Ever since the release of Zen4, it has been nearly two years. Although Zen4 boasts excellent power efficiency and commendable performance, it still suffers from issues such as heat accumulation, and its single-core performance has been outpaced by Intel's 13th and 14th generations. Therefore, ever since the exposure that Zen5 will significantly increase IPC (Instructions Per Clock), the anticipation for Zen5 has been intense.

However, the initial release of Zen5 is on the mobile platform, specifically Strix Point. Thanks to the use of TSMC's 4nm process technology, coupled with architectural improvements, Zen5's energy efficiency has seen a substantial enhancement. This has further fueled the excitement for the pure big-core desktop products, namely the Ryzen 9000 series. Through some wheeling and dealing, I managed to get my hands on AMD's latest Ryzen 5 9600X ahead of time to show everyone the performance and energy efficiency of this CPU.

The packaging of the 9600X maintains the same design as the Ryzen 7000 series, but due to the absence of a cooler, the packaging size has become much smaller, roughly the same size as a fan's packaging.

The accessories are the usual trio: a sticker, the CPU itself, and an installation manual.

The CPU itself, with its top cover, retains the same design as the previous generation. However, but! I've heard that this generation has improved thermal resistance efficiency, so it doesn't accumulate heat anymore! To test this claim, I borrowed a 360mm liquid cooler from a friend to see if it's really the case.To ensure that there are no bottlenecks during testing, I also took out the dusty ROG X670E HERO that has been sitting unused for half a year. Of course, the reason I don't use it regularly is not because it's bad, but because it's not white, which doesn't meet my PC building requirements.

The motherboard features the standard ROG design language, with a mirror acrylic cover over the I/O panel that conceals the lighting board. Both PCI-E slots are of the 5.0*8 specification.

There are 4 M.2 slots, all covered by heatsinks and equipped with pre-installed thermal pads. After all, as an X670E motherboard, it's not short on PCI-E lanes.

The CPU power supply is dual 8PIN; if testing the 9600X, one is sufficient, but when testing the 9900X later, it's likely that both will need to be fully connected.

The I/O interfaces are comprehensive, and it even offers the uncommon 40Gbps Thunderbolt ports on the X670E, and there are two of them.The motherboard comes with an additional PCI-E M.2 expansion card, which can accommodate an SSD with a bandwidth of 5.0*4. Moreover, to handle the significant heat generated by the 5.0 solid-state drive, the heatsink is quite heavy. Of course, there are downsides as well; when it's connected, the graphics card speed will be reduced to 8x.

The CPU and motherboard are combined. However, to be honest, pairing the 9600X with the X670E HERO feels a bit like a mismatch, but the 9600X has no regrets now.

Due to the inadequacy of my idle 4070S, I borrowed a three-slot-and-a-half PowerColor 7900XTX from a friend. When I received the express delivery, I noticed that it was sent with full insurance and to pay upon delivery, which made me feel quite upset, as it cost me a significant amount for shipping.

The graphics card is powered by three 8PIN connectors, and the 7900XTX is certainly not energy-efficient in terms of power consumption.I've taken out the Kingston Fury Vengeance 32GB DDR5 7600 RGB memory that I used for my previous PC build. Since the new photos didn't turn out as good as the old ones, I'm using the old ones.

I've also taken out an idle P44Pro 1T SSD, but I can't find the box!

For the water cooling, since I have a 240mm water cooler lying around and this time there's no heat accumulation, plus the PBO power is set quite high, I went and got a friend to trade me a Lian Li MJOLNIR VISION 360 all-in-one water cooler. It's definitely enough to handle the 9600X after PBO.

The packaging is still very sturdy, so no worries about any issues during transportation.The fasteners are comprehensive, supporting LGA1200, LGA1700, 20XX, and AM4/AM5, with an additional tube of TF7 thermal paste included.

The water block itself is pure white, and all three fans have RGB lighting, but only the frames are illuminated, which is suitable for players who prefer a more subtle look.

The radiator is a standard slim design, featuring the LID cooling logo.

The cold head is made of pure copper, remember to peel off the protective film first (I forgot to do this initially and only noticed after running tests). Both the radiator fans and the cold head only require a single fan cable, with RGB being universal.

The cold head display connects to the head via magnetic attachment, remember to remove it when installing the cold head. As a water cooler in the $700 price range, I believe it's quite cost-effective to offer one with a display for added flair.I just picked up an idle Seasonic Focus GX1000W power supply, a 1000W gold-rated fully modular unit with a 16PIN graphics card power supply; it's more than enough even if the 7900XTX comes along.

Thus, the overall test platform for this time is as follows:

CPU: AMD Ryzen 5 9600X

Motherboard: ASUS ROG X670E HERO

Memory: Kingston Fury Renegade 32GB DDR5 7600 RGB

SSD: Silico P44Pro 1TB

Graphics Card: PowerColor RX 7900XTX Red Devil Limited EditionCooling: Lian Li MJOLNIR VISION 360 ARGB WHITE

Power Supply: Seasonic Focus GX1000W

Although the other components may seem a bit overkill for the 9600X, the entire system ensures that the CPU will not encounter a performance bottleneck. Additionally, the memory performance of this generation is not significantly different from the previous one, so the memory is actually running at DDR5 6000 with 32-38-38-90 timings, ensuring a 1:1 ratio between UCLK and MEM. After assembling the entire platform, it is as follows. Without a test stand, the motherboard box was used as a makeshift stand.

The actual photos of the fans and the cold head are as follows, and they look quite good.

By using the TRCC software, you can also customize the display on the cold head. In addition to the built-in themes, you can also set your own images or videos.Alright, no more beating around the bush, let's get started with the testing. The CPU information is as follows: The Zen5 architecture for desktops is also synchronized with the mobile side this time, using TSMC's 4nm process technology. The R5 9600X has a specification of 6C12T, with a L2 cache of 1MB and an L3 cache of 32MB, which is consistent with the previous generation 7600X. In addition, the R5 9600X has a certain increase in base clock speed, with a default maximum Boost frequency of 5.4GHz, while the 7600X is at 5.3GHz. During testing, the graphics card used the latest 24.7.1 driver, and the motherboard BIOS was updated to the latest 2124 version of the X670E HERO, with the room temperature at 26°C.

In addition to the R5 9600X, the Zen5 family has a total of four models at launch, which are the 8-core R7 9700X, the 12-core R9 9900X, and the 16-core flagship R9 9950X.

Compared to the previous generation, Zen5's AVX512 has been directly upgraded from the previous 256bit to a full-blooded 512bit, and the front-end decoding width has also increased significantly, all of which are the main reasons for the IPC increase in Zen5.

Memory and cache benchmark results are as follows: without adjusting the small parameters, the 6000C32 memory can achieve a latency of 70ns, which is still a certain improvement compared to the previous generation.

CPU-Z single-core score is 817 points, and the multi-core score is 6355.6. In the CPU-Z score where AMD is not good at, the single-core is close to Intel's 14th generation CPU, I think the improvement is quite significant, and compared to the single-core of 7950X, it is a significant increase.In CBR23, the single-core 2168 has already caught up with the adjacent 14700K, and the multi-core 16200 has left the 12400F, which is also a 6C12T, far behind. The performance in CBR24 is similar, with a single-core score of 134, which is basically at the top level (T1) for current desktop CPU performance.

VRay 6.0 has a default score of 19,049 points.

The 7ZIP scores are as follows:

The Aida64 GPGPU scores are as follows:

The Blender 4.2 results are as follows, and the BMW render time is 150 seconds, which is an improvement compared to the 160 seconds on the TPU with the 7600X.3Dmark CPU single-core score of 1263 points, as an R5, this performance is already on par with the neighboring i7.

PCMark10 scores are as follows:

Finally, the 3Dmark TS, with a CPU score of 10,616 points.

I selected a total of 5 mainstream games, starting with the competitive Rainbow Six Siege and CS2. Testing graphics quality, R6 is at 2K maximum settings, while CS2 is at 2K high settings. With the 7900XTX, R6 achieves 550 frames, and CS2 also reaches 428 frames, which indicates that the 9600X, even when paired with a flagship card, will not be found wanting.The remaining three are AAA titles. In Shadow of the Tomb Raider, with 2K high-quality graphics, the average frame rate is 247 frames per second (fps). For Cyberpunk 2077, with 2K ultra preset graphics, the average is also 158 fps.

Next is Red Dead Redemption 2, with the following graphics settings, averaging 161 fps, which is more than enough to run smoothly even with a 2K high-refresh-rate monitor.

Moving on to the thermal tests. I mentioned from the start that this generation has improved thermal resistance efficiency and no longer has the issue of heat accumulation. So, how does it perform in reality? At idle, the temperature of the 9600X is 44.5°C. The previous generation, even with a 360mm liquid cooler, would have temperatures ranging from 50 to 60°C due to heat accumulation issues, indicating that there is indeed an improvement.

Then, under full load without PBO (Precision Boost Overdrive), the 9600X's full load power consumption is 88W, with a temperature of 62.4°C. Although the 360mm liquid cooler does have a certain effect, it is still much better than the previous generation. Moreover, the addition of AVX512 did not significantly reduce the stress test frequency but maintained the default 3.9GHz, in contrast to the blue team's... Whether AVX512 is useful or not depends on whose AVX512 it is (not serious).

In comparison, my personal R7 7800X3D, also with a 360mm liquid cooler, has a full load temperature close to 80°C, even though the power consumption is 8W lower, at only 80W. This shows that the thermal conductivity efficiency of Zen5 has indeed been significantly improved.After shutting down the stress test, the CPU temperature of the 9600X quickly drops back to the idle level, while the 7800X3D cannot, still lingering around 60°C. Although it has something to do with my fan strategy, the difference between the two is quite significant.

Of course, as an AMD processor, how can we skip testing PBO? Starting from the Ryzen 7000 series, compared to the traditional PBO, AMD has also introduced PBO2, which allows for more settings than PBO and offers higher playability. The method to enable PBO is quite simple: set Precision Boost Overdrive to enable in the BIOS. PBO2 is a bit more troublesome, as there are quite a few adjustments to make. Here, I provide my settings, which should be applicable to more than 90% of the Ryzen 7000 series (excluding major binned chips), and should work for the Ryzen 9000 series as well. These are the settings for my 7800X3D, and I will directly apply them to the 9600X for testing.

Firstly, with CPU-Z, it can be seen that the orientations of PBO and PBO2 are distinctly different. PBO primarily enhances all-core performance. PBO2 has less of an all-core enhancement, but there is a certain improvement in single-core performance.

The comparison of the default scores is organized into the following chart.In CBR23, PBO shows a significant improvement over the default multi-core score, with a 10% increase, approaching the performance of the R7 7700, while the single-core performance remains unchanged. PBO2, on the other hand, offers about a 3% increase in single-core and a 5% increase in multi-core performance, with a single-core score of 2240 points already surpassing the 14900K.

Similar performance is observed in CBR23, where with the aid of PBO, the multi-core score of the 9600X finally breaks the 1000-point milestone, which is quite an impressive achievement for a 6-core CPU.

The performance in VRay is also essentially similar to that in Cinebench.In the four tests of Blender, although there were improvements, they were not as significant as the previous ones.

The final theoretical test is the 3DMark CPU. With the aid of PBO2, the single-core score has approached the 1300-point milestone, which is more than sufficient for gaming. This also makes me look forward to the gaming performance of the 9800X3D.

In the gaming section, apart from CS2, which is sensitive to frequency and shows an improvement, the other games did not change much compared to the default settings. Therefore, I will only present the results for CS2 here. It can be seen that compared to PBO, PBO2, due to the single-core enhancement, is more suitable for friends who play games.PBO's multi-core performance offers a 10% increase compared to the default, which is definitely a substantial overclocking margin for today's CPUs. So, what's the trade-off? The answer is power consumption. Once PBO is enabled, the CPU's power limit is lifted, shifting to a temperature limit. As long as your cooling solution is robust enough, you can reach higher power consumption levels, a strategy that aligns perfectly with the blue team's approach. After enabling PBO, the power consumption of the 9600X soared to an astonishing 138W, hitting the 95°C temperature limit. At this point, the CPU's clock speed was 4.85GHz, nearly a 1GHz increase from the default 3.9GHz.

From this, we can also deduce that Zen5 has truly addressed the issue of heat accumulation. We know that Zen4, with the 7700, would reach the temperature limit around 110W, while the 9600X pushed it to the 140W range, which is a significant improvement. Solving the heat accumulation issue means that when pairing with Zen5 processors, we can consider many single-tower air coolers instead of going for water cooling. I've heard that the 9900X and 9950X can surpass 300W with PBO enabled, but I'm not sure if it's true. I'll have to test it once I get my hands on them.

After shutting down the stress test, the temperature quickly dropped back to idle levels. Although it's not as exaggerated as the blue team's post-Die shaving speeds, it's more than sufficient. I'm relieved, truly relieved. With the heat accumulation issue resolved, I believe Zen5 has no real flaws, aside from being a bit pricey.

Compared to PBO's naive increase in power consumption, PBO2 is much better. Here, I also recommend that Ryzen 7000 and Ryzen 9000 users prioritize PBO2. From the stress test power consumption, it's evident that both single-core and multi-core performance have improved, but PBO2's power consumption is quite low, remaining consistent with the default 88W, with no change. If the CPU's quality is better, the negative voltage can be adjusted even lower, which would further reduce power consumption. It's like getting stronger single-core and multi-core performance with the same power consumption as the default, who wouldn't love that? However, the full load temperature is 4°C higher than the default, at 65°C.

Well, that concludes the testing. Now it's time for a summary.

Firstly, if you ask me if Zen5 is strong, my answer is an affirmative, it's incredibly strong. With lower power consumption, it achieves single-core performance that rivals the competition's smoke-spewing capabilities, which is fantastic news for gamers. After all, this is performance that matches the 14900K in single-core, so gaming performance won't be lacking. This also makes me more excited for Zen5's X3D products; they might just be the gods of gaming. The resolution of heat accumulation and the inclusion of full AVX512 capabilities allow Zen5 to perform better in tasks like rendering. With all cores being high-performance, there's no need to worry about scheduling issues, and the 9950X might just be the first choice for personal video workstations. Furthermore, the lower power consumption also means that Ryzen 9000 users don't have to worry about stability issues and can use it with confidence.However, there are still some drawbacks. First, the price of AM5 motherboards is still too high at present, and new motherboards like the X870E, according to leaks, don't seem to offer significant improvements. Additionally, the initial price of the 9600X might be a bit high, and it's estimated that true A-series fans will have to wait until the Double 11 shopping festival to get their hands on it, which is also a traditional practice. Of course, this is just my guess, and I also hope that there will be a pleasant surprise in terms of pricing. When will there be a 9500F to replace the 7500F? That would be the real killer, the top choice for a thousand-yuan gaming CPU.

Thank you for watching.