• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

3800X build bad performance - what am I doing wrong?

Status
Not open for further replies.
Personal attacks? You make statements, as if you're an expert and then come and say you can't configure the UEFI on a board, that somewhat diminishes your previous statements. That's not an attack, that's a fact.
I had no issues whatsoever with my previous Asus X370 board. In fact, I don't ever remember having issues figuring out a UEFI. Yes, some of the layouts can be bad, but that doesn't prevent them from being used as they should. I guess you haven't looked at the advanced settings on a recent Gigabyte board? In my experience, all the board makers use different terminology and have completely different layouts, in fact, my Z270 board in my NAS, is nothing like my X570 board, yet I have no problem setting up either board.
It doesn't mean I can work any board from memory, but I know how to find the settings I need when I'm sat in front of it.

Please provide some proof for your random statement here.
I was not expecting four modules to work all that well either, but that was based on having used Intel systems in the past.
However, based both on my own experience and the link below, you don't actually lose anything with going four modules on Ryzen 3000, instead, you're likely to gain a smidgen of performance.
Also, 3600 is hardly maxing out IF clocks, as most people with a 3800X can go to 3733 or 3800MHz 1:1.

You lack the required knowledge to respond properly to the thread here pal, most of X570 motherboard use Daisy Chained for their DIMM while on Intel Z390 it is T-Topology, therefore it is better to run 2 Dimms on X570 and 4 Dimm on Intel Z390...
It's easy to tell the Motherboard OP is using is a Daisy Chain config based on the QVL, which runs better on 2 Dimms. With 2x16GB kit there is also the additional performance of Dual Rank DDR.
 
@TheLostSwede
True, Ryzen is doing much better than (amd) cpus in the past (with 4 sticks), BUT its not quad channel, so no big read/write gains (vs using 2 sticks).
when i started looking into what parts to get for my zen2 build, and questioned if i wanted 2 or 4 sticks..

reading up on reviews/test, it showed having two 16gb sticks (and both sides of each stick populated as in DS AND DR) was consistently doing better,
than same amount of ram running on 4 sticks.

and "most" is not "everyone", especially since it depends mostly on cpu.
per amd, 1800mhz is official IF for 3xxx, so i go by that.
outside the fact OP is not trying to max out perf, but get what "should be" there already,
so i rather not add another variable for possible problems (above 1800 mhz IF)
before its fixed..
 
Last edited:
You lack the required knowledge to respond properly to the thread here pal, most of X570 motherboard use Daisy Chained for their DIMM while on Intel Z390 it is T-Topology, therefore it is better to run 2 Dimms on X570 and 4 Dimm on Intel Z390...
It's easy to tell the Motherboard OP is using is a Daisy Chain config based on the QVL, which runs better on 2 Dimms. With 2x16GB kit there is also the additional performance of Dual Rank DDR.
Sorry, but that's nonsense. The board makers offers both types of boards, but they rarely tells you what is what.
Please go read the link I provided.

@TheLostSwede
True, Ryzen is doing much better than (amd) cpus in the past (with 4 sticks), BUT its not quad channel, so no big read/write gains (vs using 2 sticks).
when i started looking into what parts to get for my zen2 build, and questioned if i wanted 2 or 4 sticks..

reading up on reviews/test, it showed having two 16gb sticks (and both sides of each stick populated as in DS AND DR) was consistently doing better,
than same amount of ram running on 4 sticks.

and "most" is not "everyone", especially since it depends mostly on cpu.
per amd, 1800mhz is official IF for 3xxx, so i go by that.
outside the fact OP is not trying to max out perf, but get what "should be" there already,
so i rather not add another variable for possible problems (above 1800 mhz IF)
before its fixed..
Obviously it's not quad channel, but for whatever reason, there's still a performance advantage of going with four modules over two. The link I provided is not the only source that points at that. Sure, it's maybe 5% at the most, but it's free performance if you're not planning on upgrading your memory in the future. I wasn't planning on getting more than 16GB of RAM, but as RAM prices dropped, I took a gamble and it paid off. Sure, I see 0.7ns higher latency, but that's hardly an issue.

Care to provide a link that proves that?

Sure, not everyone is hitting 3800MHz, but I would say all 3800X CPUs are capable of 3733MHz 1:1. Keep in mind that these are some of the highest binned consumer CPUs from AMD. We're not talking the 3600 here, but the 3800X.
How do you know he doesn't want to get optimal performance from his/her system? Right now, the issue is to fix whatever weird stuff is going on, but the OP is clearly going for a high-end build. But sure, let's take one step at a time, but at the same time, why not make sure that once the problems are fixed, that there's the option to go one step further?

Here's another article showing that Intel CPUs also gain performance from going from two to four modules. It's not going to help in all scenarios, but it would appear that today's memory controllers have no issues deal with four modules, regardless of the blue or the red team.
 
Last edited:
@nguyen
the need to change more/less settings "to do the same"/having problems with different brands using different terms etc for someone messing with power limits/stuff in detail,
or ppl having a hard time to know the difference between xm,p and doc, are ppl that should leave "bios" alone, and is exactly what TLS said, not a problem with brand, but the operator.

Sadly speaking I was building systems that are way more complicated than tinkering with Bios in a short time, something like hard-tube custom water cooling rig in less than a day, so yeah had to wade through a overclocking guide on something so simple was a pain I didn't want to deal with.

Sorry, but that's nonsense. The board makers offers both types of boards, but they rarely tells you what is what.
Please go read the link I provided.


Obviously it's not quad channel, but for whatever reason, there's still a performance advantage of going with four modules over two. The link I provided is not the only source that points at that. Sure, it's maybe 5% at the most, but it's free performance if you're not planning on upgrading your memory in the future. I wasn't planning on getting more than 16GB of RAM, but as RAM prices dropped, I took a gamble and it paid off. Sure, I see 0.7ns higher latency, but that's hardly an issue.

Care to provide a link that proves that?

Sure, not everyone is hitting 3800MHz, but I would say all 3800X CPUs are capable of 3733MHz 1:1. Keep in mind that these are some of the highest binned consumer CPUs from AMD. We're not talking the 3600 here, but the 3800X.
How do you know he doesn't want to get optimal performance from his/her system? Right now, the issue is to fix whatever weird stuff is going on, but the OP is clearly going for a high-end build. But sure, let's take one step at a time, but at the same time, why not make sure that once the problems are fixed, that there's the option to go one step further?

Here's another article showing that Intel CPUs also gain performance from going from two to four modules. It's not going to help in all scenarios, but it would appear that today's memory controllers have no issues deal with four modules, regardless of the blue or the red team.

Again go read what is the difference between Daisy Chain and T-Topology...Manufacturers don't tell you what is is but reviewers usually do, and you can tell from the Memory QVL...
 
Again go read on what is the difference between Daisy Chain and T-Topology....
Dude, my board is daisy chained, NOT A PROBLEM.
You keep insisting things that aren't all that relevant to this topic. Yes, Daisy Chained topology is supposedly better for overclocking, but I'm not doing extreme overclocking here, am I? We're talking a 200MHz overclock on my RAM, well 100MHz actually and 100MHz for the IF. I'm sure the T-topology boards will do this just fine as well.
You're getting stuck on a board design technicality that isn't even a consideration at these speeds.
 
A whole lot that's been said in this thread in regard to boot loading speed is pure lunacy, e.g. random IOPS being low or UEFI/GPT as a requirement to boot. I have an outright "bad" SATA SSD which is a lot slower than what the topic starter has (both throughput and random IOPS), I have CSM enabled in BIOS and my system boots to the desktop (W10 64) in less than 20 seconds from the moment I press the power button. If I were a topic starter I'd also reinstall W10 from scratch.

People often load some crazy apps on their systems (which often run in background and use 100% of CPU for no reasons), some have RAM, registry and other "optimizers" (most such apps are a sham), some run various badly coded hacking/activation tools, etc. etc. etc. Almost always the first step in debugging your Windows issues is making it pristine, i.e. full reinstallation from scratch without installing any additional software, even drivers. Yes, drivers, because I've seen quite a lot of systems where misbehaving drivers caused slowdowns, CPU usage problems, latency/input lag issues, etc.
 
A whole lot that's been said in this thread in regard to boot loading speed is pure lunacy, e.g. random IOPS being low or UEFI/GPT as a requirement to boot. I have an outright "bad" SATA SSD which is a lot slower than what the topic starter has (both throughput and random IOPS), I have CSM enabled in BIOS and my system boots to the desktop (W10 64) in less than 20 seconds from the moment I press the power button. If I were a topic starter I'd also reinstall W10 from scratch.

People often load some crazy apps on their systems (which often run in background and use 100% of CPU for no reasons), some have RAM, registry and other "optimizers" (most such apps are a sham), some run various badly coded hacking/activation tools, etc. etc. etc. Almost always the first step in debugging your Windows issues is making it pristine, i.e. full reinstallation from scratch without installing any additional software, even drivers. Yes, drivers, because I've seen quite a lot of systems where misbehaving drivers caused slowdowns, CPU usage problems, latency/input lag issues, etc.
If you'd actually read what I said, you might've been able to make some sense of it, rather than going on a rant.
GPT is actually important for NVMe SSDs, as you lose out a lot of performance if you use MBR instead. I made this mistake with my previous system and my SSD was only at about 2/3 of the performance it was after I installed Windows 10 using GPT. I NEVER said it was a requirement to boot.
As for disabling CSM, I said it might help reduce the boot time a little bit, but it's most likely not the issue here. It saves a couple of seconds off the boot time and wasn't going to help the OP with his/her issue.
But I guess reading comprehension isn't one of your skills?

Also, the OP provided a screenshot of his SSD performance and it's clear there's something wrong with the IOPS, as they're about 1/5th of what they should be.

You did bring up one thing I didn't think about though, doesn't Samsung have their own custom driver for their SSDs? I guess the OP might not have installed it.
 
Hey guys, I appreciate all the discussion. I’m willing to put in a lot of work to get my system optimized but it is correct that I am NOT looking for maxing the performance to a point where instability could occur.
First off, I just want the system to run normal and than We can think about OCing a bit.

having that said, money isn’t the issue but ofc I wanna spend it wisely. The 4 8gig blocks g.skill royals though do look appealing to me, they have great stock timings and really fit the rgb style I have in the system.

i am well aware the 4 blocks won’t give me more than dual channel, but I will read up on the links provided.

i do think I don’t have a specific driver installed for the SSD but I have the Samsung Magician tool downloaded

i’m typing on the phone right now so I just wanted to confirm my next steps for today as suggested
  1. Trying to run cinebench in safe mode having the bios loaded with the stock optimized settings to see if it has to do with my installed stuff
  2. Trying to punch in dram 1.7.1 values - first only primary timings and try to post, second punch in all numbers and try
  3. If all fails - order the ram!
  4. For everyone, I will definitely try to reinstall windows one more time but to be honest, I do struggle installing what is really needed. can’t do that before 2 weeks either. I know I need chipset driver sound driver, and all the mobo stuff from asus support page but it is possible that I forget smth
It is mothers day so it takes 7-8h until I can test the things above.
 
Dude, my board is daisy chained, NOT A PROBLEM.
You keep insisting things that aren't all that relevant to this topic. Yes, Daisy Chained topology is supposedly better for overclocking, but I'm not doing extreme overclocking here, am I? We're talking a 200MHz overclock on my RAM, well 100MHz actually and 100MHz for the IF. I'm sure the T-topology boards will do this just fine as well.
You're getting stuck on a board design technicality that isn't even a consideration at these speeds.

You show a complete lack of knowledge here,
First you advise someone not to follow Asus memory QVL, which is a big no no.
Second you advise to use a 4 Dimms config on a cheap daisy chain board at 3800mhz, which just wont happen, even at 3600mhz and 4 Dimms you have to check the QVL very carefully.
Yeah if someone is following your advise they are just throwing money at nothing...
 
Alright peeps, keep it civil and less name calling please. We don't want to scare off OP just yet.
 
@TheLostSwede
no link. it started when i looked at ram and noticed SS/DS and SR/DR being mentioned, and wanted to know what is what.
reading about 3600/3700 reviews, it was something a couple testers saw (testing mem-controller/ram R/W output),
and once i had seen "not just one" reporting it, i just memorized which type of stick to get, without keeping any other info.


@ecopsorn
no need for the magician tool, unless you want to secure erase etc, but not needed for drive at all,
and the (missing) driver is not gonna be the cause for 4/5 performance drop, the MS one coming with win is fine (until chipset etc is installed)
i would do a clean install after loading defaults in bios.

@ngyen
QVL means tested and verified, NOT any guarantee it will work
what about older sticks getting a refresh and different chips, but model (name) stays? the new module is on the QVL,
would that now include the "older" type (with a different chip inside), just because the model name didnt change?! nope.

best example:
in the past 5y i had boards from 120-300$ (for myself/other builds) that would do jack (most the time having trouble to post) with ram listed,
yet the sticks i switched to that worked, +80% were NOT on the list, so i gave up on buying only listed ram.
its kinda physically impossible for board vendors to guarantee that, as mem controller is on the cpu (amd),
and they neither know what chip your getting (3600-39xx) nor what ends up in your rig lottery wise.
 
Last edited:
You show a complete lack of knowledge here,
First you advise someone not to follow Asus memory QVL, which is a big no no.
Second you advise to use a 4 Dimms config on a cheap daisy chain board at 3800mhz, which just wont happen, even at 3600mhz and 4 Dimms you have to check the QVL very carefully.
Yeah if someone is following your advise they are just throwing money at nothing...
Uhm, since when is daisy chained topology cheap? It allows for better overclocking, but ok...

And I guess you aren't reading my posts, just having a rant. The QVL's only mean the board makers have tested the RAM they have at hand. It doesn't mean that other RAM doesn't work just as fine or better.
As I said, none of my RAM is on the QVL and it works 1000x better than the old crap I had that was on the QVL.

I have worked with computers for most of my working life, I used to be a reviewer and have done 20+ motherboard group tests when I worked for an actual computer magazine, but yeah, I know nothing...

Hey guys, I appreciate all the discussion. I’m willing to put in a lot of work to get my system optimized but it is correct that I am NOT looking for maxing the performance to a point where instability could occur.
First off, I just want the system to run normal and than We can think about OCing a bit.

having that said, money isn’t the issue but ofc I wanna spend it wisely. The 4 8gig blocks g.skill royals though do look appealing to me, they have great stock timings and really fit the rgb style I have in the system.

i am well aware the 4 blocks won’t give me more than dual channel, but I will read up on the links provided.

i do think I don’t have a specific driver installed for the SSD but I have the Samsung Magician tool downloaded

i’m typing on the phone right now so I just wanted to confirm my next steps for today as suggested
  1. Trying to run cinebench in safe mode having the bios loaded with the stock optimized settings to see if it has to do with my installed stuff
  2. Trying to punch in dram 1.7.1 values - first only primary timings and try to post, second punch in all numbers and try
  3. If all fails - order the ram!
  4. For everyone, I will definitely try to reinstall windows one more time but to be honest, I do struggle installing what is really needed. can’t do that before 2 weeks either. I know I need chipset driver sound driver, and all the mobo stuff from asus support page but it is possible that I forget smth
It is mothers day so it takes 7-8h until I can test the things above.
Instability doesn't mean optimised for one, quite the opposite. My system have been working fine for months, not a glitch, not a hiccup, not a BSOD.

Unfortunately, computers sometimes seem to have a will of their own and it can be really hard to pin-point issues and I guess that's also why there are so many opinions on what to do and what to buy. Everyone has different experiences, even with the same parts. Then there's the fact that there are hardware revisions made by the parts makers, some you can tell, others are less obvious. It makes it quite tricky to figure out what has gone wrong sometimes and what was a solution six months ago, might no longer work.

Try the NVMe driver here, it might help improve the IOPS performance.

I hope you can figure what the issue is, but considering none of the suggestions have worked so far, or you already did the correct thing, this is a bit baffling.
 
@TheLostSwede
no link. it started when i looked at ram and noticed SS/DS and SR/DR being mentioned, and wanted to know what is what.
reading about 3600/3700 reviews, it was something a couple testers saw (testing mem-controller/ram R/W output),
and once i had seen "not just one" reporting it, i just memorized which type of stick to get, without keeping any other info.


@ecopsorn
no need for the magician tool, unless you want to secure erase etc, but not needed for drive at all,
and the (missing) driver is not gonna be the cause for 4/5 performance drop, the MS one coming with win is fine (until chipset etc is installed)
i would do a clean install after loading defaults in bios.

@ngyen
QVL means tested and verified, NOT any guarantee it will work
what about older sticks getting a refresh and different chips, but model (name) stays? the new module is on the QVL,
would that now include the "older" type (with a different chip inside), just because the model name didnt change?! nope.

best example:
in the past 5y i had boards from 120-300$ (for myself/other builds) that would do jack (most the time having trouble to post) with ram listed,
yet the sticks i switched to that worked, +80% were NOT on the list, so i gave up on buying only listed ram.
its kinda physically impossible for board vendors to guarantee that, as mem controller is on the cpu (amd),
and they neither know what chip your getting (3600-39xx) nor what ends up in your rig lottery wise.

RAM manufacturers would sometimes switch memory ICs without telling, but it's riskier to buy some kit outside the QVL unless you know exactly what ICs are in it, for example Samsung B-die are go to kit whether it's in the QVL or not. If you get some kit that is not Samsung B-die and not in the QVL, yeah that only means trouble...

Uhm, since when is daisy chained topology cheap? It allows for better overclocking, but ok...

And I guess you aren't reading my posts, just having a rant. The QVL's only mean the board makers have tested the RAM they have at hand. It doesn't mean that other RAM doesn't work just as fine or better.
As I said, none of my RAM is on the QVL and it works 1000x better than the old crap I had that was on the QVL.

I have worked with computers for most of my working life, I used to be a reviewer and have done 20+ motherboard group tests when I worked for an actual computer magazine, but yeah, I know nothing...


Instability doesn't mean optimised for one, quite the opposite. My system have been working fine for months, not a glitch, not a hiccup, not a BSOD.

Unfortunately, computers sometimes seem to have a will of their own and it can be really hard to pin-point issues and I guess that's also why there are so many opinions on what to do and what to buy. Everyone has different experiences, even with the same parts. Then there's the fact that there are hardware revisions made by the parts makers, some you can tell, others are less obvious. It makes it quite tricky to figure out what has gone wrong sometimes and what was a solution six months ago, might no longer work.

Try the NVMe driver here, it might help improve the IOPS performance.

I hope you can figure what the issue is, but considering none of the suggestions have worked so far, or you already did the correct thing, this is a bit baffling.

Most X570 board use Chaisy Chain and most Z390 board use T-Topology (Intel prefers T-Topology), even the Asus Z390 Maximum XI Hero use the T-Topology. That information can easily be extracted if you had read the Memory QVL (highest memory overclocks are available only to 4 DIMMs). I guess being a reviewer means you don't have to read the official information about a product eh ?

Furthermore using Dual Rank Rams (2x16GB kits) have the same effect as 4x8GB kit as 4 kit would simulate the effect of Dual Rank. Check out Buildzoid over at Actually Hardcore Overclock Youtube if you want to know more about best Ram configurations for Ryzen...

And lastly anecdotal evidence cannot be recommended to beginners, that's like telling people to put 1.5V into the CPU because you ran just fine...

ps: putting together a PC is like child play to me, I have been doing it as a hobby for 20+ years now lol...
 
Last edited:
Keep in mind, that you may have a dud UMC. My 3700X will not boot over 3200mhz ram. A friends 3900x will not boot over 3200mhz ram. If you are having troubles even at JEDEC speeds (I can't remember), you just have a dud.

What would be the best ram for me, speed, frequency, number of blocks if money is no object?
i would still check if you can get some CL18 kits for less (unless budget isn't an issue),

Of those, I would get the Neos. It is what I have and they work fine. Mine will do 3800 at factory timings and 1.425V.

True, Ryzen is doing much better than (amd) cpus in the past (with 4 sticks), BUT its not quad channel, so no big read/write gains (vs using 2 sticks).
when i started looking into what parts to get for my zen2 build, and questioned if i wanted 2 or 4 sticks..

There is a large difference between 2 x 8 and 4 x 8. Not so much with 2 x 16 vs 4 x 8.

Almost always the first step in debugging your Windows issues is making it pristine

That is the last step when all others have failed. Throwing a system away is not the first step. Figuring out if you have software conflicts is first. Starting from scratch skips the hard work.
 
RAM manufacturers would sometimes switch memory ICs without telling, but it's riskier to buy some kit outside the QVL unless you know exactly what ICs are in it, for example Samsung B-die are go to kit whether it's in the QVL or not. If you get some kit that is not Samsung B-die and not in the QVL, yeah that only means trouble...

Most X570 board use Chaisy Chain and most Z390 board use T-Topology (Intel prefers T-Topology), even the Asus Z390 Maximum XI Hero use the T-Topology. That information can easily be extracted if you had read the Memory QVL (highest memory overclocks are available only to 4 DIMMs). I guess being a reviewer means you don't have to read the official information about a product eh ?

Furthermore using Dual Rank Rams (2x16GB kits) have the same effect as 4x8GB kit as 4 kit would simulate the effect of Dual Rank. Check out Buildzoid over at Actually Hardcore Overclock Youtube if you want to know more about best Ram configurations for Ryzen...

And lastly anecdotal evidence cannot be recommended to beginners, that's like telling people to put 1.5V into the CPU because you ran just fine...

ps: putting together a PC is like child play to me, I have been doing it as a hobby for 20+ years now lol...
Again, you're reading WAY too much into the QVL. I have asked about the QVL's and my friends that work at various motherboard makers, as I've already explained, said that they just use whatever memory they have at hand. Also, if you read the QVL's, it doesn't even mention what CPU they tested with... As such, the QVL's are just an indication as to what may work. It's hardly something that should limit people from buying other, compatible memory, it's simply guidance.

Did you just say daisy chain was cheap? So does this mean all X570 boards are cheap? I'm confused now.
Also, I suggest you take a look at this, as most is a bit of an overstatement, as it depends a lot on the manufacturer.

I'm not sure I've said anything different. In fact, I had to point out that exact same thing to someone else here.

Anecdotal? Right...
And these CPUs still peak at 1.5V, but no, I wouldn't tell anyone to set their base Voltage to 1.5V, as that would be rather stupid. That doesn't mean the CPUs don't hit 1.5V, time and time again.

And I used to do it for a living...

Keep in mind, that you may have a dud UMC. My 3700X will not boot over 3200mhz ram. A friends 3900x will not boot over 3200mhz ram. If you are having troubles even at JEDEC speeds (I can't remember), you just have a dud.

See, this is what I thought when I built my system as well, I had no luck whatsoever going over 3,200MHz, with the same RAM that I have now. I thought I'd ended up with a dud CPU, but luckily that wasn't the case. With the UEFI and AGESA updates that came out in the following months, it all started to work as intended and then some. Have you tried clearing your CMOS after installing the latest UEFI on your system? As even after updating it, for whatever reason, not all settings are wiped properly and this can cause a problem, especially with RAM. It could also be you RAM, unless you've verified it works with a different CPU. I really don't think it's the memory controller, as AMD doesn't have that bad QC/QA on their CPUs.
 
It could also be you RAM, unless you've verified it works with a different CPU. I really don't think it's the memory controller, as AMD doesn't have that bad QC/QA on their CPUs.

The ram works absolutely fine with the 3900x and Taichi x570. The ram does not work with a 3700x and Taichi x570 or 3700x with an itx x470
 
The ram works absolutely fine with the 3900x and Taichi x570. The ram does not work with a 3700x and Taichi x570 or 3700x with an itx x470
Interesting, at least if everything except the CPU are the same. If it's the memory controller, I can't see it being a widespread issue, since it's not something that's being reported about. Does any memory work over 3200MHz with that CPU? Obviously AMD only guarantees 3200MHz support, so technically there's nothing "wrong" as such.
 
Again, you're reading WAY too much into the QVL. I have asked about the QVL's and my friends that work at various motherboard makers, as I've already explained, said that they just use whatever memory they have at hand. Also, if you read the QVL's, it doesn't even mention what CPU they tested with... As such, the QVL's are just an indication as to what may work. It's hardly something that should limit people from buying other, compatible memory, it's simply guidance.

Did you just say daisy chain was cheap? So does this mean all X570 boards are cheap? I'm confused now.
Also, I suggest you take a look at this, as most is a bit of an overstatement, as it depends a lot on the manufacturer.

I'm not sure I've said anything different. In fact, I had to point out that exact same thing to someone else here.

Anecdotal? Right...
And these CPUs still peak at 1.5V, but no, I wouldn't tell anyone to set their base Voltage to 1.5V, as that would be rather stupid. That doesn't mean the CPUs don't hit 1.5V, time and time again.

And I used to do it for a living...

See, this is what I thought when I built my system as well, I had no luck whatsoever going over 3,200MHz, with the same RAM that I have now. I thought I'd ended up with a dud CPU, but luckily that wasn't the case. With the UEFI and AGESA updates that came out in the following months, it all started to work as intended and then some. Have you tried clearing your CMOS after installing the latest UEFI on your system? As even after updating it, for whatever reason, not all settings are wiped properly and this can cause a problem, especially with RAM. It could also be you RAM, unless you've verified it works with a different CPU. I really don't think it's the memory controller, as AMD doesn't have that bad QC/QA on their CPUs.

Again stop with the BS advice like "don't bother with the QVL" as Asus has the most comprehensive memory QVL out there, they even test which Dual Kit (2x8GB) work in 4x8GB configuration and which 4x8GB or 4x16GB kit won't work with 4 DIMM installed. People might have bought a Quad kit that were tested not to work properly if they don't check the QVL.
This is included in the QVL for ASUS strix X570-E Gaming

202034e0735b-7934-4a1f-acf1-36c54e9a8e42.png


If you buy one of these 32GB and 64GB kit and they don't work at 3600mhz, that just means you are stupid. You are advising people to buy hardware based on luck there.
There are plenty of reasons why a RAM kit in the QVL might not work due to weak CPU IMC or faulty Motherboards, but it is more of an assurance when a memory is in included in the QVL rather than not...
Sure if you have the time to run back to the store and swap for a new kit why not, but other people might not have to time just to run back and forth.
 
Last edited:
I always recommend buying QVL memory for your board.
 
Again stop with the BS advice like "don't bother with the QVL" as Asus has the most comprehensive memory QVL out there, they even test which Dual Kit (2x8GB) work in 4x8GB configuration and which 4x8GB or 4x16GB kit won't work with 4 DIMM installed. People might have bought a Quad kit that were tested not to work properly if they don't check the QVL.
This is included in the QVL for ASUS strix X570-E Gaming

202034e0735b-7934-4a1f-acf1-36c54e9a8e42.png


If you buy one of these 32GB and 64GB kit and they don't work at 3600mhz, that just means you are stupid. You are advising people to buy hardware based on luck there.
There are plenty of reasons why a RAM kit in the QVL might not work due to weak CPU IMC or faulty Motherboards, but it is more of an assurance when a memory is in included in the QVL rather than not...

It's not BS advice, it's a fact, the QVL's are only for guidance, whether you want to believe that or not.
Show me which CPU they used to produce that list, otherwise it means nothing.
And no, that's not how some of it is tested, a lot of the time, they only test two modules out of a four DIMM kit, as they didn't have time to test all four modules. Even the DRAM makers, tend to test a lot of memory sticks one by one, rather than in a set, even though they sell them as a set.

Tell that to @moproblems99 I'm sure he'll agree with you...
And memory makers NEVER makes changes to their kits... :rolleyes:

I'm not advising anyone to buy anything. Please show me where I have recommended that anyone should buy anything. I mentioned what I use and it worked for me. I also said it's not the very best out there. So PLEASE stop trying to twist the truth.
 
It's not BS advice, it's a fact, the QVL's are only for guidance, whether you want to believe that or not.
Show me which CPU they used to produce that list, otherwise it means nothing.
And no, that's not how some of it is tested, a lot of the time, they only test two modules out of a four DIMM kit, as they didn't have time to test all four modules. Even the DRAM makers, tend to test a lot of memory sticks one by one, rather than in a set, even though they sell them as a set.

Tell that to @moproblems99 I'm sure he'll agree with you...
And memory makers NEVER makes changes to their kits... :rolleyes:

I'm not advising anyone to buy anything. Please show me where I have recommended that anyone should buy anything. I mentioned what I use and it worked for me. I also said it's not the very best out there. So PLEASE stop trying to twist the truth.

So Asus would test 2 DIMM out of a 4 DIMM kit, then test 4 DIMM configuration with a 2 DIMM kit, hmmm....

2020b572ff3b-d4ae-4715-82ce-05df5e58066f.png


This is of course tested on a Matisse CPU as indicated by the file name
whether they used a 3600 or a 3950X CPU for testing, does it even matter as they use the same IO die. Asus might have used a CPU with one of the best IO controller die here.
Again it is pure luck that your setup work out, that or you constantly swapped out memory kits until they work...
 
Last edited:
Again stop with the BS advice like "don't bother with the QVL" as Asus has the most comprehensive memory QVL out there, they even test which Dual Kit (2x8GB) work in 4x8GB configuration and which 4x8GB or 4x16GB kit won't work with 4 DIMM installed. People might have bought a Quad kit that were tested not to work properly if they don't check the QVL.
This is included in the QVL for ASUS strix X570-E Gaming

So for the less informed of us, can you tell me why the board affects memory when the memory controller is in the CPU? What is contained in the board besides the slot and traces that would affect memory compatibility?

Again it is pure luck that your setup work out, that or you constantly swapped out memory kits until they work...

I have never bought qvl in my life. Also, never had a problem. And no, not being able to boot my Neo over 3200 isn't a problem because that is all AMD guarantees.
 
Again it is pure luck that your setup work out, that or you constantly swapped out memory kits until they work...
Yeah, right, sure... :rolleyes:
 
What the hell happened to the thread overnight?

@ecopsorn the Royal is a more high end kit because the timings pretty much guarantee it to be B-die. 16-16-16 flat, Rev.E and DJR can't do that. With B-die like that, if you know what you're doing, chances are its OC capabilities will leave your CPU in the dust. Read: your CPU's memory controller will limit you to 3600/3733/3800, not the kit. G.skill usually reserves its best ICs for Royal that has high freq tight timings.

Your Vengeance kit has B-die too, but one is literally bottom of the scrap heap and the other is a stone's throw from the crown jewel.

Hence why it's so expensive. But if you want great performance that'll work, and maybe optimize a bit and OC to 3733, the first Neo kit should work very well for you. Timings look like it shoud be CJR or DJR. The Royal is good, but it's got the price to back that up. The other Neo kit is priced out of its own league; Neos can come with B-die, but Neo is a new line for Ryzen while quite a number of Royal SKUs are G.skill's absolute cream of the crop for Intel.

I personally have a 2 x 16GB Trident Z RGB with DJR. The XMP was originally 3600/17-19-19, a little bit looser than the common 3600/16-19-19 Tridents. I run 3600/16-19-20 at the XMP 1.35V, it runs fine. I'd expect the 3600/16-19-19 to be possibly have a bit more OC potential.
 
Last edited:
So Asus would test 2 DIMM out of a 4 DIMM kit, then test 4 DIMM configuration with a 2 DIMM kit, hmmm....

Actually, in looking at this pic of qvl, I wonder why the didn't test any 2 x 16 kits from GSkill. I don't believe for a second that none of the 2 x 16 will work and people will only get lucky. Are 2x16 kits not popular?
 
Status
Not open for further replies.
Back
Top