NEVER DO THIS TO YOUR DDR5 TIMINGS on intel (AMD works differently)

[@Laundry_Hamper]

First shalt thou jam on the Holy F10. Then shalt thou set tRDRD_dg to eight, no more, no less. Eight shall be the number thou shalt set, and the number of the setting shall be eight. Nine shalt thou not set; neither set thou seven, excepting that Intel then round up to eight. Sixteen is right out.

[@volodumurkalunyak4651]

DDR5: tRDRD_dg and tWRWR_dg - both at 8 DDR4 and lower: tRDRD_dg and tWRWR_dg - both at 4

[@bofast]

Did you by any chance post this from a KZbin app coded in Python?

[@blanchbacker]

Lmfao nice

[@clintonleonard5187]

Amen

[@abzaman77]

We are lucky that this video is free. It would be great to see a dedicated video with DD5 overclocking, like a step-by-step guide. Something like which timing should we attack first, and so on with a recommended set of kits by you. That would be a gem!

[@4fiHysteria]

I would hate to be the guy that shared this screenshot lmao

[@PowellCat745]

lmaoooo

[@hannupc8422]

But what an important lesson the person who sent that offered us 👍

[@clintonleonard5187]

Well they learned.

[@gsuberland]

At some point I need to go through the DDR5 datasheets and make a poster diagram of how all the timings fit together, so people don't make these mistakes.

[@puppieslovies]

I'd like that

[@xBINARYGODx]

hookers, blow, oral, and all that to you, kind sirs or madams

[@andrewvirtue5048]

Have you done it yet?

[@cppctek]

that would be friggin epic!!! I really need help tuning my a-die to get better latency and see what the most optimal timings would be..... 48gb 7200cl34 t-create kit, having issues running xmp on the 14900k i have but it ran perfectly on my 13600k with the same maximus hero z690 board. the 14900k is an sp97 and my 13600k was a friggin sp80.

[@meppie1922]

That would be awesome. The timings are just such a mess and very few people know how these works properly.

[@wewillrockyou1986]

Something else to keep in mind with this is that most programs tend to send lots of memory accesses bunched up together in a short amount of time, after which they will mostly run from cache for some time. So even if your program doesn't actually saturate the full bandwidth of the memory bus when averaged out, these back to back timings are still very important to quickly get through those batches of memory accesses.

[@sarinmcgee2031]

If U have a guide I would love to follow it

[@Markknightexeter]

Im surprised at the hate in comments, he knows a hell of a lot than any of you

[@bigcazza5260]

hate = idiots who had it set on 16 or 8 let those dumb shits continue to be wrong, because if someone goes out of there way to be below u then u dont have to rise above them

[@Finite-Tuning]

This is tough stuff, your explanation may have made it even tougher. But simply matching up BIOS settings, mine were set to 8 already so I'm done. Cheers 🍻

[@benjaminfrohns]

My DDR4 timings were auto set to 4. Nice. Even though I tweaked the primary timings and frequency, the board did it right. 😊 Thank you for the tip.

[@lucasrem]

DDR 4 on this 14th gen is bottleneck ! Do we the XMP 6000, where is the bottleneck ? Wait a gen now ? Please understand the basics, where is the bottleneck !

[@benjaminfrohns]

@@lucasrem I should have said that I'm on AM4. Sorry

[@christopherjackson2157]

Ur ddr timings videos got a shout out on Jeff geerlings channel. Tho I think he said "actually hardware overclocking" lol

[@Raythe]

Is there like....a dictionary on Ram timings? explaining what they are and when to mess with them? I just want a single PDF that I can Ctrl F on to find the definitions for this stuff

[@thetwitchywarlock]

I would also like this

[@simoSLJ89]

I have to say I didn't play with my DDR5 at all. Two systems, one with 12gen and one with 13gen, and both just 6400 MHz xmp. I know setting xmp is a shame, I'm a normie :D and I just wanted it to be stable. Love to follow BZ anyway, never change man!

[@dethskullcrusher]

Man you should at least test the XMP using Memtest or TestMem5 to know for a fact that they're stable, because it's easy overclock but it ia overclock in the end.

[@jtnachos16]

@@dethskullcrusher Except that isn't really what XMP is in the first place. XMP values are, memory side, tested by the manufacturer of the RAM and actively advertised to be the intended speeds and timings. The only ones who claim it is 'overclocking' are cpu/mobo manufacturers looking for excuses to avoid RMAing faulty products. If you go look up the specs sheet for a given CPU it will actively state what it's max supported RAM speeds are, and even clarify further in most cases as to specific slot configuration limits. Those listed speeds are also in excess of baseline speeds. Same goes for motherboards, where they typically have a QVL list somewhere that will actively list supported ramkits. Many of which are at more than the baseline speed for that breed of ram. So, no, enabling an XMP is NOT 'overclocking'. It is using different settings than default, but that does not equate to overclocking, as that implies parts operating BEYOND THEIR DESIGNED AND INTENDED LIMITS. A cpu that claims to support 5200mhz DDR5 in a 2 slot config, on a mobo with the appropriate socket, isn't OCing by running a 2x 5200mhz ramkit @ 5200mhz, instead of the 'default' of 4800mhz. Now, if you were to run said ramkit at 6000 mhz, THEN you would be OCing.

[@simoSLJ89]

@@dethskullcrusher of course, after assembling, ram and every other component is always tested. It's even useful to set a baseline, to check once a year if everything is still fine. And it's fun too as a weekend project.

[@stedz2000]

@@jtnachos16 is not anything faster than JEDEC overclocking?

[@jtnachos16]

@@stedz2000 No. Because Jedec standards are not the standards of the manufacturers of the parts. Jedec is a 'bare minimum' standard. I'll quote from Jedec's own site, on the article where you can purchase the document regarding the specs for DDR4 (which is nearly 300 dollars. Because god forbid information be free): "This document defines the DDR4 SDRAM specification, including features, functionalities, AC and DC characteristics, packages, and ball/signal assignments. The purpose of this Standard is to define the minimum set of requirements for JEDEC compliant 2 Gb through 16 Gb for x4, x8, and x16 DDR4 SDRAM devices. This standard was created based on the DDR3 standard (JESD79-3) and some aspects of the DDR and DDR2 standards (JESD79, JESD79-2). Committee Item 1716.78H" In particular, pay attention to the line that says 'define the minimum set of requirements'. Jedec sets the baseline speeds, structure, and voltage ranges for something to be considered a particular breed of RAM. If a manufacturer chooses to make something faster than that baseline, it is not overclocking. Again, pretty much every motherboard (via QVL list) and CPU has explicit support for RAM speeds in excess of Jedec's standard. Which means using RAM in excess of Jedec speeds is NOT inherently overclocking. It becomes overclocking when you push it outside the supported speeds of the CPU and Mobo.

[@Numfuddle]

So if the ram is actually designed for 8 cycles between reads, why even make this a changeable parameter in the first place?

[@volodumurkalunyak4651]

Those are for memory frequency world record. It also includes setting a bunch of outher timings like primaries and secondaries as high as those go.

[@lucasrem]

@@volodumurkalunyak4651 you will never get any benefit there, running what. is XMP 6000 already too fast ? bottleneck is the 14900k it self, XMP 5200 is fast enough ?

[@volodumurkalunyak4651]

@@lucasrem memory frequency world record is not about maximum performane, but macimum achieveable clock speed. Is XMP-6000 fast enought - that is whole different question. In fact answer greatly depends on bencmarl used. Cinebench - 1 stick of DDR5-4800 JEDEC memory is fast enought, y-cruncher - dual channel of manually overclocked memory at DDR5-7200 is probably not enought.

[@willjohnsonjohnson]

Nice. A useful benchmark that works on old CPUs without AVX. I hate how a lot of bench software requires AVX, even though the tests themselves don't use it.

[@jannegrey]

wow, buildzoid video under 20 minutes ;) Thanks for correcting this... not very smart mistake by someone.

[@Multimeter1]

Can we get another updated video for us casuals? Maybe not record breaking clock speeds on 14th gen, but really nice timings and clock speeds that you can be like "this is not the best, but damn this is good"

[@PolskiJaszczomb]

BZ, a protip - you can launch a new instance of a windows with your mouse middle button.

[@blackcrov840]

You know zoid post this video to win forum or discord arguments 😂

[@minn5459]

Commenting for the algorithm to spread awareness

[@andersjjensen]

7:18 "Do you see my point?".... Sometime the most frustrating part of youtube is that BZ can't hear that I actually said "Uh, yeah, you just explained it!" out loud :P

[@bigcazza5260]

lmao that made your ddr5 slower than my ddr4 lol

[@valentin3186]

Rating your DDR5 timings when? Q4 DDR5 buying guide when? :)

[@Heitorr]

70 dollar adie, that's it

[@PAIN-ot4cj]

Can you do a video on ddr5 sub timings and timings for intel 13th gen and what timings are worth moving

[@efovex]

He has done several of those.

[@martinfodor2848]

Not bad in terms of bandwidth specifically, however for better stability if you follow Jedec’s rules, WRWR should be double RDRD. It’s it written in the specs.. and it does affect the write performance. You know, there are some many ways to break the rules that make them work with a good 1DCP mobo like an apex and a good IMC. I would say it’s Pay2Win config.. if both SG are set at 8, you should try to lose / work with RRDS/L and WTRL/s..

[@ivicapavic3381]

that's a nice little subtle 14900k flex video 😂

[@pcoverthink]

Yay, you started using microbench too

[@SimpleP1D]

I'm watting for the intel video

[@Mom19]

Well, waiting on data is never really a good option unless absolutely necessary. The less wait, the better

[@emini6]

Thanks mom

[@cam_bro]

I'd like you test out a 14900k and determine if the MC is any better than the previous gen. I'm assuming no, but would still make a good video.

[@1valzo]

its not better at all

[@mathieudeshayes4316]

it's not better it's not worse. It's the same and it all comes down to how lucky you are at the lottery. My experience has been good. Out of 3 14900KF, one can run a few benchmarks at 8800 c34 48 and is stable at 8400 c36 48 48, while the other 2 are stable at 8200. All 3 of them can run benchs at 8600 c32 with enough stability to never crash (excluding Y-cruncher). Their MC SP are 69, 74 & 78. The only difference between all 3 is how much IVR TX voltage (CPU VDDQ) and MC voltage (VDD2) they need. My 13900KS (MC SP 82) and 13600k (MC SP 74) could do the same; however true stability was not possible above 8000. This is all with A-die 2x16 (G.Skill 7600 c36 46 46 121) on APEX Z790. I'm just giving the MC SP to underline the fact a higher value doesn't equal better IMC. My samples had better IMC but I know people who had worse memory controllers compared to their 13900KS.

[@puppieslovies]

the entire 14000 series is the exact same as the 13000 series with small optimizations

[@larsjrgensen5975]

He has shown that his 13600k has a better MC then his 13900K, so the chip to chip variance might be too large to make the right conclusion. A good 13th gen MC could be better then a bad 14th gen or opposite, he would need to test atleast 5 CPUes from each gen to come with a valid conclusion.

[@Tpecep]

Its way worse than last packs of 13th gen

[@gscurd75]

So sounds like someone tightened up their timings too much then started messing with stuff, saw it was stable at 8000 CL36 and thought they did a great job and had to share with everyone.

[@arkama67]

great stuff

[@bigcazza5260]

yes ddr4 does odd tertiaries at least on skylake imcs for rocketlake and up dunno

[@bigcazza5260]

it also HATES even tCWL if u are below tCL, ie -2 wont work but -3 will lol

[@blar2112]

Ive set mine to 2 so i get 75% more data per data

[@bigcazza5260]

lol

[@jihadjoe]

clobber much?

[@toonnut1]

this is a great viseo! do you have a link for microbench?

[@sharky9056]

Noob here, does this apply to ryzen cpu too? Ddr5 6000 cl 28

[@lucasrem]

XMP 6000 is more than enough for 14900k not bottlenecking it ! Play games ! or large data set needs ? Cloud compute PC ?

[@erickelly4107]

Yeah whomever posted that screenshot is probably thinking their life is a complete failure…

[@bigcazza5260]

nice both my tRDRD dg and tWRWR dgs are at 4, nice what about tRDWR? mine wont stabilise lower than 14 and thats for all of them (sg, dg and dr)

[@bigcazza5260]

u answered in vid cheers lad, low as she goes

[@volodumurkalunyak4651]

You have good tertiary (those that you showed here) timings for your DDR4 (tRDRD_dg at 4 is for DDR4, DDR3 and previous dont have such an timing, for DDR5 it won't work at

[@NMir-kd1bk]

@ActuallyHardcoreOverclocking can you share a link for the Clam cache\Mem benchmark tool been searching the internet for it and found it nowhere

[@skilletpan5674]

Hey buildzoid can you do an update on the newer amd agesa bioses? My tuf gaming x670e plus has been stable (if i set voltage for ram) since august. What will the new bios really do? After a year of instability i finally have it stable and i don't see a reason to update yet.

[@Theeletricminers]

Brother, do u hava any services to help with overclocking or just to achieve xmp speeds? I have a 13900k and ddr5 7200 gkill kit and strugling to stabilize xmp on a gigabyte z790 elite ax ... yeah, i know that mobo its a crap lol

[@qT_p13]

I love Paint

[@Syphirioth]

Hey mr overclock. I have a question and hope you can help me out. I have 4 times corsair DDR5 5600 running at 6200Mhz. Timings at 36-36-36-76 112. FCLK 2167 on AMD ryzen 7900X. But it also works on 5600mhz with timings 30-30-30-58 88. What would be the best option in your opinion to your knowledge?

[@vince4202]

I’m not buildzoid and I will not give you advice but I will say that you can download aida64 extreme and run a memory benchmark with both memory profiles and see for yourself what’s “better”.

[@roflcopter3625]

good morning sir, what timings do i have to set to what on am5 to make sure i have full bandwidth? because i don't see sg and dg.

[@user-vx4nz8lg2i]

What about dual rank ram? 32x2 gb ?

[@isaac3045x]

Can you do an easy timings video for intel 14th gen? Im trying to improve my 1% lows in games.

[@willsutton54]

ok, we have ddr4 and ddr5 and the difference is 1 and not 16 or 32 ..... my knowledge of ram is basically enabling expo but I am curious as to say in games how many fps would you get from "pushing" your ram

[@darkSorceror]

Hey Bz, could you link Microbench here please? 🥺 Having trouble finding it on the Googles

[@cppctek]

where can i download MemTweakIt ? i tried to find it on the asus page but its removed. i found a super old version from 2012 but it wont run on w11... Running a z690 maximus hero... is this a skill issue or a known issue? any one help?

[@christopherjackson2157]

I feel like the issue is probably way more the memory not being organized the way the cpu wants to read data out of it than the 14900 not having enough compute power to take advantage of the maximum bandwidth. At least in most workloads. Tho i havent tested this with a 14900, its usually a pretty good rule of thumb. If the cpu cant use the bandwidth its most likely something holding up the pipeline and the caches being full while the cpu waits for some specific data from memory

[@killerrf]

You getting an asrock z790 nova and test their claims up to 8000 ddr5 on 4 dimm board?

[@ejosephsimon]

So I'm slowly starting to realize for that average gamer, manually overclocking DDR5 is not worth it. I bought your recommended Team-group 6800MT/s kit paired with the ASUS ROG STRIX Z790-A GAMING WIFI & 13700k. Enabled XMP-I then set to 7200MT/s and get missive improvements in Read, Write & Copy as well as 54ns on latency. MemTest passes everything runs great.... So am i crazy in my assumption or am I missing something?

[@blanchbacker]

Not missing anything, you are smart. The only time people should be tweaking this stuff endlessly is if the kit they bought is shit, or their good kit doesn’t work properly at XMP. Take your results, and game!

[@cheyannei5983]

Sounds like you got a good CPU and kit with good chips. You won the silicon lottery, congrats! Take your winnings and cash out some free time!

[@Sp3LLBouND]

hi quick question, i plan to upgrade from 16gb ddr5 cl 40 4800 to 32gb ddr5 cl 36 or 40 4800 it's almost same price, i use asus f17 2022 laptop model i will get benefits from the lower timing also i checked in bios and there is no xmp or something else like most laptops have locked bios. Thanks

[@TheCrazyCanuck420]

If you question is will you benefit from the tighter timings? If so, yes you will, but I doubt you'll notice any of a difference since that's not much of a timing change and I would only expect synthetic benchmarks to show slight improvements. So if you upgrade go into it for the doubling of the density, not the speedup.

[@Multimeter1]

Was this MDIE or ADIE? Must have missed it if he said so forgive me

[@ExtremeGamer9951]

When I try to use mem tweakit it wont even start due to a driver initialisation error. How did you get past that?

[@noface_777]

Anybody have a link to a site or table that shows some of the common memory mhz, CL, timings, etc? For a beginner, it’s hard to keep up and understand all this language. Much appreciated!

[@seanpereira]

What are these timings called on AM5 ? IS it the SD (Same dimm) and DD (different dimm) timings ?

[@volodumurkalunyak4651]

SD and DD are not those. SD - works as *_dr on Intel and DD - as *_dd on Intel. There are following types of tertiaries: *_sg, *_dg (explanined in this video), *_dr, *_dd. I used Intel timing names here.

[@jeromemayhew6085]

Sc

[@magno5157]

Can RDRD_sg and WRWR_sg be 8? Or is 12 the lowest for both? What happens if they're set to 8 anyway?

[@lucasrem]

magno5157 where is the bottleneck ! need faster than XMP 6000 on this gen ? why unstable, running what ? make sense please !

[@magno5157]

@@lucasrem When have I said that I had bottleneck? When have I said that my system was unstable? You're weird dude.

[@MerseYattle]

MIND THE GAP

[@Markknightexeter]

Is there no gear down mode with ddr5?

[@marsovac]

What's the value for DDR4?

[@Sigouss]

Why can't I overclock 4x 16GB DDR5 to 6000MHz as rated?? Only stable frequency seems to be default set by motherboard which is 4533MHz,...

[@ChinchillaBONK]

Drawing the waveform out would be easier to explain what you just said.

[@N0rth0M-1]

can you not run both of these at like 4 on ddr4?

[@jamieclarke2694]

Where is memtweakit? Cant find it on ASUS or anywhere 😅

[@zlibz4582]

what would be the correct timing for ddr4 bdie ?

[@the_real_player]

dude sounds like kermit the frog

[@princetexasloaf2632]

is there a video of this for amd? ryzen 7000

[@doc_bucket1982]

It must be different with Ryzen for my tRDRDSC and tWRWRSC are set to 1.

[@ActuallyHardcoreOverclocking]

yeah AMD uses a different timing encoding scheme.

[@emperorSbraz]

HDR memory! half data rate.. 😭🤣

[@isyannnkarr]

date isn't correct 19/07/2024, or this video is from the future :)

[@jamieclarke2694]

I need to read my right to read and write...

[@BleepBlop-rh9lm]

I don't see these timings on Zen Timings. And I can't find where to download Mem Tweak it.

[@DimkaTsv]

Aren't they more of a Intel memory controller stuff? Ryzen have tRDRDSC and tWRWRSC instead

[@BleepBlop-rh9lm]

@@DimkaTsv got it, thanks

[@ActuallyHardcoreOverclocking]

AMD uses a different timing encoding scheme

[@adamh8199]

how do you get memTWEAKIT to work on a MSI board?

[@ActuallyHardcoreOverclocking]

it should start on all LGA1700 boards. If you mean the ability to change timings there should be a setting in the BIOS called something along the lines of "realtime memory timing" that you need to enable.

[@MrSamadolfo]

😃 yay!

[@onehours]

can you make avid on how to cool ram better. other than that great vid

[@Heitorr]

ram fan $10 aliexpress + noctua 140 3000 ipc $30

[@bigcazza5260]

bend a strip of aluminium, screw it somewhere in the case and screw a fan on the other end. ur good for 1.55v in aussie ambients running dual rank gskill rgb b-die without a gap between the dimms (about as bad conditions as it could be)

[@creaturedanaaaaa]

I just have a 120mm fan hanging from the top of my case from zipties lol

[@emini6]

​@@HeitorrDelta or Sanyo Denki fans are way better server grade professional fans than those from noc. Digikey and mouser got them

[@lucasrem]

Need timing on electronics levels ? Or need to know how to get advance, using the memory ! I went into the beyond setting, my OMEN system was able to do this custom too, able to do it on XMP 6000 too. Autodesk MAYA i used, big arrays of DATA, GPU and CPU beyond the limits, not willing to overvolt any parts beyond XMP profiles, needing it 24/7 operatable on a workstation i work on. XMP 6000 is too fast, you will never need faster memory on this gen. XMP 5600 should be fine ! Why clock it faster, where is the bottleneck ! not here, sorry.

[@dimwillow7113]

I dont even have these timings on Zen.. they called something else ??

[@Decki777]

Can anyone tell me what's the calculation of ddr5 timings I'm new to memory overclocking.

[@SALTINBANK]

DDR5 supports a speed of 51.2 GB/s per module and two memory channels per module and Like DDR4 you double the speed 4040 in his example show x2 = 8080 mhz

[@bofast]

How does AMD work differently? Do they just not use that setting and always run at 8 or something?

[@ActuallyHardcoreOverclocking]

different timing encoding scheme

[@chrispittmanComputing]

Does timings really make any difference with the 7800X3D performance?

[@jonahhekmatyar]

Yes

[@airzero4k]

My ryzen 7900x running at ddr5 6200 mhz

[@bowserfromsonic5517]

how is the mem controller on 14th gen? is it worth getting the 14600k over the 13600k?

[@Mom19]

It's the same architecture, its the same processing node, its the same silicon. If anything, you might be lucky when buying a 14th gen CPU extremely late in its production cycle so statistically speaking you might get a chance of better silicon quality due to process aging and improving

[@ActuallyHardcoreOverclocking]

that's definitely not worth it.

[@CS-amegmondo]

@@ActuallyHardcoreOverclocking But lot better than 12th gen not?

[@larsjrgensen5975]

A nice example of DDR5 settings that is not validated with a benchmark, instead of higher MHz = better.

[@АнатолийА-й7ж]

Rtp=12?

[@JohnDoe-tv4zd]

Thing is, what happen to change in write performance, might wanna test that to, the wait period might be for high write bandwidth

[@xtinct16]

So how "AMD work differenty"?

[@cliche_AF]

ohhhh I have mine at 7 lol

[@tommyzorera]

Twice as bad = half as good.

[@pedroteixeira4369]

what about values lower than 8, are they ok ( intel or amd )? for instance, 6 or 4?

[@ActuallyHardcoreOverclocking]

they don't work because the burst length is 8 cycles. You can't have 2 sets of data on the bus at the same time.

[@GoFasterHD]

I got my DDR 7200 running at spec now. Thanks I could not get it to run XMP or anything faster than 6800 Sure glad I didnt post any screen shots LOL

[@lucasrem]

XMP 6000 is more than enough for 14900k not bottlenecking it ! Play games ! or large data set needs ? Cloud compute PC ? XMP profiles is faster then overclock it on the wrong timing !

[@GoFasterHD]

replaced some 6400 that quit.......dont over think my issue @@lucasrem

[@rostyloco1]

+

[@kingofstrike1234]

is this just for showing off that they can do 8000mhz OC because of this and the reason you can't get it to work 😂😂😂

[@bronsondixon4747]

Btw - some of the 14th gen have decently improved IMC. I can do DDR5-6800 64Gb dual rank Hynix A-Die on a Z790 Strix-E w/ stock voltages completely stable over hours of YCruncher, OCCT & Tm5. It’s survived multiple reboots and a week of gaming too. My 13900K would error in minutes under same conditions. Would be curious to see if you could get DDR5-7200 on an Apex.

[@PowellCat745]

He already got 8000 on his 14900K

[@mircomputers]

it's normal improvement in manufacturing on the same node

[@IamMxfia]

U know they 13th and 14th gen are the same chip right? Nothing is better binned on 14th gen. I don’t get why people believe it my guess they have to justify the cost of the new cpu that they bought. It’s literally called a refresh 👌🏻

[@lucasrem]

bronsondixon4747 What did you build ? 14900k, why on the Christmas Tree Strix Z790 ? 13700k is able to run your GTX 4080 Card just as fast. Less parts is faster, 2 banks the max. Keep it on XMP profiles please, only PRO people, not the gamers, can get benefits here. Large Data sets, you ? Why you need it ? Running What ?????? Great system you did build, XMP 6000 will never be any bottleneck here ;)

[@Tpecep]

You just got lucky, mc is not better at all on 14th gen

[@bigcazza5260]

could you feasibly use 16 if u wanted to run max frequency to achieve tighter latency on the other timings? like if your just gaming could u actually use that timing to stabilise higher clocks and achieve better latency? or like u said run half clock and tighten the hell out of her

[@commanderoof4578]

6:16 No you cant it will nuke your latency and make everything worse

[@davthenub]

if your maximum theoretical bandwidth is cut in half because you can only send read commands half as often as you should be able to, no amount of tightening secondaries will fix it because in the end it's reading data from/writing data to ram that matters, not just opening a ton of rows and letting them sit there.

[@Woodzta]

That's one way of saying you didn't watch. 18:03

[@volodumurkalunyak4651]

Doing all the same adjustments with outher timings and setting those 2 at 8 (for DDR5) is just as good. The lowest latemcy - manually overclocked DDR4 Samsung 8 Gigabit B-die in Gear 1.

[@Koeras16]

Are those two timings the same as the SCL (tRDRDSCL & tWRWRSCL ) timings on AMD ?

[@iDeparture]

Yeah

[@JamesC1981]

i like that you sound like kermit the frog

[@Markknightexeter]

Maybe go join the Muppets then

[@griffon2-6]

@@Markknightexeter hello BZ's mouthpiece

[@Markknightexeter]

@@griffon2-6 haha, touche

[@Look_What_You_Did]

Losers still overclocking in 2023. Hilarious. Neato I got another 2%...

[@RiteKnight]

Standard games and standard tasks; sure I'd understand. But when you wanna play Bf2 or mod of it with 30+ bots per side and zero lag, you're probably gonna need every extra % you can get. I just get the feeling you don't play that though, must be a happy simple world you live in.