Are you using an AMD Ryzen X3D CPU with 3D V-Cache?

[Mussels]

Today I finish chapter 3 of WoT exclusively on the 12500 + igp system

That's an IGP of one of the lowest power consumption CPU's in 12/13th gen

That's got nothing in common with the 300W+ 1*600/1*700/*900 CPUs

In reality, the consumption of a ryzen is much higher.

Uhhh.... i'm gunna say no to that one.

12400F (which is quite close to the 12500 you have) is actually the most power efficient single threaded intel CPU - and that's NOT normal for the rest of their lineup

No comparison is fair if you pick one product and make broad assumptions against an entire lineup - in this case you've picked the second most efficient intel 12th gen CPU and then compared it to vague generalisations of 'ryzen' - which series? which platform?

The downside to the 12400/500 is that while they're fantastic for efficiency like the 5700g, they aren't fantastic for performance

There is a niche for IGP systems that need to be power efficient and the 12400/12500 fit that niche really well, but so does the 5700g - so if both lineups have a product that's quite equal overall (with the 5700g IGP being superior to just about any other IGP out there) where does your 'intel is better' statement come from?

 

[BetrayerX]

5800X user here. Got tempted a few times about getting a 5800X3D, but I'd probably wait to upgrade when the 8000 series arrive.

 

[kane nas]

Yeah, ASUS has been garbage as of late, probably because a lot of people just blindly buy ASUS without remembering that other names like ASRock, Biostar and Gigabyte also exist. When I worked at Tiger Direct, I saw just how overpriced that ASUS motherboards were (even then) and thus, I've never owned an ASUS motherboard.

The last five motherboard brands that I've owned have been, in order, ECS/Elitegroup (AM2+), Gigabyte (AM3+), ASRock (X370), ASRock (X570) and Biostar (A320). They all work just fine and they were a fraction of what a comparable ASUS board would've cost. That ECS board is especially impressive because I needed a cheap board to replace my MSi K9A2 Platinum (which died unexpectedly and I learned just what a-holes work at MSi support) and that board still functions flawlessly today.

Don't be concerned with brand when buying a motherboard. Always remember that noobs buy by brand while experts buy by spec. An ASUS ROG Strix motherboard may look pretty but it won't give you better gaming performance than an ASRock Phantom Gaming, it'll just cost a crapload more.

I'll tell you the truth, I'm very happy with the Asus motherboards I've bought in the last 4 years, I started with a Gigabyte x370, I made the change to ASUS and since then I haven't changed, I'm also very happy with the way you can set the bios is very easy and I've read that ASUS has had issues with motherboards for the last 2 years, but I personally haven't encountered anything that would make me change my mind.

 

[Gica]

No comparison is fair if you pick one product and make broad assumptions against an entire lineup - in this case you've picked the second most efficient intel 12th gen CPU and then compared it to vague generalisations of 'ryzen' - which series? which platfo

With all. Why not generalize if you do?
Isn't "13th" a generalization?
i5-12500/13500 have some special features:
- very cheap (12500 bought for 200 euros 14 months ago)
- the most powerful igp, comparable to xx700/900. xx400 has a much weaker version: UHD 730 (24 EU), not UHD 770 (32 EU)
- powerful enough for powerful video cards (not top, last generation).
- 13500 gets 8 E cores and fights with 7700/7700X in heavy applications, not with 7600X, processor from its range and similar price.
- super efficient in light and medium applications and here I think that no ryzen beats 12500 or 13500 if it alone consumes in idle almost as much as an entire Intel system consumes (12500 in my case, see specs) for youtube 1080p and under 25W (I repeat: the whole system!) when I read the press or the forums.

Advantages and disadvantages, but I don't understand your ironies regarding an intel processor. Do you use the system for rendering 24/24 hours, 7/7 days?
Even there it is debatable because 1, 2 ... x hours of rendering have days and even weeks of work on the project.

 

[BetrayerX]

Yeah, ASUS has been garbage as of late, probably because a lot of people just blindly buy ASUS without remembering that other names like ASRock, Biostar and Gigabyte also exist. When I worked at Tiger Direct, I saw just how overpriced that ASUS motherboards were (even then) and thus, I've never owned an ASUS motherboard.

The last five motherboard brands that I've owned have been, in order, ECS/Elitegroup (AM2+), Gigabyte (AM3+), ASRock (X370), ASRock (X570) and Biostar (A320). They all work just fine and they were a fraction of what a comparable ASUS board would've cost. That ECS board is especially impressive because I needed a cheap board to replace my MSi K9A2 Platinum (which died unexpectedly and I learned just what a-holes work at MSi support) and that board still functions flawlessly today.

Don't be concerned with brand when buying a motherboard. Always remember that noobs buy by brand while experts buy by spec. An ASUS ROG Strix motherboard may look pretty but it won't give you better gaming performance than an ASRock Phantom Gaming, it'll just cost a crapload more.

I've had similar experiences with 3 Asus motherboards and what pissed me off against them was the very crappy tech support they provide. Could go into details but my fingers will bleed explaining the out of this world garbage they spew to get away from honoring warranties.

 

[Mussels]

I'll tell you the truth, I'm very happy with the Asus motherboards I've bought in the last 4 years, I started with a Gigabyte x370, I made the change to ASUS and since then I haven't changed, I'm also very happy with the way you can set the bios is very easy and I've read that ASUS has had issues with motherboards for the last 2 years, but I personally haven't encountered anything that would make me change my mind.

Their biggest problems started with B450/x570, and then got really bad in AM5.
That's pretty much what companies do - they earn a good reputation and ride on it, with inferior products riding on the earlier products reputation.

Every brand has their ups and downs, I too quite like my Asus BIOS - but seeing the corners they cut with cooling the motherboard chipset (failed within a year, and a terrible design from the start) on their entire X570 Strix series was a major letdown - and what they did to the TUF series was a disgrace, it was so bad they were the primary cause of x570 having 'loud chipset fans' purely because they locked the chipset fan to 100% speed at all times and removed user fan control in the BIOS rather than use a better heatsink.

 

[Godrilla]

We're wondering how many people have bought a X3D processor so far

I upgraded to the 7800X3D and finally completed my itx build with 4090 suprim liquid. I am using 350 watts of total system power usage in vermitide 2 max settings 4k 120hz vsyned and frame cap on. I am getting the same rendered latency as uncapped performance but with 2/3 power usage.

New item by Aberkae Godrilla

photos.app.goo.gl

New item by Aberkae Godrilla

photos.app.goo.gl

New item by Aberkae Godrilla

photos.app.goo.gl

New item by Aberkae Godrilla

photos.app.goo.gl

New item by Aberkae Godrilla

photos.app.goo.gl

 

[Mussels]

I am getting the same rendered latency as uncapped performance but with 2/3 power usage.

Because the GPU isnt at 100%
Any time it is, the CPU will render frames ahead of the GPU and the latency will double or triple.

Yours goes from 4ms-6ms which is still great regardless - that's more like what happens in DX11 titles with Nvidias ultra low latency mode enabled

 

[Godrilla]

Because the GPU isnt at 100%
Any time it is, the CPU will render frames ahead of the GPU and the latency will double or triple.

Yours goes from 4ms-6ms which is still great regardless - that's more like what happens in DX11 titles with Nvidias ultra low latency mode enabled

Imagine wasting power to be a competitive gamer and still being a noob . All you need is 4k oled at 120hz baby! Anyone welcomed to join me at the Cataclysm Twich challenge and Deeds.

 

[Mussels]

Imagine wasting power to be a competitive gamer and still being a noob . All you need is 4k oled at 120hz baby! Anyone welcomed to join me at the Cataclysm Twich challenge and Deeds.

That wont help worth a damn if you have 50ms render latency before it even reaches the OLED - and some OLED's are slower than IPS panels.

People always seem to mix up render latency with other metrics like frametimes and display latency - they all add up together for the final result, CP2077 with a maxed out GPU is going to sit at 50-100ms for everyone due to it's shoddy design, and no display tech can shrink that.

ASUS 27" OLED
ASUS ROG Swift OLED PG27AQDM Review - RTINGS.com

These are the 60Hz values, which matter more than you might think if frame rates dip or games have FPS limits with VRR. Without VRR, just ouch - you'd get stutter/tearing pretty bad on top of this.
Rise and fall times? sure, quite solid. (Far from the 0.1ms crap marketing pulls, but still great values)

The total response however... not good. Black to anything is really slow

And the overshoot is quite bad at lower refresh rates, which includes when VRR is active and frame rates drop. If you're between 30 and 60 or 60 and 120, you're in this category - while below 60 can double into the higher rates to remove the smearing. (55FPS->110Hz, etc)

At 144Hz its good, but still falls into the 4ms category, especially with image retention (smearing) if images are dark but not black still at 5.8ms.

OLED's cant fix their issues with smearing without BFI, and you can't use BFI with variable refresh rates - which means they're not as good as people think they are.
They excel in some areas, but can't do those excellent things at the same time.



For comparison to a fast IPS panel
ASUS ROG Swift 360Hz PG27AQN Review - RTINGS.com
360Hz - note theres zero overshoot, in the 3-7.8ms range vs the 0.8 to 5.8 of the OLED. That's only a 1ms difference in the worst case results and a lot smaller than marketing have people believing.

At 60Hz this display doesnt do amazing either, but at least it doesnt have the crazy overshoot - 5.8ms of overshoot vs zero, means this will absolutely look better than an OLED in games with lots of dark items on screen.

 

[Legacy-ZA]

Wanted to, but alas, AMD took their sweet time. Maybe when they have them ready at launch with their next generation of CPUs.

 

[A Computer Guy]

I'm still considering one. I have an unused b450 tomahawk and 32GB of unused corsair LPX that will oc to 3200 and some leftover wc parts enough to make another custom loop which I said to myself 1 is enough but I've gotten an itch. Ebay has them new for $300 and I'm eyeballing a used one for $245 but it might not be worth the risk to save $50.

 

[Gica]

While I wouldn’t say TPU readership is indicative of the whole DIY market, it’s definitely representative of the ‘educated’ DIY market. Amongst those who research building a PC, Nvidia is the majority preferred brand for GPUs and AMD is the majority preferred brand for CPUs.

It's not about education, it's about something else.
I notice that AMD owners have "forgotten" the multitasking/heavy CPU usage factors since it favors Intel. The stability of the Intel systems and the DDR4 compatibility of the Intel 12/13/14 series do not seem to be an advantage in their eyes either.
As for the prices, Zen 3 was released between 2020 and... 2022. They oversold a limited series of processors for two years, releasing the rest of the series practically at the end of their life. I'm not talking about the mess with Vega, Zen 2/3 outside - Zen 1/2 inside.

Regarding X3D, the reviewers refuse to "enlighten" us on their impact depending on the video power. They used only very powerful video cards for the reviews, but we don't see anything like (I give an example) 5800X versus 5800X3D with RTX 3050/3060/3070 or RX 6500XT/6600XT/6700XT.
They are probably prohibited from proving that the impact is absolutely zero between two processors that have a huge price difference.

AMD Ryzen 7 7800X3D Review - The Best Gaming CPU

AMD's Ryzen 7 7800X3D is everything that gamers want: eight fast cores, in a single CCD, with 3D VCache stacked on top at a price point of $450. In our review we can confirm that 7800X3D is indeed the fastest gaming CPU available, beating Intel's 13900K and even the 7950X3D.

www.techpowerup.com

 

[wolf]

Currently 5800X3D which while great, the first 2 AM4 chips were a bit of a miss for me. 3700X held me back, 5900X WHEA errors and crashes bone stock. Roughly this time next year when I'm in the market to refresh my system, among other out of the ordinary options (like possibly a gaming handheld and occulink GPU), I'll very much be evaluating what both Intel and AMD have to offer at the time, with a sole focus on gaming performance. AMD have really pissed me off lately with their game sponsorship shenanigans, which I'm feeling the negative effect of despite owning AMD hardware, but to live by my own code, I'll be evaluating the performance, cost, features etc and buying hardware against only those criteria instead of factoring political reasons. Games themselves are a different story.

 

[AUTOEXEC.BAT]

None of them.

My daily Desktop is an Intel Xeon x5687
Running ZorinOS 16.1

I dont need more power.

 

[Athlonite]

Hmm tatty old Ryzen 7 5800X for me is fine for the 1080p gameplay I use it in

 

[Mussels]

I notice that AMD owners have "forgotten" the multitasking/heavy CPU usage factors since it favors Intel.

Since when?? That's been AMD's wheelhouse since zen1.
I'd love to know where the hell you're getting THAT idea from so I can make sure no one else ever views the same content you did.

The only thing intel has going for them is a small lead in single threaded performance for a massive power cost, but since almost nothing is single threaded these days and single threaded performance on a '20' core CPU is a huge waste of money and CPU, that's not a big help at all.

If you want the best gaming experience, you get an AMD 3D CPU.

If you want the best multi threading, you get an AMD 2D CPU, or one of the new hybrid ones.

The efficiency values on intel are worse too, so they're slower AND use more power most of the time.

That tiny single threaded performance advantage comes at a hell of a power cost, too - AMD isn't immune to that either, those extra cores aren't free even at idle.
The difference is that on AMD you can buy CPU's with less cores, on intel you can't - you want those 8 P cores, you can't get them without the E cores, while AMD lets you buy the 7800x3D instead of forcing everyone to buy a 7950x3D.

The stability of the Intel systems and the DDR4 compatibility of the Intel 12/13/14 series do not seem to be an advantage in their eyes either.

Why would it be? AMD systems are only unstable if people are too lazy to even check what RAM is supported and throw in random shit, on intel the exact same issues occur.
To talk to you in your own language
'Intel users don't care that AMD lets you run whatever you want, they don't even care that you're locked to low RAM speeds that cripple performance unless it's a Z board and a K CPU - and if you run a Z and K setup and it's unstable it's totally user error and not intels fault and we'll just pretend the gear issues with RAM don't exist where intel defaults to the slower setting that cripples performance'

Currently 5800X3D which while great, the first 2 AM4 chips were a bit of a miss for me. 3700X held me back, 5900X WHEA errors and crashes bone stock. Roughly this time next year when I'm in the market to refresh my system, among other out of the ordinary options (like possibly a gaming handheld and occulink GPU), I'll very much be evaluating what both Intel and AMD have to offer at the time, with a sole focus on gaming performance. AMD have really pissed me off lately with their game sponsorship shenanigans, which I'm feeling the negative effect of despite owning AMD hardware, but to live by my own code, I'll be evaluating the performance, cost, features etc and buying hardware against only those criteria instead of factoring political reasons. Games themselves are a different story.

Didn't we help you with this in the Zen garden? You're running ram that's not officially supported on any AM4 CPU, so it's not 'stock'

You're also running a PCI-E riser which I wasn't aware of, I had to look up your case... that's extremely likely to be your entire problem. 3 of my 4 are dead now, they're extremely fragile.
Did you know the riser in that case isn't rated for PCI-E 4.0?

 

[Gica]

Since when?? That's been AMD's wheelhouse since zen1.
I'd love to know where the hell you're getting THAT idea from so I can make sure no one else ever views the same content you did.

The only thing intel has going for them is a small lead in single threaded performance for a massive power cost, but since almost nothing is single threaded these days and single threaded performance on a '20' core CPU is a huge waste of money and CPU, that's not a big help at all.

If you want the best gaming experience, you get an AMD 3D CPU.
View attachment 313248


If you want the best multi threading, you get an AMD 2D CPU, or one of the new hybrid ones.
View attachment 313249


The efficiency values on intel are worse too, so they're slower AND use more power most of the time.
View attachment 313250

That tiny single threaded performance advantage comes at a hell of a power cost, too - AMD isn't immune to that either, those extra cores aren't free even at idle.
The difference is that on AMD you can buy CPU's with less cores, on intel you can't - you want those 8 P cores, you can't get them without the E cores, while AMD lets you buy the 7800x3D instead of forcing everyone to buy a 7950x3D.
View attachment 313251


Why would it be? AMD systems are only unstable if people are too lazy to even check what RAM is supported and throw in random shit, on intel the exact same issues occur.
To talk to you in your own language
'Intel users don't care that AMD lets you run whatever you want, they don't even care that you're locked to low RAM speeds that cripple performance unless it's a Z board and a K CPU - and if you run a Z and K setup and it's unstable it's totally user error and not intels fault and we'll just pretend the gear issues with RAM don't exist where intel defaults to the slower setting that cripples performance'


Didn't we help you with this in the Zen garden? You're running ram that's not officially supported on any AM4 CPU, so it's not 'stock'

You're also running a PCI-E riser which I wasn't aware of, I had to look up your case... that's extremely likely to be your entire problem. 3 of my 4 are dead now, they're extremely fragile.
Did you know the riser in that case isn't rated for PCI-E 4.0?

I notice that you write a lot and definitely waste a lot of time. I can do it with 25W, the whole system, in which this intel consumes enormously, under 1W in idle and with rare and short jumps in 10 W. At this moment, the wattmeter indicates 2.1 KW consumed this month. Not. there is no need to help me pay the bill, I think I can handle it.
All the best!

However, it's really worth $190, the difference between 13500 and 7800X3D for a 3070Ti. I bet not!
From what I have observed, with an Intel you reach the highest and most stable frequencies of the RAM memory.
And an Intel system behaves like a car: fuel, the key in the ignition and... go go go. With AMD, you risk being stuck on the road if you don't have knowledge of car mechanics.

 

[A Computer Guy]

Since when?? That's been AMD's wheelhouse since zen1.

I dunno, in particular with 13th gen it's very competitive. I've been looking at a lot of comparisons too and 13th gen doesn't seem as bad as some people make it out to be. Personally I'd like to see more clarification with Intel w/ tuned power limits and heat management compared to stock Ryzen.

As a side note the whole ecore path for Intel doesn't seem as great for VMWare and ESXI users whereas AMD users have enjoyed mostly a uniform core approach with the 3950x, 5950x, 7950x, and likely 8950x without the headaches of big.little. (edit: and threadripper too)

I'd love to know where the hell you're getting THAT idea from so I can make sure no one else ever views the same content you did.

I don't think limiting information helps but comparing and contrasting conflicting information is valuable to understanding facts, the accuracy of testing, and if certain conclusions are valid.

'Intel users don't care that AMD lets you run whatever you want, they don't even care that you're locked to low RAM speeds that cripple performance unless it's a Z board and a K CPU - and if you run a Z and K setup and it's unstable it's totally user error and not intels fault and we'll just pretend the gear issues with RAM don't exist where intel defaults to the slower setting that cripples performance'

Id also like to see more comparisons involving non-Z boards and non-K Intel CPU's including price/performance.

 

[wolf]

Didn't we help you with this in the Zen garden? You're running ram that's not officially supported on any AM4 CPU, so it's not 'stock'

I tried multiple RAM kits, bios updates, even a different PSU trying to figure it out, I even tried that 5900X in a mates rig. Plus the other two AM4 chips (3700X and 3800X3D) have exhibited zero issues whatsoever. I had little trouble getting it RMA'd. Fairly confident that was a dud, like I've heard about a fair few other Zen 3 chips. Given the exhaustive months of testing and troubleshooting, and all the testing notes and evidence gathered, I'm 100% blaming that CPU, not my RAM or any other component/configuration.

You're also running a PCI-E riser which I wasn't aware of, I had to look up your case... that's extremely likely to be your entire problem. 3 of my 4 are dead now, they're extremely fragile.
Did you know the riser in that case isn't rated for PCI-E 4.0?

I'm not using a riser, and haven't in a very long time, since before buying the faulty 5900X.

 

[Gica]

I dunno, in particular with 13th gen it's very competitive. I've been looking at a lot of comparisons too and 13th gen doesn't seem as bad as some people make it out to be. Personally I'd like to see more clarification with Intel w/ tuned power limits and heat management compared to stock Ryzen.

Consumption 12500 and 13500 in pcmark extended. No dedicated video card, just igp.
Only marked scores keep 100% of the processor. The others are influenced by the video card.

---------------------------------------------------------------------------------

In multicore applications, the i5-13500 does not fight with the 7600/7600X, it fights with the 7700X/7800X3D. In some it wins easily (CPU-Z, Cinebench R20), in others it loses by a whisker (Vray 5).
If you compare the prices, you can clearly see that 13500 destroys them all in terms of price/performance.
There are no weak or bad processors today. All have advantages and disadvantages, it matters how correctly you choose.

'Intel users don't care that AMD lets you run whatever you want, they don't even care that you're locked to low RAM speeds that cripple performance unless it's a Z board and a K CPU - and if you run a Z and K setup and it's unstable it's totally user error and not intels fault and we'll just pretend the gear issues with RAM don't exist where intel defaults to the slower setting that cripples performance'

Chipset B motherboards allow RAM frequencies above Intel specifications. There is no difference between B and Z in this chapter.
But I have a surprise for you. See below.

 

[grammar_phreak]

I don't use one personally but I bought a 5800x3d one for a rig I was going to resell. However, my brother was having problems with his Zen 2 rig and I decided to upgrade his CPU to the 5800x3d. (it turned out his Crosshair VII Hero board was having issues with the USB ports) He's now rolling with the 5800x3d, ASRock Taichi x570, and a 6900xt.... and he has no complaints about the rig.

 

[Mussels]

Chipset B motherboards allow RAM frequencies above Intel specifications. There is no difference between B and Z in this chapter.
But I have a surprise for you. See below.

Theory vs practise. See below.

You still need a K series CPU to use that memory overclocking on a B series board - the boards unlocked the speed, but without the SA voltage (that normally raises with XMP) you can't use those speeds.
The only exception is if the board changes to gear 2, so you're getting worse performance and not better.

Id also like to see more comparisons involving non-Z boards and non-K Intel CPU's including price/performance.

That's really hard to do, as you don't get given those for review samples - so you'd lose out on a lot of work to do so. You have to buy a board and CPU and then not get paid for it, vs being given parts to test and being paid. It's extremely expensive without donated parts.

You can only test synthetically by locking in TDP limits under the assumption that cheaper boards will use those wattages, and run RAM under whatever limits are common - non K CPU's cant alter SA voltage (even via XMP)

A 13400/F is capped at 3200/4800
Intel Core i513400 Processor 20M Cache up to 4.60 GHz Product Specifications

TPU has the 13400F here, but they test with DDR5 6000 on a Z board
TPU needed to run this in gear 2.
Intel Core i5-13400F Review - Force of Efficiency - Test Setup | TechPowerUp

Comparing Gear 1 vs 2, same RAM - different CPU.

Same RAM, same board. The board name only changed because AIDA was updated to detect it correctly.
The memory latency is the same, verifying the actual RAM hasn't changed - but the read values for example are absolutely tanked by gear 2 - about a 7% loss. That 7% is a huge loss.

@Taraquin here on TPU goes into it in detail
If ram oc is important, avoid locked alder lakes | TechPowerUp Forums


TLDR:
1. Artificially capping wattages doesn't cover how many trash motherboards can't even handle 125W
2. Locked SA voltages mean no synthetic testing will ever cover every combination - One person may run 4x32GB fine where another fails with 4x8GB
3. Boards change between Gear 1/Gear 2, so some may appear to run faster but downclock and lose a lot of performance.


So if you wanted that tested - what combinations would you use? How would you make it cost effective if you have to buy all the parts, and time effective that you aren't testing every single CPU on every board with a dozen different ram combinations, at even more speeds?

 

[Gica]

Theory vs practise. See below.

It can be seen that your experience with Intel is zero.
B motherboards block CPU overclocking. Don't buy K processor and B motherboard if you want overclocking.
All non-K processors have the SA voltage locked. Z motherboard can't unlock it.

Regarding your example:
North Bridge Clock (AIDA print) is not Memory Controller Frequency! In your example, Memory Controller Frequency can be 3000 MHz (Gear 1) or 1500 MHz (Gear 2). Higher than 3000 MHz cannot be for DDR5 6000MHz.
A second error of yours is that you say they tested with DDR 6000 MHz in Gear 1. It is impossible because the limit of the memory controller is somewhere around 2000 MHz even if you raise its voltage and SA above the dangerous limits.
The tests are performed with Gear 2 and the differences have two explanations:
1. Different benchmark versions.
2. Processor frequency has a strong influence on RAM performance in synthetic tests. Not only is there a 1000 MHz difference between them, but one is the weakest i5 and the other is the strongest i9 (more cores, more cache, etc,).

 

[Mussels]

It can be seen that your experience with Intel is zero.
B motherboards block CPU overclocking. Don't buy K processor and B motherboard if you want overclocking.
All non-K processors have the SA voltage locked. Z motherboard can't unlock it.

I posted links to things you seem to have not understood.

K+Z = CPU and RAM overclocking
K+B= CPU and RAM overclocking

Non-K and Z/B = Ram overclocking with locked SA voltage on 12/13/14th gen. That locked voltage doesn't force you to Gear 1 like you misunderstood, it means you can't raise voltages to stay there with higher RAM speeds.

You got 3600 working, where unlocked voltages can get you 4400 on the same CPUs.
You could be doing that with two single rank sticks as well, where dual rank sticks will refuse to work at that speed without gear 2.


Feels like you read the information and skipped everything you're not familiar with.