I contemplated that, but the 6400CL30 kits are basically the same price as the 6000s but I agree most people are probably getting 6000 kits due to that coming from AMD
@pyurologie when I bought my 9800x3d on newegg I had to get it in a bundle and it came with a 6000 32 gig kit. I'm sure a lot of people got the same thing.
@@pyurologie "...if memory speeds exceed 3000MHz (DDR5-6000), then the memory controller will fall to 1:2 with the memory frequency, which will incur a performance hit." those 400 extra Hz could lower the performance.
same secondary and tertiary timings from my 9800X3D tuning video, watch from this timestamp: ua-cam.com/video/g8uEffp5HfM/v-deo.html. it's just that these sticks are higher bin, so they can hit better ratings
@@pyurologieNot trying to be rude but 26 48 48 30 is hardly tightened timings.. Yeah you tightened the tCL (26) and tRAS (30) but your tRCD and tRP (48) are worse than a tuned 8000 kit at 34 42 42 44. Idk if you tried to copy BuildZoids timings but he doesn’t optimize his timings for gaming. You are leaving a bunch of latency and FPS on the table by running your ram like that and running your FCLK at 2200 decoupled from the memory.. But you do you 👍🏻
Would be a godsend if you made it for rust. Also if you had a cl30 6000 kit lying around that would be truly epic to include as thats usually the gold standard recommended for these newer ryzens.
Think I mentioned this in a previous video, Latency is a big part - just because it's fast doesn't mean it's the best, now finding a low Latency with tight timings and a solid speed (only thing is these kits tend to not come cheap, but worth imo).
I have a question and may a test suggestion. does your 32GB G.SKILL Trident Z5 Neo (6400 CL30) memory kit have several Expo memory presets? I know my 6400 CL34 set does. If so it would be interesting to see how smaller changes after the performance in EFT. If it is a change in just the frequency or just CL or whatever the EXPO options might be> Just a thought and great work on all of the testing so far.
Thanks for the great work. I bet that a kit of 6000 cl 30 would be so close to your 6400 kit and can be found for Under $100. Also depending on the Mobo some can't run above 6000. I understand you test what you have and still great work.
when i looked em up, 6000 rated sticks were close in price to 6400. at under $100 would be worth a look. also, motherboard memory compatibility rating accuracy can vary because they want to be "safe" on what they guarantee, but even my $85 A620 hit 6400 and 7200 with my ryzen 7000s ua-cam.com/video/2dcdaHmlpnU/v-deo.html
@@pyurologie they charge often I have 6400 clocked to 6000, my Mother board as kinda not the best and 7600 non X. The 6400 was cheaper than 6000 when I built
I ended up going with the neo royal cl28 6000mhz, and I will overclock it to 6400mhz by upping the mclk=uclk worst case, I do hope it's as good as the actual 6400mhz cl30 kit is all.
Hey. Did you tune your ram? I’ve been contemplating whether to get the G.Skill Royal 6000 CL28 or the 8000 CL38. If you managed to tune your 6000 CL28 then I might as well get that instead. There is also a 6400 CL30 option as well for the royal series.
@@nikhilnischay3955 I will be honest, I have no tuned it yet since I've been waiting on parts for awhile since I live in Canada. I am planning just to do the expo profile probably or maybe going for 2133 or 2200 and overclock that way for mclk and flck but, the fps difference isin't huge. I am sort of regretting not getting the 64gb ram sticks that were on sale since after this tarkov update, it seems 64gb is better I am not sure. Maybe if you have enough vram it's fine but, I have not tested it with my new cpu yet though.
Hi, speaking only about performance, what would be the best DDR5? I think 7200mhz 34? And is it definitely worth the jump from DDR4? Thank you very much, your channel is amazing!
Thats what i just recommended to someone on here. Decoupling the infinity fabric from the memory is t worth it for gaming, loses way too much performance. I wish the 9000 series had a good memory controller like 8700G. Mine is @ 7400/2466 but they didnt upgrade the I/O die so the 9800X3D is stuck a 6600 max
Seems like the fps increases are within margin of error at times. And even when it's outside of margin of error results I bet 90%+ of people wouldn't be able to tell the difference between 180fps vs 170fps.
yeah for sure and given the price differences between high end and basic RAM, it's good to know you're not losing out on much if you just go basic. I also did previous secondary/tertiary timing testing that showed tightening up on lower rated RAM can be even better than buying faster RAM for tarkov ua-cam.com/video/PbK0bncyh6w/v-deo.html
That is tuned... Getting lower than 8000 CL35 is extremely hard to keep stable. There isnt any point in going over 6600 CL30/2200 fclk. Ryzen cant go passed 2200fclk (DDR5 6600) and decoupling the infinity fabric loses performance, that's the reason 6400 is getting more FPS than 8000.
@@JoeLegionTV 1:1 is decoupled from the infinity fabric anyway so that's not why. The memory clock and memory controller are coupled in 1:1, but in 2:1 there is somewhat of a latency benefit to running 8000 because you can synchronize the infinity fabric at 2000. With a single CCD chip you're bandwidth limited anyway and tuned 6400 and 8000 are somewhat comparable depending on the workload, but it's really not worth the effort messing around with mem oc too much on a 9800x3d. Tuned 6400 cl26 and tuned 8000 cl38 on my 9800x3d are pretty comparable for gaming but the 8000 setup is actually lower latency and super stable so it's been fun running it that way.
@@ZWortek AM5 can run async FCLK no problem, but yeah the cost of 8000 vs 6400 or even 6000 isn't worth it unless you just like pushing stuff or messing with memory/oc
@@ZWortekI think you misunderstood what I said. I’ll start with this, there have been many tests done on “popular” channels that prove 8000 on Ryzen has a huge latency and fps penalty, not gain. Also decoupling the FCLK from the mCLK does not yield a large latency improvement. Running 6400/2133 FCLK has better latency than 6400/2200 FCLK. If you can’t divide the memory speed and FCLK by 3 to get known speeds like 6400/2133 6000/2000 7400/2466 etc.. it greatly increases latency and performance penalties. AM5s memory controllers have to be able to divide by 3 (which is gear 2) if you go into gear 4 like his 8000 test was, it ruins performance hence why it didn’t get more fps when the test wasn’t limited by the GPU.
how do u get 6400 cl26? i have 2 2x16 kits (64gb total) rated for 6000 cl30 and it does that just fine when i enable xmp (7950x3d, 4080 super; playing at 4k)
You're very likely not going to run 4x16 (waaayyyy worse) or 2x32 stable at 6400 cl26 on a 7950x3d so I wouldn't worry about it. With 4x16gb sticks your memory controller is fighting for it's life to manage any kind of stability past 6000 CL30, and even that is unstable for some people. Tune your 6000 CL30 timings as much as you can and call it a day if you care but you're going to be limited in mem oc past 2x16gb. 4 dimm slots is a design flaw in the words of the great BZ.
@@carlosferjezus I have the same kit with 14700kf and I can run it 6400 30 37 37 57 1.4v. I think it’s pretty good for a dualrank kit, but you must know that they don’t have good radiators. I reach 70 degrees and more :/
Just ordered a 9800x3d with 64gb DDR5 of Kingston Fury Beast 6400MHz (couldnt find CL rating) hope it all works well together cause i dont know jack squat about tuning ram speeds or curves😂
I built my new PC today, 9800x3D with the exact same RAM as you, it’s CL32. I only drop below 100fps on Streets when scoping in (80 was the minimum I saw near Pinwood hotel which is one of the lowest fps areas on the map, but otherwise running around the map I was anywhere within 120-200 range, 90-110 when scoped. It’s so smooth, not a single stutter in 3 hours. Blows my old build away.
Hi, I had a question. I basically follow ur 9800x3d tune guide, and im getting 130-170fps on streets, but when I try to run customs, I only got 80-100. is it cause customs has bad optimization? or is it my problem.
@@pyurologie 8000Mhz would perfom better with 2000 FCLK (4:2:2). - 4000 MCLK : 2000 UCLK : 2000 FCLK for better latency. But yeah 1:1 still is better for Tarkov, ty for the benchmarks!
I WENT WITH AMD Ryzen 7 7800X3D 4.20GHz до 5.00GHz tray × 1 Corsair Dominator Platinum RGB 32GB 6000MHz DDR5 CL36 × 1 Thermalright Frozen Prism 240 Black A-RGB LGA1851/LGA1700/AM5 × 1 Asrock B650M Pro RS AM5 I DIDN'T ASSEMBLE IT, I GAVE IT TO SPECIALIST, BUT THEY SAID THAT, THE RAM IS 6000MHZ AND THE PROCESSOR CAN'T WORK WITH IT, THE PC WON'T START OR SOMETHING, THEY TRY TO NURF THR RAM SPEED BUT THEY CAN'T, SOO WHAT DO YOU THINK, WHERE IS THE PROBLEM? They said 5600mhz need to be, high wont run they think. PS. They said also, they will try to update the bios to.
Hi can anyone help me please I think I do something really dumb I didn't konw when you use the ryzen master per core negative 20 core 1 to core 8 now I won't boot showing red light (Dram) and yellow light (CPU)
@@berserkbasshead7618 32 vs 64 on 7800X3D here too: ua-cam.com/video/lJVbaA66J_o/v-deo.html. I'm assuming results will be similar, but as usual I'm gonna add 9800X3D tests for it soon. stay tuned!
hi pyur i have 78x3d and hynix a die 32x2 ram, i tried ram over but 2nd, 3rd timing is so complicated. 2,3 timing is important in tarkov? many improvement in 1% low? or not big deal? ty for u r comparison video
I think there's a missed opportunity of having a 6000 CL30 kit since that's the recommended spec according to AMD.
I contemplated that, but the 6400CL30 kits are basically the same price as the 6000s but I agree most people are probably getting 6000 kits due to that coming from AMD
@pyurologie when I bought my 9800x3d on newegg I had to get it in a bundle and it came with a 6000 32 gig kit. I'm sure a lot of people got the same thing.
What Memory ratio was the 6400mhz ram at? You should do 6000 1:1...
@@pyurologie "...if memory speeds exceed 3000MHz (DDR5-6000), then the memory controller will fall to 1:2 with the memory frequency, which will incur a performance hit." those 400 extra Hz could lower the performance.
Some cpus can do 3200 others can only do 3000. Before you have to switch to 1:1
this was helpful! picked up a 9800x3d yesterday and went with dual 32gig 6400 CL32 gskill
congrats! good kit too!
Damn, 6400 CL 26? Would love to see a guide for that / your settings.
@@Dante_S550_Turbo nope, all tightened up 26-48-48-30 with tightened tertiaries, cmon dude
same secondary and tertiary timings from my 9800X3D tuning video, watch from this timestamp: ua-cam.com/video/g8uEffp5HfM/v-deo.html. it's just that these sticks are higher bin, so they can hit better ratings
@@pyurologieNot trying to be rude but 26 48 48 30 is hardly tightened timings.. Yeah you tightened the tCL (26) and tRAS (30) but your tRCD and tRP (48) are worse than a tuned 8000 kit at 34 42 42 44. Idk if you tried to copy BuildZoids timings but he doesn’t optimize his timings for gaming. You are leaving a bunch of latency and FPS on the table by running your ram like that and running your FCLK at 2200 decoupled from the memory.. But you do you 👍🏻
Would be a godsend if you made it for rust. Also if you had a cl30 6000 kit lying around that would be truly epic to include as thats usually the gold standard recommended for these newer ryzens.
ok a bunch of people have been asking for rust, i'll put one together
@@pyurologiedid you do one for rust?
tnx for the vid. great work as usual :)
My pleasure! thank you!
Think I mentioned this in a previous video, Latency is a big part - just because it's fast doesn't mean it's the best, now finding a low Latency with tight timings and a solid speed (only thing is these kits tend to not come cheap, but worth imo).
Thanks for the vid, Pyur
no problem, thanks for checking it out!
great video thanks for doing this!
you're welcome, thanks for checking it out!
Nice comparison as usual, didn’t expect such a noticeable gap for non 4k streets configs
yeah in 1440, the differences showed the most i feel like
Nice video it would be nice to test those on a 9700X to see how it react compared to a 9800x3d!
i didn't grab any 9000s unfortunately, but I do still have my 7700X
I have a question and may a test suggestion. does your 32GB G.SKILL Trident Z5 Neo (6400 CL30) memory kit have several Expo memory presets? I know my 6400 CL34 set does. If so it would be interesting to see how smaller changes after the performance in EFT. If it is a change in just the frequency or just CL or whatever the EXPO options might be>
Just a thought and great work on all of the testing so far.
Thanks for the great work. I bet that a kit of 6000 cl 30 would be so close to your 6400 kit and can be found for Under $100. Also depending on the Mobo some can't run above 6000. I understand you test what you have and still great work.
when i looked em up, 6000 rated sticks were close in price to 6400. at under $100 would be worth a look. also, motherboard memory compatibility rating accuracy can vary because they want to be "safe" on what they guarantee, but even my $85 A620 hit 6400 and 7200 with my ryzen 7000s ua-cam.com/video/2dcdaHmlpnU/v-deo.html
@@pyurologie they charge often I have 6400 clocked to 6000, my Mother board as kinda not the best and 7600 non X. The 6400 was cheaper than 6000 when I built
Also I wonder if the CL matters much once you are in the 6000's with a X 3D processor
I ended up going with the neo royal cl28 6000mhz, and I will overclock it to 6400mhz by upping the mclk=uclk worst case, I do hope it's as good as the actual 6400mhz cl30 kit is all.
that's a nice kit!
Hey. Did you tune your ram? I’ve been contemplating whether to get the G.Skill Royal 6000 CL28 or the 8000 CL38. If you managed to tune your 6000 CL28 then I might as well get that instead. There is also a 6400 CL30 option as well for the royal series.
@@nikhilnischay3955 I will be honest, I have no tuned it yet since I've been waiting on parts for awhile since I live in Canada. I am planning just to do the expo profile probably or maybe going for 2133 or 2200 and overclock that way for mclk and flck but, the fps difference isin't huge.
I am sort of regretting not getting the 64gb ram sticks that were on sale since after this tarkov update, it seems 64gb is better I am not sure. Maybe if you have enough vram it's fine but, I have not tested it with my new cpu yet though.
Hi, speaking only about performance, what would be the best DDR5? I think 7200mhz 34? And is it definitely worth the jump from DDR4? Thank you very much, your channel is amazing!
I have klev 6000 cl 26 nice load in Times and pretty consistent fps very happy with the choice game has been much nicer to play
very nice kit!
Is the cl26 6400 running at 1:1 ? How you achieve that oc ? Also there should be soon cl26 sticks from G Skill and Lexar but they are only 6000cl26.
yeah they are 8000cl38 rated sticks, tuned to 6400cl26 1:1
hello . I can t find 6400 mhz cl 26 in romania. Whats the difference betwen 6400 cl 26 and 6000mhz cl 30?
i had no idea the 6400 existed. i bought regular ass 6000 cl30. how does that compare"?
Literally the same 😂
gonna be very similar in actual performance dont worry
im runing 6600mhz CL30 fclk 2200mhz
Thats what i just recommended to someone on here. Decoupling the infinity fabric from the memory is t worth it for gaming, loses way too much performance. I wish the 9000 series had a good memory controller like 8700G. Mine is @ 7400/2466 but they didnt upgrade the I/O die so the 9800X3D is stuck a 6600 max
nice settings!
Seems like the fps increases are within margin of error at times. And even when it's outside of margin of error results I bet 90%+ of people wouldn't be able to tell the difference between 180fps vs 170fps.
yeah for sure and given the price differences between high end and basic RAM, it's good to know you're not losing out on much if you just go basic. I also did previous secondary/tertiary timing testing that showed tightening up on lower rated RAM can be even better than buying faster RAM for tarkov ua-cam.com/video/PbK0bncyh6w/v-deo.html
Can you tune the 8000 setup? like cl34 or even cl32 secondaries and tertiaries for savage users to see if its worth it?
That is tuned... Getting lower than 8000 CL35 is extremely hard to keep stable. There isnt any point in going over 6600 CL30/2200 fclk. Ryzen cant go passed 2200fclk (DDR5 6600) and decoupling the infinity fabric loses performance, that's the reason 6400 is getting more FPS than 8000.
@@JoeLegionTV 1:1 is decoupled from the infinity fabric anyway so that's not why. The memory clock and memory controller are coupled in 1:1, but in 2:1 there is somewhat of a latency benefit to running 8000 because you can synchronize the infinity fabric at 2000. With a single CCD chip you're bandwidth limited anyway and tuned 6400 and 8000 are somewhat comparable depending on the workload, but it's really not worth the effort messing around with mem oc too much on a 9800x3d. Tuned 6400 cl26 and tuned 8000 cl38 on my 9800x3d are pretty comparable for gaming but the 8000 setup is actually lower latency and super stable so it's been fun running it that way.
@@ZWortek AM5 can run async FCLK no problem, but yeah the cost of 8000 vs 6400 or even 6000 isn't worth it unless you just like pushing stuff or messing with memory/oc
@@pyurologie 100% agree
@@ZWortekI think you misunderstood what I said. I’ll start with this, there have been many tests done on “popular” channels that prove 8000 on Ryzen has a huge latency and fps penalty, not gain. Also decoupling the FCLK from the mCLK does not yield a large latency improvement. Running 6400/2133 FCLK has better latency than 6400/2200 FCLK. If you can’t divide the memory speed and FCLK by 3 to get known speeds like 6400/2133 6000/2000 7400/2466 etc.. it greatly increases latency and performance penalties. AM5s memory controllers have to be able to divide by 3 (which is gear 2) if you go into gear 4 like his 8000 test was, it ruins performance hence why it didn’t get more fps when the test wasn’t limited by the GPU.
how do u get 6400 cl26? i have 2 2x16 kits (64gb total) rated for 6000 cl30 and it does that just fine when i enable xmp (7950x3d, 4080 super; playing at 4k)
You're very likely not going to run 4x16 (waaayyyy worse) or 2x32 stable at 6400 cl26 on a 7950x3d so I wouldn't worry about it. With 4x16gb sticks your memory controller is fighting for it's life to manage any kind of stability past 6000 CL30, and even that is unstable for some people. Tune your 6000 CL30 timings as much as you can and call it a day if you care but you're going to be limited in mem oc past 2x16gb. 4 dimm slots is a design flaw in the words of the great BZ.
i tuned my 8000 rated sticks for this test. i've never tested 4 DIMMs, but i've heard it's tough like @ZWortek mentioned
@@ZWortek gotchu
i js wanted to get more than 100 fps without dlss on pve 😭
@@ZWortek hi brother what do you think about g skill Z5 RGB 2x32GB) 6400MT/s CL32-39-39 for a 9800x3d? Easy to tune?
@@carlosferjezus I have the same kit with 14700kf and I can run it 6400 30 37 37 57 1.4v. I think it’s pretty good for a dualrank kit, but you must know that they don’t have good radiators. I reach 70 degrees and more :/
Just ordered a 9800x3d with 64gb DDR5 of Kingston Fury Beast 6400MHz (couldnt find CL rating) hope it all works well together cause i dont know jack squat about tuning ram speeds or curves😂
I built my new PC today, 9800x3D with the exact same RAM as you, it’s CL32. I only drop below 100fps on Streets when scoping in (80 was the minimum I saw near Pinwood hotel which is one of the lowest fps areas on the map, but otherwise running around the map I was anywhere within 120-200 range, 90-110 when scoped. It’s so smooth, not a single stutter in 3 hours. Blows my old build away.
Oh I should have added I have overclocked the CPU to 5.2Ghz. I haven’t done an undervolt yet though.
@@4PipeExhaustIt can already boost up to 5.2 Ghz on stock. Do you mean 5.4 Ghz? with the +200Mhz on PBO
@@4PipeExhaustAlso what’s your GPU and resolution?
@ sorry yes :)
@pyurologie doesn’t the 9800x3d only support up too 6000 mhz ram ?
that is the "sweet spot" for price to performance, but they can go further if tuned
Great video. Is 7200 at CL 34 good?
very good yes, higher end for ddr5
what are those maps used at the beginning of the video?
just top down maps of streets and labs
Cl26 ram will be released soon already pre-ordered
from where ?
Very Intriguing.
@@pyurologiewhich one do u recommend?
G.skill trident z5 royal neo 32gb
CL28 6000 vs CL30 6400
How is the fps higher in 1440p than 1080p? (streets online) I do see the lower lows. but the average is higher?
more GPU usage probably
holy spare parts, how did you acquire that pile LOL
i go a little crazy when i get into stuff... as you can see 😂😂
How 6400 is better than 8000??!! Can someone explain it to me?
Hi, I had a question. I basically follow ur 9800x3d tune guide, and im getting 130-170fps on streets, but when I try to run customs, I only got 80-100. is it cause customs has bad optimization? or is it my problem.
Please add the FCLK Clock to the list.
stock 1800 of 4800, 2200 on 6400 and 8000
@@pyurologie 8000Mhz would perfom better with 2000 FCLK (4:2:2). - 4000 MCLK : 2000 UCLK : 2000 FCLK for better latency. But yeah 1:1 still is better for Tarkov, ty for the benchmarks!
@@olitense for AM5, FCLK can run async to UCLK and MCLK, fastest FCLK is ideal
Pyro legend
I WENT WITH AMD Ryzen 7 7800X3D 4.20GHz до 5.00GHz tray × 1
Corsair Dominator Platinum RGB 32GB 6000MHz DDR5 CL36 × 1
Thermalright Frozen Prism 240 Black A-RGB LGA1851/LGA1700/AM5 × 1
Asrock B650M Pro RS AM5
I DIDN'T ASSEMBLE IT, I GAVE IT TO SPECIALIST, BUT THEY SAID THAT, THE RAM IS 6000MHZ AND THE PROCESSOR CAN'T WORK WITH IT, THE PC WON'T START OR SOMETHING, THEY TRY TO NURF THR RAM SPEED BUT THEY CAN'T, SOO WHAT DO YOU THINK, WHERE IS THE PROBLEM? They said 5600mhz need to be, high wont run they think.
PS. They said also, they will try to update the bios to.
The 8000 seems bad considering the lows are worse and is the ones that actually matter.
Great ❤
thanks! 🙏🙏
Hi can anyone help me please I think I do something really dumb I didn't konw when you use the ryzen master per core negative 20 core 1 to core 8 now I won't boot showing red light (Dram) and yellow light (CPU)
do a cmos reset if you can't boot up anymore
@pyurologie thank you 😊
I find 6400MT/s CL26 RAM provides the best performance for my 9800X3D.
Can you make this for rust?
i'll add one, got a bunch of ppl requesting it
@@pyurologie best workbencher out there!
At the end it doesnt even matter 🤷😬
yeah, with X3D RAM speed doesn't make a huge difference
do rust pls
i'll add one, got a bunch of ppl requesting it
How about 32 vs 64 and speed
he has a 16 vs 32 vs 64 video here - ua-cam.com/video/81--TGdYIZk/v-deo.html
This is a 9800x3d cpu difrnt numbers and now we got latency killer
@@dognoodles thanks for referencing that
@@berserkbasshead7618 32 vs 64 on 7800X3D here too: ua-cam.com/video/lJVbaA66J_o/v-deo.html. I'm assuming results will be similar, but as usual I'm gonna add 9800X3D tests for it soon. stay tuned!
6000 30
200fps 😂😂😂 im running customs at 60 at best
hi pyur i have 78x3d and hynix a die 32x2 ram, i tried ram over but 2nd, 3rd timing is so complicated. 2,3 timing is important in tarkov? many improvement in 1% low? or not big deal? ty for u r comparison video