4 minutes ago, Sethos said:

That is true, I'd probably have to back off a bit but I honestly don't know what gets you the best performance, tighter timings vs 200MT/s extra.

It's something I'd be curious enough to test. I made a video a while back on the objective improvements in performance with EXPO vs non-EXPO. 

It's time consuming to do these type of tests though. I might just wait and see until 2024 releases and then do a 32GB vs 64GB test instead. 

@Cpt_Piett Hmm, bizarre. When looking at the forum overview, you're supposedly the last person to reply;

But as I'm writing this, I can't see your actual reply in the thread;

Yet as you can see at the bottom, forum is feeding me "1+ new reply" messages

But I can see the reply in your profile 🤔

Anyways, just manually quoting your reply;

Quote

It's something I'd be curious enough to test. I made a video a while back on the objective improvements in performance with EXPO vs non-EXPO. It's time consuming to do these type of tests though. I might just wait and see until 2024 releases and then do a 32GB vs 64GB test instead.

 

Would be really interesting to see, would love to see a 32 vs 64GB at least, especially when there's enough addons released to truly test the sim. Though this is one area where we really need a static in-engine benchmark with a proper result screen, so you can do 1:1 tests. Testing something like RAM timings and latency, we might be talking small differences that will just end up getting drowned in the white noise that is run-to-run variance between a non-static setup.

EDIT: And as I post this, your post shows up on a previously invisible page 18 😅

 

Edited October 1, 20241 yr by Sethos

I am getting errors in MemTest5 on every setting I have tried except for the default one on my first run.  I have tried absolut, Extreme and am just got done running DDR5 Ryzen3D and received 3 errors in Test 2 on that.

I haven't had any issues in Windows and ran MSFS for a 3 hour flight yesterday without issue.

Does anyone know what "Error in Test 2" means?  I found the following on overclock.net on a post about MemTest5 but it says it applies to default MT.cfg (no longer exists) and 1usmus_v3.cfg (have yet to try).  

Quote

Veii and KedarWolf think the default MT.cfg and 1usmus_v3.cfg tests might match these error numbers but the other configurations might not, so you might have to do some homework
Here is his post

Error 0 refresh stable error = voltage cutoff choke

Error 1=Simple Test 16mb ~ can be voltage related, can be tRFC issues,
tiny timeout issues for example tRRD_L looks like a bit awkward

Error 2, 12 is a timeout issue, somewhere something ends too quickly or you lack voltage and cells are not recharged in time
a sync issue with other words which's first culprit is voltage somewhere or resistance somewhere

Error 3 & 4 by checking the MT.cfg - are MirrorMove errors
That set shows tRFC 2 issues and this tRFC "auto predicted" ? is wrong.
tRFC is so far always even
as tRFC stepping are 32,16,8,4,2 ,

Error 5 then 6 is a timings missmatch between dimms (data mirror move)

Error 6 is purely related to the IMC , be if procODT, CLDO_VDDP or vSOC
~ it translates to "i couldn't even start transfering data, i crashed"
4-6x error 6 result in full bluescreen

Error 7, 11 are burst tests

• it will error out if if CAD_BUS is not optimal
• will error out if tRFC is too low
• mostly errors out only after time

Error 9 burst test 4mb is a voltage stability issue
If you've lowered tRP , increase vDimm a tiny bit
If you've increased tRP to longer delay, decrease vDIMM +0.01 , one tiny step

Error 10 mostly affects the first 5 main timings
- noticed it can be tRCDWR to RD. can be tRP too, but it also can be the last two tRDWR & tRDRD which don't play well with your main tRCDWR/RD

Error 13 = Simple Test 64mb, timeout while transfering big data
= full crash, nearly always related to voltage ~ as memory was not able to autocorrect it

Error 14 we know is MirrorMove Omb, a timeout issue
- it can error after the 2nd or 3rd pass if something is off my some ns and just "got lost"

 

4 hours ago, Sethos said:

That is true, I'd probably have to back off a bit but I honestly don't know what gets you the best performance, tighter timings vs 200MT/s extra.

I did go to 6200MT/s and FCLK at 2066.7MHz and went back to tCL at 30 but that gave me an error. So I'd have to loosen the timings a bit and I'm not sure I want to go down that rabbit hole.

Now I'm playing a bit with upping the tREFI and lowering the tRFC, that'll give a nice boost on DDR5 but if not, I'm just going back to what I had and locking it down. Don't want spend this much time chasing minuscule gains.

Timings look good. From what i have been seeing on a few gaming benchmarks  the increased memory speed may give a small overall perf increase ....... at the expense of 1% lows. I'm leaning going towards the tighter timings route. I want better 1% lows.

What i worry about is sending 1.4 volts on my 1.35 CL30 sticks to get to CL28 I'm sure it will post but im wary about long term issues. But if i don't try i wont know i guess and i have noted 1.4 seems to be the default voltage on a good few of the DDR5 sticks nowadays.

1 hour ago, jaxgator said:

I am getting errors in MemTest5 on every setting I have tried except for the default one on my first run.  I have tried absolut, Extreme and am just got done running DDR5 Ryzen3D and received 3 errors in Test 2 on that.

I haven't had any issues in Windows and ran MSFS for a 3 hour flight yesterday without issue.

Does anyone know what "Error in Test 2" means?  I found the following on overclock.net on a post about MemTest5 but it says it applies to default MT.cfg (no longer exists) and 1usmus_v3.cfg (have yet to try).  

 

I certainly wouldn't use a machine as my daily that threw up errors. In your case, I'd either turn up the RAM voltage purely as a test to 1.4, both VDD and VDDQ or start from the top, confirm it's not another issue, make sure the baseline is clean. Start with no EXPO or just the basic EXPO profile. Does it pass? Then add some timings. Does it pass? Add the rest etc. 

Stick to anta777's profiles, the Extreme profile is a good, commonly used test and base your testing around that.

 

31 minutes ago, Maxis said:

Timings look good. From what i have been seeing on a few gaming benchmarks  the increased memory speed may give a small overall perf increase ....... at the expense of 1% lows. I'm leaning going towards the tighter timings route. I want better 1% lows.

What i worry about is sending 1.4 volts on my 1.35 CL30 sticks to get to CL28 I'm sure it will post but im wary about long term issues. But if i don't try i wont know i guess and i have noted 1.4 seems to be the default voltage on a good few of the DDR5 sticks nowadays.

Yeah, seeing different input on what has the best effect. 1% lows are definitely the most important improvement in my book.

I honestly wouldn't worry. You got Hynix A-die sticks, just like mine. Corsair and some other manufactures commonly use 1.4v on these so they can absolutely take it. Seen people use 1.45v as their daily voltage, though I would say that is pushing it. You could at least run it as a test, see if you'd overcome the timing constraint, or just move up to 1.36, 1.37 etc.

 

I'm currently messing with tRFC and tREFI. People are saying it's the single most important timing in terms of performance, one that will make the most difference. Decreased tRFC a bit, increased tREFI to its maximum of 65535, I also decreased voltage to 1.37 (error at 1.36). Currently 2 hours into an Anta777 Extreme and still no errors. So I might have found a great combo, still maintaining tCL at 28. Though it is hitting 58.5c on the first DIMM, so it's right on the edge;

Jest upgrade ready for MSFS2026.😁

Go big with 128 Gigabytes of DDR5 RAM for its lowest-ever price | Tom's Hardware (tomshardware.com)

20 minutes ago, Sethos said:

I certainly wouldn't use a machine as my daily that threw up errors. In your case, I'd either turn up the RAM voltage purely as a test to 1.4, both VDD and VDDQ or start from the top, confirm it's not another issue, make sure the baseline is clean. Start with no EXPO or just the basic EXPO profile. Does it pass? Then add some timings. Does it pass? Add the rest etc. 

Stick to anta777's profiles, the Extreme profile is a good, commonly used test and base your testing around that.

Yeah, I'm thinking my case doesn't have enough fans and this is a heat issue.  I just reset to use only EXPO profile and am running new tests.  So far, so good.  Thanks for the suggestion on resetting to EXPO profile, test, adjust a few timings, test, etc.

2 hours ago, Sethos said:

But as I'm writing this, I can't see your actual reply in the thread;

Hmmm, that is indeed strange. I did reply earlier today at work. But it seems you got it. 

2 hours ago, Sethos said:

Would be really interesting to see, would love to see a 32 vs 64GB at least, especially when there's enough addons released to truly test the sim. Though this is one area where we really need a static in-engine benchmark with a proper result screen, so you can do 1:1 tests. Testing something like RAM timings and latency, we might be talking small differences that will just end up getting drowned in the white noise that is run-to-run variance between a non-static setup.

EDIT: And as I post this, your post shows up on a previously invisible page 18 😅

Haha 😅 Can't really explain that! But, anyway, if there's any interest for it, you're welcome to join in on a very techy RAM video post-release 🙂 

I've been using my own custom test scenario for a while now - but a standardized benchmark would be very useful. 

2 hours ago, Cpt_Piett said:

I've been using my own custom test scenario for a while now - but a standardized benchmark would be very useful. 

Yes it would.  I have tested MSFS in the past, and used a totally bone-stock sim (i.e. empty Community folder), stock Ultra settings, and stock Cessna172 G1000.   Clear weather preset to take wx out of the equation.  At stock KJFK.  At the monitor's native resolution.

I think that's about as plain-jane as you can get unless I'm missing something.

MS/Asobo could also implement a pre-programmed benchmark in the game , too, and I believe that's what you are advocating in your post.  That would indeed really help standardize things for everyone, and be a great way to compare and contrast the results of hardware changes.

I switched from 32GB 7200MHz to 96GB 6400MHz @ 6600MHz and completed a flight around KLAX with 50% AIG traffic and sceneries (IniBuilds, etc...)... Using 32GB the memory used was around 28~29GB using the X Cub... as you can see in the screenshot, with 96GB the memory use is higher with less memory paging. The icing on the cake is that I don't get the stutters I had with 32GB...

BTW... I don't have other apps open like FSRealistic, SimHaptic, etc...

Does anyone know an app that can fix the screenshot with HDR?

 

Edited October 2, 20241 yr by Dexter1979

@Dexter1979Use Nvidia screenshots.  It's either that or the game bar.  But when I do a screenshot it saves one in JXR (HDR) and another in SDR format. 

3 hours ago, Dexter1979 said:

Using 32GB the memory used was around 28~29GB using the X Cub...

3 hours ago, Dexter1979 said:

The icing on the cake is that I don't get the stutters I had with 32GB...

I don't get how you're seeing 28-29Gb with that plane even in that area.   I never see more than 18Gb anywhere.  I just flew out of KATL this morning in the FBW 320nx and 15.3Gb was it.  I'm using a 1440x3440 screen but still the difference is massive.  I'm running Self Loading Cargo, A Pilot's Life, GSX, FSLTL and frequently have Edge browser open.

 

So I was getting very smooth performance with 64 GB RAM on Windows 11 23H2. However, after updating to 24H2, I am seeing the same stutters I had back when I had 32 GB RAM.

Anyone else experience this?

Edited October 2, 20241 yr by RNAVV19R

1 hour ago, Noel said:

I don't get how you're seeing 28-29Gb with that plane even in that area.   I never see more than 18Gb anywhere.  I just flew out of KATL this morning in the FBW 320nx and 15.3Gb was it.  I'm using a 1440x3440 screen but still the difference is massive.  I'm running Self Loading Cargo, A Pilot's Life, GSX, FSLTL and frequently have Edge browser open.

 

His image is showing entire system memory usage because he is on the Performance tab.  A better depiction of how much memory MSFS itself is consuming is to use the Processes tab and then click the Memory column to sort descending as shown here.  Don't be fooled by my low memory usage in MSFS as I was only on the main page.

Edited October 2, 20241 yr by jaxgator

Quote

I don't get how you're seeing 28-29Gb with that plane even in that area.   I never see more than 18Gb anywhere.  I just flew out of KATL this morning in the FBW 320nx and 15.3Gb was it.  I'm using a 1440x3440 screen but still the difference is massive.  I'm running Self Loading Cargo, A Pilot's Life, GSX, FSLTL and frequently have Edge browser open.

The reason might be the size of the community and official folder in MSFS... I don't use a scenery manager... I'm too lazy for that...