6200MT/s as well, so if you're willing to put in the bit of effort to do so, you can max that board out very easily with a 5600MT/s kit. 

Interesting. What's so special about the capabilities of the 5800X3D in relation to it being used as a benchmark for RAM speeds?

Link to comment
Share on other sites

Link to post
Share on other sites

12 minutes ago, Definitely not BB62 said:

Interesting. What's so special about the capabilities of the 5800X3D in relation to it being used as a benchmark for RAM speeds?

RAM speed helps performance when your CPU is waiting for the RAM to return data, shortening that time with faster RAM means that its spending less time waiting and more time doing stuff. Different workloads will have different requirements from the RAM so they may be waiting around more or less time in general, and RAM speed is more or less important in those particular workloads.

 

Because the 5800X3D has such a massive cache on board, it almost never needs to wait for RAM operations to complete, and thus RAM is never the bottleneck. By comparing the performance between it and the standard 5700X/5800X, you can see how memory bandwidth bottlenecked that particular application is, and thus how much it benefits from really fast RAM. 

Link to comment
Share on other sites

Link to post
Share on other sites

In something like Aida64 5800X3D is terrible at generating bandwidth. But it does not matter as even dual rank 3200C14 is still quite strong.

AMD R7 5800X3D | Thermalright Aqua Elite 360, 3x TL-B12, 2x TL-K12
Asus Crosshair VIII Dark Hero | 32GB G.Skill Trident Z @ 3733C14
Zotac 4070 Ti Trinity OC @ 3045/1495 | WD SN850, SN850X
Seasonic Vertex GX-1000 | Fractal Torrent Compact, 2x TL-B14

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, freeagent said:

In something like Aida64 5800X3D is terrible at generating bandwidth. But it does not matter as even dual rank 3200C14 is still quite strong.

What does "dual rank" mean?

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, RONOTHAN## said:

RAM speed helps performance when your CPU is waiting for the RAM to return data, shortening that time with faster RAM means that its spending less time waiting and more time doing stuff. Different workloads will have different requirements from the RAM so they may be waiting around more or less time in general, and RAM speed is more or less important in those particular workloads.

 

Because the 5800X3D has such a massive cache on board, it almost never needs to wait for RAM operations to complete, and thus RAM is never the bottleneck. By comparing the performance between it and the standard 5700X/5800X, you can see how memory bandwidth bottlenecked that particular application is, and thus how much it benefits from really fast RAM. 

How would I go about comparing that? Is there a website I can use?

Link to comment
Share on other sites

Link to post
Share on other sites

14 minutes ago, Definitely not BB62 said:

What does "dual rank" mean?

It describes memory organization (specifically 2Rx8). On DDR4, dual rank offers a fairly significant bandwidth advantage. It's not really true on DDR5, where the difference comes from 1Rx16 and 1Rx8 based memory (again, describes organization). 1Rx16 has something like 30-40% less bandwidth than 1Rx8-based kits, so it performs worse. 

 

Currently on DDR5, 8GB DIMMs are 1Rx16, 16GB DIMMs are 1Rx8, and 32GB DIMMs are 2Rx8, so you want to go for either the 16GB or 32GB DIMMs for full DDR5 performance, otherwise you're better off sticking to DDR4. 

 

17 minutes ago, Definitely not BB62 said:

How would I go about comparing that? Is there a website I can use?

There's lots of different sites comparing performance of the different CPUs in different situations, Gamers Nexus, HardwareUnboxed, Guru3D reviews, and TechPowerUp are usually where I go for a bunch of graphs to quickly compare, with GN and HardwareUnboxed for more complete reviews, but there's a ton of other good publications that you can use instead. Just don't use UserBenchmark and you should be fine. 

Link to comment
Share on other sites

Link to post
Share on other sites

11 hours ago, RONOTHAN## said:

It describes memory organization (specifically 2Rx8). On DDR4, dual rank offers a fairly significant bandwidth advantage. It's not really true on DDR5, where the difference comes from 1Rx16 and 1Rx8 based memory (again, describes organization). 1Rx16 has something like 30-40% less bandwidth than 1Rx8-based kits, so it performs worse. 

 

Currently on DDR5, 8GB DIMMs are 1Rx16, 16GB DIMMs are 1Rx8, and 32GB DIMMs are 2Rx8, so you want to go for either the 16GB or 32GB DIMMs for full DDR5 performance, otherwise you're better off sticking to DDR4. 

 

There's lots of different sites comparing performance of the different CPUs in different situations, Gamers Nexus, HardwareUnboxed, Guru3D reviews, and TechPowerUp are usually where I go for a bunch of graphs to quickly compare, with GN and HardwareUnboxed for more complete reviews, but there's a ton of other good publications that you can use instead. Just don't use UserBenchmark and you should be fine. 

Thanks for the info, I'll look into this. Do you know a reliable "FPS calculator" website or is there no real way to do that?

Link to comment
Share on other sites

Link to post
Share on other sites

4 hours ago, Definitely not BB62 said:

Thanks for the info, I'll look into this. Do you know a reliable "FPS calculator" website or is there no real way to do that?

Those sites are usually terrible, just find a known good publication that does reviews with your particular game and look at their numbers. 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×