DDR5 memory on AM5: The Four-Way Roundup!

Luke is back today with a close look at DDR5 memory on the AM5 platform. He test four memory kits in total, with offerings from Crucial, G.Skill, Kingston and Klevv, at a variety of capacities and latencies. Read more in the written article: bit.ly/3SLFXW3
00:00 Start
00:51 Introduction to AM5 memory
01:30 Meet the Crucial 16GB
04:03 G.Skill’s Z5 Neo 64GB
06:31 Kingston Fury Beast 32GB
09:05 Klevv Cras XR5 RGB 32GB
11:27 Test setup
13:32 Performance benchmarks
17:27 Luke’s take on the four kits
22:38 Closing thoughts
Test System:
• Processor: AMD Ryzen 9 7950X at default XMP/EXPO operations
• AM5 Motherboard: Gigabyte X670E Aorus Master
• Dedicated Graphics Card: Sapphire Nitro+ Pure Radeon RX 6950 XT
• CPU Cooler: 360mm AIO liquid cooler
• Power Supply: Seasonic Prime TX-1600
• Operating System: Windows 11 Pro
Check out KITGURU MERCH over here: kitgurutech.myspreadshop.co.u...
Subscribe to our channel here bit.ly/3B5CAAs
Be sure to check out our sponsors store EKWB here: ek.tech/kit-guru
Join this channel to get access to perks: bit.ly/3xF9l4y
Discord invite link: / discord
Steam Community steamcommunity.com/groups/kit...
Be sure to support us on PATREON / kitgurutech and read our MANTRA on HONEST REVIEWS Here: bit.ly/2BopnF9
KitGuru uses a variety of equipment to produce content:
Panasonic S1 Cameras / Panasonic S5 and GH5s Cameras
Panasonic GH4 Cameras
Panasonic G7 Cameras
DJI OSMO Pocket Cameras
Canon Cameras
Various PC builds
Final output - colour grading/titling etc:
Mac Pro 2019 - 16 core, Radeon Pro Vega 2 Duo 64GB, 384GB, 120TB Raid 3xXDR monitors
iMac Pro 18 Core/Vega 64/128GB
iMac 2019 9900k Vega 48/64/1TB
Adobe Premiere Pro CC (PC)
Davinci Resolve Studio 14/15 (Mac)
iPad Pro 12.9 inch (2018) machines with LumaFusion
Final Cut Pro (Mac)
Visit our facebook page over here! / kitguru.net
Visit our Twitter page over here! kitgurupress?lang=en
#DDR5 #AM5 #EXPO

[@KitGuruTech]

Interested in DDR5 - would you pick up one of the kits shown here?

[@dominicmoass8478]

that g.skill kit is pretty good looking IMO

[@Kbux0830]

I went with the kingston fury kit on my current build, solid memory. Bought it for the expo and xmp profiles. Running an am5 system.

[@SimonZerafa]

That Kingston kit was less expensive when I ordered directly from Kingston UK. Worth checking vendor sites for pricing when ordering. May well be less expensive that the usual third party vendors 🙂🤷‍♂️

[@jakoj2098]

wish you added Corsair as its betweeen that and the GSkill trying to decide think the Gskill may have tighter timings

[@sahinyt9549]

Again really detailed look at a key part of a system build thanks guys.

[@harleylover9446]

Nice roundup luke. ❤

[@stickyfingers9710]

Some good memory here. Nice selection

[@reptilespantoso]

VERY happy with my Kingston 64 Gb 5600 kit. Can overclock to 6000 with tight timings without raising Voltage. Good quality stuff. What matters is SK Hynix chips.

[@boltsandbraces6261]

Kingston make some good memory. They are never seen as being a cool brand but they are very solid.

[@danemyburgh]

Sk hynix chips as in you should go for that over Samsung you mean?

[@Interloper322]

Kingston fury beast??

[@MrSpamcho]

Does it matter if you get straight the 6000 kit instead of the 5600?

[@jenwwielerim]

Quite like the look of that Klevv stuff.

[@fairycat]

Thanks for review

[@lucifersheadscarf6988]

Will watch this tomorrow. Hope it’s a bit better than the last luke review which he sounded bored out of his head

[@axeltroja9390]

Good video from another Video that compared multiple kits and from diffrent companies I know that it doesnt#t bring any more perforamnce to go above 5600 CL28 - it is possible to optimize sub timinings a little bit, but the G-Skill Units are already very fine tuned.

[@olliebrown89]

The Klevv ram looks so nice, I just wish I could have the G.skill performance.

[@canarychrome7012]

I’ve watched this and still not sure which came out top 😮

[@Aleph-Noll]

i got a g skill 6000 cl30 2x16 kit for 110 and its working great

[@bekironder]

Everything apart from the Crucial budget RAM is very evenly matched - may as well go for either the cheapest of the three others or the best looking depending on your build requirements.

[@KitGuruTech]

I think a lot of the RAM decisions these days do indeed come down to preference for styling and cost. Unless there are some quite specific games or workloads being used, that is. Luke

[@haremofprocessors6954]

Haven’t jumped on the ddr5 bandwagon yet. Will do soon

[@rufusmctavish1164]

I’ve always had good luck with Corsair memory. Never had a problem.

[@Sdracogotsauce]

Hello ,i just bought a combo of MSI tomahawk 650 mobo and ryzen 5 7600x . I got myself the G.skill Trident z5 NEO RGB 6000mhz 32cl . Will i have any issues with conectivity or compatibility ? Its not on the QVL list of the mobo.

[@hairsprungdutchtechnique9626]

Gskill. Always ❤

[@QuentinStephens]

Maybe I missed it but how did you control for RAM usage? Windows has a habit of making use of RAM, so if you go from 64 GB to 16 GB Windows might load extra stuff on startup (expecting 64 GB) then have to swap it out, thus impacting performance.

[@KitGuruTech]

We're not controlling specifically for RAM usage or caching etc.. We are mainly interested in how different products will run with real world use cases. So, if Windows is doing some funky trickery because there is 64GB installed, and that changes XYZ performance, that is what we are showing in the data. Kit vs kit though, we ensure there is time for the system to 'settle' after switching and start-up. I've been doing this for years as Windows used to have a habit of running virus scan or updates soon after start up, thus ruining test data. Luke

[@QuentinStephens]

@@KitGuruTech Great move to wait for a while.

[@itstheweirdguy]

I'd say my analysis of your analysis is that you don't want 4800 C40. The other 3 being about equal except in synthetic for the most part.

[@hollisnelson5925]

I thought I heard about an issue using 4 sticks of DDR5 memory on AM5 boards. Does anyone know if there is a problem using 4 vs 2 sticks?

[@poisiion]

Guy at the computer store told me the same thing, advised me to not get 4 sticks of DDR5 6000hz G Skill ram but instead get 2 sticks and then you won't have a problem

[@TheSjuris]

@@poisiion had a huge problem with certain games crashing for no reason. Ended up being due to 4 sticks of 6000hz ram. Went back to 2 sticks problem went away.

[@DrBreezeAir]

I run 6000 32Gb non RGB kit from Kingston with my 12600KF.

[@lavascoksaran6725]

Not a bad price for DDR5 memory for the Crucial stuff.

[@LSxDeviL]

Did Luke change the camera angle to a more budget friendly angle of his workspace?

[@KitGuruTech]

Luke's been out in the US and filmed this while he was over there which is why things are a bit different!

[@treyquattro]

I put G.Skill Trident Z DDR5-6000 in a 7950X system

[@KitGuruTech]

Very nice. How has it been performing for you? Luke

[@treyquattro]

@@KitGuruTech Hi Luke. Timings are 30-40-40-96. AIDA64 scores are: Read: 78779 MB/s Write: 81779 MB/s Copy: 73584 MB/s Latency: 66.3nS It's an MSI PRO X670 motherboard. EXPO is the only o/c that's been done so far. My i9-12900 with Corsair Dominator DDR5-5200 is actually faster on reads and copies (just the one test run in both cases): Read: 80014 MB/s Write: 75586 MB/s Copy: 77562 MB/s Latency: 75.8nS MSI Carbon Z690

[@KitGuruTech]

@@treyquattro That's very interesting between the Corsair and G.SKILL on Intel vs AMD. Are both kits the same capacity? Luke

[@treyquattro]

@@KitGuruTech yes, both 64GB, two sticks apiece. I expected the AMD system to be faster. The Corsair RAM is CAS 40

[@MrAckers75]

The klevv are real dark horses due to mine being Hynix a die. Got 7000mhz at a 38 latency on my z690 board

[@gulhanmseyit7696]

Love how brexit has made Amazon imports from the EU a thing. Who needs that single market with cheap prices when you can pay more for the same thing?

[@overclockingdemon7179]

It’s true ! 😂

[@jamief1263]

Tell me about it, they sent the wrong processor to me twice and then they ran out of the one I wanted

[@iordanchis2437]

Best wishes from Germany. I'd send y'all some cheap goddies over, but wait, there's this form I've to fill out first, and...what the? import tax?! Fuckin 'ell!😩

[@johnnyc.9488]

Hmmm... great video, but is there no 32GB kits that are 6000 MHz with comparable timings that don't have RGB?

[@KitGuruTech]

Yes there will likely be comparable non-RGB kits, but we don't have those in for review. Luke

[@csn2300]

What’s the best white one?

[@drekceldude]

I have Kingston Fury CL40 6000hz memory I think it's the Asus AM5 M/B bios that's coursing my problem, if I set memory at 6000hz the computer will not post, drop to 5600hz then runs just fine, I've noticed that Gigabyte M/B's tend to have better memory timings out of the box, maybe Asus just slower at putting out updated bios.

[@KitGuruTech]

Have you tried running 5600MHz using EXPO/XMP, and then power cycling it up to 6000MHz? I used to do this a lot in the early AM4 days as 3200MHz was always tough to cold boot. Luke

[@zulaerol8359]

Kingston’s Fury “Beast” naming just seems a bit silly for a stick of RAM but whatever.

[@BMBM103]

I had sick problems with corsair vengence on AM4, blocked to 2133hz cuz the XMP mode was for intels motherboards and crashing ( trying to change values from BIOS wasn't helping ).. So what brand should i chose with something like XMP mode but for intel?

[@RadialSeeker113]

The adata xpg lancer rgb 6000cl40 is my go to in terms of price : perf

[@saputrasaputra3347]

nah cl40 sucks cl 36 or 34 is the bare minimum

[@tilapiadave3234]

CL 40 ,,, no NO NO !!!!!!

[@chrisbullock6477]

Not sure why in all the testing people continue to only use DDR5 6000 on X670 and much higher on Intel. I personally picked up some DDR5 6400 and see a noticeable difference compared to 5600 especially

[@TheSjuris]

6000 is the sweet spot for Ryzen right now.

[@guccipucci69420]

Because anything over 6000 is much more unstable... This has been said in like every single video talking about it

[@humilicious]

what about 6000 cl30?

[@chrisbullock6477]

Loving my 6400mhz G.Skill Ripjaw memory.

[@64BBernard]

A purchaser of RAM may want to check their motherboard's QVL page to see if the RAM they are purchasing has been tested for stability.

[@KitGuruTech]

This is a very good suggestion! QVL is not always a complete list. But if one kit is on the QVL and the other is not, that may help sway the purchase decision. Luke

[@RiAv33]

Hi, i have a Asus MOTHERBOARD AMD X670 ROG Strix X670E-E Gaming WiFi, AM5, can i buy this memories for this motherboard? G.Skill Trident Z5 Neo RGB F5-6000J3040G32GX2-TZ5NR 64 GB 2 x 32 GB DDR5 6000 MHz MANY THANKS!!!

[@JustAFan444]

I've had multiple GSkill ddr3,ddr4, and now ddr5 memory kits... my ddr5 was, unfortunately, the only doa kit I've gotten. Disappointing to say the least.

[@dolebanana5071]

TeamGroup T-Force Vulcan DDR5 6000 vs G.skill z5 7-zip benchmark

[@iIVIakeitRainXO]

I’m no expert but surely it doesn’t make sense to test different memory sizes. Surely you should of chose say 32gb for them all to get accurate readings

[@ravenfeeder1892]

GSkill really are the kings of AM4/5 RAM, but you do pay for the privilege.

[@TheSjuris]

The Intel XmP version of the same kit is $15 more.

[@peterjansen4826]

A few suggestions for getting the most useful data when comparing RAM, as I see it. In the current setup it is hard to tell how much of the impact is the clock frequency, the timing and the amount ofr RAM. I think that it is better to use the same high-end (so at least 6400 MT/s and the lowest latency which you can get on this AM5-motherboard) modules (at the moment 1 pair of 2) and change the clock frequency and secondary timings of that to a wide range of values, that way you get the most apples to apples comparison considering that now with different modules you also have different tertiariy timings. Compare both clock frequency and timings separately (so same clock frequency and different timings and vice versa) and optionally compare to 4 modules too but considering that Wendell from L1T recommends to not go over 2 at the moment that might be less relevant. As always it is all about how much extra performance you get for the extra costs. As a computer-user I don't really care about that when the cost is double or triple and you gain 1-2% but when you spend $50 extra and gain 5-10% then it is rather relevant, especially on a $1500 system. I recommend to focus on the total extra cost of the system (excluding hardware which you move from one system to the next?) which you would pair such RAM with to evaluate whether you should also benchmark with even higher-end and more expensive RAM but this is more a point for next generations of AM5 considering that for the current generation allegedly over 6400 MT/s is not possible on a large % of the motherboards and/or CPU's.

[@KitGuruTech]

Very good comments on the testing of RAM. The main aim of this article was to review four individual kits that we had in for review, and see how they perform on the new AMD platform. For direct testing of how memory specifics (frequency, timings, and DIMM setup) affect the performance of the AMD platform, your method is indeed very useful. We did similar to this when we looked at a DDR4 vs DDR5 comparison for Intel 12th Gen because it was ideal as the rest of the test hardware could be kept completely consistent. Total extra system cost is a good way of viewing the price versus performance. Especially as RAM can usually survive many years of upgrades. However, there is the element of future improvements too. 5600MHz was the go-to just a year-or-so ago. Now, 6000MHz with nice timings is an affordable option. I'm liking the way that 2x32GB kits are heading for performance and affordability too. Luke

[@KitGuruTech]

@El Cactuar Force of habit. I prefer using MHz to be honest, as it is how RAM kits are sold and how they're communicated to the wider market (even though MT/s is perhaps technically better). Luke

[@KitGuruTech]

@El Cactuar I know how it works and what is 'correct', but it's hard to go against simple, common convention that people genuinely just understand. Kind of like how 5Gbps USB 3.0 has been renamed several times (although RAM being quoted in MHz is far less misleading than the constant USB renaming). Luke

[@KitGuruTech]

@El Cactuar The reality is that most people readily understand what 'MHz' means when it comes to RAM. After all, that's how we decide which kits to purchase and how to set them up in the BIOS. I'm not a massive fan of the confusion either, but it's probably best directing your comments to the manufacturers to change their terminology if you feel so strongly about it. New users do indeed come to reviewers to learn, so it's not ideal if reviewers use terminology that is arguably 'better' but is dissimilar to what a product page or retailer listing or BIOS profile say. Luke

[@jasonhenninger8220]

DDR5 is very sensitive to secondary timings. Different boards will run the same RAM differently on auto settings. Two kits at the same CAS will run differently on the same board with how it reads their secondary timings. DDR5 6000 30 36 36 36 96 96 (GSkill Trident Z5 Neo) DDR5 6000 30 32 32 32 76 112 (Corsair Dominator) GSkill was faster Buildzoid/ActuallyHardcoreOverclocking has a set of manual timings that should be applicable to most kits and will tighten timings and improve performance.

[@Isaac-hm5ef]

Faster by how much? Under which use case scenarios? In a way that is actually noticeable in real world use or only when viewed as a number on a chart?

[@xayzlo1545]

Another Luke review, another bamboo chopping board showcase.

[@KitGuruTech]

Wouldn't want to go for single use plastic, now would I. Luke

[@wankerka]

How come Corsair is not on this list? are they shit?

[@KitGuruTech]

Nothing wrong with Corsair RAM, we just didn't have a kit in for review that would fit into this piece. Luke

[@stefanradunovic8318]

Amd 5950 3d is broken

[@artgamer453]

The AM5 platform is going to be a good future platform but right now it still doesn’t make sense.

[@Vonklieve]

DDR5 is too expensive, but prices are slowly coming down. I would love to get some but cannot justify the bad economics. overpriced CPU's from AMD, vastly overpriced quality DDR5 motherboards and DDR5 6000, where you can say goodbye to about £180.00 from your wallet for entry level 32 gig of the stuff (16 gigabyte will not be enough soon). Until prices level from premium pricing sticking with 5800x3d, 32 gig of DDR 4 3200 and an Asus B550 motherboard. And I think many will agree. Unless AMD and Intel do something similar to what they had to try in USA, where DDR5 was given for free. Adoption of DDR5 will remain slow.

[@2009numan]

the bench marks were a bit unfair in my opinion as they were all different speeds and capacity

[@KitGuruTech]

that's kinda the point behind the video though - showing the performance on offer from different kits across the pricing spectrum

[@cloud9photography126]

G.Skill FTW

[@aboneolpls6920]

Those Crucial boards are so sparsely populated! Look at all that empty acreage.