AMD has made a tremendous leap compared to the previous generation, impacting various areas. Instead of focusing on benchmarks, interested individuals can look online (e.g., Phoronix) to witness the impressive performance of these processors.

The main reason I decided to upgrade from the 2990WX to the 3970X (in my opinion, the most significant upgrade) is the elimination of the peculiar NUMA architecture present in its predecessor (4 NUMA nodes / 2 CCX per node / 4 cores per CCX - 2 nodes with dual-channel memory each - 2 nodes with no memory channel at all). This architecture resulted in inconsistent performance in various workloads (not CPU-bound) due to memory latency. Many times, I even observed the 1700 outperforming the 2990WX. A clear example in my case was virtualization. While nested virtualization ran smoothly on the 1700, the same VMs on the 2990WX experienced significant steal time (vCPU waiting for the regular CPU to serve other vCPUs). Even after hours of optimization at the operating system, VM, and BIOS levels (e.g., memory interleaving, memory affinity, numactl, CPU pinning, etc.), I couldn't achieve the desired results. Ultimately, the only way to improve performance was to disable SMT, which was disappointing. This issue isn't limited to virtualization but affects all multicore workloads sensitive to memory latency.

To summarize, the 2990WX is a powerful processor that truly demonstrates its capabilities only in CPU-bound workloads like content creation and video editing. Those running such applications and considering an upgrade to the 3rd Gen Threadripper won't theoretically see more than a 10-15% performance increase (which matches the IPC gain between the 2nd and 3rd generations).

However, with the new Threadripper generation, all these peculiarities are a thing of the past. Besides many new features that improve performance in both single and multicore scenarios, all cores appear as a unified group (UMA) to the operating system, with equal distances between the cores and memory controllers. This architecture ensures consistent performance across all cores for every type of workload, out of the box, without additional configuration. In simple terms, it delivers true power with 32 cores/64 threads without any fine print.

In my case, with the 3970X, I no longer experience any steal time in virtualized workloads, and I've noticed significant improvements in VM deployments, kernel compilation, Spark jobs, and Gromacs, reaching up to 70% faster performance!

Regarding thermal/energy efficiency, AMD works wonders. No, the 280W TDP (notably in all-core boost) is not high; it has the highest performance-to-watt, watt/core, and watt/thread ratios in the market (see AnandTech, Phoronix). Considering it has 32 cores, almost twice as many as the 3950X, with a slightly higher base clock of 200 MHz and an additional 40 PCIe 4.0 lanes, the TDP appears modest. As for temperatures, I observe 35°C idle, 40°C average, and a maximum of 69°C under stress testing with a 560mm cooler, NFA14 3000 fans, and an EK-Velocity water block. It's important to note that these temperatures are accurate and there's no offset as with older Threadrippers (27°C). The mentioned numbers are based on stock settings alone. Anyone attempting overclocking by increasing power will realize that the performance gain is disproportionate to the power consumption and the temperatures that skyrocket. Additionally, AMD does not cover Threadripper overclocking under warranty.

AMD currently claims to have the most

Huge leap for AMD compared to the previous generation and beyond, in many areas. I won't focus so much on benchmarks, anyone who wants can look online (see phoronix) and see for themselves how well these processors perform in this generation.
The main reason I decided to switch from the 2990wx to the 3970x (in my opinion, it is the biggest upgrade) is the elimination of the peculiar numa architecture that its predecessor had (4 numa nodes / 2 CCX per node / 4 cores per CCX - 2 nodes with dual channel memory each - 2 nodes with no memory channel at all). The above architecture resulted in inconsistent performance in various workloads (not cpu bound) due to memory latency, and many times I even saw the 1700 performing better! The most obvious example for my case was in Virtualization. While nested virtualization ran without problems on the 1700, with the same vms on the 2990wx I noticed high steal time (the vcpu waits for the regular cpu to serve other vcpus). Even after many hours of optimization at the operating system, VM, or BIOS level (memory interleaving, memory affinity, numactl, cpu pinning, etc.), I couldn't achieve the desired results. In the end, the only way to improve performance was to disable SMT, which was disappointing. The previous problem is not only evident in virtualization but also in all types of multicore workloads that are sensitive to memory latency.
In summary, I conclude that the 2990wx is a very powerful processor that only shows its true power in cpu bound workloads (content creation, video editing, etc.). Those who run such applications and want to upgrade to the 3rd Gen Threadripper theoretically will not see more than a 10-15% improvement in performance (that's the IPC gain between the 2nd and 3rd generation).
However, with the new generation of Threadripper, all these peculiarities belong to the past. In addition to the many new features that improve performance in both single and multicore, all cores appear in the operating system as a unified group (UMA) with equal distances between them and the memory controllers. With this architecture, consistent performance is achieved across all cores, for every type of workload, and even out of the box without additional configuration. In simple terms, real power of 32 cores/64 threads without any compromises.
As for my case, now with the 3970x, besides the fact that the steal time is always at 0 in all the virtualized workloads I have run, I also observe a huge improvement in vm deployments, kernel compilation, spark jobs, and gromacs, reaching up to 70%!
Regarding thermal/energy efficiency, AMD works wonders again. No, the 280W it has as TDP (notably in all core boost) is not a lot, on the contrary. It has the highest performance/watt, watt/core, watt/thread ratio in the market (see anandtech, phoronix). It has 32 cores, and if you consider that it is almost 2 times the 3950x with a higher base clock by 200 MHz and 40 additional pcie4 lanes, the TDP seems small. As for temperatures, I observe 35 idle, 40 average, 69 max under stress test with a 560mm cooler, NFA14 3000 fans, and ek-velocity waterblock. It is important to mention that the temperatures are real and there is no offset like in the old Threadripper processors (27C). The above numbers refer only to stock settings. Anyone who tries to overclock by increasing the power will realize that the performance gain is completely disproportionate to the power consumption and the temperatures that actually skyrocket. Additionally, overclocking on Threadripper processors is not covered by AMD's warranty. AMD currently claims to have the most powerful desktop processor ever built. The numbers show that they are telling the truth (see guru3d, anandtech, phoronix, techspot, etc.). The only exception where Intel surpasses by far is in scientific applications that extensively use AVX-512 for matrix multiplication. I should mention that this particular processor performs equally or even better than 40,000€ multisocket systems in many multithreaded workloads (see Passmark, openbenchmarking.org, etc.). It is a huge achievement that now anyone can run/try different types of workloads at home, which in the past would require either waiting at research institutions or paying exorbitant amounts to have access to high-performance computing.

Excellent processor. It goes without saying that it's worth getting something like this only for professional use. It's a tool that will make you earn multiple times the money you spend to get it. Of course, it also handles its games without any problem. Thank you AMD for providing us with such technologies at reasonable prices.

It was placed next to a system with an Intel Xeon Gold 6152 Box worth 4,000 euros and simply demolished it. 48,000 units vs 34,000 units. The disadvantage is that it does not exist in dual motherboards in order to avoid the Epic which has double the price. Waiting for the big brother 3990X 64/128 which promises to demolish all systems with 2X Xeon costing over 20,000 euros. On a system for cad/cam with Nvidia P6000 vray renderer cpu/gpu for photorealistic models and real-time modeling.

Huge leap for AMD compared to the previous generation and beyond, in many areas. I won't focus so much on benchmarks, anyone who wants can look online (see phoronix) and see for themselves how well these processors perform in this generation.

The main reason I decided to switch from the 2990wx to the 3970x (in my opinion, it is the biggest upgrade) is the elimination of the peculiar numa architecture that its predecessor had (4 numa nodes / 2 CCX per node / 4 cores per CCX - 2 nodes with dual channel memory each - 2 nodes with no memory channel at all). The above architecture resulted in inconsistent performance in various workloads (not CPU bound) due to memory latency, and many times I would even see the 1700 performing better! The most obvious example for my case was in Virtualization. While nested virtualization ran without problems on the 1700, with the same VMs on the 2990wx, I noticed high steal time (the vCPU waits for the regular CPU to serve other vCPUs). Even after many hours of optimization at the operating system, VM, or BIOS level (memory interleaving, memory affinity, numactl, CPU pinning, etc.), I couldn't achieve the desired results. In the end, the only way to improve performance was to disable SMT, which was disappointing. The previous issue is not only present in virtualization but also in all types of multicore workloads that are sensitive to memory latency.

In summary, I conclude that the 2990wx is a very powerful processor that only shows its true power in CPU-bound workloads (content creation, video editing, etc.). Those who run such applications and want to upgrade to the 3rd Gen Threadripper theoretically won't see more than a 10-15% improvement in performance (that's the IPC gain between the 2nd and 3rd generation).

However, with the new generation Threadripper, all these peculiarities belong to the past. In addition to the many new features that improve performance in both single and multicore scenarios, all the cores appear to the operating system as a unified group (UMA) with equal distances between them and the memory controllers. With this architecture, consistent performance is achieved across all cores, for every type of workload, and even out of the box without additional parameterization. In simple terms, real power of 32 cores/64 threads without compromises.

Regarding my case, now with the 3970x, besides the fact that steal time is always at 0 in all the virtualized workloads I have run, I also observe a huge improvement in the times of VM deployments, kernel compilation, spark jobs, and gromacs, reaching up to 70%!

Regarding thermal/energy performance, AMD works wonders again. No, the 280W it has as TDP (notably in all core boost) is not a lot, on the contrary. It has the highest performance/watt, watt/core, watt/thread ratio in the market (see anandtech, phoronix). It is 32 cores and if you consider that it is almost 2 times the 3950x with a higher base clock by 200 MHz and 40 additional PCIe4 lanes, the TDP seems small. As for temperatures, I observe 35 idle, 40 average, 69 max under stress test with a 560mm cooler, NFA14 3000 fans, and ek-velocity waterblock. It is important to mention that the temperatures are real and there is no offset like in the old threadripper (27C). The above numbers refer only to stock settings. Anyone who tries to overclock by increasing the power will realize that the performance gain is completely disproportionate to the power consumption and the temperatures that actually skyrocket. Also, overclocking on threadripper processors is not covered by AMD's warranty.

AMD currently claims to have the most powerful desktop processor ever built. The numbers show that they are telling the truth (see guru3d, anandtech, phoronix, techspot, etc.). The only exception where Intel surpasses them is in scientific applications that extensively use AVX-512 for matrix multiplication. It should be noted that this particular processor performs the same or even better than 40,000€ multisocket systems in many multithreaded workloads (see Passmark, openbenchmarking.org, etc.). It is a huge achievement that now anyone can run/try different types of workloads at home, which in the past would require either waiting in research institutions or paying exorbitant amounts to access high-performance computing.