13th Gen Ram OC Question

[TalonRahl]

41 minutes ago, leclod said:

Ok, maybe but why ?

I've got 6 p-cores close in performance to yours. At least I think I do.

In my mind there's not much difference between a 12400 and 13700K until one can use/needs more than 6 p-cores. Which in my case doesn't happen 95% of the time.

The Ram is dual rank. I've got tRFC at 630 and tREFI at 25000

 

 

 

This is a off topic tangent, but there is a quite a large difference in single core performance between the two due to base and overclock frequency alone. If were just talking standard OC potential stock for 13700k is 5.3 to 5.4 where as the 12400 is 4.4. 
You then have to take into account the Lcache sizes and their relation to performance overhead. Specifically the L3 cache which is where data and instructions are stored and shared between cores. This is critical for keeping CPU performance high in multi core gaming. The 12400 has a 18mb L3 while the 13700k has 30mb. I may not know how to OC ram well as I havent done it too much for intel. But I know a thing or two about CPUs and how they work. Not to mention the overclocking capability of a none "K" intel cpu is fairly low. The faster the CPU the more performance you gain from other high end parts like better graphics cards and the like. I am positive you are aware of this already, but still. Memory is no different. I am grossly simplifying the whole ordeal specifically because I am still at work and busy, but yes there is a large difference in our CPUs. 
Plus there are a fair amount of games coming out with the capabilities of using more than 6 P cores. Warzone which is a game I play being one of them. And that list will only increase with time. Really just depends on the kinds of games you play. Most of the games I play can utilize more P cores. Plus a 12400 would bottleneck the hell out of my 4070ti super. 

[RONOTHAN##]

1 hour ago, TalonRahl said:

What do you think of this samsung b die ram? And what did you say the tRFC could be for samsung bdie? 

It's a fine kit, though pretty much anything 3600 CL14 is. I can do you one better for what timings you should be able to roughly expect:

Spoiler

Both my kits of B die do those settings at 1.45V, so it should be fairly representative (there are kits out there that need 1.6V+ for DDR4 4000 though, so don't think that's guaranteed), though the primaries may be a bit different. The tRFC timing wasn't minimized there, but when the tREFI is over 60k, tRFC ceases to matter all that much, and dropping it to 260 (where my better kit really does bottom out at 4000MT/s 1.55V) would result in basically no performance benefit. 

 

1 hour ago, TalonRahl said:

I would imagine this would also be CR1 though I wont know until I try it. Its probably dual rank rather than single. And I see conflicting reports on how that affects performance. My understanding is single rank is like going around the track once where as dual rank is going around the track twice. From simple terms like that it makes single rank seem better and yet people say dual rank is better in some forums? 

It will pretty much always default to 2T command rate unless you set it to 1T. My luck with 1T has been sketchy at best, and since it's not that much of a performance improvement I'd just stick to 2T and higher frequency. As for dual rank vs. single rank, that kit will be guaranteed dual rank as it's 16GB DIMMs (all 16GB DIMMs of B die are dual rank). The way I like to explain it is that if one rank is doing something (waiting for a pre-charge to finish, for instance), it can go to the other rank and therefore has twice the chance for a clock cycle to do something, so since it can get more operations done in a given amount of time, it does have a performance benefit, with the drawback that you can't run as high a frequency. If you're gonna stick to Gear 1, you really want to stick to dual rank, and plus you get double the capacity as well. 

[RONOTHAN##]

44 minutes ago, TalonRahl said:

Not to mention the overclocking capability of a none "K" intel cpu is fairly low.

The 12400 is actually the one exception to that, since you can overclock the crap out of them. You just need to have one of a handful of motherboards with an external clock gen and the right BIOS, then getting it to 5.1GHz is not that hard. 

 

Also, just in case you were wondering, the 12400 does have a significantly weaker memory controller, but not because it was designed to be worse. Physically they're the same, but Intel had locked it down so that there's no way to set the VCCSA voltage above 1V, so therefore getting above 3600MT/s requires quite a bit of luck. 

[TalonRahl]

36 minutes ago, RONOTHAN## said:

Both my kits of B die do those settings at 1.45V, so it should be fairly representative (there are kits out there that need 1.6V+ for DDR4 4000 though, so don't think that's guaranteed), though the primaries may be a bit different.

I know that with B die you are more able to increase the overall dimm voltage for daily use. If I were to skip on pushing 4000 and focus on 3800 instead do you think a 1.5v would allow me to keep the timings it comes with 14 15 15 35(possibly reduce to 30?) and also achieve those secondary and tertiary timings you list without crazy amounts of voltage to the VCCSA and VDDQ? Would I even have to manually adjust those past the auto settings in the motherboard xmp profile?
If so would I want to try and keep those in the same range of 1.2v VCCSA and 1.3 - 1.35v VDDQ? Would that not be enough and would I need or be able to push further to ensure stability. From what I can tell the IMC in 13700k is said to be good in most forums. 
Just to verify this is the voltage to adjust when youre referencing the dimm voltage for asus boards. (This is not my settings I just found a video showcasing the board language to verify lol 1.8 is crazy) I feel like I would never go above 1.6 even with b die. Id be scared haha. 

 

36 minutes ago, RONOTHAN## said:

It will pretty much always default to 2T command rate unless you set it to 1T. My luck with 1T has been sketchy at best, and since it's not that much of a performance improvement I'd just stick to 2T and higher frequency. As for dual rank vs. single rank, that kit will be guaranteed dual rank as it's 16GB DIMMs (all 16GB DIMMs of B die are dual rank). The way I like to explain it is that if one rank is doing something (waiting for a pre-charge to finish, for instance), it can go to the other rank and therefore has twice the chance for a clock cycle to do something, so since it can get more operations done in a given amount of time, it does have a performance benefit, with the drawback that you can't run as high a frequency. If you're gonna stick to Gear 1, you really want to stick to dual rank, and plus you get double the capacity as well. 

Your dual rank explanation makes a lot of sense. I have never owned b die ram before. The ram I am using rn is single rank. 
Just as a disclaimer I have already bought the ram. Did it just now. Will be here tomorrow lol. It was the difference of 60 bucks vs the sk hynix ram I have. And 60 bucks for easier OCing ability and more than likely better latency seems a no brainer for what I am trying to achieve. 

As for setting command rate I know how to do that for asus boards as its the 1n, 2n, N:1 yada yada. 

Though I dont know how to specify the gear. Do you know where that would be if it doesnt auto set to Gear 1 in the bios? I know its got to be in AI tweaker somewhere but where. I would want to sort of be going for 3800 to 4000mhz 14 15 15 30 GEAR1 T2 with optimized secondary and tertiary timings. At least "somewhat" optimized but still really stable through y cruncher and TestMem5 that you mentioned. 

Also while I know not all ram is the same and achievable results will differ per stick along with voltages needed. I by no means believe that what worked for you will FOR SURE work for me... but... 
Do you mind letting me know what VCCSA and VDDQ you used to achieve the clock and timings you posted above? 
Along with the app you used to showcase all of those in which you screenshot.

[leclod]

16 minutes ago, TalonRahl said:

Though I dont know how to specify the gear. Do you know where that would be if it doesnt auto set to Gear 1 in the bios?

On my MSI board it looks like that, I guess it is similar on your ASUS board

 

Edited February 14 by leclod

[RONOTHAN##]

1 hour ago, TalonRahl said:

I know that with B die you are more able to increase the overall dimm voltage for daily use. If I were to skip on pushing 4000 and focus on 3800 instead do you think a 1.5v would allow me to keep the timings it comes with 14 15 15 35(possibly reduce to 30?) and also achieve those secondary and tertiary timings you list without crazy amounts of voltage to the VCCSA and VDDQ? Would I even have to manually adjust those past the auto settings in the motherboard xmp profile?
If so would I want to try and keep those in the same range of 1.2v VCCSA and 1.3 - 1.35v VDDQ? Would that not be enough and would I need or be able to push further to ensure stability. From what I can tell the IMC in 13700k is said to be good in most forums. 
Just to verify this is the voltage to adjust when youre referencing the dimm voltage for asus boards. (This is not my settings I just found a video showcasing the board language to verify lol 1.8 is crazy) I feel like I would never go above 1.6 even with b die. Id be scared haha. 

It depends on your kit. 3800 CL14 is relatively difficult, and you need a pretty good kit for that at 1.5V. Most kits need 1.55V for it, though there are kits that will do it at just 1.5V or lower (my dual rank kit is one of them, though it's also a top bin 4400 CL17 and therefore not really representative of 3600 CL14 I consider the 4th worst good bin (out of all the good, guaranteed B die bins, it has the 4th most variance and therefore is possible to need a lot of voltage for the same settings as a higher bin kit). 

 

If you aren't pushing the frequency, then yeah, you don't need to push VDDQ and VCCSA. The 13700K's IMC is pretty decent, though as with most Intel CPUs, there's quite a lot of variance between them, and it sounds like you just got a bad chip memory controller wise. 

 

Yeah, I was talking about the DIMM voltage. 1.65V is the highest I'd be willing to tolerate for daily usage, 1.6V is the highest you can find an XMP for, though anything above 1.5V I'd want to have some sort of memory cooling setup (usually a fan sitting on the back of the GPU). There's very much a diminishing return above 1.5V though, so sticking to that for your B die OC is what I usually do (the only time I've deviated from that is when I was dailying Ryzen 5000, where I pushed 1.52V because I wanted to run 3800 CL14-13-13-21 and 1.5V was just barely not enough for that). 

 

1 hour ago, TalonRahl said:

As for setting command rate I know how to do that for asus boards as its the 1n, 2n, N:1 yada yada. 

MSI boards (where I have most of my experience) do the same thing, pretty sure that comes from Intel themselves. The reason it says N instead of T is that because they measure it in memory controller ticks, so while in gear 1 1N and 1T are the same thing, in gear 2 1N translates to 2T because the memory controller is running at half the frequency.

 

1 hour ago, TalonRahl said:

Though I dont know how to specify the gear. Do you know where that would be if it doesnt auto set to Gear 1 in the bios?

Pretty sure they don't call it "Gear" for whatever reason. Gear is just the name that Intel uses, so it's what I've defaulted to call it. Pretty sure it's "IMC to memory ratio" or something along those lines, and it will be within three settings of the memory frequency setting, pretty sure above. 1:1 means gear 1, 1:2 means gear 2, etc. 

 

1 hour ago, TalonRahl said:

Do you mind letting me know what VCCSA and VDDQ you used to achieve the clock and timings you posted above? 

Pretty sure that was 1.45V VCCSA, 1.35V VDDQ TX. That was also on an MSI board, so it likely won't translate to what you're using. 

 

As for the program, it's ASUS MemTweakIt! It's a great program, and if you set the BIOS correctly, you can actually change the timings in the OS to figure out what is stable and what isn't (with the exception of the CAS latency, command rate, and tCWL) on the fly. As for where to get it, I got it over on the HWBot forum, specifically the Z790 Apex support forum. The TurboV Core link there is also really helpful, it's basically the ASUS version of XTU that has way more settings and is way more reliable (don't know if works on non-ROG boards though). 

https://community.hwbot.org/topic/220910-asus-rog-maximus-apex-z790-apex-15/

[TalonRahl]

1 hour ago, leclod said:

On my MSI board it looks like that, I guess it is similar on your ASUS board

 

Unfortunately there is absolutely no menu even kind of similar to this for my board. Fat rip to me. 
I will just try to do some more digging for some possible CPU IMC settings. 

[TalonRahl]

46 minutes ago, RONOTHAN## said:

It depends on your kit. 3800 CL14 is relatively difficult, and you need a pretty good kit for that at 1.5V. Most kits need 1.55V for it, though there are kits that will do it at just 1.5V or lower (my dual rank kit is one of them, though it's also a top bin 4400 CL17 and therefore not really representative of 3600 CL14 I consider the 4th worst good bin (out of all the good, guaranteed B die bins, it has the 4th most variance and therefore is possible to need a lot of voltage for the same settings as a higher bin kit). 

 

If you aren't pushing the frequency, then yeah, you don't need to push VDDQ and VCCSA. The 13700K's IMC is pretty decent, though as with most Intel CPUs, there's quite a lot of variance between them, and it sounds like you just got a bad chip memory controller wise. 

 

Yeah, I was talking about the DIMM voltage. 1.65V is the highest I'd be willing to tolerate for daily usage, 1.6V is the highest you can find an XMP for, though anything above 1.5V I'd want to have some sort of memory cooling setup (usually a fan sitting on the back of the GPU). There's very much a diminishing return above 1.5V though, so sticking to that for your B die OC is what I usually do (the only time I've deviated from that is when I was dailying Ryzen 5000, where I pushed 1.52V because I wanted to run 3800 CL14-13-13-21 and 1.5V was just barely not enough for that). 

Are you effectively saying that if all I do is mess with timings and leave it at 3600 then I wont need to adjust VCCSA or VDDQ beyond auto? 
I have another 13700k in my office for a second build I have been looking to use for a stream pc specifically.
I will just slot that one in this board. See if the silicon lottery is better. 

So what about this @RONOTHAN## 1.5v DIMM and then adjust secondary and tertiary timings similar to how you set yours whilst leaving the primary timings stock xmp 14 15 15 except for tRAS adjusted to 30-32 rather than 35? All whilst leaving those VCCSA and VDDQ voltages alone. Im not overclocking the 13700k past stock 5.4 at stock voltage other than a small vcore offset undervolt.

 

54 minutes ago, RONOTHAN## said:

The TurboV Core link there is also really helpful, it's basically the ASUS version of XTU that has way more settings and is way more reliable. 

Is this somewhere on the link you sent? Or am I just gonna have to go searching? 

[RONOTHAN##]

2 minutes ago, TalonRahl said:

Is this somewhere on the link you sent? Or am I just gonna have to go searching? 

It's in that link, in the same "Tools" section. There 3 drop box links, one for MemTweakIt!, one for Turbo VCore, and one for the Worktool RPL (this is a bunch of Apex specific tools, don't worry about this one).

 

4 minutes ago, TalonRahl said:

Are you effectively saying that if all I do is mess with timings and leave it at 3600 then I wont need to adjust VCCSA or VDDQ beyond auto? 

Yes. 

 

4 minutes ago, TalonRahl said:

I have another 13700k in my office for a second build I have been looking to use for a stream pc specifically.
I will just slot that one in this board. See if the silicon lottery is better.

I wish you luck then, there's a massive amount of variance between 13th gen CPUs. The worst ones do 3733MT/s max in gear 1, while the best do 4400MT/s. Most should do 4000MT/s without too much effort, so your other one is probably better. 

 

6 minutes ago, TalonRahl said:

1.5v DIMM and then adjust secondary and tertiary timings similar to how you set yours whilst leaving the primary timings stock xmp 14 15 15 except for tRAS adjusted to 30-32 rather than 35?

That's fine, if you're going to leave it at the stock 14-15-15 I likely wouldn't even up the voltage. Those timings in the screen shot I sent were at 1.45V like the XMP (the B660M-A Pro I was using would blue screen if the memory voltage was 1.5+), and the only subtimings that scale with voltage are tRFC and sometimes tWRPRE and tRTP. Upping the voltage is mostly to get tighter primary timings and higher frequency, so if you want to get say 4133 CL15, then you would need to push the voltage. 

 

Also tRAS can always be set to 28 with B die, no need to do 30 or 32. 

[TalonRahl]

11 hours ago, RONOTHAN## said:

It's in that link, in the same "Tools" section. There 3 drop box links, one for MemTweakIt!, one for Turbo VCore, and one for the Worktool RPL (this is a bunch of Apex specific tools, don't worry about this one).

 

Yes. 

 

I wish you luck then, there's a massive amount of variance between 13th gen CPUs. The worst ones do 3733MT/s max in gear 1, while the best do 4400MT/s. Most should do 4000MT/s without too much effort, so your other one is probably better. 

 

That's fine, if you're going to leave it at the stock 14-15-15 I likely wouldn't even up the voltage. Those timings in the screen shot I sent were at 1.45V like the XMP (the B660M-A Pro I was using would blue screen if the memory voltage was 1.5+), and the only subtimings that scale with voltage are tRFC and sometimes tWRPRE and tRTP. Upping the voltage is mostly to get tighter primary timings and higher frequency, so if you want to get say 4133 CL15, then you would need to push the voltage. 

 

Also tRAS can always be set to 28 with B die, no need to do 30 or 32. 

Ram will be here tomorrow. Will be trying with brand new 13700k. 
I will try stock and just see what kind of primary, secondary and tertiary timings it gives with auto. 
You said that 3800cl14 is pretty hard to achieve at 1.45v
What about 3800 15 15 15 28 Gear 1 T2 with those secondary and tertiary timings you had for 4000cl16? 
Or you think that will require more into the VDIMM, VCCSA and VDDQ? I know you said you had 1.45v VCCSA and 1.35 VDDQ on an msi board to achieve those timings at 4000. 
Why do those not correlate to an asus board if the chip is similar? What chip did you use for that? 

Or is 3800CL15 not worth it? Its about the same latency for higher bandwidth. 
Though it may not be worth it. 
May be more worth bringing the stock xmp of 3600 14 15 15 35 down to just 3600 14 14 14 28 instead? 
And with that I would expect at the very least to have to push the voltage a tad on the VDIMM? 1.47v? Just to increase the likelihood of stability without having to specifically put in a spot fan by pushing it to 1.5v. I am not really wanting to do that and most ram fan kits are ugly as hell that you can find now a days tbh. I have good airflow rn. And there is an exhaust directly where the graphics card exhausts out. 
If all else fails. This ram will perform better than the 3600cl18 I started with and the 4000cl18 I was messing with initially. Of that I am sure. 
But at this point if I have some b die I figured why not at least "optimize slightly". 

Thanks again for joining me on this knowledge journey for me. I work in network security so my expertise is in programing mainly. Not hardware. And while I have good knowledge on how computers work as a whole I have never really messed with ram let alone for intel. 

[RONOTHAN##]

41 minutes ago, TalonRahl said:

What about 3800 15 15 15 28 Gear 1 T2 with those secondary and tertiary timings you had for 4000cl16? 

tRCD 15 might still be a little difficult depending on the kit, but it's no where near as hard as CL14. 

 

41 minutes ago, TalonRahl said:

Or you think that will require more into the VDIMM, VCCSA and VDDQ? I know you said you had 1.45v VCCSA and 1.35 VDDQ on an msi board to achieve those timings at 4000. 
Why do those not correlate to an asus board if the chip is similar? What chip did you use for that? 

These voltages depend quite heavily on how the motherboard BIOS optimizes for them, and ASUS and MSI both optimized them quite differently. More on DDR5, I've seen BIOS revisions go from 1.425V being optimal for VDDQ TX to 1.225V being optimal for the same CPU, motherboard, RAM kit, and frequency, with 1.425V not even making it into Windows. That 1.45V/1.35V setup I would only trust to work on that particular board, that particular BIOS revision, and that particular CPU. 

 

As for the CPU, it was a 13700K. 

 

41 minutes ago, TalonRahl said:

Or is 3800CL15 not worth it? Its about the same latency for higher bandwidth. 

I would try for it, since it shouldn't really be that difficult and the extra bandwidth can be helpful. Realistically though the majority of th performance improvements will come from the sub-timings, so if you don't want to put any effort in those sub-timings will do most of the job. 

 

41 minutes ago, TalonRahl said:

And with that I would expect at the very least to have to push the voltage a tad on the VDIMM? 1.47v? Just to increase the likelihood of stability without having to specifically put in a spot fan by pushing it to 1.5v.

I should note that a RAM fan only starts to become a good idea at 1.5V. I'd still be fine running a kit at that voltage without active cooling, but that's at the point where you do start to benefit from it fairly significantly (I.E. go from tRCD 16 working to 15 or even 14 without changing the voltage). 

 

41 minutes ago, TalonRahl said:

I am not really wanting to do that and most ram fan kits are ugly as hell that you can find now a days tbh

There is the solution I've gone for over the past while, just stick a standard 120mm/140mm fan on the back of the GPU, pointed directly at the memory. It works pretty well for keeping temps down and I don't personally think it looks that bad, though beauty is in the eye of the beholder. My system for reference:

Spoiler

This is a slightly old photo, I have since swapped out the fan for an Enermax fan I had with a bit better performance than the be quiet! fan in the photo, though that makes little difference. 

 

41 minutes ago, TalonRahl said:

Thanks again for joining me on this knowledge journey for me. I work in network security so my expertise is in programing mainly. Not hardware. And while I have good knowledge on how computers work as a whole I have never really messed with ram let alone for intel. 

No problem. I'm more of the opposite, great with the hardware and understanding how things work on the physical level, but while I can do some basic programming, I suck at it. As for RAM OC, you might wanna check out the channel Actually Hardcore Overclocking, since he does go through a fair bit of RAM OC and the theory/processes behind it. 

[TalonRahl]

42 minutes ago, RONOTHAN## said:

This is a slightly old photo, I have since swapped out the fan for an Enermax fan I had with a bit better performance than the be quiet! fan in the photo, though that makes little difference. 

I mean if youre watercooled then the inside of your case is already lower temp than one with an aircooler on the gpu. 
If i had more time in the day I would do a custom loop, but unfortunately with my job and wife and everything else.... Just never enough time. 

 

 

44 minutes ago, RONOTHAN## said:

1 hour ago, TalonRahl said:

Or is 3800CL15 not worth it? Its about the same latency for higher bandwidth. 

I would try for it, since it shouldn't really be that difficult and the extra bandwidth can be helpful. Realistically though the majority of th performance improvements will come from the sub-timings, so if you don't want to put any effort in those sub-timings will do most of the job. 

Well the kit is already 3600 14 15 15 35. 
The first 15 is tRCD is it not? 
Idk if you originally saw my other suggestion. 
Would 3600 14 14 14 28 primary timings be the easiest to attain with stock vCCSA and VDDQ voltages with possibly a bit more in vDIMM? Whilst adjusting the secondary and tertiary similar to what you have here. 

[RONOTHAN##]

25 minutes ago, TalonRahl said:

I mean if youre watercooled then the inside of your case is already lower temp than one with an aircooler on the gpu. 
If i had more time in the day I would do a custom loop, but unfortunately with my job and wife and everything else.... Just never enough time. 

AIOs do still work with the method I described. If you're running a dual tower air cooler like an NH-D15 or Thermalright Peerless Assassin 120, then yeah, it's just not feasible. 

 

26 minutes ago, TalonRahl said:

The first 15 is tRCD is it not? 

The 2nd value is tRCD. The order is tCL - tRCD (-tRCD_W) - tRP - tRAS (-tRC), where tRCD_W and tRC are only sometimes listed. 

 

28 minutes ago, TalonRahl said:

Would 3600 14 14 14 28 primary timings be the easiest to attain with stock vCCSA and VDDQ voltages with possibly a bit more in vDIMM?

This will heavily depend on the kit. There's a decent chance that it will just do 14-14-14 at stock voltage if they assumed that there should be some temperature safety margin, but you might need to up the vDIMM slightly in order to get it to work. 1.5V is almost certainly enough for that, though there are kits that would need slightly more. VCCSA and VDDQ TX won't need to be touched though. 

 

29 minutes ago, TalonRahl said:

Whilst adjusting the secondary and tertiary similar to what you have here. 

Yeah, there are a couple you might need to change if you end up doing that though. tCWL is somewhat tied to CAS latency, where it does anywhere between 0-2 ticks below what your CAS latency is, so at tCL 14, you should be able to lower that to somewhere between 12 and 14 depending on how lucky you are. tWRRD_sg and tWRRD_dg are affected by tCWL IIRC (it might be one of the primaries, I always adjust these two last because I forget what timings affect them and what don't) so changing the primaries will change what these need to be. tWRPRE is the same way, where IIRC tRTP and tCWL affect how stable this can be (again, it might be different timings, but those I'm fairly confident on). Also, if you're not doing active cooling, you will almost certainly need to lower the tREFI down to something less insane, 65k is usually fine if your case airflow is decent, though I've seen kits that do worse. 

 

Everything else should be exactly the same for any half-decent or better kit of B die, and those are where most of the performance comes from. Also, two timings that aren't listed in there are tRCD_W and tXP. Set those to 8 and 4 respectively, they give a small performance uplift for basically no stability penalty. 

[TalonRahl]

2 hours ago, RONOTHAN## said:

AIOs do still work with the method I described. If you're running a dual tower air cooler like an NH-D15 or Thermalright Peerless Assassin 120, then yeah, it's just not feasible. 

 

The 2nd value is tRCD. The order is tCL - tRCD (-tRCD_W) - tRP - tRAS (-tRC), where tRCD_W and tRC are only sometimes listed. 

 

This will heavily depend on the kit. There's a decent chance that it will just do 14-14-14 at stock voltage if they assumed that there should be some temperature safety margin, but you might need to up the vDIMM slightly in order to get it to work. 1.5V is almost certainly enough for that, though there are kits that would need slightly more. VCCSA and VDDQ TX won't need to be touched though. 

 

Yeah, there are a couple you might need to change if you end up doing that though. tCWL is somewhat tied to CAS latency, where it does anywhere between 0-2 ticks below what your CAS latency is, so at tCL 14, you should be able to lower that to somewhere between 12 and 14 depending on how lucky you are. tWRRD_sg and tWRRD_dg are affected by tCWL IIRC (it might be one of the primaries, I always adjust these two last because I forget what timings affect them and what don't) so changing the primaries will change what these need to be. tWRPRE is the same way, where IIRC tRTP and tCWL affect how stable this can be (again, it might be different timings, but those I'm fairly confident on). Also, if you're not doing active cooling, you will almost certainly need to lower the tREFI down to something less insane, 65k is usually fine if your case airflow is decent, though I've seen kits that do worse. 

 

Everything else should be exactly the same for any half-decent or better kit of B die, and those are where most of the performance comes from. Also, two timings that aren't listed in there are tRCD_W and tXP. Set those to 8 and 4 respectively, they give a small performance uplift for basically no stability penalty. 

I assume that you have a rig to play games and stuff. 
I know this is way off topic. 
But what kind of games you playing with your watercooled rig? 
Shooters? Single player? 
Destiny 2? Warzone? Palworld? Etc

[RONOTHAN##]

13 minutes ago, TalonRahl said:

But what kind of games you playing with your watercooled rig? 

Depends on my mood. Usually it's 3rd person over-the-shoulder story driven games (the last three main games I've played were God of War, The Last of Us, and RDR2 for the second time). I will do 1st person shooters with friends on occasion, things like Lethal Company, CS2, etc., but they're less common mostly because I suck at FPS games. 

 

I don't game as much as I probably should for the tier of hardware I have. I find tuning and benchmarking it more fun than actually using it sometimes. 

[TalonRahl]

18 hours ago, RONOTHAN## said:

Depends on my mood. Usually it's 3rd person over-the-shoulder story driven games (the last three main games I've played were God of War, The Last of Us, and RDR2 for the second time). I will do 1st person shooters with friends on occasion, things like Lethal Company, CS2, etc., but they're less common mostly because I suck at FPS games. 

 

I don't game as much as I probably should for the tier of hardware I have. I find tuning and benchmarking it more fun than actually using it sometimes. 

@RONOTHAN## All fantastic single player games listed. I didnt play last of us on PC but I have played them all on release for PS4/PS5. I heard the PC one had bugs lol. 

I am an FPS main tbh but I play all types of single player games as well if there is some sort of RPG aspect. Final Fantasy all the way to Elder Scrolls etc. 

I am finally on home stretch with the ram. I have gotten everything to post which is great. Stock xmp(which set weird settings btw) Says some weird shit like 3603 which youll see in screenshot below. 
I had to set dram frequency ratio to 1:1 manually. 
I also had to change the cpu to ramm bclk ratio to 100:100 because it auto went to 100:133? Is that normal? 

I uped some of my voltages just for testing purposes. And I reduced the tRAS to 30 rather than the 28 while I tighten everything else. 
The auto settings set the tRFC to 630 which seems high? 
I set the tREFI to 50k to start. 

Now I am looking for any recommendations on tweaking. 
Also I am having a hard time finding TM5 and ycruncher. Any help there wouuld be sick. 

 

[TalonRahl]

Also @RONOTHAN##

i am curious if there are any timings that specifically depend on one another. 
 

like if you set one timing to 4 or 14 and another to 5 then you have to set ______ to (first + second) etc?

 

or if they are all dependent on themselves. I see a lot of “formulas” online for how to figure out timings. 
and yet I am not sure they are required or used in all aspects. 
 

looking good in occt with the settings shown above. 
though I had expected to be. 
OCCT is getting my ram to 48.8 C which I feel like is cresting on getting close to as hot as I’d want the ram to be without a fan. 
 

Once I find ycruncher and tm5 I will take a look at your timing recommendations if you have them. Set them and then bench. 
none of these stress tests will harm a pc though right? Also what do you think about karhu? I have seen that listed in a forum or two. 

[RONOTHAN##]

5 minutes ago, TalonRahl said:

I also had to change the cpu to ramm bclk ratio to 100:100 because it auto went to 100:133? Is that normal? 

Yeah, that's fairly normal. That setting is referred to as the "strap" by some people, and in a lot of cases your CPU will be more stable in either 100:133 or 100:100. On 8th/9th gen especially, this was a pretty big difference, where say 4133 or 4266MT/s would be fully stable while 4100 or 4200MT/s wouldn't POST, though on 12th/13th gen this seems to be less of a factor. 

 

12 minutes ago, TalonRahl said:

The auto settings set the tRFC to 630 which seems high? 

In relation to auto settings, it's about what I'd expect (JEDEC values are designed to be very loose for compatibility with all memory chips), though compared to what it can do yeah it's really high. I'd be shocked if it couldn't do 300, and odds are you can go a little tighter, though there isn't a ton of point in going higher. For reference, the performance ends up being a ratio of the tRFC to tREFI, where tRFC/tREFI gives the amount of time your memory will be unusable due to a refresh. JEDEC of 630/11000 (I'm going off memory for JEDEC tRFC, I might be slightly off) will give ~5.7% of the memory clock cycles being useless, while if you do some more maxed out like the 333/111111 like in my timing list, you end up with 0.3% of the memory clock cycles being useless. 

 

18 minutes ago, TalonRahl said:

Now I am looking for any recommendations on tweaking. 

Absolute first thing to do would be to set tRRD_S/L to 4/4, this will get you a bandwidth and latency uplift as these are the 2nd most important timings after tREFI/tRFC. Set tFAW to 16 at the same time so that the tRRD values actually apply (tFAW is the minimum amount of time that 4 tRRD calls can be made, so ideally it's 4x tRRD_S). After that, I'd copy the dr/dd tertiary timings as those should work on every dual rank B die kit and are the most loose right now, then start copying the secondary/tertiary timings 1 by 1. 

 

23 minutes ago, TalonRahl said:

Also I am having a hard time finding TM5 and ycruncher. Any help there wouuld be sick. 

https://www.overclock.net/threads/memory-testing-with-testmem5-tm5-with-custom-configs.1751608/

http://www.numberworld.org/y-cruncher/

 

Y-Cruncher is also a benchmark that happens to be really memory sensitive, so if you want to try to compare your performance to make sure the timings are actually doing something, that would be useful (If you do want to use the benchmark, I'd use BenchMate instead). TestMem5 is a bit difficult to use as the UI is pretty clunky and requires that you edit config files to get everything to work, but once you get used to it, it's not that bad. I'd use the Anta777 Extreme1 preset on DDR4 as that's what I've had the most luck with, though some people swear by other configs. 

 

3 minutes ago, TalonRahl said:

i am curious if there are any timings that specifically depend on one another. 

Yeah, there's a couple rules like that. tRC is one such timing, as it is always tRP + tRAS, though Intel doesn't use tRC so this is only applicable to tuning AMD. tWR and tWTR are the most obvious of these timings on Intel, where in practice they don't exist but are instead calculated off a bunch of timings. Pretty sure they have something to do with CAS latency and some of the other primary timings (I do forget the exact rules and don't feel like setting up my old OC Formula to find them), but the way I treat it is to just ignore those timings until the end, then tune tWRRD_sg/dg and tWRPRE as those only affect the tWTR_L, tWTR_S, and tWR timings respectively. You can tune tWR and the tWTR timings directly in the BIOS, that should cause them to calculate the appropriate tertiary timings, though doing that usually has higher minimum values than what the memory kit is actually capable of and therefore results in slightly worse performance at the cost of convenience (if you change tCL, for instance, it will automatically adjust the tWRPRE setting to compensate). 

 

11 minutes ago, TalonRahl said:

none of these stress tests will harm a pc though right?

No, though Y-Cruncher in particular has a reputation for being so difficult to run that it can crash a CPU at stock. 

 

12 minutes ago, TalonRahl said:

Also what do you think about karhu?

It's another really good stress test. I don't use it that much since it's paid and I'm cheap (plus I get by just fine with TM5), though the couple of times I've used a friend's license it seemed to be about as effective as TM5 while having an easier to use interface. 

[TalonRahl]

1 hour ago, RONOTHAN## said:

Absolute first thing to do would be to set tRRD_S/L to 4/4, this will get you a bandwidth and latency uplift as these are the 2nd most important timings after tREFI/tRFC. Set tFAW to 16 at the same time so that the tRRD values actually apply (tFAW is the minimum amount of time that 4 tRRD calls can be made, so ideally it's 4x tRRD_S). After that, I'd copy the dr/dd tertiary timings as those should work on every dual rank B die kit and are the most loose right now, then start copying the secondary/tertiary timings 1 by 1. 

Alright so I am gonna go slow because I am sure you know the acronyms of these and also what they relate to specifically for ASUS/INTEL. Though I do not. 

When you reference tRRD_S/L to 4/4 are you referencing these values here? The names dont match "exactly" and I like to be sure. 

 

tFaw is self explanatory since it matches. 

Now for the dr/dd timings are you referencing these? 

And then should I be doing one and then rebooting and then doing the other? 


 

1 hour ago, RONOTHAN## said:

but the way I treat it is to just ignore those timings until the end, then tune tWRRD_sg/dg and tWRPRE as those only affect the tWTR_L, tWTR_S, and tWR timings respectively. You can tune tWR and the tWTR timings directly in the BIOS, that should cause them to calculate the appropriate tertiary timings, though doing that usually has higher minimum values than what the memory kit is actually capable of and therefore results in slightly worse performance at the cost of convenience (if you change tCL, for instance, it will automatically adjust the tWRPRE setting to compensate). 

 

Sort of the same thing here with these. You say tWTR_L and I am unsure what that relates to in the memtweakit columns. 

Then you stated tune tWRRD_sg/dg and tWRPRE essentially at the end? 
But what ones specifically are those if the other ones were dd and dr mentioned above? 

I mainly just want to "slightly" tighten rather than going for extreme optimal to make this as doable of an OC as possible whilst also basically insuring stability. 
But I see huge variances in the values specifically like my tWRPDEN is 42 but on your screenshot yours was 4? 
Like thats insane? Should I be even be adjusting it that low unless I am going for an extreme overclock? 

Here is a screenshot with numbered boxes of sort of how I am understanding what you are telling me to do. 
The numbers in the box that arent near the values would be considered the steps I would take where I would adjust and then restart. With the question marks sort of being me asking "is this right?" 

Also what is the performance gain from changing tRAS from 30 to 28? Is it required with having my timings not as tight as 14 14 14 14 ? I only have mine at 14 15 15 15. 

 

 

1 hour ago, RONOTHAN## said:

(if you change tCL, for instance, it will automatically adjust the tWRPRE setting to compensate). 

tCL as in CAS latency? 

 

 

1 hour ago, RONOTHAN## said:

You can tune tWR and the tWTR timings directly in the BIOS, that should cause them to calculate the appropriate tertiary timings, though doing that usually has higher minimum values than what the memory kit is actually capable of and therefore results in slightly worse performance at the cost of convenience

Which specific settings are those on that chart? tWR and tWTR? 

On your own screenshot your tWR is empty. I see mine is 23. 
I do not though see tWTR. I am probably just blind. 

 

 

You DID say my bdie is the 4th worst. 
Which is why I ask all of this because it makes me wonder if I will even be able to achieve some of these settings with the sticks I have. 
For example that insanely lower tWRPDEN of 4 in comparison to mine being 42 right now. 

 

I was having some odd bsod issues before I inevitably set my VCCSA to 1.25 and my VDDQ to 1.35. 
And to be honest I didnt creep them up either. I just pushed them to that and now everything is running cool and stable so the likelihood of me reducing them needlessly is low since I know those voltages arent like dangerous for 24/7 use from my understanding. 
All makes me believe that either my IMC is slightly finnicky in the chip I put in. 
Or the sticks are. 
Or the install was just wonky. I did reinstall after adjusting those settings though and I was having issues with that USB boot drive. So it COULD have been that as I grabbed a different one when I reinstalled. 
So many possibilities. 
Do you think those voltages I set are good for just "making sure" stuff stays stable during all of this testing or did I over do it?

 

1 hour ago, RONOTHAN## said:

It's another really good stress test. I don't use it that much since it's paid and I'm cheap (plus I get by just fine with TM5), though the couple of times I've used a friend's license it seemed to be about as effective as TM5 while having an easier to use interface. 

I will probably just get it to be honest. Looks like its 9 bucks. And I may not like throwing money everywhere but ease of use is ease of use. 
9 bucks aint bad for that. Plus I dont mind supporting people that make this kind of thing easier for people like me that are fresh to it and learning. 
 

[RONOTHAN##]

10 minutes ago, TalonRahl said:

When you reference tRRD_S/L to 4/4 are you referencing these values here? The names dont match "exactly" and I like to be sure. 

Yeah, those are the timings I'm referencing. tRRD_sg is equal to tRRD_L and tRRD_dg is equal to tRRD_S. Everyone just uses a different name for the timings, and I tend to forget which manufacturer uses which. Pretty sure S/L is the AMD naming convention, though I want to say Gigabyte uses it as well and MSI does a similar thing where they label them as tRRD and tRRD_L.

 

14 minutes ago, TalonRahl said:

Now for the dr/dd timings are you referencing these? 

The dr/dd timings are the bottom half of those, the ones that end in dr and dd. Those are the timings specific to dual rank memory configs, and they tend to train very loose for whatever reason. You can do all the tertiary timings at the same time if you'd rather, though looking at the screen shot, most of the tertiary timings already trained about as tight as they'll get. 

 

15 minutes ago, TalonRahl said:

And then should I be doing one and then rebooting and then doing the other? 

Yes, that's what I would do. 

 

16 minutes ago, TalonRahl said:

Sort of the same thing here with these. You say tWTR_L and I am unsure what that relates to in the memtweakit columns. 

MemTweakIt doesn't actually give a tWTR readout because that's a timing that doesn't exist. It's a timing that exists in the JEDEC spec and on the memory kit itself, so the motherboard needs to be able to interpret it into the tWRRD timings, but it's not one that software itself necessarily has to read. 

 

Basically, just ignore tWTR and focus on tWRRD_sg and tWRRD_dg. 

 

21 minutes ago, TalonRahl said:

But I see huge variances in the values specifically like my tWRPDEN is 42 but on your screenshot yours was 4? 
Like thats insane? Should I be even be adjusting it that low unless I am going for an extreme overclock?

The reason my tWRPDEN was 4 is because this timing is for power down mode. Setting it to 4 effectively disables power down mode and gets slightly better performance theoretically. If you don't want to set this that low, that's fine, set this equal to tWRPRE. Those 4 bottom timings in the tertiaries are all like that. 

 

24 minutes ago, TalonRahl said:

tCL as in CAS latency? 

 

Yeah, tCL is CAS Latency. All of the timings have an expanded out name, tRRD_sg for instance is Read to Read Delay Same Group, so I just stick to the acronyms since they're easier to remember. 

 

26 minutes ago, TalonRahl said:

On your own screenshot your tWR is empty. I see mine is 23. 

It's empty because ASUS for whatever reason calculates tWR based on tWRPDEN, and since I set that to 4, it doesn't exist. 

 

28 minutes ago, TalonRahl said:

You DID say my bdie is the 4th worst. 
Which is why I ask all of this because it makes me wonder if I will even be able to achieve some of these settings with the sticks I have. 
For example that insanely lower tWRPDEN of 4 in comparison to mine being 42 right now. 

I probably should figure out a way to rephase that. It's the 4th worst of the good bins, meaning it is still really good B die, it's just the 4th least consistent out of the good bins and therefore I wouldn't expect it to do super tight primary timings or super high frequency. The only bins that are really worse are 3200 CL15, 3600 CL17, 3200 CL14/3600 CL16 (those two are tied). If you have one of those bins, you do have good B die, it's just that it's likely to do slightly worse timings than say a 4000 CL14 kit. 

 

All of those timings in that screen shot work on both the B die kits I currently own, which are a top spec 2x16GB G.Skill 4400 CL17 kit (someone was selling this for next to nothing on FB Marketplace, that's the only reason I bought it), and a 2x8GB ADATA 3600 CL17 kit (this kit weirdly enough does tighter primary timings than the G.Skill kit, though the secondary timings are ever so slightly worse). I'd be shocked if those timings just didn't work for you. The reason I set that up is because I had access to a DDR4 motherboard for a week (friend asked me to do a system build for them, and I was waiting on the case for a week or two) and I wanted to compare DDR4 to DDR5 in games when overclocked to the max. That overclock was what I could throw together in 20 minutes, so they aren't actually tightened to the maximum, more just what I automatically input on any of my past Intel systems while adjusting the frequency and primary timings until they passed stress tests. 

 

B die is a very inconsistent memory chip, but the areas where it's inconsistent are the primary timings, tRFC, and the maximum frequency at a given voltage. The remaining secondary and tertiary timings work pretty much regardless of who made the kit or what speed bin it is. The only exception to that is OEM B die straight from Samsung, as those kits don't tighten the secondary timings for whatever reason. 

 

I see no reason for your kit to not do those timings listed above. They're about as generic of B die timings as you can find. 

 

43 minutes ago, TalonRahl said:

Also what is the performance gain from changing tRAS from 30 to 28? Is it required with having my timings not as tight as 14 14 14 14 ? I only have mine at 14 15 15 15. 

Basically nothing, tRAS barely affects performance. I only say to change it to 28 as on AMD setups you can reliably set this to 25 or lower (a decent number of kits will do as low as 21, including every kit of B die I've used), so 28 is pretty safe while also being the lowest value supported on Intel CPUs. 

 

45 minutes ago, TalonRahl said:

Yeah, that's roughly the order of the timings I'd set. As for tRTP, the best kits do 6 while the worst kits do 10. Generally with this timing, if it boots it's stable, so if you really want to be sure about it, lower it until the system stops booting, then raise it one tick above that (I.E. if it boots 7, doesn't boot 6, set it to 8) and you should be good. tWRRD_dr and tWRRD_dd should do 7 as well, though they don't affect performance quite all that much. 

 

 

 

Let me know if I missed anything, that was quite a bit to get through. 

[TalonRahl]

@RONOTHAN## I will come back to your reply with a larger one in general, but first. Why are there two dram refresh cycles? Do they both get set to the same value? 
 

 

[RONOTHAN##]

8 minutes ago, TalonRahl said:

Why are there two dram refresh cycles? Do they both get set to the same value? 

There's a few different refresh modes that exist. I'm 95% sure that Intel DDR4 only uses the first refresh mode, so you can leave the other two on auto and just set the first (the one you currently have highlighted). 

[TalonRahl]

1 hour ago, RONOTHAN## said:

There's a few different refresh modes that exist. I'm 95% sure that Intel DDR4 only uses the first refresh mode, so you can leave the other two on auto and just set the first (the one you currently have highlighted). 

I am more than likely going to need a fan. So I bought a small grey noctua 80mm. 
With things like OCCT and its CPU/MEMORY AVX2 test etc I can start to see temps of 49C for the ram itself. And thats only with tRFC at 400 and tREFI at 45k. It isnt erroring thankfully, but I want to tune those more and I know NOW that they are temp sensitive so I will more than likely wait till tomorrow for when this fan arrives and zip tie it somewhere. It at least matches my inner case aesthetic and we will see how ugly it feels after it is in lol. And then I will tighten those values some more and then mess with the tertiary timings.

I do want to say that even with those two timings not tightened and none of the tertiary timings tightened I can FEEL the difference in the performance of the computer from all the other adjustments. 
It may seem negligble to some, but due to working from home and gaming etc. I am in front of my computer at a minumum 10-12 hours a day 5 days a week. 
When you spend that much time with a beast like this one you can tell when things run faster. 

AND.... cherry on the cake. Mind you I have only tested a couple games so far. And I still have some tightening to do on some of the timings... BUTTTTTT... that little slight amount of stutter I had? 

Well it SEEMS as if its gone. I mean I am not putting all my eggs in the basket of "SOLVED" BUTTTT I am hopeful I am good. Again what I had wasnt the typical stutter that people complain about or is EASY to see. This stutter was variable and only truly able to be seen when panning the camera slowly. The fluidity was not there. And I could ACHIEVE fluidity by capping frames at like 120 or whatever etc but I hadnt had that issue before. So when I got the stutter it was noticable. 
Well. Bye bye stutter. Hello samsung b die fast ram. 

@RONOTHAN## I honestly cant thank you enough. I know I am not done. Tomorrow and the coming days of stability testing will be fun. I may will still have a couple questions left. 
But I just wanted to say. Damn. What a ride. 

 

Firstly. Can I buy you a coffee? I mean do you like coffee?  From like starbucks or bigby or anything that has online gift cards lol. I myself have two spectrums. I love black coffee. BUT I also like them nice lattes from starbucks every once in awhile. 
If you were comfortable giving me an email address. It can be a burner one idc lol. One you have access to. And if you care to have it. I would be more than willing to send you a coffee. I always buy a coffee for people that help me in my REAL life. Either that or an IOU and its hard to give IOUs to people online. Tradition mind you. And as I said when I started this post. I had never really reached out to forums like this before. This whole debacle was my first entering into the world of forums other than "reading them" obviously. And out of the 5 forums I tried you were the only one to actually give me REAL data and REAL help. 

Second. 
I have a discord. It is a community discord. It does relate to my stream which has been on hiatus for two weeks because of all of this(I have only been streaming for like 8 months but its been doing good). 
Besides all that. It is also where I contact my friends. Talk about games. Talk about anime. Horror stuff too as I am a horror enthusiast and play those types of games a lot for stream. 
IF you wanted. I could send you an invite. Open one. Throw it in the trash if you like. Wont hurt my feelings.  
Though after all of this and what I have learned I think I might dedicate a piece of my server to information like this. To help others. Slowly amassing more things. 
I also have my server setup so that things are invisible or visible based on the interests you choose when joining through the server questionaire onboarding (Plus you can silence stream notifications through same questionaire because a lot of my friends arent joining the server for that bologna lol.) Which is nice for all the people who dont want to see a huge server filled with COD, DESTINY, HORROR or whatever when really they are just their for anime or just to chat etc. I like minimalism.

I am soapboxing at this point. 
But yeah. 
I am pretty proud of this server. Its recent and new as is my stream. But I put a lot of work in. And I have a really good group growin. 30 years young! Plan to in 10 years hopefully have something worthwhile. 
And this moment or at least these last few days have become a core memory. Hallelujah the stutter is gone. 

https://discord.gg/VKRKTYBs

 

[RONOTHAN##]

2 hours ago, TalonRahl said:

With things like OCCT and its CPU/MEMORY AVX2 test etc I can start to see temps of 49C for the ram itself. And thats only with tRFC at 400 and tREFI at 45k.

To be fair, at least when I tested it on DDR5 (though I don't see why it shouldn't be applicable to DDR4 as well), the tREFI and tRFC timings don't affect the memory temperature. Going from tREFI 6k to 128k gave the same ~60C without any cooling at 1.35V. The voltage, case ambient temp, and memory usage are the only things that seem to affect the memory temps. 

 

Getting a fan is still a nice idea though, especially since then you could push the voltage a bit and maybe try for 3800 CL14 or 4000 CL15 as well as push the tREFI sky high.

 

2 hours ago, TalonRahl said:

Well. Bye bye stutter

Glad the stutter is gone. 

 

2 hours ago, TalonRahl said:

Firstly. Can I buy you a coffee? I mean do you like coffee?

Don't think that you have to, I mostly stay on here because I enjoy it, not because I want to get things in return. I'm not much of a coffee drinker anyway. 

 

2 hours ago, TalonRahl said:

IF you wanted. I could send you an invite.

I can join if you want me to, though don't expect me to be too active. 

[TalonRahl]

6 minutes ago, RONOTHAN## said:

To be fair, at least when I tested it on DDR5 (though I don't see why it shouldn't be applicable to DDR4 as well), the tREFI and tRFC timings don't affect the memory temperature. Going from tREFI 6k to 128k gave the same ~60C without any cooling at 1.35V. The voltage, case ambient temp, and memory usage are the only things that seem to affect the memory temps. 

 

Getting a fan is still a nice idea though, especially since then you could push the voltage a bit and maybe try for 3800 CL14 or 4000 CL15 as well as push the tREFI sky high.

 

I mean more so because these parts of the ram timings degrade in high heat that I would want to maximize them truly when the heat of the ram itself would allow them to perform at their best and not error. 

 

 

7 minutes ago, RONOTHAN## said:

Don't think that you have to, I mostly stay on here because I enjoy it, not because I want to get things in return. I'm not much of a coffee drinker anyway. 

I mean.. you have to like.... something? lol 
I definitely understand you werent helping me for something in return. Sometimes though in life youre gonna find people like me that really enjoy paying people back for their kindness. 

 

 

8 minutes ago, RONOTHAN## said:

I can join if you want me to, though don't expect me to be too active. 

You dont have to be active. 
But you never know. Might end up being a game or something else in my server you find interesting. 
Plus who knows. Maybe one day I could have another computer hardware question in the future. At least in my server you wont be pestered by people. Its not huge lol. 
Plus I have a lot of people in my server that know a plethora of different things. From programming, to hardware, home repair, etc. 
I hope to grow that too. Its nice having a cool place to come to where people know a bunch of random shit and are kind enough to help you and is a bit more personal than a forum. 

As I said though. Dont have to. The offer was just put out there. Im a people person and in my own real life I am the outgoing and helpful type. And I like gathering friends the same way my wife likes to collect her rocks...  I mean crystals lol. 
Cant ever have too many friends.