Are RAM Timings THAT Important In Star Citizen?

32GB of 6000Mhz CL30 RAM vs CL36 on Ryzen 7000 CPU's! Is there much difference in performance in Star Citizen? One of the kits has Samsung memory and the other has SK Hynix- can you squeeze more fps out of them by manually tuning?
🟣5% OFF PREDATOR MOUNTS LINK: www.predatormo...
(any order will support the channel)
US Parts To Buy (Amazon Affiliate Link): www.amazon.com...
UK Parts SCAN (Affiliate): tidd.ly/3TdYCJz
DISCORD: / discord
Become a member to support the channel:
/ @tenpoundfortytwo
New to Star Citizen? Use a referral code for some extra credits: robertsspacein...
Donate to directly to support the channel: paypal.me/tenp...
Test Setup
Star Citizen Alpha 3.22
Resolution 1080p
Main System
CPU: 7800X3D
GPU: RTX 4080
RAM: 32GB 6000MHZ CL36 VS 32GB 6000MHZ CL30
#StarCitizen​​ #SpaceSim​​ #PCGaming​ #Squadron42​ #S42

[@Burty117]

It's worth mentioning, if you play more than just Star Citizen, you can see more performance improvements in other games, Battlefield or really any game running on the Frostbite engine can see considerable gains with the better timings.

[@tenpoundfortytwo]

Yep very good point!

[@ssilver88]

Thanks for another great video. I had been meaning to give Buildzoid's timings a try since I swapped to DDR5 and your vid finally gave me the push to do it. Picked up about 10fps across the board compared to stock 6000cl30 expo timings testing in Area 18 at 1080p with 7800x3D, 32gb Hynix A die using Buildzoids tuned timings and rtx3070

[@JSmith73]

Nice, I'm going to try the same!

[@ubermk3]

Thanks as always for your rigorous testing to helps us squeeze every bit of performance out of our hardware. I'll take an extra 8 frames in SC any day. In some areas of the game it can make a noticeable difference in smoothness.

[@tenpoundfortytwo]

No problem! Yep 8 frames is definitely not bad!

[@fnunez]

Reminder to anyone experimenting with this: if your PC starts blue-screening, game crashes on startup or any other unusual behavior, the first thing you need to try is to put all your settings back to stock. In my experience this usually fixes the problem. Then you can try less aggressive settings.

[@opreax2145]

Exactly, not like some poeple immediately blame AMD drivers.

[@B.D.E.]

AMD stability is more sensitive to RAM overclocks than Intel though.​@@opreax2145

[@cmdr_thrudd]

That's pretty cool, didn't realise the chip manufacturer would have such a difference. Thanks for doing the testing :)

[@tenpoundfortytwo]

No problem!

[@JSmith73]

Glad I got the C30s then, G.Skill, 64gb with my 7800x3d. It all adds up to the total cost, but then again years pass between each upgrade. Still have to tune it, just running one of the 6000 Expo profiles "out of the box"! (All major voltages set to manual, conservative ones). The Samsung 990 Pros were my other splurge of luxury! :-) EDIT 30/12: My 2x32GB SK Hynix modules (G.Skill f5-6000j3040g32g) are dual rank, so no can do on Buildzoids easy timings which caters for the single rank crowd only. (It'd take me a fair while to actually learn them, so will pass for now. Any pointers much welcomed!)

[@B.D.E.]

How are your 0.1% lows with that setup? Have you tried doing a proper benchmark with CapFrame?

[@JSmith73]

@@B.D.E. 0.1% lows in Star Citizen will always be pretty bad, until the new servers with higher frame rates are out! Overall it's pretty darned good, especially when compiled - and you get used to a bit of SC stutter anyway. I can't be bothered recording frame rates, with the limited time I have for entertainment I prefer to just play. (It's pretty darned good!) Nonetheless, I'll take this as a reminder to spend an hour on ram timings now that I've got a day off!

[@kaisersolo76]

Glad you made this. Its basically a 5-9% difference when tuned between the two set s of modules. Its Not really worth the extra cost unless budget is not limited. I picked up another set of cl36 for £80 so i know have 64gb.

[@commanderoof4578]

Hope you aint in AM5 because you would shoot yourself in the foot and gimp to around 5200MHz as 4 sticks and AM5 aint refined

[@kaisersolo76]

@@commanderoof4578 well i did and i've been running 6000 cl36 4x16 since i got them.

[@commanderoof4578]

@@kaisersolo76 check with zentimings that its actually running them as chances are high it did a fallback and is running lower 4 sticks at 6000 does not work 95% of the time

[@cpt.tombstone]

That 11% uplift you had with Buildzoid's timings is avout the same uplift I saw moving from a 7900X to a 7800X3D. Buildzoid's timings also work for hynix ICs at 6400 MT/s without changing anything, and that should an additional 6% improvement, if the CPU can achieve 6400 MT/s. I didn't see you mention it or showing it in the video, but what was the fclock you were using in the tests? With single CCD Ryzen chips, the infinity fabric is limiting factor to both bandwidth and latency with the memory.

[@tenpoundfortytwo]

I didnt touch fclock at all, but yeah, I'll have to take a look at 6400 for a future video!

[@jnthn.x]

Thank you for this. I've been thinking if Buildzoid's timing works for 6400Mhz. I'm running Kingston Renegade 32GB 6400Mhz on A-XMP profile 1. I was surprised that it worked right away without any problem when I enabled it. May I ask what is your VSOC? Mine currently at 1.3V from the XMP profile 1 and I'm a bit worried since a lot people saying that 1.3V is unnecessary and can degrade the cpu life faster.

[@cpt.tombstone]

@@jnthn.x I'm running 1.215V on the SoC, but I'm also not running the memory at 6400MT/s, as my CPU's memory controller cannot handle that without throwing memory errors. Although at 1.3V Vsoc, it throws a lot fewer errors, so it's close to stable at that voltage. When I had a 7900X I had to run 1.34V on it to have 6400 MT/s stable. 1.3V should be fine, but that's the higher end for x3ds for a reason. You will also see much higher idle power consumption with a higher SoC voltage. And it seems that you can set a higher fclock with a lower SoC voltage as well, which might be more beneficial at some point than running higher frequency memory. Try 1.25V and see if you have errors. If not, run it at 1.25V

[@chris.mc3b]

keep up the good work man

[@tenpoundfortytwo]

Thanks!

[@The_Opinion_of_Matt]

With DDR5 the motherboard seems to matter too. My ASUS TUF Gaming X670E-Plus WiFi seems to struggle with faster memory and timings compared to other motherboards if you look at the manufacturer's approved memory lists. I think the only thing that is really good about this motherboard is the power phases. Running a Ryzen R9 7900X with 170w TDP.

[@tenpoundfortytwo]

Good info, thanks!

[@nathanforrester5140]

thanks for the video. I normally do not do much in the way of memory tuning but it might be fun to mess around when when I upgrade my system to DDR5.

[@tenpoundfortytwo]

Yeah it's not something I normally do either to be honest, but it's a pretty decent boost in performance with the hynix stuff!

[@nathanforrester5140]

I will keep it in mind when the time comes for my new build. I am on ddr4 atm with a 5800x3d so I am in no hurry. @@tenpoundfortytwo

[@royvanempel]

Nice compact video! Some day i might have more then 30 fps in lorville. Hopefully not to far into the future. I am really considering upgrading from my 5900x to a 7000 x3d chip seeing that SC performance!

[@Blackcat70_]

something we used to understand back before the ddr days was that you never got a big bump from any upgrade, you got a bunch of lilttle bumps that added up to real performance. Now, I think people forget how important a little bump like 3 fps can be.. you add that to other 1-2% increases and THAT'S how we pwned in the old days. Little bumps are worth it.

[@VonSpud]

Great stuff... I've had this discussion with clients and colleagues... Ram GHz numbers are a bit overhyped for the actual game performance improvement out of the box. Did not know that Hynix was the top dog for Ram tuning.

[@B.D.E.]

I'd love to see the data for RAM, GPU, and CPU, focusing purely on the 0.1% and 1% lows. Because right now, no matter what hardware you have, that's what ruins the 'smoothness' of the experience.

[@corbett666]

You’ll probably find a bigger difference with intel processors, I have an i9 14900 and I’m getting well into the 100’s fps doing the standard test run on max settings with tuned ram. I can’t get anywhere close to that with my AMD rig and the difference is way smaller vs stock. I think it’s due to the cache, amd seems to rely more on the beastly cache than the ram, but SC eats the cache so quickly it needs ram.

[@luffy435]

Always love these performance based videos for SC. Currently have a 2060 Super and a Ryzen 5600x. Was thinking of upgrading my GPU to a 6800, but at the same time I don't mind waiting out another generation too! Not sure how trustworthy UK Second hand market is through Ebay.

[@tenpoundfortytwo]

For SC I think you'd probably be better picking up a 5800x3d to replace the 5600x! I've always had pretty decent luck on ebay, but there is always an element of risk.

[@Whoadayson]

Ive always wondered this. Thank you! Real quick, do you think your results would be different on AM4?

[@tenpoundfortytwo]

It will be different on am4 but you can still tune ddr4- should be plenty of videos on youtube (will be a bit of a rabbit hole!).

[@Sn1PeRZzBLOOD]

I’ve been running 32gb ddr5 Hynix A die @6000 with Buildzoid timings. However, there was one option that should not be paired with an other option cuz it would just cause bluescreens. Anyhow, next up is PBO settings tuning for cpu. :)

[@B.D.E.]

I went for 6800mhz cl32 😊

[@parkerxgps]

guessing like ddr4, we'll be talking much more once/if we get down below 20.

[@acli9704]

memtest86 - from test No6. to end is critical... pc may work for browsing but more demanding may restart or get blue error screen

[@PredatorMounts]

You're doing the Lord's work good sir. Happy Holidays!

[@tenpoundfortytwo]

Thanks! And to you!

[@festersmith8352]

I watching this out of curiosity. But I came to your channel hoping to find what your Nvidia Control Panel settings might be. I have learned to trust you, and you specialize in SC. I couldn't find a video for Nvidia settings. Please!

[@tenpoundfortytwo]

To be honest, I've never really found anything in the NVC panel that really makes much of a difference!

[@festersmith8352]

@@tenpoundfortytwo Thank you! That is good to know. I'll take your word for it, since i have come to really trust you!

[@macunknown]

Check out karhu Ram Test to verify your memory tunings stability

[@ImreBertalan86]

Woa, woa, woa! What about CL30 vs 36 with a bottlenecked PC? Both either CPU, or GPU side, or both...

[@tenpoundfortytwo]

Well these results were at 1080p so we were cpu/ram/engine bottlenecked. If you were gpu bottlenecked there might still be scenarios where you'd see some benefit from tuned ram, but it wouldn't be as noticeable.

[@ImreBertalan86]

@@tenpoundfortytwo Cheers for the reply. I'm wondering still, how would a weaker CPU or a weaker GPU deal with the better latency timed RAM. Are the differences constant along all the "low-end, mid-range, high-end" CPUs, or low-end CPU-s benefit more or less from a better CL RAM? That is my question. But as you've said, it, in 1080p, we are almost bottlenecked in every possible way. But can you (pretty) please, check just out of my own, selfish curiosity if for example, a Ryzen 3600 benefits, the same as the 7800X3D? For testing, I would compare the "G.SKILL Ripjaws V DDR4 3200MHz CL14 16GB Kit2 (2x8) " with the "G.SKILL Ripjaws 16GB (2x8GB) DDR4 3200MHz F4-3200C22D-16GRS " as they are the same brand, same dedicated speed and both are dual-kit. But one is LC14 and the other is CL22. This is a fascinating test that I've never knew I want to know 😀

[@BboyCorrosive]

I can only assume 10pound works at some pc building place 😊

[@tenpoundfortytwo]

If only! I buy things and then try to sell them on without losing too much money!

[@Pernicuz]

i dont mind the low fps at some places so much, most of th etime the low fps is at cities not where fps is vital. what i don tlike is the "lag" when loading and jumping from planets. how can i get rid of that if possible.

[@Kevin-fl7mj]

You can't get rid of it completely,you can minimize it by tuning your Ram,get the fastest SSD available,keeping your shaders reasonably clean etc.

[@Pernicuz]

@@Kevin-fl7mj ok, yea my ssd and ram is not close to the fastest. So it will be my next upgrade. Thanks.

[@fatDOX8]

i suspect gains would be higher for chips with less cache, did u test it with any other CPU?

[@lozm8308]

I know this question is off video topic but which is more efficient for star citizen a 7800 xt or a Rtx 4070?

[@tenpoundfortytwo]

Efficient in terms of performance or power?

[@bad_token]

I love this

[@TheJudgeo7]

Big question here: is 32gb enough or should I go 64gb? Building a PC to handle super ultrawide with over4k res (with either 5090 or next gen AMD card and 7800x3d for now)

[@tenpoundfortytwo]

If you're going to be playing the game with lots of stuff running in the background then there might be a case for 64GB- but for most people I'd recommend 32GB for now! I will eventually test 32 vs 64 when prices come down a bit!

[@TheJudgeo7]

@@tenpoundfortytwo thanks a lot mate, much appreciated!

[@TheJudgeo7]

Hey mate, have a look at that - guy took Ollie’s footage with optimized RAM, up to 80% faster in SC! kzbin.info/www/bejne/f3Waln-dZd2Ghcksi=xkJ4lwX8nJWaL95S

[@terry7342]

Do you have similar results with intel? Considering jumping up from 4800mhz to 6000, or is there some higher that is viable? Still rocking i9-12700k

[@tenpoundfortytwo]

Unfortunately I don't have my intel ddr5 motherboard anymore. But from all reports, the 12700k should be perform quite a lot better with faster ram. You'll have to look into your motherboard, but you should be able to go faster than 6000 from what I understand.

[@terry7342]

@@tenpoundfortytwo Thanks, it’s the little bit that counts in some cases. But a new PC every 5 years is also a thing haha. Great content as always 42.

[@Tainted-Soul]

cool If i was brave enough I may but like you say when you tune / change setting it can also add a stability issues too. On a different subject LOL but still on FPS with SC when traveling in QT when in the pilots seat in a Vulture ( I said the ship because it may make a difference ) I was getting 157fps and my GPU was running at 85% ish then I pressed F4 and my FPS shot up to over 215FPS !!!!! and my GPU usage went down to 70% ish. SO why would if make the much difference inside the ship in Frist person to outside the ship ? is it just the UI ? and is it different between ships LOL this is using the R_display 2 in game but they change. hope you had a great Christmas

[@tenpoundfortytwo]

Thanks yeah I had a great Chirstmas, hope you did too! I think it's all of the ships displays- I think they are pretty heavy performance wise. You could try just turning the power on and off in a hangar and see how much performance changes. Its a strange situation where different ships with different HUDs perform differently. Hopefully CIG will optimise the screens at some point.

[@Tainted-Soul]

@@tenpoundfortytwo yes I had a great Christmas thanks, I would have thought a small difference but over 60 fps difference is like is this the big FPS problem we get in combat ? I do think its a % more than raw fps but when your doing a bounty and your fps is like 60 this could give you an extra 20 or so

[@smokeclouds8]

My new unopened kit is G.skill flare X5 (2x24gb) ddr5 6000MT/s CL40 Is this a good one for star citizen? I see a lot of CL30, is that better?

[@tenpoundfortytwo]

Cl30 would be better, but we're not taking a lot. Up to you whether it's worth it!

[@mlwg9317]

I have a 12700k and z690 UnifyX. I tried a 7200 Hynix A and managed 7400, and the latency on latency mon was amazing...8-12 sitting idle...using a 8200 48gb Hynix M die I could run at 7800 and had 20-40 idle at desktop. Hynix A die, 7800 is coming tomorrow...going to see what I can hit with it and the idles with Latency Monitor. Anyone else getting higher latency with the Hynix M die? EDIT: I ended up switching to the 7800 A-Die, it ran stable at 7600 and the latency was 12-14

[@rinkaemina]

Tuning ram is scary somehow, my best just fast ram xmp l'm just casual

[@Nommicus]

Ohh posted on your 16 vs 32 vid from 3 years ago, then fiund this. Illmpost again and apologise for the cut and paste: Cheers for this got 32 of CORSAIR VENGEANCE DDR5 RAM 32GB (2x16GB) 4800MHz CL40 Intel XMP iCUE Compatible Computer Memory - Black (CMK32GX5M2A4800C40) Comeing later to replace my 16 of ddr4. My question is: I have a GeForce GTX 2070 super. I can't afford to replace the GPU atm but am hoping the new ram will still help, how much of a bottle neck could this cause and is it worth it? Other than futurproofing for when I do upgrade my GPU.

[@tenpoundfortytwo]

Hey, that's a fairly slow ddr5 kit, what CPU have you got? 2070 is still a good GPU, should be good for a while yet!

[@Nommicus]

@@tenpoundfortytwo I replied but it went missing, I made a mistake. My m-board does not support that ram config is DDR 5. I had to Google and used manufacturers recommended and went with Kingston FURY Beast 32GB (2x16GB) 2666MHz DDR4 CL16 Desktop Memory Kit of 2 KF426C16BB1K2/32 Cost 2 quid less on Amazon... I think original post may have been deleted as posted Amazon page link? Still I apologize for posting twice mate. It was expensive porchase for me atm lol

[@sphaera3809]

I reverted my memory back to default EXPO settings… I think the gains are not that evident, and actually might cause some extra stutter and instability in SC (hard to be conclusive given it’s alpha nature and differences from one server to another). But my SC game play feels smoother without any custom tuning (7800X3D+4080). Also, for AMD CPUs, anything above 6000 MHz is a waste.

[@fnunez]

I also noticed this (7800X3D and 32GB of DDR5-6000 using EXPO). The average FPS can go up a little bit when you mess with the timings, but it's less smooth overall.

[@vulcan4d]

7700x + tight timings = 7800x3d + stock timings

[@echo5827]

Definitely not

[@MehlstaubtheCat]

Never! :D

[@zeroxception]

not worth the effort/time/cost unless min/maxing is your hobby

[@jackwood6214]

Yes lol

[@SP95]

Man just ruins the sole purpose of gamedaytoday1's channel 😂 That is not nice

[@tenpoundfortytwo]

Well I don't normally dip into tuning- it's a bit too in depth for most people!

[@commanderoof4578]

Didn't ruin anything... In fact what they did was either nothing or pushed people towards looking into actually sitting down and tuning memory XMP is a very rubbish standard that changes the bear minimum of values and leaving it as is works but in games like star citizen it kneecaps performance Ive seen some AM4 motherboards for example... set tFAW to flipping 40 when it has a fixed value of 16 anything higher is wrong because tRRDS should always be 4 and such for DDR4... unsure for DDR5 You can even have tRDRDSCL which doesn't affect stability it just changes the wait time between doing 1 thing and another and will either boot or it wont... and should be 2 or 4 for DDR4 or 4, 6, 8 for DDR5 depending what boots