RandomX CPU Hashing: Ryzen 9 7950X vs Threadripper 7000

Introduction

RandomX is one of those mining algorithms that makes CPU performance matter again. Unlike algorithms designed around specialized ASIC hardware, RandomX is built around general-purpose processors and makes heavy use of CPU resources, memory, and modern instruction sets.

That makes processors such as AMD's Ryzen 9 7950X interesting for mining. But what happens when we move from a high-end desktop CPU to AMD's much larger Threadripper 7000 platform?

The difference in core count is enormous. The Ryzen 9 7950X has 16 cores and 32 threads, while the Threadripper 7000 family goes as high as 64 cores and 128 threads with the Ryzen Threadripper 7980X. AMD also gives Threadripper a quad-channel DDR5 memory controller, compared with the dual-channel memory controller on the 7950X.

On paper, that sounds like an easy win for Threadripper. In mining, however, the answer is not quite that simple.

Why RandomX Is Well Suited to CPU Mining

RandomX is used by Monero and several other CPU-oriented cryptocurrencies. The algorithm is designed to make efficient CPU execution important, so simply having a high clock speed is not enough.

Memory performance, thread count, CPU architecture, cache, and the way the processor is configured can all affect the final result.

This is also why CPU mining is not simply a race for the processor with the largest number of cores. A processor needs to deliver enough hashing performance without turning the electricity bill into the largest part of the mining operation.

For that reason, the useful number is not just kH/s. We also want to know how much power is required to produce that hash rate.

Ryzen 9 7950X Monero Mining: Best Settings and Hashrate

The Ryzen 9 7950X is a Zen 4 desktop processor with 16 cores and 32 threads. It has a 4.5 GHz base clock, up to 5.7 GHz boost, 64 MB of L3 cache, and a default TDP of 170 W. It also supports AMD EXPO memory overclocking and Precision Boost Overdrive.

For a CPU miner, those specifications make the 7950X particularly interesting because it offers a large number of cores without requiring a workstation-class platform.

The practical test we are using here gives us a much better idea of what that means in a mining environment.

The 7950X was configured at a fixed 4.5 GHz clock, with the memory running at DDR5-6000 using EXPO. High Bandwidth Support was enabled, and the CPU voltage was reduced to approximately 0.95 V.

That last part is particularly important.

Instead of simply pushing the processor as hard as possible, the goal was to find a voltage low enough to reduce power consumption while maintaining stability. The approach was simple: lower the voltage until the system became unstable, then raise it slightly and test the configuration over time.

This is exactly the kind of tuning that makes sense for a CPU mining system.

Real-World Ryzen 9 7950X Power Consumption in Mining Rigs

With this configuration, the 7950X produced roughly 22.7–22.9 kH/s on RandomX.

The measured power consumption was around 175–177 W at the wall, while the CPU temperature was approximately 55°C.

That is an impressive result considering the cooling solution used in the test was a relatively inexpensive cooler costing around 23 Canadian dollars.

The low temperature was helped by an ambient temperature of around 20°C, while the mining system itself was operating at a relatively low voltage.

The cooler was also extremely quiet. The measured sound level was only about one decibel above the ambient room level when the measurement was taken close to the cooler.

So the 7950X was not being treated as a maximum-performance benchmark machine. It was being tuned as a practical mining system.

That distinction matters.

A CPU that produces slightly more hash rate but consumes significantly more electricity may not be the better mining processor.

What the Public XMRig Benchmarks Show

The 7950X result above comes from a real-world configuration, but we can also look at the public XMRig benchmark database to put the result into a broader context.

XMRig currently shows multiple RandomX results for the Ryzen 9 7950X. One of the higher recent submissions records 25.4 kH/s, while other submissions are around 22–23 kH/s. This spread shows why CPU mining benchmarks should not be treated as a single fixed number. Configuration, software version, memory, operating system, cooling, and tuning can all change the result.

The practical 22.7–22.9 kH/s result therefore sits comfortably within the range seen in public RandomX testing.

More importantly, it demonstrates that the 7950X can deliver roughly 23 kH/s while staying around 175 W at the wall in the particular configuration tested.

That is the number we should keep in mind when moving to Threadripper.

Enter Threadripper 7000

AMD brought Threadripper back to the high-end desktop market with the Zen 4-based Threadripper 7000 family.

There are three main non-Pro models:

All three use four-channel DDR5 memory and provide 48 PCIe 5.0 lanes. AMD launched them at $1,499, $2,499, and $4,999 respectively.

That immediately changes the economics.

The 7950X is a high-end desktop processor. Threadripper 7000 is a workstation-class platform that happens to be available as an overclockable high-end desktop product.

And RandomX gives us an interesting workload with which to test the difference.

Threadripper 7960X vs Ryzen 9 7950X

The first Threadripper 7000 model is the 24-core, 48-thread 7960X.

Public XMRig RandomX results show the 7960X reaching more than 32 kH/s in some validated submissions. One result recorded 32.4 kH/s using all 48 threads, while another recorded 29.3 kH/s.

That puts the 7960X clearly ahead of the approximately 22.7–22.9 kH/s achieved by the tuned 7950X in the practical test.

But there is an important catch.

The 7960X carries a 350 W TDP, more than twice the 7950X's official 170 W TDP.

That does not mean RandomX will necessarily consume exactly 350 W. TDP is not the same thing as actual wall power. But it tells us something important about the class of processor we are dealing with.

The Threadripper has considerably more performance available, but it is designed for a much higher power envelope.

Threadripper 7970X: More Cores, More Hashrate

The 7970X doubles the core count of the 7950X.

With 32 cores and 64 threads, it has twice as many physical cores and twice as many threads as the Ryzen 9 7950X. AMD rates it at up to 5.3 GHz and 350 W TDP.

The RandomX results reflect that extra hardware.

XMRig's public benchmark database contains validated 7970X results above 43 kH/s, with one recent submission reaching 43.5 kH/s using all 64 threads. Other submissions fall into the high-30s and low-40s kH/s range.

Again, these are not directly comparable laboratory measurements with identical hardware and settings. But they clearly show the performance potential.

Compared with the approximately 22.7 kH/s practical result from the 7950X, a 7970X can potentially deliver close to twice the RandomX hash rate.

The important word is potentially.

Hash rate alone does not tell us whether the 7970X is the better mining investment.

Threadripper 7980X Takes It Even Further

Then there is the 7980X.

This is where Threadripper becomes almost absurd compared with a conventional desktop processor.

The 7980X has 64 cores and 128 threads, 256 MB of L3 cache, a 3.2 GHz base clock, a boost clock of up to 5.1 GHz, and a 350 W TDP.

XMRig's public RandomX database currently shows the 7980X reaching approximately 85.7 kH/s in its highest listed result, with several other results above 80 kH/s.

That is dramatically higher than the 7950X.

But there is a reason this comparison needs some perspective.

AMD launched the 7980X with a $4,999 suggested price. The 7950X belongs to a completely different price class and platform.

If the goal is simply to build the fastest possible CPU RandomX system, the 7980X is obviously in another league.

If the goal is to make money from mining, however, the question becomes much harder.

Core Count Is Not the Whole Story

The Threadripper results demonstrate something important about RandomX.

More cores can produce substantially more total hash rate, but the scaling is not perfectly linear.

The 7950X has 16 cores and a practical result of roughly 22.7–22.9 kH/s.

A 24-core 7960X can reach more than 32 kH/s in public benchmarks.

A 32-core 7970X can reach more than 43 kH/s.

And the 64-core 7980X can exceed 85 kH/s.

The relationship is strong, but not perfectly proportional.

There are several reasons.

RandomX does not operate in a vacuum. Memory behavior, cache, thread scheduling, software configuration, and CPU architecture all influence the final number.

This is where Threadripper's quad-channel memory controller becomes relevant.

The 7950X uses two memory channels, while Threadripper 7000 provides four. AMD says the Threadripper memory controller can provide twice the memory bandwidth of typical dual-channel desktop systems.

That extra bandwidth is useful for heavily threaded workloads and helps explain why Threadripper is more than simply "a 7950X with more cores."

Memory Matters for RandomX

The 7950X test used DDR5-6000 memory with EXPO enabled.

The BIOS also provided high-bandwidth and low-latency options, and the high-bandwidth setting was preferred because it produced better hash-rate results in the test.

This is an important lesson for CPU miners.

Memory should not be treated as an afterthought.

Two systems using the same processor can produce different RandomX results if their memory configuration, timings, or thread settings are different.

Threadripper also has an advantage here because its platform is designed around four memory channels.

That does not mean that simply installing more RAM automatically increases RandomX performance. Capacity and bandwidth are two different things. But when many CPU cores are working simultaneously, the additional memory bandwidth gives the processor more room to keep those cores fed.

How to Calculate Hashes Per Watt (H/W) for CPU Mining Profitability

This is where the Ryzen 9 7950X becomes much more interesting.

The practical test produced about 22.7–22.9 kH/s at approximately 175–177 W from the wall.

Using the midpoint of 22.8 kH/s and 176 W gives roughly:

22.8 ÷ 176 ≈ 0.13 kH/s per watt

or about 130 H/s per watt.

That is the type of number a miner should pay attention to.

We cannot calculate the same figure for the Threadripper XMRig submissions because those benchmark entries do not provide a standardized wall-power measurement. And this is exactly why simply comparing the public hash-rate numbers can be misleading.

A 7980X producing 85 kH/s is impressive.

But if the machine costs several thousand dollars more and requires a much larger power budget, that does not automatically make it a better mining machine.

The 7950X Also Has a Major Platform Advantage

The Ryzen 9 7950X uses AMD's mainstream AM5 desktop platform and has integrated Radeon graphics. AMD specifies two memory channels and support for up to 128 GB of memory.

That makes building and testing a 7950X mining system relatively straightforward.

The integrated graphics are particularly convenient when testing a headless mining system. You do not necessarily need to install a separate GPU simply to get display output during setup.

This was one of the practical advantages highlighted in the original 7950X testing.

Threadripper, by comparison, requires the much more specialized sTR5 platform.

That means the processor is only one part of the investment. The motherboard, memory configuration, cooling, and power delivery all become more serious considerations.

Safe Undervolting Settings for AM5 CPU Mining Rigs

Both platforms are unlocked.

The 7950X test demonstrates a useful approach for mining: do not automatically chase the highest clock speed.

The processor was locked around 4.5 GHz and then undervolted to approximately 0.95 V. If the system remained stable, the voltage could be reduced further. Once instability appeared, the voltage could be increased slightly and the system tested for longer periods.

This is a much more sensible strategy for a 24/7 mining machine than simply pushing the CPU to its maximum possible frequency.

The Threadripper 7000 processors are also unlocked, and AMD specifically positions the series as an overclockable high-end desktop platform.

But there is a major difference in power headroom.

In the Threadripper testing described earlier, enabling PBO and increasing the performance limits caused power consumption in Cinebench to climb dramatically, reaching roughly 500 W during the test.

That does not mean RandomX will draw 500 W.

Cinebench and RandomX are different workloads, and the measurement should not be transferred directly from one to the other.

What it does demonstrate is how quickly the power requirements of a high-end Threadripper system can increase when the processor is taken off its stock limits.

For mining, that is something to watch carefully.

Budget-Friendly CPU Air Coolers for 24/7 RandomX Mining Rigs

The 7950X test also shows that good CPU mining performance does not necessarily require an enormous cooling system.

At approximately 55°C, the processor was running comfortably below its 95°C maximum operating temperature, despite delivering around 22.7–22.9 kH/s.

The relatively inexpensive cooler was also quiet.

Threadripper is a different challenge.

A 350 W processor naturally demands more serious attention to cooling, motherboard power delivery, airflow, and long-term operating conditions.

This becomes even more important when multiple systems are deployed in the same room.

A single CPU drawing an extra 100 or 200 W may not sound dramatic.

Ten or twenty machines change the equation completely.

Ryzen 9 7950X vs Threadripper 7000: The Practical Picture

The comparison can be summarized like this:

Official TDP is not equivalent to actual mining power consumption.

*Public XMRig results are submitted by different users under different configurations and should not be treated as a controlled head-to-head laboratory test. The 7950X figure in the first row refers specifically to the practical test configuration described above; Threadripper figures are selected public XMRig results.

The table tells us something useful.

Threadripper wins the absolute hash-rate contest.

That is not surprising.

The 7980X has four times the core count of the 7950X, while the 7970X has twice the core count. The hardware is simply much larger.

But the 7950X makes a much stronger argument when efficiency, hardware cost, cooling, and simplicity are included.

Best CPU for RandomX Mining: ROI Comparison (Ryzen vs Threadripper)

If your only objective is maximum RandomX hash rate from a single CPU, Threadripper 7000 is the obvious winner.

The 7960X, 7970X, and especially the 7980X can produce substantially more RandomX hash rate than the 7950X in public XMRig testing.

But mining is rarely about maximum hash rate alone.

For a smaller operation, the Ryzen 9 7950X may be the more sensible choice.

It offers 16 cores and 32 threads, a strong Zen 4 architecture, support for fast DDR5 memory, a much lower official TDP, and a mainstream desktop platform. With careful undervolting and memory tuning, the practical test shows that it can deliver around 22.8 kH/s while consuming roughly 176 W at the wall.

That combination is attractive for a 24/7 mining system.

Threadripper becomes more interesting when the objective is different.

If the same machine will also be used for rendering, video encoding, scientific workloads, virtualization, or other heavily threaded applications, the enormous core count can justify the platform.

This is exactly where Threadripper makes more sense than simply buying it for RandomX.

Final Verdict

The Ryzen 9 7950X and Threadripper 7000 represent two very different approaches to CPU mining.

The 7950X is a powerful desktop processor that can be tuned into a surprisingly capable RandomX miner. Its combination of 16 cores, fast DDR5 memory, relatively modest power requirements, and an affordable desktop platform makes it an attractive option for miners who care about efficiency.

Threadripper 7000 takes the opposite approach.

AMD essentially built a CPU platform around massive parallel processing. With 24, 32, or 64 cores and four-channel DDR5 memory, these processors can push RandomX hash rate far beyond what the 7950X can deliver. Public XMRig results show just how large that difference can become.

But that extra performance comes with a much higher platform cost and a much larger power and cooling envelope.

So the answer depends on what "better" means.

For maximum RandomX hash rate, Threadripper 7000 wins.

For a simpler and potentially more efficient mining system, the Ryzen 9 7950X remains a very compelling option.

And for miners, that distinction is more important than the biggest number on the benchmark chart.

The fastest CPU is not necessarily the CPU that makes the most money.

Sources

• XMRig Benchmark Database — Public RandomX benchmark results used to compare CPU mining performance across different processors and configurations.

• AMD Ryzen 9 7950X — Official Specifications — Official specifications covering the Ryzen 9 7950X architecture, core and thread count, clock speeds, cache, TDP, and memory support.

• AMD Ryzen Threadripper 7000 Series — Official Specifications — Official AMD information on the Threadripper 7960X, 7970X, and 79