4x4GB vs. 2x8GB, Intel & AMD Dual-Channel Gaming Benchmark
ฝัง
- เผยแพร่เมื่อ 10 ม.ค. 2020
- Support us on Patreon: / hardwareunboxed
Merch: crowdmade.com/hardwareunboxed
DDR4 Memory: amzn.to/2nM6BDy
TeamGroup T-Force Vulkan Z: amzn.to/37W0tKZ
Read the feature on TechSpot: www.techspot.com/article/1971...
4x4GB vs. 2x8GB, Intel & AMD Dual-Channel Gaming Benchmark
Disclaimer: Any pricing information shown or mentioned in this video was accurate at the time of video production, and may have since changed
Disclosure: As an Amazon Associate we earn from qualifying purchases. We may also earn a commission on some sales made through other store links
FOLLOW US IN THESE PLACES FOR UPDATES
Twitter - / hardwareunboxed
Facebook - / hardwareunboxed
Instagram - / hardwareunboxed
Music By: / lakeyinspired - วิทยาศาสตร์และเทคโนโลยี
![[ไฮไลต์] แบดมินตันหญิง "เมย์ รัชนก" ไทย vs เบลเยี่ยม รอบแบ่งกลุ่ม | โอลิมปิก 2024](http://i.ytimg.com/vi/qfMROCoiKMY/mqdefault.jpg)
Buildzoid talks about motherboard topology here: th-cam.com/video/D_Nb9CohSBM/w-d-xo.html
I saw the video almost the same time you posted the comment, great :-)
Great complementary video.
#HWUB Hi Steve, please can you do some investigation and testing on whether changing VRM and Chipset on motherboard thermal pads with better quality ones, graphite pads(increase mounting pressure) or thermal paste. Does this improve vrm temps and overclocking. All hail Steve the Benchmark King.
A Lot of people will get triggered but .... how about a 4K version of these benchmarks? I'd like to see if the FPS are within reason for playing games at 4K with different modules configuration potentials. Also You do have a 2080 afterall....
Hey mate, just so you are aware, Techspot.com stole your content and is passing it off as their own...........if that was me I would be pissed.
link here
www.techspot.com/article/1971-more-ram-modules-better-for-gaming/?fbclid=IwAR3YwC0AWbo3zCJlYTPimKLfw3LKGB52RIApDoxnSX_8l-_MuAbdBDleHa4#allcomments
More dimms = more RGB = more FPS
Science
The human eye can only see 2 dimms.
S T O N K S
More money too!
Science bitch!
DIMM-wit.
It might be interesting to repeat this using 2 sticks of dual rank memory against 4 sticks of single rank.
Came to comments to say exactly this :)
I was wondering the same. Sounds like there would be minimal difference between 2x16gb vs 4x8gb, except for maybe memory stability like he spoke about at the end. Would like to see this tested as well.
Yes, please do this.
Did you guys watch to the end? Steve said this and it may be a video for another day.
@@techforge-Nate I started reading the comments lol, thanks for pointing that out. I will be keeping an eye out.
I'm very glad that one of the bigger PC tech channels has finally talked about memory rank interleaving. This is a massive thumbs up from me.
@@ArteriusSaren yeah we're going to learn a Slavic language just to watch one video... And you call us arrogant?
There's a Wikipedia article called "Memory rank". It was first written just over ten years ago and it's in English. It's a little lacking in the citation department but it does have a section about the performance impact of multi-rank modules. Unfortunately it's not a video so you have to read it in order to learn from it. People with language skills could contribute by helping to translate it..
@@ArteriusSaren maybe if you fucks wouldn't be such nationalistic pricks people would like you more.
@@ArteriusSaren Only 3 years ago... memory bank interleaving and it's improvements to performance has been a thing since at least 32-bit 72 pin SIMMs... at least that's when I learned about it back in the 90s.
@Maxim Wow, such ultranationalist bullshit political garbage. Not everyone is obligated to learn other languages, twat.
Steve: "Are 4 sticks faster than 2?"
Me: **laughing in mini itx*
Same here, but it's more of a crying in ITX
@@alexp3577 These findings are only true of identical timings and speeds. Two modules still give you more stable overclocks. 2 sticks at 3600 MT/s is faster than 4 sticks at 3200.
#thatitxlife
Most people use only 2 sticks anyway so theres no difference there, Plus 2x dual rank dimms=4 single ranked dimms (which nearly no one does).
So being ITX has no downside unless you want more than 64GB of ram.
You can just run dual rank DIMMs and have the same result on your Mini ITX motherboard.
I never expected you to do an actual video on interleaving, when I asked you for it on the last RAM test video. Thank you so much. You're the best! Now we finally have some solid data on a good set of games.
Thanks you for your knowledge and love your channel I always learn something new and important for the PC.
Thanks guys, you always do an amazing job or testing and explaining the results. Keep up the great work.
1:53 - also to consider the ICs (and some ICs also perform worse in larger varieties for example 4gbit AFR vs 8Gbit AFR)
Would have been neat to see 2 dimm single rank vs 2 dimm dual rank vs 4 dimm dual rank in at least one title where the scaling was seen.
I believe 2 dims dual rank would perform 100% identical to 4 dims single rank as he said at the start. So the first part he more or less already tested. The 4x2 rank would have been interesting, though. But you might struggle to find boards capable of going past 3000mhz with 8 ranks hammering the IMC.
Well that's why I said one game. No need to spend a ton of time on it but just to be able to show it for the sake of science.
A relatively cheap little trick on the AMD FX CPUs was to use 4 dual rank RAMs to squeeze a little more power out of them.
In theory, it shouldn't matter, if you are having 2 ranks of memory on a single module or on two separate RAM module.
But in practice, you can reach the IMC's or the motherboard's limitsts, if you are overlocking, or using high-end memory sticks. You may have to lower the clock or increase timings, if you want a dual or quad rank memory configuration. The overall performance will be very similar, whatever you go for more ranks or for better frequency/timings, because an other hardware component is bottlenecking the memory performance (either the MoBO or the IMC).
Thank you , I do appreciate the time it must take for you to do all this testing. Graham from the uk😀
Thanks Steve and all at Hardware Un-boxed. I actually LEARNED something today in regards to memory!
Brilliant!!!
Was waiting for this cheers
Ohhh those benchmarks, such goodness
How did I not discover your channel sooner? This is very helpful and so are your other videos.
Thank you so much for this video.
It's both informative and highly interesting.
Would love to see the dual rank dims test.
Thumbs up.
From the different sources I've seen it would seem that things work this way with Ryzen 3xxx: 4 SR modules can net you some extra performance over 2 SR modules especially if you use the XMP profile of your memory. If you use DR modules and try to optimize their timings and frequency you are more likely to get better results with 2 modules instead of 4.
This is mostly true, as I understand it. With quad ranked memory (two dual-ranked sticks in each channel) the memory controller is often not able to keep up, and the DIMMs themselves are unlikely to support high frequencies. Quad ranked memory is better than dual ranked in a hypothetical situation where the memory can run at the same speed and the memory controller can handle it, but this hypothetical situation doesn't often apply to desktop PCs (though it can sometimes be relevant to servers with high-capacity, low-frequency DIMMs).
Is it true that the Ryzen 1 and 2 series of CPUS work better with 2 DIMMs still? I read that 4 DIMMs are too demanding on the CPU's memory controller.
I have my 3950x with 4x16gb (DR) working perfectly at 3800 DTR (1900mhz) with the IF at 1 to 1 ratio (1900mhz) and this with optimised timings, I got a tiny bit less than 60gbit of bandwidth which seems to be the limit with Zen 2 (cf buildzoid). With Intel the memory controller take a big hit with 4 DR module but with Zen 2 the limit will still probably be the IF frequency...if you aim at the 1 to 1 ratio.
@@Sonikks More physical DRAM chips attached to the memory bus do present a more difficult load for the memory controller to drive, which is why servers use registered (buffered) memory. Motherboard manufacturers usually specify the maximum recommended speed for the various combinations of ranks per DIMM and DIMMs per channel. For example, for 1000 series Ryzen it was typically 2666 MT/s for one single rank DIMM per channel, 2400 for either one dual rank DIMM per channel or two single rank DIMMs per channel and 2133 MT/s for two dual rank DIMMs per channel. Anything greater than this was, technically, an overclock and not guaranteed.
The more sticks of ram you put in, the worst its gonna get stability wise. I have a kit of 4x8gb and when i use only 2 DIMMs i get it to run at 4000mhz. If i put all 4 the highest i can get is 3300mhz.
Really interesting subject, and great analysis. Love your channel.
Thank you for your videos. I've been researching a lot over the past few weeks on what direction to go with my PC build and you have genuinely helped me, thanks for that. Decided to go with a Ryzen 9 3900x on a Asus TUF Gaming Plus w/ 32GB Trident Z Royal memory and keeping my EVGA GTX 1070 Hybrid until the next gen GPU's are released.
Wow, didn't expect a flashback from Computer Architecture when I clicked on this video. Excellent video, very informative and well spoken.
9:11 Thank you, happy merchant.
nice timestamp
Gewel ✔ nice checkmark
There is also something about how the bus is laid out on the PCB that affects 2 vs 4
Thanks for the video! I've been curious about this for a while now. It would be great if you guys could follow up with a similar comparison for non-gaming, professional workloads. Cheers!
Finally someone covered it! I've been asking multiple channels to test this for a long time because there's too little info about dual rank memory and channel interleaving in case of 4 dimms. I'm personally running 2x16 DR on 3466c14 with tweaked timings on my 3700x :P so far happy with it and been since first gen
Kudos to you guys for covering the 4 sticks of ram versus 2 way before gamers nexus apparent recent discovery on Zen3!
The human eye can't see any more than dual ranked modules anyway
Its called retina modules
*My vision is augmented*
Damnit!!!
Still funny.
I am in the process of planning my next build, which will be used primarily for gaming, so I found this video extremely useful. Thank you guys!
Thx 4 the overview Steve... I would be interested a look at Dual Ranked Memory Modules... 👍😎
I would definitely like to see this done with dual-rank modules as well. I'm sure other ITX system owners would, too.
7:27 For results.
thank you for helping me finally understand this rank stuff! great video
You guys are just the best. Keep up the incredible work!
I remember asking this in 2017 dayum
That's actually pretty interesting. I've wondered this in the past when buying memory. I would of expected the opposite results
Yeah, me too. Didn't even heard about memory ranks and interleaving. So glad that now I know about it.
Thanks for the testing and the explanation... it also helps explain the results I saw when going from 16Gb to 32Gb by adding another two 8Gb sticks (of Corsair LPX with the same model number, but the cheap sticks I got from Amazon in the black friday sales were an older version than the ones I've had since 2017) on my Ryzen 5 1600 @ 3900 Mhz - my unscientific testing in five games at 3440x1440 got the same results (less than 0.6% difference at most, averaged over five runs) with 32Gb at 3000 MHz and XMP timings (CL16 instead of the CL15 when using only two sticks) versus 16Gb at 3066 MHz CL15 with optimized sub-timings... I had expected the 32Gb to trail a bit until I tweaked the timings but with these results I haven't yet bothered to do so.
What ever you will change in the future on Hardware Unboxed, please never change your outro song. I love it!
Appreciate the quality content 👍
I remember back in the 80s spending almost $1,000 to back fill the memory on my Intel 8088 Columbia Data Products PC from 128kb to 640kb. It's been fun watching the PC grow from that to what I use now.
When I first became interested in computers and attended tech school memory was one hundred dollars per megabyte. I have three million two hundred thousand in my new desktop!
Prices are insanely cheap! Back in Jan. spent $1200 for 192 GB of RAM for my three Threadrippers. (64 GB each)
Very interesting! I've pretty much watched every video from you, GN and Linus over the last 18 months, but none has ever covered this. RAM still is a mystery to me in many regards and I now have more questions than before. Time to dig into it!
Thanks for all the hard benchmarking work!
Great content, looks like you got your work cut out for you with the alternate memory configurations:
Is it better to have 4x single rank vs 2x dual rank?
Is it better to have 2x single rank at 3200 vs 2x dual rank at 3000?
Any advantages of having 4x dual rank over 2x dual rank?
Who knows? the videos aren't out yet!
I've been wondering about this for some time, can't seem to get an answer anywhere. Plus I don't have the funds to support this kind of testing.
You need to just think in ranks, instead of DIMMs. On a dual channel motherboard, you want 4 ranks for interleaving. Having more ranks will not benefit you anymore.
The next thing is that the less DIMMs you have, the less the memory controller is strained, which might net you a little tiny bit better overclockability.
And that's it.
Is it better to have 4x single rank vs 2x dual rank? The latter.
Is it better to have 2x single rank at 3200 vs 2x dual rank at 3000? Depends on the timings and if your application is more sensitive to latency or bandwidth.
Any advantages of having 4x dual rank over 2x dual rank? I don't think so.
@@xnoreq Until I actually see some field testing I consider this to be speculative or expected behavior. I'm curious for these to be qualified and quantified with blue and bluer bars. If it's not in a pivot table, then is it really data? XD
@@mattroy3154 Nothing speculative. All of these things have been tested and benchmarked before, even specifically for Ryzen setups.
More RAM testing yes please.
Quite a nice video, you are diving into some fairly deep technical topics that I wouldn't have expected.
Generally speaking, the issue comes down to how much of the memory bus bandwidth is being used by the program. Games do not tend to exercise main memory by a whole lot. Try testing with some things that are more memory intensive (that aren't games).
-Matt
Great video!! Yes please test 2 dual rank vs 4 dual rank modules and add them to the results of this test.
In ddr3 systems like IVY Bridge and AM3+ systems, this was a myth. Running 2 dimms often allowed the ability to use 1T command rate over a 2T command rate, which was a huge impact on performance in some scenarios. As much as 8%. 2 small low density dimms were always the easiest to get tighter timings when tuning.
Quad or Dual...
That's something I wanted to now, thanks !
@@wykeless Yeah but Ryzen kind of shaked up the whole RAM debate, it really requires different considerations, so this video definitely has it's place.
@@wykeless And I happened to never have search about it, at all. That just the way it is. So when this video poped up, I simply wanted to know. Please don't be mad at me :\
This is not quad channel
@@frosty98412 quad channel need to be tested on threadripper from amd
and your not gonna buy those for just gaming only
from intel hedt = skylake x or coffe lake x or cascade lake x
@@shredderorokusaki3013 This comment hurt me physically in multiple ways.
Thanks for this test. I’ve been trying to run cinebench scores between a 2x8 kit and a 4x8 kit and they seem to be similar. Same with gaming.
I was waiting for this video. Thanks.
How people saying difference between 2 and 4 sticks is not noticeable, but there's an easy 20 fps difference at times?
Depends on IMC and topology of motherboard i think...
Depends on the quality if your ram, motherboard and cpu
Sure, daisy chain is faster compared to T topology but I assume they could all reach at least the highest JEDEC standard. Zen2 needs higher frequency for better performance, above JEDEC.
@@PhreakDarkSoul Daisy is better for 2-DIMM configs - though not as good as a 2-slot board - while T-Topo is better for 4-DIMM configs.
cheers
Shredder Orokusaki no, I don’t think I will
Ryzen 3rd imc from what I’ve heard is outstanding. And most mobos (like 90 percent) are Daisy chain.
Finally someone made a video about it, and that too pretty explanatory. Great thanks.
Love this content , could we see a comparison on memory timings and scaling ? the Cl 16 ,14 stuff is really super expensive ( double the price in Canada for a 32 gig kit $200 vs $400) would be nice to see a comparison of cl 16,14, vs 17,18,19 kits if possible , keep up the good work.
The 68 people who disliked this only have memory in single - channel
Kody Tourelle how do you download ram?
I would love to see more 4K benchmarks in general, seriously miss seeing those!
4k lowers fps, and as such makes the margin of error higher, especially with stuff like this where you need as much precision as possible
@@jaykoerner
..and would tell that there's no real difference which one you choose If you play at 4K
They are benchmarking cpu peformance, no point in doing that at 4k.
4K benchmarks only make sense for GPU tests.
@@kimnice depending the game settings and card there's no difference between a dual core at 2ghz and a 5ghz 16core, what matters is what your trying to test, "4k is the new standard" doesn't mean anything when your testing something that has nothing to do with the gpu, in most games the resolution has little to almost no effect of the maximum fps(where the gpu is not the bottleneck), if you get 100 fps at 480p your going to only get 100fps max at 4k, just they way things are
this is crazy ive ALWAYS heard the opsite for YEARS now so cool i love these test
This is the most brilliant explanation of dimms in relation to mobo memory controllers and ranks, thank you
Just when I think I know how RAM works, I see this video! Great video bud, keep em coming.
"Thaiphoon Burner" is a great program for reading RAM dimms and there memory IC's. Even the brand, rank, PCB's memory IC layout and die (nm) that the manufacturers used.
I've been saying that a 4x4GB kit simaler the one used here (3200MHz CL16, etc.) or even DDR3 1866MHz kit OC'ed to 2400MHz+ would net a 5% to 10% performance boost over a 2x8GB kit at the same clocks and timings, on Intel's platform since the z77 3770K days. Also going back to the Core2 Quads with a 4x1GB or 2GB DDR2 1200MHz kit.
Same with a 4x8GB kit vs a 2x16GB kit.
I tested this with a 2080 but i'm sure simaler results can be seen using a lower-end GPU like a RX 5700/XT.
With more dimms, it can be harder to hit higher overclocks and lower timings on certain daisy-chain memory traced motherboards. ASUS likes to use this "daisy-chain" memory trace layout on there Strix boards. I'm guessing they figure more consumers will be buying/running a 2x8GB kit or a 2x16GB kit.
Overclocking a 4x(X)GB kit can depend on the specific memory IC's, PCB and module layout as well as the memory trace layout on the motherboard. You'll hear alot of overclockers favor an A-0 IC layout and Samsung B-die in a 2x8GB kit for example. If the motherboard is using a T-Topology memory trace layout, like all of Gigabyte's z370-z390 Aorus line, then this has no effect. In fact having 4 dimms in a T-Topology memory traced motherboard will run faster and more stable than a motherboard with a daisy-chain memory trace. A T-Topology memory trace just means that all of the boards dimm slots are wired (equally) the same distance to the CPU socket. A daisy-chain memory trace is just like it sounds. Each dimm is basically tied to the next one. That's why on most boards you'll see a recommended 2 dimm slot configuration when using 2 sticks. On Gigabyte's boards, any 2 dual channel dimm slots can be used because there all wired the same. The closer the dimm slots the better the overclock. That's how GSkill was able to hit 6GHz on MSI's iTX 2dimm motherboard.
The board used here is Gigabyte's z390 Aorus Ultra and it uses a T-Topology memory trace layout. The results shown is going to be the best case scenario. I have a 4x4GB kit of 3200MHz CL16 running at 3800 CL17 and its faster than a 2x8GB kit of 3200MHz CL14 Samsung B-die running at 4000MHz CL17. Not by much but the difference is there. Using Gigabyte's Aorus boards, i would go for a 4x(X) kit because of its T-Topology memory trace layout. For most other boards like ASUS, MSI and ASRock a 2x(X) kit would be fine. Though, for the price of a 2x8GB kit of 3200MHz CL16 is anywhere from $60 to $80 and 2 2x4GB kits (4x4GB) is about $88. So i guess it depends on the budget and timing.
Some people would buy a 2x4GB kit or 2x8GB kit and add another 2 dimms a year or so later. The only problem is, that the 2 newer dimms might not have the same exact memory IC's in it. Which could cause instability or even the XMP profile not to work.
I've done this a few times and have had no problem with matching 2 kits that were manufactured a year apart. Just be on the lookout for specific kits and which IC's they use. "Thaiphoon Burner" is a great little program that can help tell which IC's a manufacturer is using.
Don't waste performance and choose 3600 sticks at least for Zen 2. ;)
thanks for the informative comment, but what exactly you mean by "The closer the dimms the better the overclock", you mean ram used in the ram slots closest to the cpu (if you only have 2x2) or the population on the ram slots itself for example 2x2 vs 4x4
@@smolcocky6466 th-cam.com/video/D_Nb9CohSBM/w-d-xo.html
Very interesting angle, thanks a lot for that test!
Great video, thank you! And hay, I got the Tforce memory for my build, awesome!! lol
Does it effects APU's performances?
More ranks are likely to have a more pronounced benefit for APUs, since their integrated GPU adds more pressure on the system memory. APUs also greatly benefit from faster memory. That's being said, at some point it becomes worth spending the money on a discrete GPU instead of better memory.
Now THIS is quality content
Now THIS is no quality coment
Steve..great work per usual mate !
Wow that was a great video, thanks!
Earned my subscription!
Can you repeat with 32 GB DDR4 3600 CL16? 2x16 GB and 4x8GB?
16gb modules are almost always double ranked, so it's the same thing, at least you have 8gb Dr. The 8gb Dr are extremely expensive, so...
just reading the title I already know this is gonna get a like out of me
And I already can say the answer is No, at least because it's easier to oc two sticks of ram.
@Ziv Zulander don't really care, I just know it's better to have 2 sticks in dual channel.
EXCELSIUS Then why are you here
@@goldash17 to say that I know it. Other questions?
You don't smash the like button on every single video you watch? pathetic.
Great video, thanks for doing this thoroughly
Great topic! Often wondered this myself, but I heard two sticks perform better than 4, but see the video. When AM4 first came out there was definitely a benefit using 2 vs 4 sticks, but looks like that's all been worked out.
What about using four dual-ranked memory sticks?
More memory sticks would probably work better on T-topology motherboards
I was just thinking about that and the video is here!
Doesnt this also depend on if your motherboard rocks a t-topology or daisy chain setup? T-topology is better for 4 dimms and daisy chain for 2.
Only the ASRock X570 Taichi and Creator are using t-top from the reviews that I've seen for X570 boards, no idea for Intel. Topology has nothing to do with interleaving, though, only electrical balance in the circuit which governs highest reliable frequency when using 1 DIMM per channel or 2.
You noobz don't understand simple facts, 4×4 configuration was faster coz it was red. It would have been even more faster if it had RGB.
Both good points.
Are you Trying to remind me that "I m poor" ?🤤
Hi, thank you for this great review! So if I need 16Gb in total, the best will be 2 sticks 8Gb each, but should each stick be 1R or 2R? Thank you!
Loving you videos, can you guys do an optimisation guide for monster hunter world iceborne on pc? Your past guide helps alot
Memory interleaving...
M e m o r y i n t e r l e a v i n g
Mmeemmoorryy iinntteerrlleeaavviinngg
This is a reason you can't compare benchmarks and reviews from different sources even if they seemingly use the same components such as a generic description of "3200Mhz CL16" ram.
Finally the first video that clearly explains the difference between dual channel and quad channel slots. Thanks for the info!
First time I've ever heard of "ranks" so I'll have to pay more attention to that in the future if I can find out how know. Thanks for the education, while being cheaper than university too! ;)
love gamers nexus as well but everyone's on the band wagon thinking there first to break this news when you guys made this ages ago you deserve more credit
Yeah it's weird that GN Steve presented this as if it's brand new information. Maybe it's just that Zen 3 benefits more from dual rank RAM than previous platforms idk
Of course the 4 dimm sticks were faster than the two, they were red. Everyone knows red makes things go faster.
That's so yesterday!
Recent patch is :
Red = +10%STR
Blue = +10%INT
Green = +10%AGI
So red are no longer faster, but green does!
Also makes you hungry
This is amazing. Thanks for explanation. Great job guys!
Superb! This complexity is why I depend on reviewers like you for tech information.
I just upgraded to 16gb dual channel...and you uploaded it😀
Dual vs single is notably faster
But does it affect the GPU performance of an APU?
This!
Yeah, that would have been interesting to see on the GPU side. I'd still imagine the overclock to 3600 with a pair of rev-E Ballistix 3000 would beat quad channel at lower clocks.
@@eugkra33 unfortunately, the current Apu generation is not going to hit 3600 that easily. The 3200-3400 range with tight timings will be just as interesting tho!
@@TheSilviu8x there have been people hitting 3600 with B-die, and I believe Micron Rev-E as well on Zen and Zen+.
APUs require lots memory throughput, which is exactly what dual rank per channel helps. I wouldn't be surprised to see a 15% improvement in FPS. I hope Hardware Unboxed does the benchmarking.
Another interesting video Steve.
How about using ram in the AMD system that's optimised for the Ryzen 3000 series chips, eg: G.Skill Trident Z Neo. I'm wondering if it'll make a difference compared to the ram you used for the testing in this video. Just a thought.
You got a sub, needed this video, I bought 4 dimms for my 9900k & read everywhere that 2 dimms was what I should have, something about the memory controler being stressed with 4
I heard it can mess with the xmp profiles and you can’t hit them. Did it work for your 9900k? I have a 10700k and want to do this to it. 16gb isn’t enough for me.
memory is probably the most enigmatic thing in a system
Hardware Unboxed: Actually more modules can improve performance in certain games
Me: RAM IS RAM
Lesser, greater, middling.
mate I've seriously been wandering this for like 5 years and I never got a satisfying answer until now
So glad to see these real results. Yesterday I bought 2 DIMMs to add to my corsair 2x8 3200 (dominator platinum). Seeing these slight boosts for gaming makes me feel a bit better for the high price of that memory module 🤣
This will be useful for competitive gaming where ppl play on low resolution and many times are CPU bound. I actually learned something today, thank you. Please test dual rank modules vs each other too.
Have to made a video testing 2 vs 4 dual rank sticks yet. You mentioned it in the last minute of this video.
Did you set timings and subtimings manually, or verify their concistancy during testing?
Thanks. It was so hard to find this balanced answer
Dude your background just makes me subscribe to you!!!
Very informative Steve! 👍🏻
Might also be related to the memory traces in the mobo, depending if it's t topology or not
Great video! Could you please test this again but with the iGPU? I wonder if the 3400G would gain any performance by using 4 DIMMs/dual rank vs 2 DIMMs/single rank.
Thanks!
Having more than 2 total ranks will benefit as long as your board supports it properly, in the past they compared dual rank vs single rank at the same timings and capacity, dual rank always came out ahead when mem intensive stuff were used igp is one such case, what is however interesting is if the board doesnt implement a proper setup it can all be for naught, aida64 mem tests were always faster on dual rank sticks vs single rank