Low Time Spy Scores - RTX 4090 STRIX OC - 30K Graphics

GoodBoy said:
A few things:

Try 100% power target if you haven't yet already.

Low Timespy scores can be CPU related, even on the GPU test. So anything that can affect the CPU performance could affect the score:
Ensure Ram is configured for dual-channel or quad channel, whatever mobo supports/how many modules you have.
This could be a case of a bios issue where the settings appear correct, but are not actually applied. And this is settings specifically concerning RAM. So load default values on the Bios and then Save the changes. Then reconfigure everything, including the RAM XMP settings, Rebar.
If that doesn't help, turn off XMP and try manual DDR settings at 1 step slower clocks. So if it's 6000 Mhz now for that XMP profile, Try 5667Mhz If that's what the next slowest DDR5 modules would run at.
If that doesn't help, I would roll back the Bios to the previous version you were running.

Did you try yet with cpu virtualization optimizations turned off?
Click to expand...
Yup.

The key thing here is I have 3 other rigs similarly configured with RTX 4090s that have scores that are normal:

B650E
https://www.3dmark.com/spy/44583772
https://valid.x86.fr/ksnuwi

X670E [low score]
https://www.3dmark.com/spy/44597975
https://valid.x86.fr/16jfxx

X670E2
https://www.3dmark.com/spy/44584750
https://valid.x86.fr/j27n1r

X670E3
https://www.3dmark.com/spy/44597107
https://valid.x86.fr/g54i81

I also took a known good 4090 from one of the "normal" rigs and tried it in the "low score" rig - same results. OS has been wiped. BIOS has been reset. XMP has been verified.

Starting to wonder if it's the new BiOS with security mitigations. Although X670E3 has the ProArt and it has the same ASUS BIOS as the GENE and it is performing normally. Shrug.
 
Seems the motherboard or CPU might be the culprit. Assuming you already tried memory swap.
 
KickAssCop said:
Seems the motherboard or CPU might be the culprit. Assuming you already tried memory swap.
Click to expand...
Good idea. I haven't swapped RAM, yet. I have the same 2 sticks on all 4 rigs so I wrote it off. But something to try!

I'll run an exhaustive mem test first to make sure nothing is wrong with the sticks. I feel like swapping isn't needed if they pass the test.

EDIT: nothing found with the latest PassMark MemTest:
Screenshot 2024-01-20 204417.png


EDIT2: also with HCL MemTest Pro:

memtest_pro.png
 
Last edited:
Furious_Styles said:
Cyberpunk, tomb raider are two I recall off the top of my head, I'm sure other people might be able to chime in.
Click to expand...
No clue if this is good or bad since this isn’t a standard resolution for the tech sites:

Cyberpunk 2077

IMG_4967.jpeg
 
Last edited:
Make sure your PSU's 12v output isn't dropping under load.

Install GPU-Z, go to sensors tab, click on all of the voltage fields until they show "Min", then run 3dmark, or play cyberpunk for 10 minutes. Then exit and go back and review the minimum observed voltages and report back what you see.
1705952532452.png

This test will rule out the PSU. Also, hover over the perfcap reason and report what it shows. Blue = vRel on mine.
 
GoodBoy said:
Make sure your PSU's 12v output isn't dropping under load.

Install GPU-Z, go to sensors tab, click on all of the voltage fields until they show "Min", then run 3dmark, or play cyberpunk for 10 minutes. Then exit and go back and review the minimum observed voltages and report back what you see.
View attachment 629371
This test will rule out the PSU. Also, hover over the perfcap reason and report what it shows. Blue = vRel on mine.
Click to expand...
Thank you! Here - 3DMark Time Spy run - AX1600i FWIW

gpuz.png
 
Your voltages look good.

How many NVME's do you have? It could be that your card is running on only 8 lanes, pcie bus speed running at Gen 2 or Gen 3 speeds... But even that wouldn't likely make a huge difference.

You might go into the Bios, and check all of the CPU/Mobo options. I wonder if there is an app that can view PCIe Gen versions and number of lanes in use by various devices..
 
sk3tch said:
Thank you! Here - 3DMark Time Spy run - AX1600i FWIW

View attachment 629380
Click to expand...
I'd have to run a Timespy on my 4090 with this open to compare, but assuming your 2 graphic runs are the ones with the most GPU usage, you can clearly see the GPU on "average" is probably sitting at 60~80% on both tests. I know Timespy on a 4090 is NOT the most demanding, but it IS indicating your GPU is being limited, likely by the CPU and/or memory. I could try and run a test a little later to compare if you want. I assume this was just "Timespy" and not the Extreme?
 
GoodBoy said:
Your voltages look good.

How many NVME's do you have? It could be that your card is running on only 8 lanes, pcie bus speed running at Gen 2 or Gen 3 speeds... But even that wouldn't likely make a huge difference.

You might go into the Bios, and check all of the CPU/Mobo options. I wonder if there is an app that can view PCIe Gen versions and number of lanes in use by various devices..
Click to expand...
Just one SN850X - I will check it out just in case. In my OP I show x16 for PCI Gen4 so should be good?

Thank you all! Yes, Time Spy regular; not Extreme.

Realized that the Cyberpunk 2077 benchmark stats that I found were for the new DLC (Phantom Liberty) - so I can't compare lol. I just have the vanilla Cyberpunk.

EDIT: just updated my main 1080p/360Hz rig to a crazy beta BIOS (I can't help myself) - still performs as expected: https://www.3dmark.com/3dm/106140850.
 
Last edited:
sk3tch said:
Thank you! Here - 3DMark Time Spy run - AX1600i FWIW

View attachment 629380
Click to expand...
So as promised, I ran Timespy with GPUz open so you can compare. I limited my GPU voltage to 1.07V to match closer to you (as my chip is unlocked to 1.1V). I think we found your problem buddy... something is limiting your GPU on that setup, and I'd venture to guess the CPU and/or memory as that would feed the GPU data to process. Either something is not configured right in your bios, memory timings are too loose, or something weird with the 7950X3D combo chip and how it behaves with 3dmark...

You can see the 2 graphics test runs pretty solidly on the GPU Load graph, damn near locked at 100%, and that's on a 5950X w/ DDR4. :)

1705961558375.gif
 
III_Slyflyer_III said:
So as promised, I ran Timespy with GPUz open so you can compare. I limited my GPU voltage to 1.07V to match closer to you (as my chip is unlocked to 1.1V). I think we found your problem buddy... something is limiting your GPU on that setup, and I'd venture to guess the CPU and/or memory as that would feed the GPU data to process. Either something is not configured right in your bios, memory timings are too loose, or something weird with the 7950X3D combo chip and how it behaves with 3dmark...

You can see the 2 graphics test runs pretty solidly on the GPU Load graph, damn near locked at 100%, and that's on a 5950X w/ DDR4. :)

View attachment 629421
Click to expand...
Thanks!

Completely stock - so strange. Can’t be a bad CPU?
 
sk3tch said:
Thanks!

Completely stock - so strange. Can’t be a bad CPU?
Click to expand...
Doubt it unless u are failing prime95, bad CPUs cause crashes, not bad benchmarks.

I'm not familiar with the 7950X3D past specs, but are you able to "all core OC" those chips like the 5xxx series? I'm sure you would have 2 different clocks with the X3D CCD vs normal CCD, but maybe ur PBO boosting is all jacked up, or maybe your OS has no clue whether or not to use the X3D cores or not?

I'm assuming you have updated the AMD chipset drivers and are using the latest bios and actually reset that bios after flashing?

If you run something like Cinebench, where do you land relative to other 7950X3D chips?
 
sk3tch said:
I am not good with computer. Haha. Thank you. Duh on my part.

Here’s 1440p Cyberpunk 2077 bench:

View attachment 629198

EDIT: omg disabled iGPU and things are way smoother now…may be getting somewhere…
EDIT2: nope - same Time Spy GPU 30k... https://www.3dmark.com/3dm/106099465
Click to expand...

I ran this same bench to compare. This Bench pre-dates Phantom Liberty and isn't affected by it. You are on latest 2.1 patch, so it's an identical test.


Changed my res from 3440x1440 to 2560x1440
Changed to the RayTracing Overdrive preset from Custom. (I usually set DLSS to Quality instead of Auto, and disable depth of field and motion blur)
Right side settings are identical to yours. I'm running same Windows version, 64Gb ram, same Driver version.
I do have the game's HDR Support enabled. I should test again with HDR off, not idea if that affects performance.

My score was a little less than yours, but I know I am CPU limited. I'm still on an Intel i9-10900x 10 core 20 thread processor. I'm on an ASUS Rampage VI Extreme Encore on it's latest bios, Rebar is enabled. My Ram is XMP 3466, so significantly slower than XMP 6000. The board is also only PCIe Gen 3.

Intel Core i9-10900x CPU @ 3.70Ghz
Average FPS 135.66
Min FPS 37.13
Max FPS 265.50
Time 64.24
Number of Frames 8709


The point of concern with your result is the Min FPS, 5.86, that's really bad. My Bench runthru was very smooth with maybe 1 place that seemed to drop a frame: for an instant the video motion seemed to jump, it's right after coming out of the door of the club as the camera turns to the right outside in the alley. I do have G-Sync enabled.
I have the game installed on an 8Tb Sabrent NVME.

So, next suggestion is to install CrystalDiskMark 8 64bit, and run the All test against the drive where your game is installed. My Sabrent gets: (My mobo is only PICE Gen 3 so the NVMe is running at half of its capable speed).
1705965876781.png


Also look at how full that NVMe drive is. SSD/NVMe drives write performance takes a massive (s)hit once they are 85% or more full, due to the auto wear-leveling. Probably an article someplace that explains it.

An example of this:
1705966277522.png

Write performance drops to about 8% of what it likely was before he went over 85%.
 
Last edited:
GoodBoy said:
I ran this same bench to compare. This Bench pre-dates Phantom Liberty and isn't affected by it. You are on latest 2.1 patch, so it's an identical test.


Changed my res from 3440x1440 to 2560x1440
Changed to the RayTracing Overdrive preset from Custom. (I usually set DLSS to Quality instead of Auto, and disable depth of field and motion blur)
Right side settings are identical to yours. I'm running same Windows version, 64Gb ram, same Driver version.
I do have the game's HDR Support enabled. I should test again with HDR off, not idea if that affects performance.

My score was a little less than yours, but I know I am CPU limited. I'm still on an Intel i9-10900x 10 core 20 thread processor. I'm on an ASUS Rampage VI Extreme Encore on it's latest bios, Rebar is enabled. My Ram is XMP 3466, so significantly slower than XMP 6000.

Intel Core i9-10900x CPU @ 3.70Ghz
Average FPS 135.66
Min FPS 37.13
Max FPS 265.50
Time 64.24
Number of Frames 8709


The point of concern with your result is the Min FPS, 5.86, that's really bad. My Bench runthru was very smooth with maybe 1 place that seemed to drop a frame: for an instant the video motion seemed to jump, it's right after coming out of the door of the club as the camera turns to the right outside in the alley. I do have G-Sync enabled.
I have the game installed on an 8Tb Sabrent NVME.

So, next suggestion is to install CrystalDiskMark 8 64bit, and run the All test against the drive where your game is installed. My Sabrent gets:
View attachment 629434

Also look at how full that NVMe drive is. SSD/NVMe drives write performance takes a massive (s)hit once they are 85% or more full, due to the auto wear-leveling. Probably an article someplace that explains it.

An example of this:
View attachment 629441
Write performance drops to about 8% of what it likely was before he went over 85%.
Click to expand...
Thank you!

Gave it a test:
CrystalDiskMark_20240122184122.png


Also re-did the Cyberpunk 2077 bench (since I disabled iGPU - same low drop :()

IMG_4998.jpeg
 
Furious_Styles said:
Wonder if you're on to something. Single digit min seems too low.
Click to expand...
Yeah definitely gonna try Cinebench later to see what’s up. No tweaks at all on CPU outside of disabling iGPU. Latest AMD drivers installed…fresh OS.

EDIT: reinstalled AMD drivers and it fixed Cyberpunk 2077 - but 3DMark Time Spy still has between 40% and 80% GPU utilization!

IMG_5001.jpeg


EDIT2: haven't solved 3DMark, yet - but Fortnite is back to smooth as butter when I use Process Lasso Pro and Park Control Pro to give affinity to cores 0-15 (vcache ones) and real-time priority.

So yeah - this is a CCD1 (vcache) vs. CCD2 issue. My other 3 rigs have 7800X3D so only one CCD to deal with. Makes me wanna dump this 7950X3D since I only game, anyway.
 
Last edited:
Yeah this might just be the way that the 7950X3D works.
Your NVMe speeds are good. We've ruled out the GPU, PSU, and NVMe, and you ruled out the OS when trying a reinstall. So Mobo, CPU, or maybe a DDR timing is all that's left. Since the CPU does differ from your other systems, that is the next thing to try.
If you could move a 7800X3D over to this system and test, would be a pita but at least you would know.

And really, even if you do determine that it is the 7950X3D, I don't think it would be a malfunction, but more likely that that's just the way those work. Maybe with some app that controls thread placement you could improve the performance. Chances are the performance is good enough either way.
 
How many (and what kind) of USB devices do you have hooked up, and how many hard disks do you have total? Bad I/O handling can easily destroy your entire system's performance. I have had a windows computer that had an SSD that was going bad just outright refuse to even start, even though it wasn't the boot drive and wasn't even being accessed.

I'm not sure if anyone has suggested it already, but start unhooking peripherals, maybe disconnect your motherboard's USB-C header, take off additional disks, etc. If that solves the issue, then start hooking them back up one by one and seeing which one starts giving you beef. That's just my thought since you've ruled out so many things already.
 
pendragon1 said:
if everything but 3dmark is working properly, get back to gaming....
Click to expand...
Totally agree. However, it has helped me learn that the 7950X3D is not worth it for me. I don't want to deal with this BS. Gonna swap out the CPU and if that fixes it - then the 7950X3D is gone for a 7800X3D. The issue was in game, the comparison across all platforms was 3DMark Time Spy (very difficult to take the game that I main - Fortnite - and figure out consistent performance due to the immense amount of variables within online multiplayer).

I'm too old to tweak my systems to death now. Ain't nobody got time for that. I need consistent, reliable performance. You can tell from my "good" 3DMark scores - they're not exactly chart toppers but they're within spec. :)

StoleMyOwnCar said:
How many (and what kind) of USB devices do you have hooked up, and how many hard disks do you have total? Bad I/O handling can easily destroy your entire system's performance. I have had a windows computer that had an SSD that was going bad just outright refuse to even start, even though it wasn't the boot drive and wasn't even being accessed.

I'm not sure if anyone has suggested it already, but start unhooking peripherals, maybe disconnect your motherboard's USB-C header, take off additional disks, etc. If that solves the issue, then start hooking them back up one by one and seeing which one starts giving you beef. That's just my thought since you've ruled out so many things already.
Click to expand...
Good call. I will look into this. I have every USB port used but they're for devices like cameras and stuff no storage or anything like that.
 
Last edited:
Wow, what a rollercoaster. Went out to Micro Center and bought a new 7800X3D combo (I mean, why not?). Tossed in the 7800X3D - orange light and C5 Q-code (memory) - reset CMOS, did a ton of things - finally got to swapping RAM sticks (ugh) - even though they passed full memory tests ONE STICK IS BAD! Put one stick in - it works. Put both in - doesn't work. Put one stick in - doesn't work. Gonna hit up G.Skill.

I'll boot to the single DIMM and see if my graphics score has improved.

EDIT: all fixed. WOW. https://www.3dmark.com/3dm/106195500 - this is running a 4800MHz and a single DIMM too lol...

a single bad DIMM...undiagnosed by multiple memory testing tools. whoa.

THANK YOU ALL FOR COMING WITH ON THIS JOURNEY AND FOR YOUR HELP!!! :)
 
Last edited:
sk3tch said:
Wow, what a rollercoaster. Went out to Micro Center and bought a new 7800X3D combo (I mean, why not?). Tossed in the 7800X3D - orange light and C5 Q-code (memory) - reset CMOS, did a ton of things - finally got to swapping RAM sticks (ugh) - even though they passed full memory tests ONE STICK IS BAD! Put one stick in - it works. Put both in - doesn't work. Put one stick in - doesn't work. Gonna hit up G.Skill.

I'll boot to the single DIMM and see if my graphics score has improved.

EDIT: all fixed. WOW. https://www.3dmark.com/3dm/106195500 - this is running a 4800MHz and a single DIMM too lol...

a single bad DIMM...undiagnosed by multiple memory testing tools. whoa.

THANK YOU ALL FOR COMING WITH ON THIS JOURNEY AND FOR YOUR HELP!!! :)
Click to expand...
Not to spoil the fun, but are you absolutely sure its the RAM and not a bad channel on the IMC of the new 7800X3D you picked up? I would think it would have failed with the 7950X3D too as they have the same I/O die (and memory controller).
 
III_Slyflyer_III said:
Not to spoil the fun, but are you absolutely sure its the RAM and not a bad channel on the IMC of the new 7800X3D you picked up? I would think it would have failed with the 7950X3D too as they have the same I/O die (and memory controller).
Click to expand...
lol you sure did...man. I used a known good set from another rig (same exact memory) and it goes orange and C5 Q-code just like before. Now you have me wondering if it is a bad memory slot on the board!? The slot closest to the CPU (there are only two on the X670E Gene) is the one that seems to cause issues with memory installed.

So now...swap CPU again or swap board?! lol what a mess
 
You tried the suspected bad stick in single slot config and had issues right? If so that's likely the culprit. I take it these 2 ram sticks were from the old system?

Ram timings + mobo's can be tricky. It's always a good idea to check your specific mobo's QVL Qualified Vendor List for ram that they confirm is compatible. Heck all you might have needed to do was set Ram to Custom settings and bump the voltage by .02 volts to maybe .05 volts. My ram is supposed to be at 1.35v but when it was on auto the measured reading in the bios would be 1.34. Do manual on that value and bump it a little bit, can clear up stability issues. Most of the time the automatic + XMP and it's fine and everything works. Trying a slower speed/loosen the timings can help identify issues. You probably have a single chip on one dimm that is barely out of spec causing all of your problems.

Be a good test to try 2 new sticks in the old system with the 7950X3D and see how it performs. If that fixed it at least you would know that the mobo and cpu are ok.
 
sk3tch said:
Wow, what a rollercoaster. Went out to Micro Center and bought a new 7800X3D combo (I mean, why not?). Tossed in the 7800X3D - orange light and C5 Q-code (memory) - reset CMOS, did a ton of things - finally got to swapping RAM sticks (ugh) - even though they passed full memory tests ONE STICK IS BAD! Put one stick in - it works. Put both in - doesn't work. Put one stick in - doesn't work. Gonna hit up G.Skill.

I'll boot to the single DIMM and see if my graphics score has improved.

EDIT: all fixed. WOW. https://www.3dmark.com/3dm/106195500 - this is running a 4800MHz and a single DIMM too lol...

a single bad DIMM...undiagnosed by multiple memory testing tools. whoa.

THANK YOU ALL FOR COMING WITH ON THIS JOURNEY AND FOR YOUR HELP!!! :)
Click to expand...
Glad you got it sorted but mem sticks can only be tested using swap method. Any other method is quite unpredictable. These tests are outdated and horseshit for checking for real errors. Can possibly detect issues with overclock / some basic stability. At least my 2 cents. That is why I also don't trust when people post these tests and say they are running their mem at thousands of MHz because I know a few games that will insta crash those systems while memtest or prime show stability.

Sorry I couldn't come to this thread sooner to reinforce my earlier message about swapping mem sticks and not relying on tests. If you still have issues I suggest:
1) Swap with known good sticks. If ram slot is an issue just run one known good stick in the slot that you suspect is causing the error. Will be faster to detect than to swap the full motherboard.
2) Swap processor
3) Swap motherboard

In that order. You have enough systems to do this and unless you are on custom water cooling, the above should not take more than 10-15 mins to find out the culprit. Don't rely on software based tests except the one you are trying to maximize (3dmark).
 
Last edited:
KickAssCop said:
Glad you got it sorted but mem sticks can only be tested using swap method. Any other method is quite unpredictable. These tests are outdated and horseshit for checking for real errors. Can possibly detect issues with overclock / some basic stability. At least my 2 cents. That is why I also don't trust when people post these tests and say they are running their mem at thousands of MHz because I know a few games that will insta crash those systems while memtest or prime show stability.

Sorry I couldn't come to this thread sooner to reinforce my earlier message about swapping mem sticks and not relying on tests. If you still have issues I suggest:
1) Swap with known good sticks. If ram slot is an issue just run one known good stick in the slot that you suspect is causing the error. Will be faster to detect than to swap the full motherboard.
2) Swap processor
3) Swap motherboard

In that order. You have enough systems to do this and unless you are on custom water cooling, the above should not take more than 10-15 mins to find out the culprit. Don't rely on software based tests except the one you are trying to maximize (3dmark).
Click to expand...
Yeah...eye opening for me.

So, here's what happened:

- 7950X3D installed - issues with maximizing graphics performance in 3DMark and stuttering in games.
- 7800X3D installed - immediately only one DIMM works. The same slot every time. Memory was not changed at all (left in place and had been in place for months) - just CPU swapped.
- Swapped sticks from within system, tried both slots. Both sticks always work in the secondary slot (the one away from the mobo - GENE only has two slots)
- Took known good sticks from another system - same behavior as above.
- So now I'll see if both sticks work with the 7950X3D (will swap back) - then I know I have a bad IMC on the new 7800X3D. If not, may be that the mem slot finally gave out (I had intermittent booth issues with the 7950X3D in place) and I need to start an RMA with ASUS for the X670E GENE board. It works with one stick just fine for gaming so I won't swap out mobos until I get a replacement (GENE boards are hard to find new right now).

Does that jive?
 
Makes sense. Once you rule out memory, I would think it is a board issue but to confirm you will need to slot back the 7950X3D. IMC issues are rare.
 
KickAssCop said:
Makes sense. Once you rule out memory, I would think it is a board issue but to confirm you will need to slot back the 7950X3D. IMC issues are rare.
Click to expand...
Thanks for the validation! Hopefully I can submit this thread to ASUS RMA...lol
 
sk3tch said:
Thanks for the validation! Hopefully I can submit this thread to ASUS RMA...lol
Click to expand...
Hopefully they don't give you trouble. I can only imagine them asking 50 questions after not reading your message about how you have multiple systems and tested this thoroughly.
 
sk3tch said:
Yeah...eye opening for me.

So, here's what happened:

- 7950X3D installed - issues with maximizing graphics performance in 3DMark and stuttering in games.
- 7800X3D installed - immediately only one DIMM works. The same slot every time. Memory was not changed at all (left in place and had been in place for months) - just CPU swapped.
- Swapped sticks from within system, tried both slots. Both sticks always work in the secondary slot (the one away from the mobo - GENE only has two slots)
- Took known good sticks from another system - same behavior as above.
- So now I'll see if both sticks work with the 7950X3D (will swap back) - then I know I have a bad IMC on the new 7800X3D. If not, may be that the mem slot finally gave out (I had intermittent booth issues with the 7950X3D in place) and I need to start an RMA with ASUS for the X670E GENE board. It works with one stick just fine for gaming so I won't swap out mobos until I get a replacement (GENE boards are hard to find new right now).

Does that jive?
Click to expand...

I don't see anywhere where you tried another motherboard in this list. If you do have the spare systems, like Cop said, you might wish to try that.

To be frank, I think I have seen more than one issue with ASUS in this gen, for AM5. I would not pick ASUS as my choice of boardmaker for this chipset. I will be curious to see how your RMA goes, too. I have mainly heard bad things regarding ASUS, but we'll see.
 
StoleMyOwnCar said:
I don't see anywhere where you tried another motherboard in this list. If you do have the spare systems, like Cop said, you might wish to try that.

To be frank, I think I have seen more than one issue with ASUS in this gen, for AM5. I would not pick ASUS as my choice of boardmaker for this chipset. I will be curious to see how your RMA goes, too. I have mainly heard bad things regarding ASUS, but we'll see.
Click to expand...
Absolutely. That's the nuclear option for me - but I am ready. The benefit of getting that 7800X3D in a combo from Micro Center is that I have a brand new Gigabyte B650 Gaming X AX mobo ready. So when I put my 7950X3D back and am able to verify that both 3DMark works and both DIMM slots (or one DIMM slot) works - if it's a bad DIMM slot I will swap the board out while I wait for my RMA'd GENE to get back.
 
sk3tch said:
Absolutely. That's the nuclear option for me - but I am ready. The benefit of getting that 7800X3D in a combo from Micro Center is that I have a brand new Gigabyte B650 Gaming X AX mobo ready. So when I put my 7950X3D back and am able to verify that both 3DMark works and both DIMM slots (or one DIMM slot) works - if it's a bad DIMM slot I will swap the board out while I wait for my RMA'd GENE to get back.
Click to expand...
One thing I've found out while repairing/testing some AM5 boards is that many simply won't work if you use certain dimm slots. Like A2B2 works properly but A1B1 simply won't post. A1A2 and B1B2 works but at reduced speed sometimes. I don't have 4 memory sticks atm but the manuals all say they should work fine.
 
Furious_Styles said:
One thing I've found out while repairing/testing some AM5 boards is that many simply won't work if you use certain dimm slots. Like A2B2 works properly but A1B1 simply won't post. A1A2 and B1B2 works but at reduced speed sometimes. I don't have 4 memory sticks atm but the manuals all say they should work fine.
Click to expand...
Yeah, bad thing with the GENE is it has only two memory slots. So one not working doesn't exactly allow for a workaround. :)
 
sk3tch said:
Yeah, bad thing with the GENE is it has only two memory slots. So one not working doesn't exactly allow for a workaround. :)
Click to expand...
Yes in that case there's no doubt about it. As cop said earlier memory problems blow, I too have never found any real test that works consistently for it.
 
Swapped the 7950X3D - wow, I want off of this ride...memory problems confirmed - one DIMM slot bad on mobo. 3DMark Time Spy - still with a graphics score of around 30K (7800X3D scored 38K with one DIMM - same config).

Not even sure what to do at this point. I guess I will start the RMA process with ASUS for the board - but how do I handle the CPU situation? Is the 7950X3D faulty somehow? I'd hate to sell it to someone to recoup my costs and have them get this experience. But it hasn't exhibited any issues at all outside of this strange behavior. Maybe the board is just cursed. I'll run Cinebench to see...
 
sk3tch said:
Swapped the 7950X3D - wow, I want off of this ride...memory problems confirmed - one DIMM slot bad on mobo. 3DMark Time Spy - still with a graphics score of around 30K (7800X3D scored 38K with one DIMM - same config).

Not even sure what to do at this point. I guess I will start the RMA process with ASUS for the board - but how do I handle the CPU situation? Is the 7950X3D faulty somehow? I'd hate to sell it to someone to recoup my costs and have them get this experience. But it hasn't exhibited any issues at all outside of this strange behavior. Maybe the board is just cursed. I'll run Cinebench to see...
Click to expand...
Did 7950X3D produce 38K with single dimm?
 
KickAssCop said:
Did 7950X3D produce 38K with single dimm?
Click to expand...
Sorry if that wasn't clear - no. Still 30K.

Did Cinebench R23 - now mind you - only one DIMM at EXPO I (CL30 6000MHz) - but it's 5K off what I saw is "normal" (around 38K):

EDIT: May not be too far off especially since it’s one DIMM and I have a 280MM AIO. No throttling occurred but it hit the 90C ceiling quickly.

cinebench_r23_7950x3d.png
 
Last edited:
You must log in or register to reply here.
Back
Top