I’ve noticed the same with my 9950X - you make a minor change in the BIOS, (sometimes not even a memory related change) and ZenTimings can show all the timings are identical to a configuration that’s 20000% Karhu stable and all of a sudden it won’t get past 100%. Even reloading ASUS User Profiles that were stable are not stable all of a sudden and the temps are the same. Drives me mad and puts me off playing in the BIOS.
Nothing exclusive to AMD or ASUS. On my 13700K, I spent freaking hours to get a known good 7000 and then 7200 on that godawful, garbage IMC. I had the known-good profiles SAVED on BIOS, SAVED as a file on USB. But if I tinkered and went back, same problem. Clear CMOS/Load Optimized, and then load the profile, no help. I used GIGABYTE and ASUS and the same issues happened across many generations and platforms. ASUS is by far the best at BIOS overall, not that they don't make mistakes, but NO ONE comes close to not breaking everything constantly. Everyone uses AMI BIOS and I swore the day Phoenix/Award BIOS got tossed out that this kind of crap would happen, because AMI was always known to be weird and now, no competition. But what does everyone do? Take everything out on ASUS. Everyone should be hounding GIGABYTE for being the king of hardware with a sucky, constantly breaking BIOS, but no, let's hound the only Taiwanese ODM that has a good history of BIOS and was the former GOD of mobos. ASUS' chief BIOS engineer and GIGABYTE's chief hardware engineer need to swap companies and learn from the others.
@@lePurpleDragon I used to think the speeds i could achieve on my 14900k were imc related, after swapping to a 2dimm asrock board I realized that's not the case. Apparently it was the signal integrity of the motherboard all along.
@@Jaejgaren 1.) If it's a problem with sub-7600 (Intel users say sub-7200 to cover up for their trash IMC, but it's actually up to 7600 on basically any board), that's basically the IMC. Period, the end. If it's a problem with a lack of a GDM-equivalent, which is a big, big issue on Intel, esp. Gear 2 on Rocket Lake/Alder/Raptor Lake, that's the IMC. 2.) Yes, on Intel, you can get a $500-800+ dollar board and run higher speeds (which is largely impractical). With this absurd investment, you can finally match or come close to almost EVERYTHING AMD does at 6000/MTs. Again, another fault in the DESIGN of the entire Intel chip, and also, of the IMC. I hope Intel pays you to meatshield for them.
@@Jaejgaren Nope, it just makes you an Intel meatshielder without pay. And do I know you've stability tested that? No. And even if it is stable, how do I know it performs better? I don't. Fun fact: almost ANY board can POST and boot (or like you worded it: do, can DO 8000) 8000 (and likely 8200) with a degree of timings play. Just like almost ANY motherboard can POST and boot DDR4 4000. Does posting high speeds mean anything? No. It means nothing. Rather, is it stable, and does it perform better? Likely, no. Why? Because even if your timings are right (I doubt it, you probably actually run XMP), there's going to be a sh*t ton of error correction on that mess. How do I know? Because unless you've spent HOURS learning a new memory controller than almost no one knows because almost no one uses this dead-end, paper launch platform on a sub-par motherboard, you likely don't have optimal timings and haven't worked around your board's quirks. So very much like the benchmarks Hardware Unboxed put out, your 8200 is going to be slower than 7200, and slower than 6000 on AMD. And even if you actually, actually, actually do know what you are doing AND somehow have faster DDR5 8200, you still don't run faster than AMD on 6000. P.S: Quit while you're ahead. You (actually) bought Core Ultra, you failed the test.
@@WmJamesWofford what? you have no idea what you are talking about, buildzoid has done streams where people submit their timings and there were quite a few 9800X3Ds at 8000 not to mention all the posts on overclocking forums, ive even seen a few a 8200. stop talking, you dont know anything.
Non-enthusiast gaming buddy of mine got a 7800X3D on my recommendation, built via microcenter, and went with a 7900XTX GPU. Passed GPU benches, but games would crash with driver timeout, no XMP enabled. Swapped the GPU, same crash issue. In the end the CPU was swapped to fix the issue, presumably the trashiest of IODs. Feels bad to have him had to deal with that on my recommendation.
I know the pain. Got a launch day 5600x with golden cores for frequency overclocking. Running stock the memory controller couldn't handle anything over 2133Mhz on a bunch of boards and memory kits without corrupting everything.
Well yeah, it's rare that zen 3 can get over 2000mhz, my 5800x only runs at 1867mhz, it doesn't bother me much as I've got it down to about 55ms in aida.
@@Markknightexeter As cool as it could be that my infinity fabric could hit 2133 unfortunately I meant the maximum frequency of any RAM kit could hit (1066 fabric).
Hi. If my VDDQ and VDDIO is exactly tied to my DRAM Voltage. Should I manually lower than to not be above 1,430V. Or can I set my DRAM to 1,6V. (So does the VDDQ and VDDIO) and still perfectly fine ?
This is exactly the same thing that happens with my 7800x3d. It'll run all day every day at 7800mhz just peachy, 1950 fclk, 1.0 vsoc, but throw 8000mhz in, after 1-5 minutes in prime95 it'll just hard reset with no actual errors reported.
My 9800x3d is the same way. I bought the same kit just to see if the expo timings would help as I've got the same XMP kit. Same results, pretty much. Every now and then I could get maybe an hour of karhu to pass. Sometimes it would fail within seconds.
Guys I just discovered the channel. Not someone that ever did overclock. If you can suggest some intro videos to ddr5 timings cas latency explanation, why they matter and what impact they have for gaming performance or else please do. I just got a kit of DDR5 2x16gb Corsair Vengeance RGB CL36 and I was wondering if it's a bad choise mainly for gaming. I got it as part of a combo with a R7 7700X and an Asus TUF B650-E Motherboard. The thing is that there was another combo with a TeamGroup T-FORCE VULCAN 32GB (2x16GB) DDR5 6000MHz CL30 Black 1.35V that was 20$ cheaper. I think I messed up but I might be able to swap them at least I could try and go to the store. Do you guys think it's worth the try or I won't noticed much of difference?
im using ddr 5 8000 cl40 working on my asrock x870e had a hard time getting it to work but got it done. Gskill ddr5 8000 cl40 trident z gold 2x24 gigs expo just gives me jdec speeds here is the sque F5-8000j0448F24GX2-TR5G had to use xmp but working and must be "128 and lower" tryed 3 times but it locked.
@@nyuuuuuuuiI feel you. Unfortunately you should not trust what’s on the box as it is all marketing. My Z790 Asus gaming F could not go past 7200mhz with a 13900kf even if Asus stated that it could do 7800mhz +. You should rely on real life testing (like the latest x870 board review of hardware unboxed) and buyers feedback and even that does not guarantee the same result at home as there are too many parameters to consider (and so many are outside of your control). I now have a 9800x3d and x870 tomahawk (goated board) and heard it can reliably hold 8000mhz… which mine effectly does with a 2x24go 8200mhz m die kit.
I bought the 9950x the day of that impossible output to remain stable at 6000MT while on my 7950x in 6400MT/2200 Stable on a gigabyte X670E Aorus master I directly returned the processor for reimbursement Now I am on an MSI X870-E Carbon still on my stable 7950X 8000/2200 still with the same ram kit My memory kit G.Skill 32 Go DDR5-6600 F5-6600J3440G16GX2-TZ5RK
I got a 7800X3D with really bad IOD on it. Cannot pass even 6000C32 and it crashed even with 4800MT JEDEC settings. I've seen they're many people on the Internet complain about their poor AM5 system cannot run EXPO/XMP default settings even their MEM kits is on the QVL list. I guess there's a big problem with AMD tolerance that tons of AM5 cpu do not have good/acceptable IOD. AMD just keep quite on their fault.
@JomarsYT Mine can boot on 8000MT as well but cannot pass the TM5 test. The system is kind of stable though, it can run CB 23 without problem. But it cannot pass the mem test. I've tested the performance difference between 7600C36, 7800C36, 8000C36, all using ASUS mem timing preset (I'm on a X670E-A, entry level strix board ), I see no difference among the AIDA 64 result. As long as the system is happy with 7600C36 and it pass the mem test, I just go for that config. My mem kit is Trident Z5 RGB 7800C36.
Hey BZ fellow tester here. Is there no support for 8400MHz kits on Zen 3 7800X3D? I can do as high as 8000. I wouldn't know because the QVL of the 8400 kits that do work isn't what I have (G.Skill F5-8400J4052G24GX2-TZ5RK). I notice when I manually key the XMP values, the BIOS hits a limitation when setting some values, so it's not even accepting the latency timings. Mobo is the MSI X870 Tomahawk Wifi w/Latest BIOS (Late november).
I’m using the same motherboard and ram with a 9800x3d just fine with expo enabled. Would be interested to see you set timings for it to see if any difference vs the 9700x.
As a current owner of a 9800X3D myself, all I can say is that you gotta get your foot in the door early. In either case I hope you get yours soon. Happy gaming 🫡
@@Unprotected1232 As the future owner of a 9800X3D (assuming USPS actually delivers the damn thing) I can say all you have to do is somehow wake up randomly in the middle of the night and check your phone while half asleep just as it comes in stock. True story btw... after I spent about 2 days with the Restocker notification thing open... Actually thought I was dreaming until I checked Amazon in the morning.
Press F to pay respects to this dirt/💩 sample of an IMC. Seriously, not tryna brag (especially as my first Ryzen which was a 1700 failed to exceed 3.7 GHz below 1.4V allcore; definitely not for lack of me trying) but I'm here chilling with a 9800X3D and untuned EXPO 6000 CL30-38… which I just upped to 1.37V and 6400 MT/s for the ‘virtual’ 3:3:2 configuration with FCLK @2133 MHz. Life's good now as I don't really need a 2nd CCD or any E-Core tomfoolery on this machine.
I like my MSI TUF X870 Tomahawk, but it hasn't been the easiest board to get running with my 9800x3D. I still have some messing around to do, but I'm almost certain that I should have gone with my usual choice of an ASUS mobo instead. Time will tell.
Anyone ever had the following strange experience with memory context restore on AM5? When booting into Windows 10 with Memory Context Restore disabled, I'd get bizzare beeping and lag when I shake the mouse for a few seconds. With memory context restored this jank is gone. It's on my older Ryzen 7900X rig with an Asus B650-PLUS TUF board. The very same board I've got trust issues with ever since the toasted 7800X3D debacle.
is it possible that because AMD ships xx50 tier CPU with a good bin CCD and a bad bin CCD there would be inherently higher chance that the memory controller is worse? I remember 7950x also had good bin CCD and bad bin CCD
Maybe there is a worse chance that you can't get a stable high clock speed for the infinity fabric that well with a bad binned CCD, but the memory controller itself is on the IO die and should not be impacted by whatever CCDs are attached to it.
this is so trash, the 8k marketing behind the 7000 and 9000 series is so stupid, IT SHOLD work if even there are kits 8000 expo, i dont remember a generation of cpus being a FULLY LOTERRY.
@@kasimirdenhertog3516 AMD really are a stupid, the imc of 8000 is straight better than 7 and 9, why an Apu have better imc than a desktop top tier chip?. Who knows xF
Hello buildzoid, can you explain ddr5 timings. Formula to calculate values as per jedec. What is the upper limit and lower limit of each timings, how the values are derived, which timings effect read, which timings effect write etc.. you did video for ddr 4 , video for ddr5 is long pending.. a request from one enthusiast to another 🥂
As someone who does not overclock my ram, getting 4 Crucial 48gb modules to work without random BSODs and data loss has been painful. And that is running the modules at 3600. Is it AMDs memory controller at fault?
Its a combination of 3 things... Quality of the motherboard Quality or lack of... of the memory controller Incorrectly installing the RAM sticks [which also isnt likely a thing you can fix easily] you put both sticks from 1 kit into the 2 slots from THE SAME CHANNEL and the other 2 sticks into the other channel Memory stick kits are binned and bundled to run on THE SAME CHANNEL not on separate ones So if you can remember what sticks are what get the matching pairs back together It will be CPU > 1-1 > 1-2 > 2-1 > 2-2 With 1-1 and 1-2 being kit 1 and then with 2-1 and 2-2 being kit 2 And assuming you can remember what sticks where what kit you may want to try them the other way around as well as 1 kit might have different impedance and having the kits 1 way around or the other can have a frequency ceiling change by being slightly less demanding on the controller
@ another thing to try is the bios versions... Starting from the newest progressively getting old load XMP or EXPO and set frequency to 4800 if you find a bios that boots it you can either use that one or you can mark it down as runs it and make a list and then keep increasing by 200 till that list becomes 1 or 2 at what ever frequency and then use that one But if you cant unmix the memory kits this method is basically not gonna work as you will be lucky if you can get 4000 with memory kits mixed up across different channels instead of on the same channel as each other
I owned a 7950x3d that did the same thing with 6400mhz ram and was never 100% stable. The memory controller in the CPU finally died and forced replacement of the CPU.
No I removed the lid for water cooling but that's not what damaged the memory controller. If I down clocked the RAM it stabilized for a while then became unstable so I would down clock more till stable. Rinse and repeat to finally outright failure.
My 9950X won't go past 7200 1:2 on my Taichi B650 board. But will run 1:1 6400 all day. My issue is PBO and CO... will pass stress testing 25hrs with settings then will fail after 7hrs on same setting if I test again .. frustrating.
I got a really turd 1700 once though. Every1 was basically saying ”Just overclock to 4 GHz and if that isn't 💯% try 3.9. GG; eZ.“ Yeah, mine literally failed 3.725 GHz @1.39V in some tests on a PRIME X370-Pro. I flipped it for the same SKU just for sanity and easily reached 3.9 GHz that time.
I'd also say that you probably have a higher chance of getting a golden or at least better sample if you don't buy them right when they come out. I don't really have enough CPUs to make any scientifically sound statistical claims about this, but it would make sense given that process nodes improve over time and maybe AMD even has less constraints with the bins once they've produced a bunch. My limited sample of CPUs though includes worse overclockers that I bought right after release, and better ones when I bought them much later in their release cycle.
for some reason i can boot and run 8200 c38 m die on my 9950x but if i even try 7800 on the same ram on my 9800x3d i cant even boot. its on the same board tho so its def. the cpu's that are the problem
it doesn't say on the box that it can run reliably run 8000MT/s memory. The officially supported max speed for zen5 is DDR5-5600, anything beyond that is technically an overclock. AMD did for some reason decide though to imply that x870 Mobos can do 8000 in their marketing of them, which most (but not all) of them can, but only if the CPU memory controller can too. And of course not an officially supported configuration.
Thx for video i have f5-8000j4048f24gx2-tz5rk INTEL XMP on 9800X3D aorus x870e pro ice amd work just set XMP on 8000MT 40 48 48 127 1.35V. Aida in normal win 67ns y-prime 2.0 7.3ns. I can use 38cl if i add voltage not worth increased temperature. Hynx M 2x24 I try se ram to 6000 1:1 28cl or 6400 30cl fabric 2000 to 2166 still best is 8000mt controler2000 fabric 2000 for latency. New agesa add 8ns delay now i hope gigabyte add feature to reverse this like other manufacturers do.
Well even if you have a capable memory controller, a B650 motherboard will be in the way, especially if you are running more than 16GB per memory stick.
Well his raptor lake CPU that he has also turned out a dud and way more cumbersome. He tried those Intel fanboys approved "totally runs stable, easy" memory speeds, and couldn't even get 7200 to run actually stable. The worst thing about it is that it doesn't just die and reset like Ryzens tend to do when they can't do it, but ran seemingly stable for a while and only crapped out after hours or days - which just makes it much harder to actually validate that your settings are indeed stable.
X870 hate 8000 mts. You need x870e to get that even close to stable. AM5 memory controller was intended to go 8000 but cpus aren't stable enough and missing Cudimm usability on many boards make it not easy. Going below CL40 and 8000 is hard and not recommended on Udimm
I’ve noticed the same with my 9950X - you make a minor change in the BIOS, (sometimes not even a memory related change) and ZenTimings can show all the timings are identical to a configuration that’s 20000% Karhu stable and all of a sudden it won’t get past 100%. Even reloading ASUS User Profiles that were stable are not stable all of a sudden and the temps are the same. Drives me mad and puts me off playing in the BIOS.
Nothing exclusive to AMD or ASUS. On my 13700K, I spent freaking hours to get a known good 7000 and then 7200 on that godawful, garbage IMC. I had the known-good profiles SAVED on BIOS, SAVED as a file on USB. But if I tinkered and went back, same problem. Clear CMOS/Load Optimized, and then load the profile, no help. I used GIGABYTE and ASUS and the same issues happened across many generations and platforms. ASUS is by far the best at BIOS overall, not that they don't make mistakes, but NO ONE comes close to not breaking everything constantly. Everyone uses AMI BIOS and I swore the day Phoenix/Award BIOS got tossed out that this kind of crap would happen, because AMI was always known to be weird and now, no competition. But what does everyone do? Take everything out on ASUS. Everyone should be hounding GIGABYTE for being the king of hardware with a sucky, constantly breaking BIOS, but no, let's hound the only Taiwanese ODM that has a good history of BIOS and was the former GOD of mobos. ASUS' chief BIOS engineer and GIGABYTE's chief hardware engineer need to swap companies and learn from the others.
@@lePurpleDragon I used to think the speeds i could achieve on my 14900k were imc related, after swapping to a 2dimm asrock board I realized that's not the case.
Apparently it was the signal integrity of the motherboard all along.
@@Jaejgaren 1.) If it's a problem with sub-7600 (Intel users say sub-7200 to cover up for their trash IMC, but it's actually up to 7600 on basically any board), that's basically the IMC. Period, the end. If it's a problem with a lack of a GDM-equivalent, which is a big, big issue on Intel, esp. Gear 2 on Rocket Lake/Alder/Raptor Lake, that's the IMC.
2.) Yes, on Intel, you can get a $500-800+ dollar board and run higher speeds (which is largely impractical). With this absurd investment, you can finally match or come close to almost EVERYTHING AMD does at 6000/MTs. Again, another fault in the DESIGN of the entire Intel chip, and also, of the IMC.
I hope Intel pays you to meatshield for them.
@@lePurpleDragon How come I'm doing 8200 with a-die on a board that is currently 250 on newegg? I guess that makes everything you just claimed false?
@@Jaejgaren Nope, it just makes you an Intel meatshielder without pay. And do I know you've stability tested that? No. And even if it is stable, how do I know it performs better? I don't. Fun fact: almost ANY board can POST and boot (or like you worded it: do, can DO 8000) 8000 (and likely 8200) with a degree of timings play. Just like almost ANY motherboard can POST and boot DDR4 4000. Does posting high speeds mean anything? No. It means nothing. Rather, is it stable, and does it perform better? Likely, no. Why? Because even if your timings are right (I doubt it, you probably actually run XMP), there's going to be a sh*t ton of error correction on that mess. How do I know? Because unless you've spent HOURS learning a new memory controller than almost no one knows because almost no one uses this dead-end, paper launch platform on a sub-par motherboard, you likely don't have optimal timings and haven't worked around your board's quirks. So very much like the benchmarks Hardware Unboxed put out, your 8200 is going to be slower than 7200, and slower than 6000 on AMD. And even if you actually, actually, actually do know what you are doing AND somehow have faster DDR5 8200, you still don't run faster than AMD on 6000.
P.S: Quit while you're ahead. You (actually) bought Core Ultra, you failed the test.
You know what: it is actually hardcore overclocking!
my 9950x also has a pants memory controller. your not alone
we need a 9800x3d @ ddr5 8800 Mhz test, lets go :D
Wont run it … just simple not in the mem table for AMD cpus….
@@WmJamesWoffordit works and stable there are some users with 8800/2200
@@WmJamesWofford what? you have no idea what you are talking about, buildzoid has done streams where people submit their timings and there were quite a few 9800X3Ds at 8000 not to mention all the posts on overclocking forums, ive even seen a few a 8200. stop talking, you dont know anything.
@ there is a huge difference between 8800 and 8200 Even with cherry picking ….. Check yourself before your reck yourself
@@thelasthallow 8000 is possible, 8800 is insane
Non-enthusiast gaming buddy of mine got a 7800X3D on my recommendation, built via microcenter, and went with a 7900XTX GPU. Passed GPU benches, but games would crash with driver timeout, no XMP enabled. Swapped the GPU, same crash issue. In the end the CPU was swapped to fix the issue, presumably the trashiest of IODs. Feels bad to have him had to deal with that on my recommendation.
The 7600X3D😅
Have you try ...lets say tight 7800 c34-36? Isnt better than 8000 c38-40? For gaming and all these
was just 0:25 going to mention how you doing finance wise. here is a comment for the algo
I know the pain. Got a launch day 5600x with golden cores for frequency overclocking. Running stock the memory controller couldn't handle anything over 2133Mhz on a bunch of boards and memory kits without corrupting everything.
Well yeah, it's rare that zen 3 can get over 2000mhz, my 5800x only runs at 1867mhz, it doesn't bother me much as I've got it down to about 55ms in aida.
@@Markknightexeter As cool as it could be that my infinity fabric could hit 2133 unfortunately I meant the maximum frequency of any RAM kit could hit (1066 fabric).
Hi. If my VDDQ and VDDIO is exactly tied to my DRAM Voltage. Should I manually lower than to not be above 1,430V. Or can I set my DRAM to 1,6V. (So does the VDDQ and VDDIO) and still perfectly fine ?
It’s normally more difficult on dual ccd chips to run high speeds. It seems this particular sample you have has an intolerance to VDDP.
This is exactly the same thing that happens with my 7800x3d. It'll run all day every day at 7800mhz just peachy, 1950 fclk, 1.0 vsoc, but throw 8000mhz in, after 1-5 minutes in prime95 it'll just hard reset with no actual errors reported.
My 9800x3d is the same way. I bought the same kit just to see if the expo timings would help as I've got the same XMP kit. Same results, pretty much. Every now and then I could get maybe an hour of karhu to pass. Sometimes it would fail within seconds.
i just settled for 7800c34 on my 9800x3d since 8000 refuses to work
I was going to ask if you can't adjust other voltages that could help, but then I saw your VDDQ already at 1.45V
Guys I just discovered the channel. Not someone that ever did overclock. If you can suggest some intro videos to ddr5 timings cas latency explanation, why they matter and what impact they have for gaming performance or else please do. I just got a kit of DDR5 2x16gb Corsair Vengeance RGB CL36 and I was wondering if it's a bad choise mainly for gaming. I got it as part of a combo with a R7 7700X and an Asus TUF B650-E Motherboard. The thing is that there was another combo with a TeamGroup T-FORCE VULCAN 32GB (2x16GB) DDR5 6000MHz CL30 Black 1.35V that was 20$ cheaper. I think I messed up but I might be able to swap them at least I could try and go to the store. Do you guys think it's worth the try or I won't noticed much of difference?
Your ability to buy bad CPU’s is incredible when was the last good cpu u bought?
12600KF
im using ddr 5 8000 cl40 working on my asrock x870e had a hard time getting it to work but got it done. Gskill ddr5 8000 cl40 trident z gold 2x24 gigs expo just gives me jdec speeds here is the sque F5-8000j0448F24GX2-TR5G had to use xmp but working and must be "128 and lower" tryed 3 times but it locked.
How would you get gold version of their ram? Can find any gold version selling on Amazon, Newegg, and Micro Center.
You don't have any luck at all when it comes to high end processors....
Lack of skill.
You've said 24GB should be easier to run right? Does it do the same with a 24GB 8000 kit?
I have the same problem even with 24gb dimms on a 9800X3D.
Can run 6400 1:1, but not 8000 1:2. Both my ram and motherboard are 8000+ rated.
I've tried A LOT of memory kits with this CPU they ALL do this.
@@nyuuuuuuuiI feel you.
Unfortunately you should not trust what’s on the box as it is all marketing. My Z790 Asus gaming F could not go past 7200mhz with a 13900kf even if Asus stated that it could do 7800mhz +.
You should rely on real life testing (like the latest x870 board review of hardware unboxed) and buyers feedback and even that does not guarantee the same result at home as there are too many parameters to consider (and so many are outside of your control).
I now have a 9800x3d and x870 tomahawk (goated board) and heard it can reliably hold 8000mhz… which mine effectly does with a 2x24go 8200mhz m die kit.
I bought the 9950x the day of that impossible output to remain stable at 6000MT while on my 7950x in 6400MT/2200 Stable on a gigabyte X670E Aorus master I directly returned the processor for reimbursement
Now I am on an MSI X870-E Carbon still on my stable 7950X 8000/2200 still with the same ram kit
My memory kit G.Skill 32 Go DDR5-6600 F5-6600J3440G16GX2-TZ5RK
I got a 7800X3D with really bad IOD on it. Cannot pass even 6000C32 and it crashed even with 4800MT JEDEC settings. I've seen they're many people on the Internet complain about their poor AM5 system cannot run EXPO/XMP default settings even their MEM kits is on the QVL list. I guess there's a big problem with AMD tolerance that tons of AM5 cpu do not have good/acceptable IOD. AMD just keep quite on their fault.
When it crashes on JEDEC settings, then it is broken.
@derdaniel8261 Yeah. So I got a new one from dealer and this one is much better. At least it boots on 7600C36, and passed TM5 absolute for 12 hours.
if crash on stock, the cpu is just broken, mine can do 8000mhz, need voltge, but it works, i was really lucky with this 7800x3d.
@JomarsYT Mine can boot on 8000MT as well but cannot pass the TM5 test. The system is kind of stable though, it can run CB 23 without problem. But it cannot pass the mem test. I've tested the performance difference between 7600C36, 7800C36, 8000C36, all using ASUS mem timing preset (I'm on a X670E-A, entry level strix board ), I see no difference among the AIDA 64 result. As long as the system is happy with 7600C36 and it pass the mem test, I just go for that config. My mem kit is Trident Z5 RGB 7800C36.
@@ati4280 try 1.45 iod, 1.2 or up to 1.3 soc, also my ram need 1.6 to make it.
So, around 8000 and up CUDIMMs are the way to go huh?
If having a clockgen on-stick helps so much, perhaps clock drive strength is key?
Hey BZ fellow tester here. Is there no support for 8400MHz kits on Zen 3 7800X3D?
I can do as high as 8000. I wouldn't know because the QVL of the 8400 kits that do work isn't what I have (G.Skill F5-8400J4052G24GX2-TZ5RK).
I notice when I manually key the XMP values, the BIOS hits a limitation when setting some values, so it's not even accepting the latency timings.
Mobo is the MSI X870 Tomahawk Wifi w/Latest BIOS (Late november).
Just a question im going to buy 9000 series is x870 tuf enough for xpo 8000 or i need to higher as you said at this speed mobo is the limiting factor
For wath you want 8000 mhz DDR?
@ionutortian5885 why not its cool i like overclocking and in 2 to 1 at 8000 you can sync the clock with does give you a great boost
Better buy a better board you are limited on that one better to go 670e actually
I’m using the same motherboard and ram with a 9800x3d just fine with expo enabled. Would be interested to see you set timings for it to see if any difference vs the 9700x.
So when they say that they will produce 9000+ MTS RAM sticks now. What's the use if no mobo and CPU can utilise the speed?
That is only for CU-DIMMs and AMD isn't supporting CU-DIMMs yet. Essentially 9000MHz+ is only realm of 245k/265k/285k
@piotrj333 can intel support them
This is going to be my 9800X3D for sure :\. Can't check though because USPS is currently having fun playing ping pong with it...
As a current owner of a 9800X3D myself, all I can say is that you gotta get your foot in the door early.
In either case I hope you get yours soon. Happy gaming 🫡
@@Unprotected1232 As the future owner of a 9800X3D (assuming USPS actually delivers the damn thing) I can say all you have to do is somehow wake up randomly in the middle of the night and check your phone while half asleep just as it comes in stock.
True story btw... after I spent about 2 days with the Restocker notification thing open... Actually thought I was dreaming until I checked Amazon in the morning.
Reminds me of my 6700k which can overclock easily but when you try anything with xmp it just simply crashes
Press F to pay respects to this dirt/💩 sample of an IMC. Seriously, not tryna brag (especially as my first Ryzen which was a 1700 failed to exceed 3.7 GHz below 1.4V allcore; definitely not for lack of me trying) but I'm here chilling with a 9800X3D and untuned EXPO 6000 CL30-38… which I just upped to 1.37V and 6400 MT/s for the ‘virtual’ 3:3:2 configuration with FCLK @2133 MHz. Life's good now as I don't really need a 2nd CCD or any E-Core tomfoolery on this machine.
Is that an arctic p12 max fan with its frame cut off for your ram cooler? Looks like their fan design.
It’s a stock AMD fan that came with some of their older AM4 chips.
No, that's most likely the fan from an AMD cpu cooler.
Looks like a fan off a ryzen cpu
I like my MSI TUF X870 Tomahawk, but it hasn't been the easiest board to get running with my 9800x3D. I still have some messing around to do, but I'm almost certain that I should have gone with my usual choice of an ASUS mobo instead. Time will tell.
What’s wrong with it, it works fine for me
Do you think dual ccd's make a difference?
If your 9950x isn't performing as advertised, can't you return it and get a new one?
Anyone ever had the following strange experience with memory context restore on AM5?
When booting into Windows 10 with Memory Context Restore disabled, I'd get bizzare beeping and lag when I shake the mouse for a few seconds. With memory context restored this jank is gone.
It's on my older Ryzen 7900X rig with an Asus B650-PLUS TUF board. The very same board I've got trust issues with ever since the toasted 7800X3D debacle.
is it possible that because AMD ships xx50 tier CPU with a good bin CCD and a bad bin CCD there would be inherently higher chance that the memory controller is worse? I remember 7950x also had good bin CCD and bad bin CCD
Maybe there is a worse chance that you can't get a stable high clock speed for the infinity fabric that well with a bad binned CCD, but the memory controller itself is on the IO die and should not be impacted by whatever CCDs are attached to it.
this is so trash, the 8k marketing behind the 7000 and 9000 series is so stupid, IT SHOLD work if even there are kits 8000 expo, i dont remember a generation of cpus being a FULLY LOTERRY.
I have a 13900K that can't even do 7200 XMP reliably.
I believe I have hit The lottery with Mine!
@@angeluorteganieves2539 I also, I can do 8k with 7800x3d.
Yep, better get Ryzen 8000 😎
@@kasimirdenhertog3516 AMD really are a stupid, the imc of 8000 is straight better than 7 and 9, why an Apu have better imc than a desktop top tier chip?. Who knows xF
Quick question that's unrelated to the video! But how can i stop my i9 14900k from degrading? I want it to last more than a few months
You lower the voltage through BIOS -25mv
so what you consider the sweetspot to run?
Hello buildzoid, can you explain ddr5 timings. Formula to calculate values as per jedec. What is the upper limit and lower limit of each timings, how the values are derived, which timings effect read, which timings effect write etc.. you did video for ddr 4 , video for ddr5 is long pending.. a request from one enthusiast to another 🥂
msi finally changed their bios >....YES!
700$ Processor Mine its giving Me Monster benches!
I feel like xmp/expo speed ratings are for the most part just a marketing thing
As someone who does not overclock my ram, getting 4 Crucial 48gb modules to work without random BSODs and data loss has been painful. And that is running the modules at 3600. Is it AMDs memory controller at fault?
Bruh do you really need that much ram ? Going 2 dimm per slot is a huge ask for both teams, especially team red
@night_light2867 Photoshop can use well over 100GB in some of my projects. Having lots of headroom is nice!
Its a combination of 3 things...
Quality of the motherboard
Quality or lack of... of the memory controller
Incorrectly installing the RAM sticks [which also isnt likely a thing you can fix easily] you put both sticks from 1 kit into the 2 slots from THE SAME CHANNEL and the other 2 sticks into the other channel
Memory stick kits are binned and bundled to run on THE SAME CHANNEL not on separate ones
So if you can remember what sticks are what get the matching pairs back together
It will be CPU > 1-1 > 1-2 > 2-1 > 2-2
With 1-1 and 1-2 being kit 1 and then with 2-1 and 2-2 being kit 2
And assuming you can remember what sticks where what kit you may want to try them the other way around as well as 1 kit might have different impedance and having the kits 1 way around or the other can have a frequency ceiling change by being slightly less demanding on the controller
@@commanderoof4578 Thank you for the information, I will look into this. I have an MSI Tomahawk x670e WIFI board, and a 7900X.
@ another thing to try is the bios versions...
Starting from the newest progressively getting old load XMP or EXPO and set frequency to 4800 if you find a bios that boots it you can either use that one or you can mark it down as runs it and make a list and then keep increasing by 200 till that list becomes 1 or 2 at what ever frequency and then use that one
But if you cant unmix the memory kits this method is basically not gonna work as you will be lucky if you can get 4000 with memory kits mixed up across different channels instead of on the same channel as each other
I owned a 7950x3d that did the same thing with 6400mhz ram and was never 100% stable. The memory controller in the CPU finally died and forced replacement of the CPU.
expo? and the memory controller just died? did you get it replaced under warranty?
No I removed the lid for water cooling but that's not what damaged the memory controller. If I down clocked the RAM it stabilized for a while then became unstable so I would down clock more till stable. Rinse and repeat to finally outright failure.
My 9950X won't go past 7200 1:2 on my Taichi B650 board. But will run 1:1 6400 all day. My issue is PBO and CO... will pass stress testing 25hrs with settings then will fail after 7hrs on same setting if I test again .. frustrating.
ddr5 8k? or no.
is 6000 cl 30 that bad that you have to oc to 8000?
can u put RAM serial number in description? easy to copy paste
@buildzoid your living the the future 2045 lol 😂😂
i'd RMA it and point on AMDs marketing slides.
Solicon lottery losing CPU, sounds like an upgrade to the daily system
Jesu fuck ddr5 is getting insane 😅
I don't think, that BZ got bad CPUs in the past. He has got normal CPUs that everybody else gets also. Everything more is pure luck and not garanteed.
I got a really turd 1700 once though. Every1 was basically saying ”Just overclock to 4 GHz and if that isn't 💯% try 3.9. GG; eZ.“
Yeah, mine literally failed 3.725 GHz @1.39V in some tests on a PRIME X370-Pro. I flipped it for the same SKU just for sanity and easily reached 3.9 GHz that time.
I'd also say that you probably have a higher chance of getting a golden or at least better sample if you don't buy them right when they come out. I don't really have enough CPUs to make any scientifically sound statistical claims about this, but it would make sense given that process nodes improve over time and maybe AMD even has less constraints with the bins once they've produced a bunch. My limited sample of CPUs though includes worse overclockers that I bought right after release, and better ones when I bought them much later in their release cycle.
everyone @ 8:50 thank me later
thanks @ironsteal !!
would love to see a test with 9800x3d if 8000mt/s is possible, and if you can manage to run it at 5,6/5,7 ghz OC
i got a 7800x3d that wont do 6000 at soc 1.3 and i have it at 1.15 soc at mem 5800
my 7800x3d can do 6400 1:1 but cant do 8000 too, sadly
for some reason i can boot and run 8200 c38 m die on my 9950x but if i even try 7800 on the same ram on my 9800x3d i cant even boot. its on the same board tho so its def. the cpu's that are the problem
If it doesnt do what is says on the box its a scam.
it doesn't say on the box that it can run reliably run 8000MT/s memory. The officially supported max speed for zen5 is DDR5-5600, anything beyond that is technically an overclock.
AMD did for some reason decide though to imply that x870 Mobos can do 8000 in their marketing of them, which most (but not all) of them can, but only if the CPU memory controller can too. And of course not an officially supported configuration.
Thx for video i have f5-8000j4048f24gx2-tz5rk INTEL XMP on 9800X3D aorus x870e pro ice amd work just set XMP on 8000MT 40 48 48 127 1.35V.
Aida in normal win 67ns y-prime 2.0 7.3ns.
I can use 38cl if i add voltage not worth increased temperature.
Hynx M 2x24 I try se ram to 6000 1:1 28cl or 6400 30cl fabric 2000 to 2166 still best is 8000mt controler2000 fabric 2000 for latency.
New agesa add 8ns delay now i hope gigabyte add feature to reverse this like other manufacturers do.
Im still trying to get 4 sticks to run faster than 3200 on my asus x670e hero
Have u tried asking nicely?
@@WillFuI lmao
A-die?
not ui..gui (goooeeee)
lu b z
Well even if you have a capable memory controller, a B650 motherboard will be in the way, especially if you are running more than 16GB per memory stick.
you do realize X870E is literally just 2 B650 chipsets glued together. Also the chipsets don't have a memory controller.
ironic we own the same ssd.. but get a dram sata/nvme drive for better user experience
I feel like someone constantly reminds us that intel IMC's are just better.
Well his raptor lake CPU that he has also turned out a dud and way more cumbersome. He tried those Intel fanboys approved "totally runs stable, easy" memory speeds, and couldn't even get 7200 to run actually stable. The worst thing about it is that it doesn't just die and reset like Ryzens tend to do when they can't do it, but ran seemingly stable for a while and only crapped out after hours or days - which just makes it much harder to actually validate that your settings are indeed stable.
X870 hate 8000 mts. You need x870e to get that even close to stable. AM5 memory controller was intended to go 8000 but cpus aren't stable enough and missing Cudimm usability on many boards make it not easy. Going below CL40 and 8000 is hard and not recommended on Udimm