We overclock the AMD Ryzen Threadripper 7960X up to 5715 MHz with the GIGABYTE TRX50 Aero D motherboard and EK custom loop water cooling.

https://www.youtube.com/watch?v=IhSo9r-QTwU

It is my second Ryzen Threadripper 7000 overclocking guide. Despite the overclocking process being very similar to the 7980X, I learned some new things about Ryzen Threadripper 7000 overclocking. I hope you enjoy the blog post!

AMD Ryzen Threadripper 7960X: Introduction

The Ryzen Threadripper 7960X is the lowest-spec processor of AMD’s Zen 4-based Ryzen Threadripper 7000 high-end desktop series codenamed “Storm Peak.” The Storm Peak processors were announced on October 19, 2023, and entered the market as two distinct segments. The Ryzen Threadrippers and the TRX50 chipset are designed for high-end desktop. In contrast, the Ryzen Threadripper PRO and the WRX90 chipset are designed for workstation.

There are a couple of differences between the Threadripper and Threadripper PRO processors:

• Threadripper only supports up to 8 CCDs, so 64 cores, whereas Threadripper Pro supports up to 12 CCDs, up to 96 cores.
• Threadripper only supports 4-channel memory, whereas Threadripper Pro has 8 memory channels.
• Threadripper supports up to 48 PCIe Gen 5 lanes, whereas Threadripper Pro has 112 PCIe Gen 5 lanes.

The TRX50 and WRX90 chipsets are both Promontory 21, identical to the B650 chipset.

The CCDs on the Ryzen Threadripper 7000 are the same as those on the Ryzen 7000 processors … there’s just more of them. That means Zen 4 core architecture, 5nm TSMC, DDR5, and PCIe 5 support.

The CPUs come with a new TR5 or SP6 socket. While there’s a significant difference in pin count versus the Threadripper 5000 predecessor, thus making them incompatible, the physical size of the socket is the same. That means you can repurpose your old TR4/SP3 thermal solutions … as long as they can handle the 350W TDP, of course!

The Ryzen Threadripper 7960X has 24 cores with 48 threads. The base frequency is 4.2 GHz, and the listed maximum boost frequency is 5.3 GHz. However, the programmed Fmax is 5.65 GHz for one CCD and 5.35 GHz for the others. The C-State Boost limiter restricts the core frequency to 4.8 GHz when 8 or more cores are active. The TDP is 350W, and the TjMax is 95 degrees Celsius.

In this guide, we will cover four different overclocking strategies:

• First, we enable Precision Boost Overdrive 2 and enable DOCP.
• Second, we use GIGABYTE’s PBO Enhancement.
• Third, we tune the frequency using the Precision Boost Overdrive 2 toolkit.
• Lastly, we try a manual overclock.

However, before we jump into overclocking, let us quickly review the hardware and benchmarks used in this guide.

AMD Ryzen Threadripper 7960X: Platform Overview

The system we’re overclocking today consists of the following hardware.

Item	SKU	Price (USD)
CPU	AMD Ryzen Threadripper 7960X	1500
Motherboard	GIGABYTE TRX50 Aero D	600
CPU Cooling	EK-Quantum Magnitude sTRX4 EK-Quantum Kinetic FLT 240 D5 EK-Quantum Surface P480M – Black EK-Furious Meltemi 120 (x4)	340 240 160
Fan Controller	ElmorLabs EFC-SB	50
Memory	G.SKILL Zeta R5 Neo DDR5-6800 128GB	1060
Power Supply	XPG FUSION 1600 Titanium	680
Graphics Card	ASUS ROG Strix RTX 2080 TI
Storage	AORUS RGB 512 GB M.2-2280 NVME	60
Chassis	Open Benchtable V2	200

ElmorLabs EFC-SB & EVC2

The Easy Fan Controller SkatterBencher Edition (EFC-SB) is a customized EFC resulting from a collaboration between SkatterBencher and ElmorLabs.

I explained how I use the EFC-SB in a separate blog post on this website. By connecting the EFC-SB to the EVC2 device, I monitor the ambient temperature (EFC), water temperature (EFC), and fan duty cycle (EFC). I include the measurements in my stability test results.

I also use the ElmorLabs EFC-SB to map the radiator fan curve to the water temperature. Without going into too many details, I have attached an external temperature sensor from the water in the loop to the EFC-SB. Then, I use the low/high setting to map the fan curve from 25 to 40 degrees water temperature. I use this configuration for all overclocking strategies.

AMD Ryzen Threadripper 7960X: Benchmark Software

We use Windows 11 and the following benchmark applications to measure performance and ensure system stability.

BENCHMARK	LINK
Geekbench 6	https://www.geekbench.com/
Cinebench 2024	https://www.maxon.net/en/cinebench/
CPU-Z	https://www.cpuid.com/softwares/cpu-z.html
V-Ray 5	https://www.chaosgroup.com/vray/benchmark
Corona 10 Benchmark	https://corona-renderer.com/benchmark
AI-Benchmark	https://ai-benchmark.com/
Y-Cruncher	http://www.numberworld.org/y-cruncher/
Blender	https://opendata.blender.org/
3DMark Night Raid	https://www.3dmark.com/
Handbrake	https://handbrake.fr/
Shadow of the Tomb Raider	https://store.steampowered.com/app/750920/Shadow_of_the_Tomb_Raider_Definitive_Edition/
Final Fantasy XV	http://benchmark.finalfantasyxv.com/na/
OCCT	https://www.ocbase.com/

AMD Ryzen Threadripper 7960X: Stock Performance

Before starting overclocking, we must check the system performance at default settings. The default Precision Boost 2 parameters for the Ryzen Threadripper 7960X are as follows:

• PPT: 350 W
• TDC: 175 A
• EDC: 235 A
• THM: 95 C
• VID: 1.475 V
• FMAX: 5650 / 5350 MHz
• FIT: 2395
• C-State: 8 cores / 4800 MHz

I also want to spend a few minutes highlighting a unique aspect of the Ryzen Threadripper 7000 processors related to overclocking. As I mentioned in my Storm Peak launch post, while all Ryzen Threadripper 7000 processors can overclock, the user must first formally enable overclocking. When the user unlocks the processor, it leaves an irreversible, permanent mark indicating the processor was unlocked for overclocking.

The benefit of unlocking the overclocking capabilities is not just unlocking the overclocking tools but also extra performance. The maximum CPU core frequency is limited to 5350 MHz by default, which is also the default specification on the AMD website. But after unlocking, the Fmax of CCD0 increases by 300 MHz to 5650 MHz.

So, how can you know whether your CPU is unlocked? There are several ways:

1. First, you can try setting an overclock in the BIOS to see if it works. If the overclock is not applied, then it’s possible that overclocking has not been unlocked.
2. Second, you can install Ryzen Master and try opening the software. If your processor is not unlocked for overclocking, the software will ask you to unlock it before showing you the settings.
3. You can also check the BIOS on this GIGABYTE TRX50 Aero D motherboard. If “Processor OC Mode” is disabled under the System Info tab, your processor is locked for overclocking.

For this guide, I present the performance figures after unlocking the processor. Thus, with a maximum boost frequency of 5650 MHZ for CCD0.

Here is the benchmark performance at stock:

• Geekbench 6 (single): 3,000 points
• Geekbench 6 (multi): 23,292 points
• Cinebench 2024 Single: 121 points
• Cinebench 2024 Multi: 2,836 points
• CPU-Z V17.01.64 Single: 777.7 points
• CPU-Z V17.01.64 Multi: 21,961.9 points
• V-Ray 5: 40,591 vsamples
• Corona 10: 17.79 MRays/s
• AI Benchmark: 8,501 points
• Y-Cruncher Pi MT 10B: 144.648 seconds
• Blender (Monster): 349.28 fps
• Blender (Classroom): 186.81 fps
• 3DMark Night Raid: 75,710 marks
• Handbrake (H265, 1080P): 55.42 fps
• Tomb Raider: 194 fps
• Final Fantasy XV: 193.86 fps

Here are the 3DMark CPU Profile scores at stock

• CPU Profile 1 Thread: 1,122
• CPU Profile 2 Threads: 2,216
• CPU Profile 4 Threads: 4,244
• CPU Profile 8 Threads: 7,552
• CPU Profile 16 Threads: 13,977
• CPU Profile Max Threads: 19,066

When running the OCCT CPU AVX2 Stability Test, the average CPU core effective clock is 4632 MHz with 1.152 volts. The average CPU temperature is 72.7 degrees Celsius. The average CPU package power is 349.5 watts.

When running the OCCT CPU SSE Stability Test, the average CPU core effective clock is 4787 MHz with 1.225 volts. The average CPU temperature is 75.4 degrees Celsius. The average CPU package power is 346.0 watts.

Of course, we can increase the maximum power consumption limit using Precision Boost Overdrive. That’s what we’ll do in our first overclocking strategy.

However, before we get going, make sure to locate the CMOS Clear button. Pressing the Clear CMOS button will reset all your BIOS settings to default, which is helpful if you want to start your BIOS configuration from scratch. The Clear CMOS button is located at the bottom of the motherboard.

OC Strategy #1: PBO + EXPO

In our first overclocking strategy, we take advantage of enabling Precision Boost Overdrive 2 and AMD EXPO.

Precision Boost Overdrive 2

With the launch of Zen 3, AMD introduced an improved version of the Precision Boost Overdrive 2 toolkit, allowing for manual tuning of the parameters affecting the Precision Boost frequency boost algorithm. Precision Boost Overdrive 2 builds on the PBO implementation of Zen 2. In addition to the overclocking knobs from Zen+ (PPT, TDC, EDC) and Zen 2 (Boost Override and Scalar), Precision Boost Overdrive 2 also introduced Curve Optimizer.

There are essentially 3 levels of Precision Boost Overdrive

1. AMD’s stock values, which can be set by disabling PBO
2. The motherboard vendor values, which are programmed into the BIOS to match the motherboard VRM specification and can be set by enabling PBO
3. Custom values, which can be programmed by the end-user

I most recently took a deep dive into the Precision Boost Overdrive 2 toolkit in my Ryzen 7000 launch content. If you want to learn more about the impact of each of these settings, I suggest you check out that article.

In this overclocking strategy, we’re just enabling Precision Boost Overdrive, whereas, in the following strategies, we’ll explore tuning the parameters. By enabling Precision Boost Overdrive, we rely on the motherboard pre-programmed PBO parameters. We find that the following values have changed:

EXPO – Extended Profiles for Overclocking

EXPO stands for AMD Extended Profiles for Overclocking. It is an AMD technology that enables ubiquitous memory overclocking profiles for AMD platforms supporting DDR5 memory.

EXPO allows memory vendors such as G.SKILL to program higher performance settings onto the memory sticks. If the motherboard supports EXPO, you can enable higher performance with a single BIOS setting. So, it saves you lots of manual configuration.

AMD introduced EXPO with the launch of the Ryzen 7000 and its transition from DDR4 to DDR5 memory. Even though the Ryzen Threadripper 7000 is only compatible with RDIMM, it still supports the EXPO specification.

BIOS Settings & Benchmark Results

Upon entering the BIOS

• In Easy Mode
  • Set XMP/EXPO Profile to EXPO 1
• Switch to Advanced Mode
• Enter the Advanced CPU Settings submenu
  • Enter the Precision Boost Overdrive submenu
    • Set Precision Boost Overdrive to Enabled

Then save and exit the BIOS.

We re-ran the benchmarks and checked the performance increase compared to the default operation:

• Geekbench 6 (single): +0.53%
• Geekbench 6 (multi): +5.06%
• Cinebench 2024 Single: +0.83%
• Cinebench 2024 Multi: +1.94%
• CPU-Z V17.01.64 Single: +1.08%
• CPU-Z V17.01.64 Multi: +0.28%
• V-Ray 5: +0.36%
• Corona 10: +1.32%
• AI Benchmark: +3.49%
• Y-Cruncher Pi MT 10B: +7.88%
• Blender (Monster): +2.40%
• Blender (Classroom): +1.34%
• 3DMark Night Raid: +0.75%
• Handbrake (H265, 1080P): +0.22%
• Tomb Raider: +2.06%
• Final Fantasy XV: +2.18%

Here are the 3DMark CPU Profile scores

• CPU Profile 1 Thread: +0.09%
• CPU Profile 2 Threads: +0.05%
• CPU Profile 4 Threads: +0.12%
• CPU Profile 8 Threads: +0.25%
• CPU Profile 16 Threads: +0.14%
• CPU Profile Max Threads: +0.03%

As you’ll see in a minute, this 24-core processor barely reaches the 350W TDP limit, even in an all-core workload. Hence, we don’t get any benefit from unlocking the power limits. However, enabling EXPO does provide a minor performance improvement in all-core workloads that are somewhat memory-sensitive. The Geomean performance improvement is +1.46%, and we get a maximum improvement of +7.88% in Y-Cruncher.

When running the OCCT CPU AVX2 Stability Test, the average CPU core effective clock is 4784 MHz with 1.241 volts. The average CPU temperature is 75.8 degrees Celsius. The average CPU package power is 358.0 watts.

When running the OCCT CPU SSE Stability Test, the average CPU core effective clock is 4786 MHz with 1.217 volts. The average CPU temperature is 72.5 degrees Celsius. The average CPU package power is 333.6 watts.

We check this processor’s boost behavior and per core maximum effective clock frequency. The boost frequency at 1 active thread is about 5561 MHZ and trails off to 5083 MHz when seven cores are active. Then, when 8 cores are active, the C-State Boost Limiter kicks in and limits the frequency of all cores to 4.8 GHz. Despite Cores 0 to 5 of CCD0 having a programmed Fmax of 5650 MHz, none of the cores achieve an effective clock higher than 5.5 GHz. For the other CCDs, only three cores achieve over 5.2 GHz even though the programmed Fmax is 5350 MHz.

OC Strategy #2: GIGABYTE PBO Enhancement

In our second overclocking strategy, we use the GIGABYTE PBO Enhancement feature integrated into the GIGABYTE BIOS.

GIGABYTE PBO Enhancement

PBO Enhancement, formerly Performance Bung, is a new performance-oriented technology in the GIGABYTE AM5 and TR5 motherboard BIOSes. Essentially, it’s a list of preset profiles to configure Precision Boost Overdrive. The profiles configure two things specifically:

1. The maximum allowed temperature
2. An all-core Curve Optimizer setting

For example, this strategy’s “90 Level 1” profile will set the maximum temperature to 90 degrees Celsius and an all-core Curve Optimizer of -10.

You can select your preferred profile based on the maximum temperature you’re comfortable with and how well your CPU can undervolt. A lower maximum temperature will yield lower CPU frequency in all-core workloads. A higher negative Curve Optimizer will push the CPU frequency higher at a similar voltage.

Note that PBO Enhancement does not adjust the Precision Boost power limits. So, you want to ensure the PPT, TDC, and EDC are also set correctly to maximize the performance.

BIOS Settings & Benchmark Results

Upon entering the BIOS

• In Easy Mode
  • Set XMP/EXPO Profile to EXPO 1
• Switch to Advanced Mode
• Set Precision Boost Overdrive(PBO) Enhancement to 90 Level 1
• Switch to the Settings menu
• Enter the AMD Overclocking submenu
  • Click Accept
  • Enter the Precision Boost Overdrive submenu
    • Set Precision Boost Overdrive to Advanced
    • Set PBO Limits to Motherboard

Then save and exit the BIOS.

We re-ran the benchmarks and checked the performance increase compared to the default operation:

• Geekbench 6 (single): +2.10%
• Geekbench 6 (multi): +5.65%
• Cinebench 2024 Single: +0.83%
• Cinebench 2024 Multi: +3.14%
• CPU-Z V17.01.64 Single: +1.68%
• CPU-Z V17.01.64 Multi: +0.41%
• V-Ray 5: +0.87%
• Corona 10: +1.42%
• AI Benchmark: +3.99%
• Y-Cruncher Pi MT 10B: +8.29%
• Blender (Monster): +2.85%
• Blender (Classroom): +1.40%
• 3DMark Night Raid: +0.92%
• Handbrake (H265, 1080P): +0.34%
• Tomb Raider: +2.06%
• Final Fantasy XV: +2.30%

Here are the 3DMark CPU Profile scores

• CPU Profile 1 Thread: +1.25%
• CPU Profile 2 Threads: +1.99%
• CPU Profile 4 Threads: +2.05%
• CPU Profile 8 Threads: +1.20%
• CPU Profile 16 Threads: +0.12%
• CPU Profile Max Threads: +0.11%

While PBO Enhancement provides additional performance headroom with the minor Curve Optimizer improvement, it doesn’t lift the maximum frequency. Hence, we don’t expect significant performance uplifts. The Geomean performance improvement is +2.03%, and we get a maximum improvement of +8.29 % in Y-Cruncher.

When running the OCCT CPU AVX2 Stability Test, the average CPU core effective clock is 4782 MHz with 1.242 volts. The average CPU temperature is 81.1 degrees Celsius. The average CPU package power is 350.3 watts.

When running the OCCT CPU SSE Stability Test, the average CPU core effective clock is 4787 MHz with 1.186 volts. The average CPU temperature is 66.5 degrees Celsius. The average CPU package power is 309.0 watts.

We check this processor’s boost behavior and per core maximum effective clock frequency. The boost behavior with increasing active threads is almost identical to the behavior we saw in the previous OC Strategy. Also, the per-core maximum effective clock is essentially the same.

OC Strategy #3: PBO Tuned

In our third overclocking strategy, we tune the CPU’s Precision Boost dynamic frequency technology using the Precision Boost Overdrive 2 toolkit. In addition, we also try to optimize the performance of the infinity fabric, memory and memory controller.

PBO 2: Fmax Boost Override

Fused maximum frequency, or Fmax, is one of the Precision Boost infrastructure limiters constraining the CPU performance. The limiter determines the maximum allowed processor frequency across all CPU cores inside your CPU.

Boost Clock Override or Fmax Override is one of the overclocker tools available in the PBO 2 toolkit. It allows the user to override the arbitrary clock frequency limit between -1000 MHz and +200 MHz in steps of 25 MHz.

We need to make three important notes about Boost Clock Override:

1. The override only adjusts the upper ceiling of the frequency and doesn’t act as a frequency offset. Ultimately, the Precision Boost 2 algorithm still determines the actual operating frequency.
2. Each CCD can have its own programmed Fmax limit. For example, on the Ryzen Threadripper 7960X, CCD0 has an Fmax of 5650 MHz, whereas CCD1 to CCD3 can only boost up to 5350 MHz.
3. The Boost Clock Override also affects the C-state boost limiter which prevents the CPU from boosting over 4.8 GHz when eight or more cores are active.

When we increase the Fmax boost limit by 200 MHz, CCD0’s new Fmax is 5850 MHz, and the other CCDs have a new boost limit of 5550 MHz. In addition, it also lifts the C-State Boost Limiter frequency in an all-core workload by 200 MHz.

PBO 2: Curve Optimizer

Curve Optimizer is an important new feature of Precision Boost Overdrive 2.

Curve Optimizer allows end-users to adjust the factory-fused VFT or voltage-frequency-temperature curve for each CPU core separately. The VFT curve is a unique curve for each core inside your CPU that defines the required voltage for a given frequency at a given temperature. Higher frequencies or higher operating temperatures require higher voltage.

Curve Optimizer adjusts the VFT curve by offsetting the voltages of the factory-fused VFT curve. By setting a positive offset, you increase the voltage point. Conversely, you decrease the voltage point by setting a negative offset. You can offset the entire curve by up to 50 steps in a positive or negative direction. Each step represents approximately 5mV.

The traditional overclocking approach for AMD Ryzen CPUs is to set a negative curve optimizer. When you adjust the VFT curve with a negative point offset, two things happen.

1. First, you tell the CPU that it needs less voltage for a given frequency. And, as a consequence, at a given voltage, it can apply a higher frequency.
2. Second, the CPU temperature will be lower because you use less voltage at a given frequency. That extra thermal headroom will also encourage the Precision Boost algorithm to target higher voltages and frequencies.

In my AMD Raphael overclocking launch content, I explored the ins and outs of curve optimization with Precision Boost Overdrive 2. If you want to learn more about the impact of Curve Optimizer, I suggest you check out that blog post.

On the Ryzen Threadripper 7000, Curve Optimizer is available per CPU, per CCD, or per core. The new feature, Per CCD Curve Optimizer tuning, offers extra flexibility. However, mixing and matching with per-core tuning is impossible. So, you tune Per CPU, Per Core, or Per CCD, but not some chiplets per CCD and others chiplets per core.

Like in the past, per-core tuning offers a real benefit as it provides some cores with a lot more frequency headroom.

Curve Optimizer Tuning Process

As people following me on X might have noticed, I tried the Ryzen Master automatic Curve Optimizer with this processor. The process lasted a long time – I think I had the system run for an entire afternoon – and the results were odd. Ryzen Master recommended -48 for almost all cores, which wasn’t very stable.

So, I switched to my usual manual tuning process. The manual tuning process for Curve Optimizer can become quite convoluted since it affects the CPU core voltage in all scenarios ranging from very light single-threaded workloads to heavy all-core workloads. Usually, I spend a lot of time on per-core curve optimization. I wanted to rely on the OCCT toolkit for this guide to find the optimal Curve Optimizer settings. Here’s my broad approach.

1. I started with a negative curve optimizer of -10 for each core and the Fmax Boost Override set to +200
2. Then, I used the OCCT CPU Stability test with small data set, extreme mode, and steady load type. I cycle through each core every 10 seconds with 2 operating threads, starting with AVX-512, then AVX2, and then SSE. I do this no more than three times for each core.
   1. I increased the negative curve optimizer magnitude by 5 steps for the cores that didn’t crash.
   1. I reduced the negative curve optimizer magnitude by 5 steps for the cores that did crash.
3. Once all cores were dialed in, I ran another OCCT AVX2 and SSE stability test to pass for 30 minutes without any core failing or clock stretching. Again, when a core fails, I reduce the negative curve optimizer magnitude by 5 steps.

Once all that’s done, I run through the usual benchmark suite and stability tests used in this guide. If there are no instabilities, I consider my Curve Optimizer settings to be stable.

FCLK – Infinity Fabric

The Infinity Fabric, sometimes called Scalable Data Fabric, is the primary means of communication between all the chiplets on the Storm Peak package. Every chiplet on the package has a set of GMI links, short for Global Memory Interconnect, via which the CCDs communicate with the IOD.

The infinity fabric frequency and performance are most relevant for performance tuning regarding the memory subsystem performance. Since the DDR5 memory controllers are located on the IO die, if the CPU cores need to store or retrieve data to and from the DDR5 memory, it has to go via the infinity fabric to the IO die.

The Fabric frequency, or FCLK, is generated by the SOC PLL, derived from a 100 MHz reference clock input. The reference clock is multiplied by the FCLK ratio, which you can configure in the BIOS.

The Infinity Fabric frequency is now decoupled from the system memory and CPU memory controller frequency. So you can set it independently. In my case, I further increased the fabric clock frequency to 2200 MHz, an increase of +200 MHz.

If you want to play around with voltages, you should know that the VDDG voltage supply, derived via an integrated voltage regulator from the VDDIO voltage rail, provides the fabric voltage. Eight VDDG voltage rails are available for manual adjustment: two for each pair of CCD-IOD connections.

• CCD0-CCD VDDG: signals sent from CCD0 to IOD are sent at this voltage
• CCD0-IOD VDDG: signals sent from IOD to CCD0 are sent at this voltage
• CCD1-CCD VDDG: signals sent from CCD1 to IOD are sent at this voltage
• CCD1-IOD VDDG: signals sent from IOD to CCD1 are sent at this voltage
• CCD2-CCD VDDG: signals sent from CCD2 to IOD are sent at this voltage
• CCD2-IOD VDDG: signals sent from IOD to CCD2 are sent at this voltage
• CCD3-CCD VDDG: signals sent from CCD3 to IOD are sent at this voltage
• CCD3-IOD VDDG: signals sent from IOD to CCD3 are sent at this voltage

Note that the VDDG voltage does not adjust automatically with VDDIO. So, if you need to increase VDDG, for example, to support higher memory frequency, you need to change it manually.

MCLK – System memory

The DDR5 memory frequency, or MCLK, is derived from the UMCCLK, one of the SOC PLLs. The UMCCLK is driven by a 100 MHz reference clock. The memory controller drives the system memory frequency, which we’ll get to in a minute.

While we still rely on the EXPO settings for memory timings and voltages, we also slightly raise the memory frequency from DDR5-6400 to DDR5-6600 for this OC Strategy. This was the maximum stable frequency we could reach without voltage tuning.

UCLK – Memory Controller

The Unified Memory Controller frequency, or UCLK, is also derived from the UMCCLK, one of the SOC PLLs. The memory controller frequency is tied directly to the system memory frequency. It can run either at the same or half its frequency.

You’d usually prefer to run the memory controller and system memory at the same frequency for performance reasons. However, if your memory can run very high frequencies, then it’s possible running the memory controller at half the frequency still provides better performance.

We run the memory controller frequency in sync with the system memory for this OC Strategy. At DDR5-6600, the memory controller frequency is then 3300 MHz.

Should you wish to adjust the memory controller voltage, look at the VDD_11 voltage rail, which provides the external power for the VDDP_DDR internal voltage regulator. VDDP is the voltage for the DDR bus signaling or DRAM PHY.

As a rule, the external VDD_11 should always be higher than the internal VDDP_DDR + 100mV. Furthermore, the external VDDCR_SOC voltage rail should be lower than the external VDD11 + 100mV. When memory overclocking, you may need to manually increase the VDDP voltage as it does not automatically adjust when changing the VDD_11 voltage.

GIGABYTE High Bandwidth & Low Latency

A last quick setting I want to discuss before moving on to the BIOS settings is the High Bandwidth and Low Latency memory options. I used these before in previous guides, and they seem to provide additional performance benefits, but GIGABYTE is pretty tight-lipped on what it does.

Enabling the settings may cause instability, but it’s up to you to see if your memory is stable when enabled. In my case, I enabled the High Bandwidth and Low Latency options and saw a positive performance impact in benchmarks like Aida64, Geekbench 6 Multi, and AI Benchmark.

BIOS Settings & Benchmark Results

Upon entering the BIOS

• In Easy Mode
  • Set XMP/EXPO Profile to EXPO 1
  • Enable XMP/EXPO High Bandwidth Support
  • Enable Low Latency Support
• Switch to Advanced Mode
• Set System Memory Multiplier to 66.00
• Set Infinity Fabric Frequency and Dividers to 2200 MHz
• Set UCLK DIV1 Mode to UCLK=MEMCLK
• Switch to the Settings menu
• Enter the AMD Overclocking submenu
  • Click Accept
  • Enter the Precision Boost Overdrive submenu
    • Set Precision Boost Overdrive to Advanced
    • Set PBO Limits to Motherboard
    • Set CPU Boost Clock Override to Enabled (Positive)
    • Set Max CPU Boost Clock Override(+) to 200
    • Enter the Curve Optimizer submenu
      • Set Curve Optimizer to Per Core
      • Set Core 0 to 23 Curve Optimizer Sign to Negative
      • Set Core 0 to 3 Curve Optimizer Magnitude to 5
      • Set Core 4 and 5 Curve Optimizer Magnitude to 10
      • Set Core 6 to 23 Curve Optimizer Magnitude to 50

Then save and exit the BIOS.

We re-ran the benchmarks and checked the performance increase compared to the default operation:

• Geekbench 6 (single): +3.50%
• Geekbench 6 (multi): +10.45%
• Cinebench 2024 Single: +2.48%
• Cinebench 2024 Multi: +8.67%
• CPU-Z V17.01.64 Single: +2.25%
• CPU-Z V17.01.64 Multi: +4.47%
• V-Ray 5: +5.56%
• Corona 10: +6.56%
• AI Benchmark: +9.99%
• Y-Cruncher Pi MT 10B: +14.50%
• Blender (Monster): +7.43%
• Blender (Classroom): +5.87%
• 3DMark Night Raid: +3.21%
• Handbrake (H265, 1080P): +4.64%
• Tomb Raider: +3.61%
• Final Fantasy XV: +4.18%

Here are the 3DMark CPU Profile scores

• CPU Profile 1 Thread: +1.52%
• CPU Profile 2 Threads: +0.72 %
• CPU Profile 4 Threads: +2.29%
• CPU Profile 8 Threads: +4.54%
• CPU Profile 16 Threads: +4.49%
• CPU Profile Max Threads: +4.57%

As we try to squeeze the most performance out of the Precision Boost algorithm, we expect some performance bumps. However, unfortunately, there’s not that much more headroom available. The Geomean performance improvement is +5.20%, and we get a maximum improvement of +14.50% in Y-Cruncher.

When running the OCCT CPU AVX2 Stability Test, the average CPU core effective clock is 4963 MHz with 1.296 volts. The average CPU temperature is 95.1 degrees Celsius. The average CPU package power is 375.1 watts.

When running the OCCT CPU SSE Stability Test, the average CPU core effective clock is 4971 MHz with 1.323 volts. The average CPU temperature is 90.7 degrees Celsius. The average CPU package power is 353.9 watts.

We check this processor’s boost behavior and per core maximum effective clock frequency. The boost frequency at 1 active thread is about 5532 MHZ and trails off to 5450 MHz when 7 cores are active. Then, once again, the C-State boost limiter kicks in and locks all cores to a maximum frequency of 5 GHz. We only see a slight improvement of the Fmax for cores in CCD0, where the Cuve Optimizer setting is pretty low. That said, we have three cores that achieve over 5.5 GHz. The most impressive improvement, however, is with the cores in CCDs 1 to 3. With a maximal Curve Optimizer of -50, all cores boost over 5.5 GHz.

OC Strategy #4: Manual Overclocking

In our fourth overclocking strategy, I will try manual overclocking. As with the 7980X, I focused on stability in our worst-case scenario: OCCT CPU Stress Test with AVX2.

One could question the use-case for manual overclocking of the Ryzen Threadripper CPUs. Just like with all Ryzen processors, the major downside of manual overclocking is that you lose the benefits of Precision Boost 2 technology in low-threaded benchmark applications. So, whereas this Ryzen Threadripper 7960X can boost up to 5650 MHz with Precision Boost, it will be limited to your highest manual frequency when manually overclocking.

However, I believe there’s a good use case for manual overclocking on the Ryzen Threadripper, and that’s for maximizing workload-specific performance. As I demonstrated in SkatterBencher #43 with the Zen 3 Ryzen Threadripper, optimizing a specific workload can yield another 10 to 20 percent more performance than relying on a tuned Precision Boost 2 algorithm.

Furthermore, AMD introduced the C-State Boost Limiter on Ryzen processors, limiting the maximum frequency in an all-core workload. As we saw in OC Strategy #3, the CPU hits the maximum allowed frequency of 5 GHz in our OCCT stress tests. This indicates there might be more headroom available in all-core workloads.

Let’s look at its architecture in more detail to better understand the performance tuning opportunities embedded in the Ryzen Threadripper 7960X processor.

CPU Core Frequency & Voltage

The Ryzen Threadripper 7960X is AMD’s Storm Peak Zen 4 architecture and features five chips on the package: four (4) CCDs and a single IO die.

CCD stands for Core Chiplet Die and is a die on a Ryzen CPU with CPU cores. The Zen cores are packed together in a CCX, or Core Complex. A Zen 4 CCX consists of up to eight individual cores, each with its L1 and L2 cache and a shared 32MB of L3 cache. On the Ryzen Threadripper 7960X, however, only six out of the eight available cores are enabled.

The frequency of the CPU cores is driven by a 100 MHz reference clock input. Each CCX has its own PLL and thus can run an independent frequency. The cores within a CCX share the same PLL, so they’ll run at the same frequency. That means, for the Ryzen Threadripper 7960X, we can set an independent frequency for each of the four CCXs.

The voltage of the CPU cores is provided by not one but two VDDCR_CPU voltage rails. That’s significantly different than before. It can be challenging to figure out which cores are powered by what voltage rail. Counterintuitively, CCDs 0 and 3 are powered by the first power rail for this motherboard and CPU. The other two are powered by the second power rail.

• VDDCR_CPU0 = CCD0, CCD3
• VDDCR_CPU1 = CCD1, CCD2

The voltage can be configured in two ways: either you set the target VID by programming the CPU registers or directly set the target voltage of the voltage regulator. Only the latter option is available in the main BIOS menus of this motherboard. However, you can find the option to program the CPU directly in the AMD Overclocking menu.

Choosing the proper manual voltage is always a matter of finding a suitable trade-off between increased overclocking potential, the thermal challenges that come with increased voltage, and, of course, consideration of CPU lifespan.

Per-CCX Frequency Tuning Process

With that last thought, we kick off the most tedious or exciting aspect of Ryzen Threadripper overclocking: per-CCX frequency tuning.

Our limiting factor will ultimately be the cooling solution as power consumption increases exponentially with operating voltage and temperature scales (somewhat) linearly with power consumption.

The maximum voltage will be determined by the application we’re tuning for. So, the first step in our tuning process would be to set a fixed CPU ratio and check the maximum temperature when running our workload. If there’s thermal headroom left, increase the operating voltage.

Once we know the maximum voltage, we can tune the CPU ratio of each CCX. Increase the CPU ratio of one CCX until the application shows signs of instability, then back off. Do this for each CCX sequentially, and you’ll end up with the maximum stable per-CCX frequency for a given voltage.

We do this for the applications we include in our manual overclocking strategy. In this table, you can find the maximum ratio for each CCX.

As a reminder, the values in this table are for my specific system and a specific stability test. Your CPU may have wildly different values depending on the CPU sample, cooling and motherboard, and your chosen stability test.

Now that we know the ins and outs of Ryzen Threadripper 7000 manual overclocking let’s jump into the BIOS. I will provide the BIOS settings of the OCCT AVX2 manual overclocking scenario.

BIOS Settings & Benchmark Results

Upon entering the BIOS

• In Easy Mode
  • Set XMP/EXPO Profile to EXPO 1
  • Enable XMP/EXPO High Bandwidth Support
  • Enable Low Latency Support
• Switch to Advanced Mode
• Set CPU Ratio Mode to Per CCX
  • Set CCD0 and CCD1 Ratio to 52.50
  • Set CCD2 and CCD3 Ratio to 52.00
• Set System Memory Multiplier to 66.00
• Set Infinity Fabric Frequency and Dividers to 2200 MHz
• Set UCLK DIV1 Mode to UCLK=MEMCLK
• Set CPU Vcore 0 to 1.200
• Set CPU Vcore 1 to 1.225

Then save and exit the BIOS.

We re-ran the benchmarks and checked the performance increase compared to the default operation:

• Geekbench 6 (single): -2.10%
• Geekbench 6 (multi): +14.45%
• Cinebench 2024 Single: -4.13%
• Cinebench 2024 Multi: +12.66%
• CPU-Z V17.01.64 Single: -4.73%
• CPU-Z V17.01.64 Multi: +9.62%
• V-Ray 5: +9.78%
• Corona 10: +11.19%
• AI Benchmark: +12.99%
• Y-Cruncher Pi MT 10B: +22.34%
• Blender (Monster): +11.76%
• Blender (Classroom): +10.62%
• 3DMark Night Raid: +5.85%
• Handbrake (H265, 1080P): +9.53%
• Tomb Raider: +3.09%
• Final Fantasy XV: +3.29%

Here are the 3DMark CPU Profile scores

• CPU Profile 1 Thread: -5.79%
• CPU Profile 2 Threads: -4.15 %
• CPU Profile 4 Threads: -0.92%
• CPU Profile 8 Threads: +6.02%
• CPU Profile 16 Threads: +9.57%
• CPU Profile Max Threads: +9.15%

With our manual overclock, we lose the benefit of the high Precision Boost single-threaded frequencies. We can see the effect clearly in the single-threaded benchmarks, where the performance is significantly lower than the stock settings. However, the benefit of the manual overclock is that our all-core frequency is much higher. The Geomean performance improvement is +7.45%, and we get a maximum improvement of +22.34% in Y-Cruncher.

When running the OCCT CPU AVX2 Stability Test, the average CPU core effective clock is 5218 MHz with 1.272 volts. The average CPU temperature is 88.6 degrees Celsius. The average CPU package power is 425.0 watts.

When running the OCCT CPU SSE Stability Test, the average CPU core effective clock is 5218 MHz with 1.264 volts. The average CPU temperature is 81.8 degrees Celsius. The average CPU package power is 376.5 watts.

We check this processor’s boost behavior and per core maximum effective clock frequency. As expected, the boost behavior and per core maximum effective clock are between 5250 and 5200 MHz. The most exciting thing about the manual overclock is that beyond 7 active cores, or 14 active threads, the operating frequency is 200 MHz higher than with a tuned precision boost overdrive.

AMD Ryzen Threadripper 7960X: Conclusion

All right, let us wrap this up.

When I put together this system, I was excited to try a GIGABYTE motherboard for high-end desktop as it was a first. However, I wasn’t particularly enthusiastic about the Ryzen Threadripper 7960X. However, the overclocking experience has taught me some things I didn’t know about Ryzen Threadripper 7000.

First, I think it’s excellent that GIGABYTE makes it easy to determine whether the processor in your system is enabled for overclocking. While it shouldn’t affect the warranty, some people may not want a fused-for-overclocking processor.

Second, I learned that unlocking the processor for overclocking also unlocks an additional 300 MHz for CCD0. That’s not only with the Ryzen Threadripper 7960X but with all Ryzen Threadripper 7000 processors.

Third, the C-State Boost Limiter is unfortunate and perhaps unnecessary for the 24-core Ryzen Threadripper 7960X. I see little point in limiting the performance in lighter workloads to 5 GHz. If AMD is adamant it’s mandatory for processor longevity, then perhaps a Precision Boost Overdrive option to disable the C-State Boost Limiter is an option.

Lastly, tied to the previous point, manual overclocking is a viable option for these processors thanks to the C-State Boost Limiter. My manual overclock increased the operating frequency by 200 MHz in all-core workloads and brought a 6 to 10-degree Celsius reduction in operating temperature.

Anyway, that’s it for this blog post. I want to thank my Patreon supporters for supporting my work. If you have any questions or comments, please drop them in the comment section below. ‘Till the next time!