New Ryzen 7 9700X, RTX 4080, 9700X build - EXPO not working

[Mr_Engineer]

I just built a new system that boots fine at default memory speed (4800) but crashed when enabling EXPO. I confirmed that the memory is on the supported QVL list for the motherboard.

Motherboard: Asus Rog Strix X870-F Gaming WIFI
Memory: G.Skill Trident Z5 Neo RGB DDR5-6000 CL30-40-40-96 1.40V (2x32GB) (F5-6000J3040G32GX2-TZ5NR)

One thing I tried was to select EXPO, memory at 6000Mhz, but use CAS 32 instead of CAS 30, but the system would still not boot.

Any advice from the experts please

 

[oxrufiioxo]

I would rma or get a replacement kit. Beyond that the only thing that could help is bios updates but I wouldn't wait if this is a new build.

Try each stick individually with expo enabled If you haven't already if they both boot and test fine could be the board or cpu as the culprit. Inspect the cpu socket for bent pins is another option although that's probably unlikely.

 

[Mr_Engineer]

I forgot to add that I am using the latest Motherboard BIOS Version 0706 dated 05/12/2024 and AGESA 1.2.0.2b

I will try your suggestions and testing each memory stick separately.

Thanks for the answer.

 

[oxrufiioxo]

I forgot to add that I am using the latest Motherboard BIOS Version 0706 dated 05/12/2024 and AGESA 1.2.0.2b

I will try your suggestions and testing each memory stick separately.

Thanks for the answer.

There's a slim chance it could be the density as well. More memory is typical harder on the memory controller but 2x dimms should be fine unless you lost the silicon lottery pretty badly on the cpu.

If you can order a 2x16 6000CL30 and it boots fine that could be it, preferably from somewhere that has a good return policy.


Other than that hopefully some of the more experienced members chime in with some ideas.

 

[AleXXX666]

There's a slim chance it could be the density as well. More memory is typical harder on the memory controller but 2x dimms should be fine unless you lost the silicon lottery pretty badly on the cpu.

If you can order a 2x16 6000CL30 and it boots fine that could be it, preferably from somewhere that has a good return policy.


Other than that hopefully some of the more experienced members chime in with some ideas.

QVL shows FULL compatibility of this kit and Ryzen 9000 series. F.ck ASUS...

 

[oxrufiioxo]

QVL shows FULL compatibility of this kit and Ryzen 9000 series. F.ck ASUS...

I haven't looked at a qvl in a decade and I've never had any issues on either Intel or AMD other than some really old 2014 corsair LPX on AM4 that was notorious for giving people issues but somthing being on or not on the qvl doesn't guarantee a whole lot in my book.

You have to remember they test likely with one cpu and probably only that it boots with very minimal stress testing. Same with the qvl on the actual memory website.

 

[AleXXX666]

I haven't looked at a qvl in a decade and I've never had any issues on either Intel or AMD other than some really old 2014 corsair LPX on AM4 that was notorious for giving people issues but somthing being on or not on the qvl doesn't guarantee a whole lot in my book.

You have to remember they test likely with one cpu and probably only that it boots with very minimal stress testing. Same with the qvl on the actual memory website.

yeah, have had a sh*t recently with K1ngston RAM and H610 MSI (lol). RAM is perfectly in QVL, then.. boom, Shitgston decided to change RAM dies brand, and in QVL there was another die manufacturer for my sticks.. glad it started and worked somehow at the end lol

 

[Taraquin]

You have ram in slot 2 and 4, not 1 and 3? What is soc voltage and vddp set to when enabling expo?

 

[Mr_Engineer]

Yes, the RAM is in slots 2 and 4.

Just wanted to confirm that I got the memory to run even higher than rated speed, @ 6200MHz CAS 30

I used Buildzoid timings, and the memory ran without issue.

I am going to try to tweak a bit further tomorrow, the memory latency seems a bit high for some reason.

Thanks to everyone for the responses and help, really appreciate it.

 

[ir_cow]

I suggest not applying someone else's timings unless you know what they do, the voltages required (and potential damage it can cause) and are willingness to stress test. Just because it boots, is NOT a indication of stability.

 

[rv8000]

Yes, the RAM is in slots 2 and 4.

Just wanted to confirm that I got the memory to run even higher than rated speed, @ 6200MHz CAS 30

I used Buildzoid timings, and the memory ran without issue.
View attachment 375110View attachment 375111

I am going to try to tweak a bit further tomorrow, the memory latency seems a bit high for some reason.

Thanks to everyone for the responses and help, really appreciate it.

tRC should be 66; tRC = tRP+tRAS

Leaved your rtt settings at auto, you shouldnt need to change these.

FCLK with ddr5 6200 should be set to 2066 sync and latency benefits.

Seems like an excessive amount of vddio, 1.3-1.35 should suffice and test for stability.

Also, at 1.4v vdd/vddq you can likely turn GDM off, will improve latency considerably alongside sync’d fclk.

 

[oxrufiioxo]

I suggest not applying someone else's timings unless you know what they do, the voltages required (and potential damage it can cause) and are willingness to stress test. Just because it boots, is NOT a indication of stability.

My personal theory was it was voltage related but I'm not a huge fan of having people mess around with memory voltages or recommending that regardless of what is listed as safe online.

It's always preferable to have a kit that works out of the box.

 

[Taraquin]

Yes, the RAM is in slots 2 and 4.

Just wanted to confirm that I got the memory to run even higher than rated speed, @ 6200MHz CAS 30

I used Buildzoid timings, and the memory ran without issue.
View attachment 375110View attachment 375111

I am going to try to tweak a bit further tomorrow, the memory latency seems a bit high for some reason.

Thanks to everyone for the responses and help, really appreciate it.

Maybe soc or vddp-voltage is too low giving you throttling. Latency should have been about 60. try fclk 2066 since fclk gets in a 2/3 ratio with mclk. It will improve latency. You may be able to run RCD and RP at 37, 36 or 35.

 

[AusWolf]

There's a slim chance it could be the density as well. More memory is typical harder on the memory controller but 2x dimms should be fine unless you lost the silicon lottery pretty badly on the cpu.

If you can order a 2x16 6000CL30 and it boots fine that could be it, preferably from somewhere that has a good return policy.


Other than that hopefully some of the more experienced members chime in with some ideas.

That's what I thought as well.

64 GB kits are usually double sided, running 6000 MHz on them could be a bit too much to ask. I'd check the motherboard's web page to see if there's any info on this.

 

[oxrufiioxo]

That's what I thought as well.

64 GB kits are usually double sided, running 6000 MHz on them could be a bit too much to ask. I'd check the motherboard's web page to see if there's any info on this.

From the op being able to run 6200 with manually setting voltages it seems to me the motherboard is playing it probably too safe for 6000CL30 at the 64GB density. Still I'd look for a kit that just works out of the box vs having to manually set it up after every bios update. There is no guarantee bios to bios that whatever is manually set will work and actually stress testing memory is a headache it can pass every test and still crash during gaming etc.

 

[Mr_Engineer]

I'll perform the suggestions today regarding voltages and FCLK and retest.

Also, I'll probably just settle on 6000MHz with tweaked timings, as you guys had mentioned, 64GB is already stressful on the IMC and Motherboard. Stability and longevity are more important to me than a 1-2% more performance. And I won't know how stable the PC is for sure until I use it for a while for both productivity and games.

If you have any suggestions for software to stress test and also for benchmarking performance, please let me know. I'll post the results here for reference.

Note: The above timings are for Hynix die DDR5 memory, so before applying them, I ensured that this what I have in my G.Skill memory. These same timings worked well for some on Ryzen even with 128GB (DDR5 Hynix die) (32GBx4) @ 6000MHz.

But ofc as always with memory YMMV. Cheers.

 

[AusWolf]

I'll perform the suggestions today regarding voltages and FCLK and retest.

Also, I'll probably just settle on 6000MHz with tweaked timings, as you guys had mentioned, 64GB is already stressful on the IMC and Motherboard. Stability and longevity are more important to me than a 1-2% more performance. And I won't know how stable the PC is for sure until I use it for a while for both productivity and games.

If you have any suggestions for software to stress test and also for benchmarking performance, please let me know. I'll post the results here for reference.

Note: The above timings are for Hynix die DDR5 memory, so before applying them, I ensured that this what I have in my G.Skill memory. These same timings worked well for some on Ryzen even with 128GB (DDR5 Hynix die) (32GBx4) @ 6000MHz.

But ofc as always with memory YMMV. Cheers.

There are various programs made for stress testing RAM (some are included in Windows and Linux as well as a separate bootable option), although some, or even none of them may give you errors while even regular use could. There's no single solution that beats all. So I'd say, do some gaming, work, or whatever you need your PC for, and look out for freezes, bluescreens, or random restarts.

 

[Taraquin]

I'll perform the suggestions today regarding voltages and FCLK and retest.

Also, I'll probably just settle on 6000MHz with tweaked timings, as you guys had mentioned, 64GB is already stressful on the IMC and Motherboard. Stability and longevity are more important to me than a 1-2% more performance. And I won't know how stable the PC is for sure until I use it for a while for both productivity and games.

If you have any suggestions for software to stress test and also for benchmarking performance, please let me know. I'll post the results here for reference.

Note: The above timings are for Hynix die DDR5 memory, so before applying them, I ensured that this what I have in my G.Skill memory. These same timings worked well for some on Ryzen even with 128GB (DDR5 Hynix die) (32GBx4) @ 6000MHz.

But ofc as always with memory YMMV. Cheers.

Try testmem5 with usmus config. You should direct a fan abive ram atleast while testing.

 

[LittleBro]

Pardon me, did I understand it right? You tweaked timings and then complain the system fails to boot?
Timings you sent before can't be set from G.Skill, they would NEVER sell a kit with such low tRAS and tRC, let alone tRFC.

Never consider someone else's hardware settings to be compatible/suited for your hardware even where you identical SKUs. It's a silicon lottery, after all.

I'd also recommend HCI MemTest for stress testing. Combine with testmem5.

 

[Wasteland]

Testing memory is a bit of a rabbit hole. There are several ways of doing it, but in my experience the best singular all-around option--in terms of speed and thoroughness--is stressapptest:

Stressful Application Test (or stressapptest, its unix name) is a memory interface test. It tries to maximize randomized traffic to memory from processor and I/O, with the intent of creating a realistic high load situation in order to test the existing hardware devices in a computer. It has been used at Google for some time and now it is available under the apache 2.0 license.

 

[Mr_Engineer]

Pardon me, did I understand it right? You tweaked timings and then complain the system fails to boot?
Timings you sent before can't be set from G.Skill, they would NEVER sell a kit with such low tRAS and tRC, let alone tRFC.

Never consider someone else's hardware settings to be compatible/suited for your hardware even where you identical SKUs. It's a silicon lottery, after all.

I'd also recommend HCI MemTest for stress testing. Combine with testmem5.

Hey Bro, initially I just enabled EXPO profile from the BIOS and Windows stopped booting, it only boots at the default 4800 speed.

The memory is already rated to run at 6000 and I confirmed this on both ASUS and G.Skill QVL sites:

I probably should have done what the guys had suggested and get an RMA for the memory, but decided to try and tweak instead in the name of science

Currently I am running some stress tests at 6000 CAS 30, which is what the memory is rated for anyway.

Honestly the issue could be with the Ryzen (shitty) memory controller, as I have tweaked the memory timings but also voltages etc, which may have helped it run at the rated speed.

 

[LittleBro]

Tightening timings will put even more stress on the IMC. You probably solved the problem with raising voltage for IMC.
You can test it quite easily - lower voltage to default and you'll see.

 

[GerKNG]

QVL shows FULL compatibility of this kit and Ryzen 9000 series. F.ck ASUS...

ASUS is not responsible for your IMC.
and you should try EXPO with 1.3V SoC and if it boots try lower it to 1.25V. maybe the board defaults to something very low due to a bug? who knows if you can't get into windows.

 

[_roman_]

I would rma or get a replacement kit. Beyond that the only thing that could help is bios updates but I wouldn't wait if this is a new build.

I probably should have done what the guys had suggested and get an RMA for the memory, but decided to try and tweak instead in the name of science

A few hours ago I checked the datasheet for the ryzen 9800X3d. The supported speed is 5600MT/s. I doubt the 9700X support faster speeds according to amd. Anything else is overclocking.

Why should there be a rma when the user overclocks?

--

@ topic poster

I highly recommend stable testing your memory with your "settings".

 

[Marcus L]

A few hours ago I checked the datasheet for the ryzen 9800X3d. The supported speed is 5600MT/s.

Why should there be a rma when the user overclocks?

That's AMD's officially supported speed though both the motherboard and memory MFR have confirmed this kit to run at 6000 cl30 with Ryzen 9000, the same as 2933/3200 is the max officially rated speed by AMD on am4 though there are kits and motherboard qvl that have ram speeds up to 3600/3800+- etc

The main issue however is trying to find out if it's the CPU IMC, mobo or ram itself that is the culprit, RMA may not be needed anyway as in the UK and EU we have consumer laws that allows you to return any items purchased within 14 days, outside of this it could be tricky if they test the ram and it runs at its rated speed and the RMA is denied, but there's a lot of assuming going on now so I'll just stfu