Ryzen 5000 Memory Performance Guide

In this article we'll be searching for Zen 3's retention sweet spot and looking at DDR4 retention performance with the new Ryzen 5000 CPU series. Only before we get into information technology, it'due south of import we establish a reference based on previous testing and insights.

Earlier this twelvemonth nosotros compared Intel's Cadre i9-9900K and Ryzen ix 3900X using two 8GB modules versus iv 4GB modules using the verbal same memory timings. We found that when using rather slow DDR4-3000 CL16 memory, we were looking at a 3-5% performance uplift for the average frame rate at 1080p using an RTX 2080 Ti, but also gains every bit high equally 10% when looking at the one% low results. This was seen with both AMD and Intel processors. Certainly non massive gains, but this is relatively dull memory and we expected that the margins would abound a little with faster retentiveness.

Now before we leap into our updated testing, here'southward a very cursory caption of why 4 sticks are faster than two...

It boils down to how the retentivity is configured, or rather the memory "rank." For those of you unaware, the term 'rank' means the number of 64-bit memory banks on a module. Near consumer-class memory features a unmarried rank, though college chapters modules are usually dual rank, while server-course memory is often quad-rank.

Identifying if your memory is dual or single rank tin be difficult as software doesn't always read modules correctly and not all memory manufacturers note the rank in the modules' ID. Typically, single rank modules feature all the retentiveness chips on one side of the PCB, while dual rank retentiveness places chips on both sides of the PCB. However, that'south not ever the case. A module with chips on both sides of the PCB is really just dual-sided, and can yet be a single ranked module, so information technology'south a bit confusing.

DIMM Module Rank Configurations

Where things can become even more disruptive is when you innovate more retentivity sticks or modules. A system populated with more than two unmarried ranked modules volition actually human action as if dual ranked modules are installed. In fact, in that location's very picayune difference betwixt i dual ranked module and two single-rank modules when connected to the same memory controller, even though the retentivity fries reside on dissimilar PCBs.

So when using two single ranked modules for dual-channel performance, the memory is configured as a unmarried rank. However, when using iv single rank modules for dual-channel operation, the retentiveness is now configured as a dual rank.

This tin requite the 4 DIMM module configuration an advantage as it allows several open DRAM pages in each rank. Although the ranks can't be accessed simultaneously, they can be accessed independently, and this means the controller tin can send write data to 1 rank, while it waits for read information previously selected from another rank and as you lot'll see presently, this profoundly increases memory bandwidth.

How much of an impact this has on performance depends on the application and the memory controller'due south ability to take reward of open pages. But what all this ways is, yes, it's possible for four modules to improve performance over two modules in a dual-aqueduct system.

Our test arrangement has been equipped with the GeForce RTX 3090 and here comes our comparison of the Ryzen 9 5900X, Ryzen 9 3900X and Core i9-10900K when using ii DDR4-3200 modules vs. four DDR4-3200 modules.

Benchmarks

We've only tested using two games every bit that's all the data we demand to clear up any misconceptions you might have regarding four sticks with Zen 3 versus other CPUs.

Testing with Shadow of the Tomb Raider, which is a very CPU enervating game, we notice that the 3900X saw a 14% increase in average frame rate when going from 2 modules to 4, that's a very meaning improvement, though notation we're using an extreme GPU at a low resolution.

Moving to the Core i9-10900K, we once more see a big functioning uplift when using four sticks, this fourth dimension a 15% performance boost. Now with the Ryzen ix 5900X, we're looking at a like performance increase, this time 12% and it'due south possible we're starting to meet a GPU limitation equally Ampere scales poorly at 1080p, but the point is all iii CPUs encounter a similar double digit functioning uplift with 4 sticks, and then this isn't some kind of special feature unique to Zen three.

Hitman ii is another CPU/retention sensitive game, and nosotros're looking to be GPU bound with the 5900X. Let'due south focus on the 1% low data for our comparisons. With the 3900X we're looking at an 8% operation increment with four sticks. However, using the 10900K shows a far more than pregnant 29% operation increase, though it's but a 13% increase for the boilerplate frame charge per unit.

Hitman 2 can be a bit odd and we suspect with the 3900X we're looking at a performance bottleneck that'south more related to cadre to core latency than DRAM functioning. Anyway, with the 5900X we're also witnessing a massive increase in 1% depression functioning with 4 sticks, this time a 21% operation boost. Simply over again we suspect we're running into a GPU limitation for the average frame rate. The point is, the 10900K and 5900X come across a similar increment to performance with four memory modules, and non something that is unique to Zen 3.

DRAM Benchmarks

At present let'due south cheque out how the Zen 3 architecture behaves using dissimilar memory modules, frequencies, and timings. Nosotros're using the RTX 3090 once more, but we besides have information with a more than mainstream GPU subsequently on.

At this point in time, none of our Ryzen 5000 processors work using a 2000 MHz FCLK, which limits us to DDR4-3800 retention. Evidently new BIOS will brand a 2000 MHz FCLK more than likely, though it didn't assistance in our instance. The skilful news is that all worked perfectly at a 1900 MHz FCLK, and that wasn't the case with Zen two.

For much of this testing we used G.Skill's TridentZ 3600 CL14 retentivity which we manually tuned, raising the CL to sixteen, only aggressively tightening the secondary and third timings which massively improves performance for all Ryzen processors. This configuration will be tested at DDR4-4000, 3800, 3600, and an underclocked 3000 config.

We've likewise included some stock XMP loaded memory configurations, one using Corsair's Dominator Platinum DDR4-3600 CL18-nineteen-19 retention, some other using ADATA XPG Spectrix D50 DDR4-3600 CL18-xx-20, simply with two different configurations: one using 2 dual rank modules and the other two single rank modules.

We're also including our 1000.Skill TridentZ F4-3200 C14 4x8GB test configuration used in our reviews and we'll add together in a 2x8GB config for single rank testing equally well. We've washed our best to label this data equally uncomplicated as possible, but we empathise that for some of you it will be a bit confusing. Finally, we're sticking with the Ryzen nine 5900X, though note these results apply to all Zen 3 processors, fifty-fifty the Ryzen 5 5600X.

Death Stranding is a game where Zen 3 processors went like a bat out of hell and despite that this isn't a particularly memory-sensitive championship. For example, if you compare our review configuration using iv DDR4-3200 CL14 memory modules, ranking the memory upwards to 3800 with tightly tuned timings simply boosted functioning by 3 FPS, a mere 1.iii% increase.

Moreover, we're looking at 7 to ten% difference between the fastest and slowest retentiveness configurations tested, which isn't that significant. It also looks like dual ranked DDR4-3600 CL18 memory is comparable to our dual rank DDR4-3600 CL14 memory, at least in this title.

Moving on to F1 2022, we're seeing very little uplift over our examination system configuration when using manually tuned DDR4-3800 memory. This fourth dimension nosotros're seeing a ~five% performance delta betwixt the fastest and slowest memory configurations tested. Therefore F1 2022 is another game that isn't particularly sensitive to retention performance.

We know Far Weep New Dawn is a memory sensitive game, specially latency sensitive and here we're seeing quite a more notable 12% performance gain with the tuned DDR4-3800 retentiveness over our 3200 test configuration.

If we accept the aforementioned 3800 spec and just increase the frequency to DDR4-4000, which correct now in the absenteeism of 2000 MHz FCLK support decouples from a 1:1 ratio with the Infinity Fabric, nosotros really intermission operation a little, dropping down to the tuned 3600 spec.

Horizon Zero Dawn like F1 2022 and Death Stranding isn't peculiarly sensitive to memory and we see just a 5% difference again betwixt the fastest and slowest retentivity. There'due south also little to no difference between unmarried and dual rank memory configurations.

Another game that isn't heavily influenced by memory functioning is Rainbow Half-dozen Siege. Hither we're looking at less than a 2% change between the top and bottom configurations tested.

Retentiveness performance makes a reasonable difference in Watch Dogs Legion. Nosotros're looking at a 6% functioning increase past simply calculation two more DDR4-3200 modules. And so by overclocking and tuning upwardly, performance was increased by a further 3%. Certainly not something that will materialize at higher resolutions where GPU limitations kick in, but it is a measurable difference.

Every bit seen earlier, Hitman 2 is very memory and CPU sensitive. Tuning upwards your memory can make a big difference in this game, though that will just be the case when performance is CPU express.

Nosotros come across that performance using stock DDR4-3600 CL18 kits like the Corsair Dominator Platinum RGB or ADATA XPG Spectrix D50, in a single rank configuration is pretty horrible relative to what we see with the dual rank or manually tuned configurations. Shockingly we're looking at a ~23% reduction in 1% depression performance.

Also, if we expect at the DDR4-3200 configurations using ii and 4 sticks, we come across that our 4 stick test configuration would be upward to 17% slower if we removed two of the modules.

Nosotros're also seeing Spectrix D50 16GB DDR4-3600 modules offering a nice performance boost every bit they were vi% faster than our 3200 CL14 exam configuration. The power of dual-rank functioning is strong, and unfortunately nosotros were not able to test the manually tuned DDR4-3800 configuration in the dual-rank way as nosotros don't take plenty of those modules.

In Hitman, with memory at or above the 3200 spec we're seeing upward to a 21% difference in functioning and upwardly to a 12% increase over our test configuration.

That said, if we increase the resolution to 1440p that 21% margin is reduced to 14%, which is still substantial, just it is heavily reduced by increasing the GPU load. Moreover, whereas the manually tuned DDR4-3800 memory was 12% faster than our DDR4-3200 examination configuration at 1080p, at 1440p it's just four% faster. Even so, dual rank memory makes a large difference in this title.

Like Hitman 2, Shadow of the Tomb Raider is a CPU demanding game that's too sensitive to memory performance. Looking at our test configuration we run into that using 4 TridentZ DDR4-3200 CL14 modules improved performance by a whopping 12% when compared to but two modules. It'south also faster than the single ranked DDR4-3600 CL18 configurations.

We're looking at like performance from the ADATA XPG 32GB DDR4-3600 kit, and of class, that uplift isn't explained past the actress chapters, but the dual rank configuration. Across that though, we're not gaining much with the manually tuned DDR4-3600 and 3800 retentiveness.

We besides ran some 1440p tests with Shadow of the Tomb Raider and this is what nosotros believe you can expect to see in nigh games, even with something equally extreme as an RTX 3090.

Whereas we saw a 19% divergence between the fastest and slowest tested configurations at 1080p, at 1440p that margin is reduced to just 4% and with memory running at or in a higher place the AMD base spec, we're talking about a 0.7% departure.

8 Game Average

If we average the 1080p data across the viii games tested, this paints a clear motion picture of the kind of performance divergence you can expect to see with an extreme GPU at a depression-ish resolution.

In more CPU express gaming scenarios, a manually tuned DDR4-3800 configuration will net you some 7% more operation when compared to a stock retentiveness kit, like the Corsair Dominator Platinum RGB, for example. When compared to our review setup, nosotros're looking at only a 3% boost on average.

AIDA64 Retentiveness Bandwidth

For those of you lot wondering, hither's a look at the memory bandwidth performance of these various configurations. Although the operation of DDR4-4000 was boilerplate given nosotros couldn't run at a 2000 MHz FCLK, the bandwidth is still very impressive, sustaining 55 GB/southward.

The manually-tuned DDR4-3800 memory managed 53 GB/south, which wasn't much faster than the ADATA Spectrix 32GB kit which achieved almost 52 GB/s. Our test configuration was good for almost 47 GB/s, which is about the most you can hope for from DDR4-3200 retentiveness.

Hither's why the 55 GB/s DDR4-4000 configuration didn't dominate the gaming benchmarks: the latency is rather unimpressive at 60 ns and that's only marginally ameliorate than our DDR4-3200 test setup. The tuned DDR4-3800 memory reduced DRAM latency by 9%, hence why it did then well in memory sensitive titles such as Hitman two, Far Weep New Dawn, and Shadow of the Tomb Raider.

DDR4 Frequency Scaling

Speaking of Hitman two, here's a expect at retentivity frequency scaling, and so we're using the aforementioned memory and timings, with the only changes fabricated to the memory frequency and the FCLK which has been kept at a ane:1 ratio for optimal performance, with the exception of the DDR4-4000 configuration.

In Hitman 2, we see fairly consistent scaling as the memory bandwidth and/or latency is improved, right up to DDR4-3800. If we could get a 2000 MHz FCLK working, we await y'all'd see a further 3% performance heave for the DDR4-4000 configuration.

Shadow of the Tomb Raider is a more inline with other enervating games and here we find that DDR4-3600 is the sweet spot, every bit was the case with Zen 2. For those of you wanting as much performance as possible without going overboard on memory prices, DDR4-3600 CL16 looks like the fashion to go.

RTX 2070 Super Benchmarks

Merely wait, at that place's more. Here'south the same retention configuration we but saw in the scaling benchmarks, but this time with using the RTX 2070 Super...

Fifty-fifty in Hitman at 1080p, with an RTX 2070 Super yous're going to be well-nigh entirely GPU bound and if you happen to be gaming at 1440p, well, y'all'll be entirely GPU bound, and no matter how much y'all spend on your retentiveness, or how many modules yous have, performance is going to be the aforementioned.

We run into exactly the same thing in Shadow of the Tomb Raider, even at 1080p we're looking at equalized functioning beyond the board due to the GPU clogging and the 2070 Super is no slouch, nosotros're talking virtually Radeon RX 5700 XT-like performance here, so very solid mid-range GPU functioning. Of course, increasing the resolution only neutralizes the results further and at present it doesn't matter if you're running DDR4-2800 or 3800, performance will be the aforementioned in a very CPU demanding title.

What Nosotros Learned

Bottom line, if you lot're a gamer wanting to maximize performance, if and when you see CPU limited situations, and want the best bang for your cadet, then nosotros recommend getting DDR4-3600 CL16 memory. For most we doubtable 16 GB will be fine, merely if yous tin can afford more than, 32GB is squeamish and it ways if you buy two 16GB kits you lot'll also take the advantage of dual ranked performance.

Correct now something like Crucial'southward Ballistix 16GB DDR4-3600 CL16 kit looks great and costs only $75. Should you want, they will offer a loftier degree of tunability. M.Skill also offers an affordable DDR4-3600 CL16 kit for around $80.

Assuming we were able to get a 2000 MHz FCLK working with future BIOS revisions, we don't think it'south worth spending over $100 on those kits. Chances are, you'll never spot the difference. That stuff is better reserved to overclockers wanting to become a bigger 3DMark scores or whatever leaderboard it is that gets them excited these days.

As for the argue regarding ii sticks vs 4 sticks of memory. There's null new to report since our ain test almost a year dorsum. The operation uplift for Zen 3 is no dissimilar to that of Zen 2 or competing Intel processors. The margins volition also depend on the quality settings used, and of class, the hardware. If you lower the quality settings in games, you're going to exaggerate the margins further and that does get you further away from the reality for virtually gamers.

We realize that most of you will simply install your memory and become gaming, and bluntly unless yous enjoy playing with this stuff, spending hours tuning and tinkering with memory timings for what volition likely amount to very little real-world gains, is just a waste of time.

This tin be more of import stuff for us reviewers trying to carry scientific testing, and sometimes we do become a bit carried away with isolating a specific component in order to come across what performance differences at that place might exist, just it really is important to remind readers that for the most role they're unlikely to see these gains under realistic gaming weather, so keep that in listen.