Overclock.net banner
Cancel
Create thread New Forums

[Official] Zen 5 X3D Owner's Club (9800x3D / 9900x3D / 9950x3D)

Tags
3d vcache 9700x3d 9800x3d 9950x3d zen 5
20 reading
Jump to Latest
5,501 - 5,520 of 9,494 Posts
#5,501 · (Edited)
I mean in my small testing so far you are paying $250 for about 1% boost in performance. It can go up to 5750MHz with PBO +200 but rarely ever stays there. It is usually bouncing between 5575MHz and 5400MHz. The clocks are not stable which is the complete opposite of my 9800X3D. The voltage is also higher and a negative curve offset does not help with the clock stretching. Surprisingly so far neither does a positive curve offset but I have only tried up to +10. This is with CCD 1 disabled. So to sum it up right now the clocks bounce around a lot. There is clock stretching and compared to the 9800X3D performance is about 1% better give or take.
 
#5,502 · (Edited)
Sam_Oslo said:
It's really selling for $699 USD, but the price is too close to 9800x3d and 9950X for AMD's own good. Nobody would buy any of these any longer.

This is going to affect the price of the whole range, I guess.
Click to expand...
I don’t know as I have one and this is definitely not a gaming first chip so I think the prices are fair. The extra clock speed is not really represented in the actual performance because the clocks don’t hold in the games I would need the extra CPU boost like UE5 games. They bounce around constantly for some reason even with CCD 1 disabled. I am using my RTX 5090 and even in Port Royal my best score with my 9800X3D was 41,099. With this CPU I can’t get past 40,760. The highest peaks for the clock speed are indeed higher than the 9800X3D but the average is lower. HWINFO64 still sees the disabled CCD and it is giving off heat apparently sitting at 25-27 degrees Celsius.
 
#5,503 ·
cta62 said:
I mean in my small testing so far you are paying $250 for about 1% boost in performance. It can go up to 5750MHz with PBO +200 but rarely ever stays there. It is usually bouncing between 5575MHz and 5400MHz. The clocks are not stable which is the complete opposite of my 9800X3D. The voltage is also higher and a negative curve offset does not help with the clock stretching. Surprisingly so far neither does a positive curve offset but I have only tried up to +10. This is with CCD 1 disabled. So to sum it up right now the clocks bounce around a lot. There is clock stretching and compared to the 9800X3D performance is about 1% better give or take.
Click to expand...
What are you cooling it with?
 
Save
Reply Quote Rep+ Rep+
#5,505 · (Edited)
From what I can see the reviews that showed a bigger gap in performance didn’t have PBO enabled on the 9800X3D. Stock against stock the 9950X3D beats it by about 5-6 percent. While that isn’t amazing it is something. The gap closes when both are PBO +200 because the 9950X3D can’t hold a steady 5750MHz. This causes the 9800X3D to catch up and the difference is only about 1% or tied. One positive is it should be an XOC overclocking dream. I am not comfortable increasing the voltage more than +10 on CO but if someone doesn’t mind blowing up their chip you can pump more voltage into it and have it hold the clocks better. When I added +10 on CO it held 5575MHz solidly but also ran 7 degrees Celsius warmer. For the average enthusiast without some elaborate water cooling loop this CPU is a waste of time so far. It is starved of voltage out of the box for the clocks it is trying to achieve so you have clock stretching. If you can keep it cool and don’t mind pumping a lot of voltage into it you are rewarded with higher clocks. Just like the 9800X3D with eCLK. If you aren’t comfortable doing so and are trying to keep voltages reasonable you basically have a 9800X3D that does PBO +250 instead of +200. If you need the extra cores you have a solid CPU. For those like me hoping the higher clocks on the cache CCD would of added something to the gaming performance in CPU hungry titles? This does not appear to be it unless you have a strong water cooling setup and are not afraid to add voltage to the CPU. This CPU has strong XOC potential though with the right cooling.
 
#5,506 ·
cta62 said:
From what I can see the reviews that showed a bigger gap in performance didn’t have PBO enabled on the 9800X3D. Stock against stock the 9959X3D beats it by about 5-6 percent. While that isn’t amazing it is something. The gap closes when both are PBO +200 because the 9950X3D can’t hold a steady 5750MHz. This causes the 9800X3D to catch up and the difference is only about 1% or tied. One positive is it should be an XOC overclocking dream. I am not comfortable increasing the voltage more than +10 on CO but if someone doesn’t mind blowing up their chip you can pump more voltage into it and have it hold the clocks better. When I added +10 on CO it held 5575MHz solidly but also ran 7 degrees Celsius warmer. For the average enthusiast without some elaborate water cooling loop this CPU is a waste of time so far. It is starved of voltage out of the box for the clocks it is trying to achieve so you have clock stretching. If you can keep it cool and don’t mind pumping a lot of voltage into it you are rewarded with higher clocks. Just like the 9800X3D with eCLK. If you aren’t comfortable doing so and are trying to keep voltages reasonable you basically have a 9800X3D that does PBO +250 instead of +200. If you need the extra cores you have a solid CPU. For those like me hoping the higher clocks on the cache CCD would of added something to the gaming performance in CPU hungry titles? This does not appear to be it unless you have a strong water cooling setup and are not afraid to add voltage to the CPU. This CPU has strong XOC potential though with the right cooling.
Click to expand...
Thanks for sharing your observations. Would disabling the non-X3D CCD allow higher sustained boost clocks than a 9800X3D?
 
Save
Reply Quote Rep+ Rep+
#5,507 ·
I did all of my testing with the other CCD disabled only having the cache one enabled. Despite disabling the other CCD in the BIOS it still idles around 27 degrees Celsius according to HWINFO64. Windows obviously does not see it anymore though.
 
#5,508 ·
cta62 said:
When I added +10 on CO it held 5575MHz solidly but also ran 7 degrees Celsius warmer.
Click to expand...
That's not the proper way to overclock these CPUs. You have use negative CO (preferably -50 on some) to reduce the voltage and increase the clock on good cores.

You are jumping to conclusions too early. . Take your time for a proper tuning, with a good per-core CO and Curve Shaper, RAM overclock, etc., and do a proper test before wining about hardware problems.
 
Save
Reply Quote Rep+ Rep+
#5,509 · (Edited)
If you read what I wrote I already tried a negative CO offset. All that rewarded me with was more clock stretching. I can do -27 CO stock and it holds 5475MHz-5500MHz no problem. It has a crazy thirst for voltage past 5525MHz. To reduce clock stretching when you enable PBO +200 you need to change to a positive CO offset to reduce the clock stretching. I have played with it all day today I know the limitations of the chip. It behaves like a good binned 9800X3D. There is no magic here that we were all hoping for. The same wall is there and after 5525MHz this chip wants and desires more voltage. If you are comfortable with that you can run higher clocks. Just like with the way you would overclock a 9800X3D using eCLK. Also not sure why you said I am whining by sharing my observations. I am pointing out where this chip excels and where it leaves more to be desired. It is a great chip for people who do other things besides just gaming. If you were hoping to gain something from the higher clock speed ceiling you need to feed it voltage.
 
#5,510 ·
cta62 said:
I can do -27 CO stock and it holds 5475MHz-5500MHz no problem.
Click to expand...
Are you using -27 on all cores? Then one dud core is enough to render the whole CPU dud. That's a dud setup when you have so many cores.

When you have so many cores, you have to do a proper per-core tuning. When you have 2 CDD, you have to do a proper Curve shaping, etc. etc. . Did you do all this?

Don't jump to convolutions before a proper tuning!
 
Save
Reply Quote Rep+ Rep+
#5,511 · (Edited)
I have done all testing with CCD 1 disabled. I have no use for the frequency CCD. I have not tested individual cores more than about 10-15 minutes each so far. It still requires increasing the voltage when pushing the PBO clocks higher unless you are willing to disable a bad core if that turns out to be the case but I am sure it is not. I am going to do an individual core test now actually to see how low I can take the curve without it crashing with the fMAX at 5550MHz. For higher frequencies you are talking about a positive CO offset no matter how much you don’t want to believe it. That is the reality. You are free to purchase one as well and see how it does.
 
#5,513 ·
Save
Reply Quote Rep+ Rep+
#5,514 ·
cta62 said:
I did all of my testing with the other CCD disabled only having the cache one enabled. Despite disabling the other CCD in the BIOS it still idles around 27 degrees Celsius according to HWINFO64. Windows obviously does not see it anymore though.
Click to expand...
Did you disables CCD1 cores through the AMD overclocking menu in the BIOS?

You shouldn't be seeing those cores in HWInfo64.
 
Save
Reply Quote Rep+ Rep+
#5,515 ·
AdamK47 said:
Did you disables CCD1 cores through the AMD overclocking menu in the BIOS?

You shouldn't be seeing those cores in HWInfo64.
Click to expand...
I did and it was the same on the 7950X3D I had. You don’t see them anywhere else for usage or anything. Just temperature. Are you sure that is the case or just assuming? Have you had a 7950X3D before? I mean I guess it could be my ASUS X670E Extreme board but I highly doubt it.
 
#5,516 ·
What's the highest safe voltage for SOC and CPU core voltage for 9950X3D?
 
Save
Reply Quote Rep+ Rep+
#5,519 ·
cta62 said:
I did and it was the same on the 7950X3D I had. You don’t see them anywhere else for usage or anything. Just temperature.
Click to expand...
If they are disabled, then I don't think they are generating any heat. Those temperature you are seeing comes from the other CCD, somehow, some trough the heat sink and such.

cta62 said:
That is why I said this CPU is going to be an XOC dream. Memory overclocking is going to be where this thing shines. The extra clock speed on the cache not so much.
Click to expand...
That's really good news. So there is something good with it, after all ;)
I bet there is more goodies, you just need to get to know this new beast and take your time to tune it.
 
Save
Reply Quote Rep+ Rep+
#5,520 · (Edited)
domdtxdissar said:
Ever wondered whats needed to beat synced 6600/2200 in pure latency ?
Wonder no longer, all it takes is synced 8800/2200 🤣
  • 9950X3D ES
  • New G.SKILL 6000 CL26 @ 8800MT/s CL32 💪
  • 1.74 VDD / 1.55 VDDQ needed for these settings
  • 1,500 CPU VDDIO
  • 1.1 CLDO VDDP
Pyprime 2B
View attachment 2701747
Click to expand...
Good stuff man! What voltage does your chip use with PBO +200 and motherboard limits in one run of CB R23? Also CCD 1 disabled and no CO offset. You said with +200 your sample would be fMAX 5550MHz which is what mine is stock. Would be cool to compare.
 
5,501 - 5,520 of 9,494 Posts
About this Discussion
9.5K Replies
519 Participants
Last post:
Tatili
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling