• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

What is most realistic all core overclock I could get on a Ryzen 7 7700X and 7950X or 13900K/13700K with e-cores off using Noctua NH-D15

ambient cooling using a custom waterloop- all core overclock without modification to the IHS or removable. Intel 13th Gen (5.7 GHz), Ryzen 7000 (5.3-5.4 GHz). NH-D15 probably will be 200-400 MHz lower. After all the top clocks needs more voltage.
 
ambient cooling using a custom waterloop- all core overclock without modification to the IHS or removable. Intel 13th Gen (5.7 GHz), Ryzen 7000 (5.3-5.4 GHz). NH-D15 probably will be 200-400 MHz lower. After all the top clocks needs more voltage.


Would those be with rock solid stability or just game stable do you think that would crash Linpack XTREME or Prime95??

And if I use the Thermalright bracket for LGA 1700 or AMD AM5, would that help the NH-D15S get better contact for better temps and better overclocking headroom
 
It's unwise to expect much variance in the silicon lottery in today's semiconductor marketplace.

These companies have many avenues of properly diverting parts to places that will net higher profits whether it be another retail SKU or an OEM-only part for a system builder (Dell, HP, Lenovo, etc.) or a datacenter customer.
 
Thermalright bracket for LGA 1700
My motherboard has a backplate, I have good temps with my i7, not using such bracket.

I think if you buy a motherboard with backplate you might not need that bracket.
 
It's unwise to expect much variance in the silicon lottery in today's semiconductor marketplace.

These companies have many avenues of properly diverting parts to places that will net higher profits whether it be another retail SKU or an OEM-only part for a system builder (Dell, HP, Lenovo, etc.) or a datacenter customer.


So are you saying variances in silicon between chips is very little now?? And there is no more lottery really? All chips are pretty much the same??
 
Would those be with rock solid stability or just game stable do you think that would crash Linpack XTREME or Prime95??

And if I use the Thermalright bracket for LGA 1700 or AMD AM5, would that help the NH-D15S get better contact for better temps and better overclocking headroom
ambient cooling using a custom waterloop- all core overclock without modification to the IHS or removable. Intel 13th Gen (5.7 GHz), Ryzen 7000 (5.3-5.4 GHz). NH-D15 probably will be 200-400 MHz lower. After all the top clocks needs more voltage.
You're not going to get those 5.7 GHz all core frequencies with a D15 as Ircow has said, this means any all core OC will lose at least 200 MHz ST from the stock boost configuration.

If I were you I would aim for a per core not an all core OC.

The chip already boosts to 5.5 GHz all core if it's not too hot, with 5.8 on two cores. I would go for 5.8 on two cores, 5.7 on four, 5.6 on eight. You'll probably start thermal throttling anyway even with that setting.
 
You're not going to get those 5.7 GHz all core frequencies with a D15 as Ircow has said, this means any all core OC will lose at least 200 MHz ST from the stock boost configuration.

If I were you I would aim for a per core not an all core OC.

The chip already boosts to 5.5 GHz all core if it's not too hot, with 5.8 on two cores. I would go for 5.8 on two cores, 5.7 on four, 5.6 on eight. You'll probably start thermal throttling anyway even with that setting.


Are you referring to the Ryzen 7950X cores?? And correct me if I am wrong but there is no way to static overclock individual cores? Though with PBO I could do it. But I could static set each CCD I believe.

How about the 7700X?? Is it a worse binned CCD than the best 8 core CCD in a 7950X usually??

And could I get good and even better boosting doing the DeBauer delid:


And could I use the NH-D15 with it
 
So are you saying variances in silicon between chips is very little now?? And there is no more lottery really? All chips are pretty much the same??

The delta is a lot smaller than years ago. Chip manufacturers are putting a lot of the overclockable capacity in the boost clock.

Why sell a chip that has a 1500 MHz base clock/2000 MHz boost clock that can overclock to 2500 MHz when they can charge more for one that has a boost clock of 2400 MHz (that overclocks to 2500 MHz)?
 
Last edited:
  • Like
Reactions: Kei
Got my 7700x all core overclock on air with ai overclocking to 5.65 ghz fyi.
 
The screen shot below from CPUID HWMonitor shows my new 7950X with a Noctua NH-D15 Chromax Black running a video stabilization program. Motherboard is Asus Prime X670 WiFi running stock settings. No PBO, overclock or anything else applied. I haven't run any stress tests yet.
 

Attachments

  • Video Stabilize.jpg
    Video Stabilize.jpg
    151.6 KB · Views: 706
The screen shot below from CPUID HWMonitor shows my new 7950X with a Noctua NH-D15 Chromax Black running a video stabilization program. Motherboard is Asus Prime X670 WiFi running stock settings. No PBO, overclock or anything else applied. I haven't run any stress tests yet.
Your CPU in that screenshot is only using 85 watts. Put a real load on it :)
 
  • Like
Reactions: Kei
That's true. The instantanenous power consumption of my 7950X was only 87.08W when the screenshot was taken. Peak power was 136.61W. Unfortunately I cannot perform an Aida64 stress test for the next 2 days and 16 hours, until the program has finished stabilizing the 1 hour video shot from a moving vehicle. N.B. Most of the video processing is taking place on the NVidia GPU, leaving the CPU just ticking over in the background. I'll try a CPU stress test when it has finished.
 
Not sure if this video will help some but this guy achieved 7.747 ghz all core oc on 7700x with 8% delta gains
Also cpu Id shows all core clocks as high as 5.999 ghz. FYI.
 
Were these overclocks with PBO and curve optimizer or static frequency set in BIOS meaning all core all the time?
 
I'm still running in a new 7950X system after loading the most recent BIOS into an Asus Prime X670-P WiFi motherboard with 64GB DDR5 4800MHz RAM. The cooler is a Noctua NH-D15 Chromax Black. The intended use for this computer is video conversion and photo editing.

The Puget Systems web site shows virtually no improvement applying 'Core Performance Boost' or 'Precision Boost Overdrive' for content creation software. It seems pointless running the CPU up to 95C for several days on end, just to shave a few minutes off a particularly long video conversion. I'm more interested in long term stability than absolute speed.


The chart below shows 7950X clock speeds and core temperatures at stock BIOS settings (no CPB or PBO) and no manual overclock, running AIDA64 Stress Test.

1671530647312.jpeg
 
  • Like
Reactions: Kei
1) turning off e cores is ridiculous
2) if talking about multithreaded workloads the question isnt whether you can overclock, but whether you can even run stock clocks without throttling.
4) i have a u12a and it hits 95c on 330watts running ycruncher at stock. There is no such thing as overclocking for mt workloads, you cannot cool it unless we are talking about chillers
5) for games an oc is much more manageable. I can run 5.8 on cyberpunk, but then the problem is the cpu is drawing over 200 watts doing that.

The best course of action for a 13900k is leaving it at stock (with ecores on for gods sake), maybe clock your ring (4.5 to 5ghz ring is 1.5ns of latency) and tune the crap out of your ram.
 
1) turning off e cores is ridiculous
2) if talking about multithreaded workloads the question isnt whether you can overclock, but whether you can even run stock clocks without throttling.
4) i have a u12a and it hits 95c on 330watts running ycruncher at stock. There is no such thing as overclocking for mt workloads, you cannot cool it unless we are talking about chillers
5) for games an oc is much more manageable. I can run 5.8 on cyberpunk, but then the problem is the cpu is drawing over 200 watts doing that.

The best course of action for a 13900k is leaving it at stock (with ecores on for gods sake), maybe clock your ring (4.5 to 5ghz ring is 1.5ns of latency) and tune the crap out of your ram.


Turning off e-cores is a good thing and very valid and more thermal headroom and some do not like the e-cores and hybrid arch crap. Of course P cores take more power than e0cores, but e cores take a lot and reduce thermal footprint and better all core overclock for P cores on air

Intel has no option to buy an 8 core CPU without e-cores so turn them off and you get the best 8 core 16 thread CPU in existence so another valid reason to turn off the e-cores.

I can run 5.6HGHz ALL CORES 1.32 VCore LLC6 ON NH-D15S with low 90s load temps CInebench R23 and mid 80s CPU-Z stress test with 2 140mm fans onky 1000 RPM.

Turning off e-cores is not ridiculous and stop trying to tell others otherwise.
 
Turning off e-cores is a good thing and very valid and more thermal headroom and some do not like the e-cores and hybrid arch crap. Of course P cores take more power than e0cores, but e cores take a lot and reduce thermal footprint and better all core overclock for P cores on air

Intel has no option to buy an 8 core CPU without e-cores so turn them off and you get the best 8 core 16 thread CPU in existence so another valid reason to turn off the e-cores.

I can run 5.6HGHz ALL CORES 1.32 VCore LLC6 ON NH-D15S with low 90s load temps CInebench R23 and mid 80s CPU-Z stress test with 2 140mm fans onky 1000 RPM.

Turning off e-cores is not ridiculous and stop trying to tell others otherwise.
Low 90s huh? Very impressive, turning off ecores really helps there
 

Attachments

  • 41161.JPG
    41161.JPG
    225.8 KB · Views: 320
Low 90s huh? Very impressive, turning off ecores really helps there


They run very hot no way around it especially on air as you said when clocks get high. First CPU to hit mid 5GHz or even higher without throttling on a large air cooler with excellent IPC. There is no way I would have been able to get anywhere near not throttling with 16 e-cores as well on in Cinebench with P cores set to 5.6GHz on air cooler. With the e-cores off I can do it in 90 to 91C very low and mid 80s full load CPU-Z stress test. No chance that happens with e-cores on and being used. Even good AIOs struggle with all cores active without manual tuning.

Clocks would have had to be much lower with e-cores on to not throttle/hit 100C in Cinebench with a large air cooler instead of AIO or custom loop.
 
Throwing perf away when you turn off those e-cores. Say you can go 5.8GHz all core with just P enabled and with both p/e 5.6. that 200 mhz will not make up for a bunch of e-cores at 4.3Ghz.

If your strictly talking about gaming, neither scenario will yeild more fps since 99.9% likely your GPU bound anyways.
 
Throwing perf away when you turn off those e-cores. Say you can go 5.8GHz all core with just P enabled and with both p/e 5.6. that 200 mhz will not make up for a bunch of e-cores at 4.3Ghz.

If your strictly talking about gaming, neither scenario will yeild more fps since 99.9% likely your GPU bound anyways.

If Intel had an 8 core only Raptor Cove with no e-cores and 36MB of L3 cache I would have gotten it.

I do not and never wanted the e-cores.
 
Back
Top